diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last-checkpoint/trainer_state.json"
@@ -0,0 +1,90256 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.015,
+  "global_step": 15000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.931,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6816,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.2796,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.119,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 0.8744,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.1561,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 1.0062,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1478,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-06,
+      "loss": 0.5617,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.5931,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.6567,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.6785,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.0113,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.8829,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-06,
+      "loss": 0.8225,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1247,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 1.168,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 1.0633,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-06,
+      "loss": 1.0441,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.2362,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.05e-05,
+      "loss": 1.1747,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.9891,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.8767,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.9538,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9734,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.839,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.7487,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.9824,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.45e-05,
+      "loss": 1.0037,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.32,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.55e-05,
+      "loss": 0.9389,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9922,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.65e-05,
+      "loss": 0.9843,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.0919,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.75e-05,
+      "loss": 0.8617,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 1.0183,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.85e-05,
+      "loss": 0.9565,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 0.79,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 1.2117,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.111,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.05e-05,
+      "loss": 1.8326,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 1.7969,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.15e-05,
+      "loss": 1.0872,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.0279,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.25e-05,
+      "loss": 1.2225,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.9716,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.35e-05,
+      "loss": 1.2187,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 0.871,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.45e-05,
+      "loss": 0.8476,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8047,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 1.2494,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.0532,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 1.0323,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.0223,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.9494,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.0272,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 1.2829,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-05,
+      "loss": 1.2682,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.95e-05,
+      "loss": 1.6736,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 2.3082,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.05e-05,
+      "loss": 1.1302,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1e-05,
+      "loss": 0.7822,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.15e-05,
+      "loss": 0.9861,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.9721,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.6753,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-05,
+      "loss": 1.047,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.35e-05,
+      "loss": 1.0422,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.9134,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.45e-05,
+      "loss": 1.5718,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 1.0625,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.55e-05,
+      "loss": 1.0779,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 1.1702,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.65e-05,
+      "loss": 1.1024,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7e-05,
+      "loss": 1.0382,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.0676,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-05,
+      "loss": 1.2035,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.85e-05,
+      "loss": 1.0576,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.1175,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 1.0663,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.2545,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.05e-05,
+      "loss": 0.7244,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1e-05,
+      "loss": 1.1455,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.15e-05,
+      "loss": 1.1514,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-05,
+      "loss": 1.0883,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.25e-05,
+      "loss": 0.8759,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3e-05,
+      "loss": 1.0519,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.35e-05,
+      "loss": 1.1014,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.918,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 1.0174,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 1.0974,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.55e-05,
+      "loss": 1.0889,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.0881,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 0.8923,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7e-05,
+      "loss": 1.0121,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.75e-05,
+      "loss": 1.3273,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 1.0579,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.85e-05,
+      "loss": 1.149,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9e-05,
+      "loss": 1.4682,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 1.4034,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 0.8914,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999999987661e-05,
+      "loss": 1.0526,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999999950642e-05,
+      "loss": 1.2849,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999999888945e-05,
+      "loss": 1.2292,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999999802569e-05,
+      "loss": 1.0693,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999996915136e-05,
+      "loss": 0.952,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999999555779e-05,
+      "loss": 1.1392,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999993953664e-05,
+      "loss": 1.0248,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999999210274e-05,
+      "loss": 1.2939,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999990005034e-05,
+      "loss": 1.2429,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999998766053e-05,
+      "loss": 1.1394,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999998506924e-05,
+      "loss": 1.3022,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999998223116e-05,
+      "loss": 0.9869,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999979146295e-05,
+      "loss": 1.2054,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999997581464e-05,
+      "loss": 1.1226,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999997223619e-05,
+      "loss": 1.0491,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999996841095e-05,
+      "loss": 0.8425,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999964338926e-05,
+      "loss": 1.305,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999996002011e-05,
+      "loss": 1.0233,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999999554545e-05,
+      "loss": 0.5656,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999950642105e-05,
+      "loss": 1.0478,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999945582926e-05,
+      "loss": 1.191,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999940276956e-05,
+      "loss": 1.3638,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999993472419e-05,
+      "loss": 0.355,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999928924634e-05,
+      "loss": 1.095,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999922878296e-05,
+      "loss": 1.151,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999991658516e-05,
+      "loss": 1.4362,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999991004524e-05,
+      "loss": 1.0988,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999990325853e-05,
+      "loss": 0.8029,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999896225036e-05,
+      "loss": 1.2147,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999988894474e-05,
+      "loss": 1.1023,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999881417667e-05,
+      "loss": 0.9649,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999998736438e-05,
+      "loss": 1.059,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999986562314e-05,
+      "loss": 1.4863,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999998573557e-05,
+      "loss": 1.4881,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999984884146e-05,
+      "loss": 1.0374,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999840080426e-05,
+      "loss": 1.0476,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999983107262e-05,
+      "loss": 1.0538,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999982181801e-05,
+      "loss": 0.8671,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999981231661e-05,
+      "loss": 1.1986,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999802568433e-05,
+      "loss": 1.2,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999979257346e-05,
+      "loss": 1.1648,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999997823317e-05,
+      "loss": 1.0579,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999977184314e-05,
+      "loss": 1.0379,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999997611078e-05,
+      "loss": 0.5544,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999975012567e-05,
+      "loss": 0.9992,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999973889675e-05,
+      "loss": 1.2599,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999972742104e-05,
+      "loss": 1.3262,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999715698544e-05,
+      "loss": 1.3029,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999970372925e-05,
+      "loss": 1.238,
+      "step": 149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999969151317e-05,
+      "loss": 1.1567,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999996790503e-05,
+      "loss": 1.1385,
+      "step": 151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999666340646e-05,
+      "loss": 1.068,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999996533842e-05,
+      "loss": 1.2784,
+      "step": 153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999964018096e-05,
+      "loss": 0.7641,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999962673094e-05,
+      "loss": 0.6886,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999961303412e-05,
+      "loss": 1.1172,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999959909052e-05,
+      "loss": 0.847,
+      "step": 157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999958490012e-05,
+      "loss": 0.8203,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999957046294e-05,
+      "loss": 1.3131,
+      "step": 159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999955577897e-05,
+      "loss": 1.4838,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999995408482e-05,
+      "loss": 1.1396,
+      "step": 161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999525670646e-05,
+      "loss": 1.3872,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999951024631e-05,
+      "loss": 1.9415,
+      "step": 163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999949457518e-05,
+      "loss": 1.5973,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999478657254e-05,
+      "loss": 1.2385,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999462492545e-05,
+      "loss": 1.2302,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999446081045e-05,
+      "loss": 1.08,
+      "step": 167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999942942276e-05,
+      "loss": 1.3935,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999941251768e-05,
+      "loss": 1.2663,
+      "step": 169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999395365814e-05,
+      "loss": 1.3094,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999377967155e-05,
+      "loss": 1.2008,
+      "step": 171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999936032172e-05,
+      "loss": 0.8156,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999342429474e-05,
+      "loss": 4.8589,
+      "step": 173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999324290446e-05,
+      "loss": 6.0248,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999305904634e-05,
+      "loss": 5.9313,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999928727202e-05,
+      "loss": 5.9243,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999926839264e-05,
+      "loss": 5.7681,
+      "step": 177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999924926645e-05,
+      "loss": 5.6746,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999229893476e-05,
+      "loss": 5.7433,
+      "step": 179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999921027372e-05,
+      "loss": 5.7031,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999190407165e-05,
+      "loss": 5.5989,
+      "step": 181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999917029382e-05,
+      "loss": 5.24,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999149933695e-05,
+      "loss": 5.0161,
+      "step": 183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999912932677e-05,
+      "loss": 4.415,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999910847306e-05,
+      "loss": 5.2371,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999908737256e-05,
+      "loss": 5.5717,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999066025276e-05,
+      "loss": 5.6388,
+      "step": 187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999990444312e-05,
+      "loss": 5.6072,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999902259033e-05,
+      "loss": 5.6016,
+      "step": 189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999999000502676e-05,
+      "loss": 5.5708,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999897816823e-05,
+      "loss": 5.5852,
+      "step": 191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999895558699e-05,
+      "loss": 5.572,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999893275897e-05,
+      "loss": 5.529,
+      "step": 193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998909684156e-05,
+      "loss": 5.5539,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999888636255e-05,
+      "loss": 5.563,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998862794154e-05,
+      "loss": 2.3107,
+      "step": 196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999883897897e-05,
+      "loss": 1.4454,
+      "step": 197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998814917e-05,
+      "loss": 1.4917,
+      "step": 198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999879060824e-05,
+      "loss": 1.0891,
+      "step": 199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998766052684e-05,
+      "loss": 0.7431,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999874125034e-05,
+      "loss": 1.0081,
+      "step": 201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999871620121e-05,
+      "loss": 1.4269,
+      "step": 202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999869090529e-05,
+      "loss": 1.1632,
+      "step": 203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998665362585e-05,
+      "loss": 0.9132,
+      "step": 204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999863957309e-05,
+      "loss": 0.708,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999986135368e-05,
+      "loss": 1.159,
+      "step": 206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999858725372e-05,
+      "loss": 1.2212,
+      "step": 207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998560723855e-05,
+      "loss": 1.1131,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999985339472e-05,
+      "loss": 0.9072,
+      "step": 209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999850692375e-05,
+      "loss": 1.3964,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999847965352e-05,
+      "loss": 1.2253,
+      "step": 211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999845213649e-05,
+      "loss": 0.8362,
+      "step": 212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999842437267e-05,
+      "loss": 0.6663,
+      "step": 213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999839636207e-05,
+      "loss": 1.006,
+      "step": 214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998368104686e-05,
+      "loss": 1.6042,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998339600496e-05,
+      "loss": 1.1701,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999831084953e-05,
+      "loss": 1.1874,
+      "step": 217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999828185177e-05,
+      "loss": 1.1069,
+      "step": 218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999825260722e-05,
+      "loss": 0.9765,
+      "step": 219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998223115875e-05,
+      "loss": 0.8295,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999819337774e-05,
+      "loss": 1.012,
+      "step": 221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999816339282e-05,
+      "loss": 1.1255,
+      "step": 222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999813316111e-05,
+      "loss": 1.2156,
+      "step": 223
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999998102682616e-05,
+      "loss": 1.0133,
+      "step": 224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999807195733e-05,
+      "loss": 1.111,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999804098525e-05,
+      "loss": 1.0167,
+      "step": 226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999800976639e-05,
+      "loss": 1.2424,
+      "step": 227
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999797830073e-05,
+      "loss": 1.1881,
+      "step": 228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999794658828e-05,
+      "loss": 1.2338,
+      "step": 229
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999791462905e-05,
+      "loss": 1.277,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999788242302e-05,
+      "loss": 1.1273,
+      "step": 231
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999784997021e-05,
+      "loss": 1.1403,
+      "step": 232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999978172706e-05,
+      "loss": 1.1259,
+      "step": 233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999778432421e-05,
+      "loss": 1.1319,
+      "step": 234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999775113103e-05,
+      "loss": 1.1755,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999771769106e-05,
+      "loss": 0.8809,
+      "step": 236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999768400429e-05,
+      "loss": 0.8313,
+      "step": 237
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999765007075e-05,
+      "loss": 0.4808,
+      "step": 238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999761589041e-05,
+      "loss": 0.3866,
+      "step": 239
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999758146328e-05,
+      "loss": 0.4625,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999754678936e-05,
+      "loss": 1.005,
+      "step": 241
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999751186866e-05,
+      "loss": 0.478,
+      "step": 242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999747670115e-05,
+      "loss": 0.4184,
+      "step": 243
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999744128687e-05,
+      "loss": 0.3674,
+      "step": 244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999740562579e-05,
+      "loss": 0.3536,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999736971793e-05,
+      "loss": 0.3212,
+      "step": 246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999733356327e-05,
+      "loss": 0.3954,
+      "step": 247
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999997297161825e-05,
+      "loss": 0.5134,
+      "step": 248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999726051359e-05,
+      "loss": 0.2877,
+      "step": 249
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999722361857e-05,
+      "loss": 0.3694,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999718647676e-05,
+      "loss": 0.4496,
+      "step": 251
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999997149088154e-05,
+      "loss": 0.2565,
+      "step": 252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999711145276e-05,
+      "loss": 0.3561,
+      "step": 253
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999707357058e-05,
+      "loss": 0.3123,
+      "step": 254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999970354416e-05,
+      "loss": 0.2779,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996997065845e-05,
+      "loss": 0.6054,
+      "step": 256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999969584433e-05,
+      "loss": 1.317,
+      "step": 257
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999691957395e-05,
+      "loss": 1.1632,
+      "step": 258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996880457834e-05,
+      "loss": 1.1206,
+      "step": 259
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999684109491e-05,
+      "loss": 1.2338,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996801485205e-05,
+      "loss": 1.0508,
+      "step": 261
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999676162871e-05,
+      "loss": 1.3887,
+      "step": 262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996721525424e-05,
+      "loss": 1.1023,
+      "step": 263
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999668117534e-05,
+      "loss": 1.3006,
+      "step": 264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996640578476e-05,
+      "loss": 1.1455,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999659973482e-05,
+      "loss": 1.1464,
+      "step": 266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996558644376e-05,
+      "loss": 1.1131,
+      "step": 267
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999651730715e-05,
+      "loss": 1.2394,
+      "step": 268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996475723124e-05,
+      "loss": 1.2348,
+      "step": 269
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996433892314e-05,
+      "loss": 1.2294,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996391814705e-05,
+      "loss": 0.9849,
+      "step": 271
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999634949032e-05,
+      "loss": 1.1667,
+      "step": 272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999630691914e-05,
+      "loss": 1.28,
+      "step": 273
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996264101164e-05,
+      "loss": 1.1602,
+      "step": 274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999962210364e-05,
+      "loss": 1.2452,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996177724864e-05,
+      "loss": 1.4849,
+      "step": 276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999613416652e-05,
+      "loss": 1.1768,
+      "step": 277
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999609036139e-05,
+      "loss": 1.0437,
+      "step": 278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999604630947e-05,
+      "loss": 0.8973,
+      "step": 279
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999996002010766e-05,
+      "loss": 1.0841,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995957465276e-05,
+      "loss": 0.7193,
+      "step": 281
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999591267299e-05,
+      "loss": 1.3937,
+      "step": 282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999586763391e-05,
+      "loss": 1.2132,
+      "step": 283
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995822348044e-05,
+      "loss": 1.2343,
+      "step": 284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995776815395e-05,
+      "loss": 1.1092,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995731035955e-05,
+      "loss": 1.2969,
+      "step": 286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995685009716e-05,
+      "loss": 0.8917,
+      "step": 287
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999956387367e-05,
+      "loss": 0.9051,
+      "step": 288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995592216884e-05,
+      "loss": 1.0114,
+      "step": 289
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995545450284e-05,
+      "loss": 0.9757,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999549843689e-05,
+      "loss": 1.237,
+      "step": 291
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999545117671e-05,
+      "loss": 1.1462,
+      "step": 292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999540366974e-05,
+      "loss": 1.1132,
+      "step": 293
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999535591598e-05,
+      "loss": 1.1785,
+      "step": 294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999530791544e-05,
+      "loss": 1.0634,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995259668096e-05,
+      "loss": 1.3001,
+      "step": 296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999521117397e-05,
+      "loss": 1.7609,
+      "step": 297
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999516243306e-05,
+      "loss": 0.939,
+      "step": 298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999995113445354e-05,
+      "loss": 1.0552,
+      "step": 299
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999506421086e-05,
+      "loss": 0.9572,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999501472957e-05,
+      "loss": 0.925,
+      "step": 301
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999994965001495e-05,
+      "loss": 0.8796,
+      "step": 302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999491502664e-05,
+      "loss": 0.9126,
+      "step": 303
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999486480498e-05,
+      "loss": 1.2216,
+      "step": 304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999481433654e-05,
+      "loss": 1.1434,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999947636213e-05,
+      "loss": 1.1265,
+      "step": 306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999471265929e-05,
+      "loss": 1.1126,
+      "step": 307
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999466145048e-05,
+      "loss": 1.0995,
+      "step": 308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999994609994876e-05,
+      "loss": 0.9912,
+      "step": 309
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999994558292486e-05,
+      "loss": 1.1196,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999994506343305e-05,
+      "loss": 1.2063,
+      "step": 311
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999445414734e-05,
+      "loss": 1.1179,
+      "step": 312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999440170459e-05,
+      "loss": 1.292,
+      "step": 313
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999434901503e-05,
+      "loss": 1.1431,
+      "step": 314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999942960787e-05,
+      "loss": 1.0945,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999994242895574e-05,
+      "loss": 1.1347,
+      "step": 316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999418946566e-05,
+      "loss": 1.0005,
+      "step": 317
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999994135788956e-05,
+      "loss": 1.0945,
+      "step": 318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999994081865456e-05,
+      "loss": 1.1316,
+      "step": 319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999402769517e-05,
+      "loss": 1.1023,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999939732781e-05,
+      "loss": 1.2254,
+      "step": 321
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999391861424e-05,
+      "loss": 1.1501,
+      "step": 322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999386370359e-05,
+      "loss": 0.8656,
+      "step": 323
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999380854615e-05,
+      "loss": 1.197,
+      "step": 324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999993753141917e-05,
+      "loss": 1.109,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999993697490896e-05,
+      "loss": 1.1346,
+      "step": 326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999364159309e-05,
+      "loss": 1.0394,
+      "step": 327
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999358544849e-05,
+      "loss": 1.27,
+      "step": 328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999935290571e-05,
+      "loss": 1.2734,
+      "step": 329
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999347241893e-05,
+      "loss": 1.0557,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999341553396e-05,
+      "loss": 1.1114,
+      "step": 331
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999933584022e-05,
+      "loss": 1.4432,
+      "step": 332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999993301023654e-05,
+      "loss": 0.9807,
+      "step": 333
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999324339832e-05,
+      "loss": 1.1951,
+      "step": 334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999318552619e-05,
+      "loss": 1.0012,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999993127407284e-05,
+      "loss": 1.1595,
+      "step": 336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999993069041584e-05,
+      "loss": 1.1289,
+      "step": 337
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999301042909e-05,
+      "loss": 1.2733,
+      "step": 338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999295156981e-05,
+      "loss": 1.1768,
+      "step": 339
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999289246373e-05,
+      "loss": 1.2367,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999283311088e-05,
+      "loss": 1.0992,
+      "step": 341
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999277351122e-05,
+      "loss": 1.0069,
+      "step": 342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999992713664784e-05,
+      "loss": 0.9432,
+      "step": 343
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999265357155e-05,
+      "loss": 1.1886,
+      "step": 344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999259323154e-05,
+      "loss": 1.1029,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999253264473e-05,
+      "loss": 0.7799,
+      "step": 346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999247181113e-05,
+      "loss": 1.2012,
+      "step": 347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999241073074e-05,
+      "loss": 0.8971,
+      "step": 348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999234940357e-05,
+      "loss": 1.1842,
+      "step": 349
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999922878296e-05,
+      "loss": 1.114,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999222600885e-05,
+      "loss": 1.3876,
+      "step": 351
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999921639413e-05,
+      "loss": 1.2368,
+      "step": 352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999992101626966e-05,
+      "loss": 1.2026,
+      "step": 353
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999992039065844e-05,
+      "loss": 1.1897,
+      "step": 354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999197625794e-05,
+      "loss": 1.0612,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999191320324e-05,
+      "loss": 1.0526,
+      "step": 356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999184990174e-05,
+      "loss": 1.1144,
+      "step": 357
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999178635346e-05,
+      "loss": 1.3094,
+      "step": 358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999172255839e-05,
+      "loss": 0.4662,
+      "step": 359
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999165851653e-05,
+      "loss": 0.2525,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999991594227885e-05,
+      "loss": 0.8735,
+      "step": 361
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999991529692444e-05,
+      "loss": 1.0231,
+      "step": 362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999991464910226e-05,
+      "loss": 1.2666,
+      "step": 363
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999139988121e-05,
+      "loss": 1.1313,
+      "step": 364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999913346054e-05,
+      "loss": 1.2996,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999912690828e-05,
+      "loss": 1.1301,
+      "step": 366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999120331342e-05,
+      "loss": 1.1243,
+      "step": 367
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999113729724e-05,
+      "loss": 1.3049,
+      "step": 368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999107103428e-05,
+      "loss": 1.3117,
+      "step": 369
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999100452453e-05,
+      "loss": 1.1549,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999093776798e-05,
+      "loss": 1.0233,
+      "step": 371
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999087076465e-05,
+      "loss": 1.3463,
+      "step": 372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999080351453e-05,
+      "loss": 1.0907,
+      "step": 373
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999073601762e-05,
+      "loss": 0.9733,
+      "step": 374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999066827392e-05,
+      "loss": 1.4368,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999060028343e-05,
+      "loss": 5.9099,
+      "step": 376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999990532046156e-05,
+      "loss": 3.7277,
+      "step": 377
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999990463562086e-05,
+      "loss": 1.1872,
+      "step": 378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999039483123e-05,
+      "loss": 1.0093,
+      "step": 379
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999990325853577e-05,
+      "loss": 1.0568,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999990256629145e-05,
+      "loss": 0.831,
+      "step": 381
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999018715792e-05,
+      "loss": 0.9143,
+      "step": 382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999999011743991e-05,
+      "loss": 1.6355,
+      "step": 383
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999900474751e-05,
+      "loss": 1.4826,
+      "step": 384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999899772635e-05,
+      "loss": 1.0458,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998990680512e-05,
+      "loss": 1.1318,
+      "step": 386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999989836099946e-05,
+      "loss": 1.0718,
+      "step": 387
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999989765147994e-05,
+      "loss": 0.8531,
+      "step": 388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998969394924e-05,
+      "loss": 0.9618,
+      "step": 389
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999989622503695e-05,
+      "loss": 1.2255,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998955081137e-05,
+      "loss": 1.3624,
+      "step": 391
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998947887225e-05,
+      "loss": 1.1627,
+      "step": 392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998940668634e-05,
+      "loss": 1.2721,
+      "step": 393
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998933425365e-05,
+      "loss": 1.1621,
+      "step": 394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999989261574154e-05,
+      "loss": 1.2259,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998918864788e-05,
+      "loss": 1.1092,
+      "step": 396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998911547481e-05,
+      "loss": 1.205,
+      "step": 397
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999989042054954e-05,
+      "loss": 1.164,
+      "step": 398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998896838831e-05,
+      "loss": 1.083,
+      "step": 399
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998889447488e-05,
+      "loss": 1.0954,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998882031466e-05,
+      "loss": 0.9265,
+      "step": 401
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999988745907645e-05,
+      "loss": 0.8312,
+      "step": 402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998867125384e-05,
+      "loss": 0.8345,
+      "step": 403
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998859635325e-05,
+      "loss": 0.7232,
+      "step": 404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999988521205864e-05,
+      "loss": 0.6289,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999884458117e-05,
+      "loss": 0.9681,
+      "step": 406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999988370170735e-05,
+      "loss": 1.2402,
+      "step": 407
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998829428299e-05,
+      "loss": 1.1044,
+      "step": 408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998821814845e-05,
+      "loss": 0.9407,
+      "step": 409
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998814176713e-05,
+      "loss": 1.0668,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998806513901e-05,
+      "loss": 1.1896,
+      "step": 411
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999987988264104e-05,
+      "loss": 1.1691,
+      "step": 412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998791114241e-05,
+      "loss": 1.2159,
+      "step": 413
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998783377392e-05,
+      "loss": 1.3894,
+      "step": 414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998775615865e-05,
+      "loss": 1.329,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999987678296585e-05,
+      "loss": 1.0454,
+      "step": 416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998760018773e-05,
+      "loss": 0.9511,
+      "step": 417
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998752183209e-05,
+      "loss": 0.4759,
+      "step": 418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999987443229654e-05,
+      "loss": 0.8928,
+      "step": 419
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999987364380435e-05,
+      "loss": 0.9391,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999987285284424e-05,
+      "loss": 1.1387,
+      "step": 421
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998720594162e-05,
+      "loss": 1.1293,
+      "step": 422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999987126352034e-05,
+      "loss": 1.2704,
+      "step": 423
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998704651566e-05,
+      "loss": 1.137,
+      "step": 424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998696643249e-05,
+      "loss": 0.9722,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999986886102536e-05,
+      "loss": 1.2422,
+      "step": 426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998680552579e-05,
+      "loss": 1.0776,
+      "step": 427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998672470225e-05,
+      "loss": 0.8233,
+      "step": 428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998664363192e-05,
+      "loss": 1.2957,
+      "step": 429
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998656231481e-05,
+      "loss": 1.0972,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998648075091e-05,
+      "loss": 1.2265,
+      "step": 431
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998639894021e-05,
+      "loss": 1.3584,
+      "step": 432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998631688273e-05,
+      "loss": 1.9795,
+      "step": 433
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998623457846e-05,
+      "loss": 1.4203,
+      "step": 434
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999861520274e-05,
+      "loss": 1.595,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998606922955e-05,
+      "loss": 0.6731,
+      "step": 436
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998598618491e-05,
+      "loss": 1.2655,
+      "step": 437
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999985902893484e-05,
+      "loss": 1.1741,
+      "step": 438
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999985819355266e-05,
+      "loss": 1.0423,
+      "step": 439
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998573557026e-05,
+      "loss": 1.1478,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998565153846e-05,
+      "loss": 1.044,
+      "step": 441
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998556725987e-05,
+      "loss": 0.7539,
+      "step": 442
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998548273449e-05,
+      "loss": 0.685,
+      "step": 443
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998539796233e-05,
+      "loss": 5.6911,
+      "step": 444
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998531294338e-05,
+      "loss": 5.9401,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998522767764e-05,
+      "loss": 6.034,
+      "step": 446
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999851421651e-05,
+      "loss": 5.9444,
+      "step": 447
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998505640578e-05,
+      "loss": 5.9153,
+      "step": 448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998497039967e-05,
+      "loss": 5.8,
+      "step": 449
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984884146765e-05,
+      "loss": 5.7652,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984797647076e-05,
+      "loss": 5.5309,
+      "step": 451
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984710900596e-05,
+      "loss": 5.6267,
+      "step": 452
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984623907324e-05,
+      "loss": 1.8665,
+      "step": 453
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984536667274e-05,
+      "loss": 1.5152,
+      "step": 454
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984449180425e-05,
+      "loss": 0.5595,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984361446786e-05,
+      "loss": 1.2043,
+      "step": 456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998427346636e-05,
+      "loss": 1.1112,
+      "step": 457
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984185239145e-05,
+      "loss": 1.4293,
+      "step": 458
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998409676514e-05,
+      "loss": 1.0783,
+      "step": 459
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999984008044345e-05,
+      "loss": 1.049,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998391907677e-05,
+      "loss": 1.029,
+      "step": 461
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998382986239e-05,
+      "loss": 0.8213,
+      "step": 462
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998374040123e-05,
+      "loss": 1.2395,
+      "step": 463
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999983650693274e-05,
+      "loss": 1.3614,
+      "step": 464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998356073854e-05,
+      "loss": 1.1796,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999983470537e-05,
+      "loss": 1.2114,
+      "step": 466
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998338008869e-05,
+      "loss": 0.973,
+      "step": 467
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998328939358e-05,
+      "loss": 0.9211,
+      "step": 468
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999983198451684e-05,
+      "loss": 1.0192,
+      "step": 469
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998310726299e-05,
+      "loss": 1.1802,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998301582752e-05,
+      "loss": 0.9868,
+      "step": 471
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998292414525e-05,
+      "loss": 1.195,
+      "step": 472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999828322162e-05,
+      "loss": 1.1628,
+      "step": 473
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998274004035e-05,
+      "loss": 0.9806,
+      "step": 474
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998264761772e-05,
+      "loss": 1.0956,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999825549483e-05,
+      "loss": 1.3023,
+      "step": 476
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999982462032085e-05,
+      "loss": 1.203,
+      "step": 477
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998236886908e-05,
+      "loss": 1.5488,
+      "step": 478
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998227545929e-05,
+      "loss": 1.3148,
+      "step": 479
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998218180271e-05,
+      "loss": 1.2602,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999982087899346e-05,
+      "loss": 1.1743,
+      "step": 481
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998199374918e-05,
+      "loss": 0.9299,
+      "step": 482
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998189935224e-05,
+      "loss": 0.7564,
+      "step": 483
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999981804708504e-05,
+      "loss": 1.0182,
+      "step": 484
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998170981798e-05,
+      "loss": 1.1082,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998161468066e-05,
+      "loss": 1.3019,
+      "step": 486
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998151929656e-05,
+      "loss": 0.8937,
+      "step": 487
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998142366566e-05,
+      "loss": 0.8715,
+      "step": 488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998132778798e-05,
+      "loss": 0.6235,
+      "step": 489
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998123166351e-05,
+      "loss": 0.5427,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998113529224e-05,
+      "loss": 0.9652,
+      "step": 491
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998103867419e-05,
+      "loss": 1.0241,
+      "step": 492
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998094180935e-05,
+      "loss": 1.2874,
+      "step": 493
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998084469772e-05,
+      "loss": 1.1236,
+      "step": 494
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999980747339305e-05,
+      "loss": 1.2312,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998064973409e-05,
+      "loss": 1.1679,
+      "step": 496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999980551882095e-05,
+      "loss": 0.7927,
+      "step": 497
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999980453783306e-05,
+      "loss": 0.7081,
+      "step": 498
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998035543773e-05,
+      "loss": 1.2299,
+      "step": 499
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998025684536e-05,
+      "loss": 1.1899,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.1330914497375488,
+      "eval_runtime": 85.2447,
+      "eval_samples_per_second": 16.247,
+      "eval_steps_per_second": 4.071,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998015800621e-05,
+      "loss": 0.9488,
+      "step": 501
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998005892026e-05,
+      "loss": 0.9233,
+      "step": 502
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997995958753e-05,
+      "loss": 1.0061,
+      "step": 503
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999979860008006e-05,
+      "loss": 0.8307,
+      "step": 504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999797601817e-05,
+      "loss": 0.9544,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999979660108597e-05,
+      "loss": 0.5917,
+      "step": 506
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999979559788704e-05,
+      "loss": 0.5971,
+      "step": 507
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997945922202e-05,
+      "loss": 1.3349,
+      "step": 508
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997935840855e-05,
+      "loss": 1.0993,
+      "step": 509
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999792573483e-05,
+      "loss": 1.0986,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999979156041255e-05,
+      "loss": 1.0709,
+      "step": 511
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997905448741e-05,
+      "loss": 1.1506,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997895268679e-05,
+      "loss": 0.9141,
+      "step": 513
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997885063937e-05,
+      "loss": 1.5013,
+      "step": 514
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997874834517e-05,
+      "loss": 0.9547,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997864580418e-05,
+      "loss": 1.09,
+      "step": 516
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997854301639e-05,
+      "loss": 1.1179,
+      "step": 517
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997843998182e-05,
+      "loss": 1.0737,
+      "step": 518
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997833670046e-05,
+      "loss": 0.7466,
+      "step": 519
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997823317231e-05,
+      "loss": 0.9024,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997812939737e-05,
+      "loss": 0.9933,
+      "step": 521
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997802537564e-05,
+      "loss": 0.8115,
+      "step": 522
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997792110712e-05,
+      "loss": 0.8077,
+      "step": 523
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999977816591816e-05,
+      "loss": 0.6865,
+      "step": 524
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997771182972e-05,
+      "loss": 0.5826,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999977606820834e-05,
+      "loss": 1.1181,
+      "step": 526
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997750156515e-05,
+      "loss": 1.4692,
+      "step": 527
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997739606269e-05,
+      "loss": 1.0833,
+      "step": 528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997729031344e-05,
+      "loss": 1.1319,
+      "step": 529
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997718431739e-05,
+      "loss": 1.252,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997707807456e-05,
+      "loss": 1.0724,
+      "step": 531
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997697158494e-05,
+      "loss": 1.2738,
+      "step": 532
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997686484853e-05,
+      "loss": 0.9913,
+      "step": 533
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997675786533e-05,
+      "loss": 1.1738,
+      "step": 534
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997665063534e-05,
+      "loss": 1.3109,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997654315855e-05,
+      "loss": 1.3323,
+      "step": 536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997643543499e-05,
+      "loss": 1.1117,
+      "step": 537
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999976327464635e-05,
+      "loss": 1.0758,
+      "step": 538
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997621924749e-05,
+      "loss": 1.029,
+      "step": 539
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997611078355e-05,
+      "loss": 1.2414,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997600207282e-05,
+      "loss": 1.1829,
+      "step": 541
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999975893115315e-05,
+      "loss": 1.037,
+      "step": 542
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999975783911016e-05,
+      "loss": 0.9917,
+      "step": 543
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997567445992e-05,
+      "loss": 0.9406,
+      "step": 544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997556476204e-05,
+      "loss": 1.1532,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997545481736e-05,
+      "loss": 0.9999,
+      "step": 546
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999975344625906e-05,
+      "loss": 1.0536,
+      "step": 547
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999975234187656e-05,
+      "loss": 1.1719,
+      "step": 548
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999975123502615e-05,
+      "loss": 1.0755,
+      "step": 549
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997501257079e-05,
+      "loss": 0.1063,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997490139217e-05,
+      "loss": 0.0709,
+      "step": 551
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997478996676e-05,
+      "loss": 0.2467,
+      "step": 552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997467829457e-05,
+      "loss": 0.2245,
+      "step": 553
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999974566375585e-05,
+      "loss": 0.1978,
+      "step": 554
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997445420981e-05,
+      "loss": 0.1696,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999974341797254e-05,
+      "loss": 0.1546,
+      "step": 556
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999974229137894e-05,
+      "loss": 0.1919,
+      "step": 557
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997411623175e-05,
+      "loss": 0.2029,
+      "step": 558
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997400307882e-05,
+      "loss": 0.1537,
+      "step": 559
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999738896791e-05,
+      "loss": 0.1449,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999973776032594e-05,
+      "loss": 0.7279,
+      "step": 561
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999736621393e-05,
+      "loss": 0.6953,
+      "step": 562
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997354799921e-05,
+      "loss": 0.6295,
+      "step": 563
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997343361233e-05,
+      "loss": 0.5171,
+      "step": 564
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999973318978664e-05,
+      "loss": 0.486,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997320409821e-05,
+      "loss": 1.2449,
+      "step": 566
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997308897097e-05,
+      "loss": 1.4136,
+      "step": 567
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999972973596935e-05,
+      "loss": 1.2475,
+      "step": 568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997285797611e-05,
+      "loss": 1.3221,
+      "step": 569
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999972742108495e-05,
+      "loss": 1.1538,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999972625994095e-05,
+      "loss": 0.8241,
+      "step": 571
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997250963291e-05,
+      "loss": 1.1044,
+      "step": 572
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999972393024926e-05,
+      "loss": 1.2559,
+      "step": 573
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997227617016e-05,
+      "loss": 1.0469,
+      "step": 574
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999721590686e-05,
+      "loss": 1.2542,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997204172025e-05,
+      "loss": 1.1862,
+      "step": 576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997192412512e-05,
+      "loss": 1.3234,
+      "step": 577
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999971806283185e-05,
+      "loss": 0.8016,
+      "step": 578
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997168819447e-05,
+      "loss": 1.3645,
+      "step": 579
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997156985897e-05,
+      "loss": 1.1961,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997145127668e-05,
+      "loss": 1.2315,
+      "step": 581
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999971332447594e-05,
+      "loss": 1.0403,
+      "step": 582
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997121337172e-05,
+      "loss": 1.1351,
+      "step": 583
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999971094049066e-05,
+      "loss": 1.1116,
+      "step": 584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997097447961e-05,
+      "loss": 1.2778,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997085466337e-05,
+      "loss": 1.1626,
+      "step": 586
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997073460034e-05,
+      "loss": 1.0976,
+      "step": 587
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999970614290527e-05,
+      "loss": 1.3518,
+      "step": 588
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997049373392e-05,
+      "loss": 1.1483,
+      "step": 589
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997037293053e-05,
+      "loss": 1.1493,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997025188034e-05,
+      "loss": 1.1073,
+      "step": 591
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997013058337e-05,
+      "loss": 1.3351,
+      "step": 592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999970009039604e-05,
+      "loss": 1.0137,
+      "step": 593
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996988724905e-05,
+      "loss": 1.159,
+      "step": 594
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996976521171e-05,
+      "loss": 1.0414,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996964292758e-05,
+      "loss": 1.2767,
+      "step": 596
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999969520396664e-05,
+      "loss": 1.0204,
+      "step": 597
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999969397618954e-05,
+      "loss": 0.8231,
+      "step": 598
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996927459445e-05,
+      "loss": 1.2339,
+      "step": 599
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999969151323165e-05,
+      "loss": 1.1286,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996902780509e-05,
+      "loss": 1.3251,
+      "step": 601
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999968904040224e-05,
+      "loss": 1.1275,
+      "step": 602
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996878002857e-05,
+      "loss": 0.9398,
+      "step": 603
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999968655770124e-05,
+      "loss": 1.1224,
+      "step": 604
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996853126489e-05,
+      "loss": 1.0039,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999968406512865e-05,
+      "loss": 1.0953,
+      "step": 606
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996828151405e-05,
+      "loss": 1.0682,
+      "step": 607
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996815626846e-05,
+      "loss": 0.981,
+      "step": 608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996803077607e-05,
+      "loss": 1.135,
+      "step": 609
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996790503689e-05,
+      "loss": 2.0639,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999967779050915e-05,
+      "loss": 1.208,
+      "step": 611
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996765281816e-05,
+      "loss": 0.9878,
+      "step": 612
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999967526338615e-05,
+      "loss": 1.0357,
+      "step": 613
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996739961228e-05,
+      "loss": 1.0447,
+      "step": 614
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999967272639156e-05,
+      "loss": 1.0037,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996714541924e-05,
+      "loss": 1.256,
+      "step": 616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999967017952544e-05,
+      "loss": 1.0181,
+      "step": 617
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996689023905e-05,
+      "loss": 0.8774,
+      "step": 618
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999966762278766e-05,
+      "loss": 1.0415,
+      "step": 619
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999966634071696e-05,
+      "loss": 1.1236,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999966505617835e-05,
+      "loss": 1.1345,
+      "step": 621
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996637691718e-05,
+      "loss": 1.103,
+      "step": 622
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999966247969745e-05,
+      "loss": 0.8407,
+      "step": 623
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996611877552e-05,
+      "loss": 1.2822,
+      "step": 624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996598933451e-05,
+      "loss": 1.3973,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999965859646704e-05,
+      "loss": 1.1597,
+      "step": 626
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996572971211e-05,
+      "loss": 1.9025,
+      "step": 627
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999965599530726e-05,
+      "loss": 1.2152,
+      "step": 628
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996546910255e-05,
+      "loss": 1.157,
+      "step": 629
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996533842759e-05,
+      "loss": 1.3811,
+      "step": 630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996520750584e-05,
+      "loss": 1.2883,
+      "step": 631
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999650763373e-05,
+      "loss": 1.4924,
+      "step": 632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999964944921974e-05,
+      "loss": 1.0566,
+      "step": 633
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996481325985e-05,
+      "loss": 1.3231,
+      "step": 634
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996468135094e-05,
+      "loss": 1.3814,
+      "step": 635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996454919524e-05,
+      "loss": 1.0348,
+      "step": 636
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999964416792764e-05,
+      "loss": 1.1868,
+      "step": 637
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996428414349e-05,
+      "loss": 1.2078,
+      "step": 638
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996415124743e-05,
+      "loss": 1.1857,
+      "step": 639
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999964018104574e-05,
+      "loss": 1.083,
+      "step": 640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996388471493e-05,
+      "loss": 1.2147,
+      "step": 641
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999637510785e-05,
+      "loss": 3.2249,
+      "step": 642
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996361719528e-05,
+      "loss": 3.0262,
+      "step": 643
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996348306527e-05,
+      "loss": 1.0029,
+      "step": 644
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996334868847e-05,
+      "loss": 1.5653,
+      "step": 645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999963214064885e-05,
+      "loss": 1.0408,
+      "step": 646
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999963079194505e-05,
+      "loss": 1.037,
+      "step": 647
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996294407734e-05,
+      "loss": 1.2011,
+      "step": 648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999962808713385e-05,
+      "loss": 1.1739,
+      "step": 649
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996267310264e-05,
+      "loss": 1.3935,
+      "step": 650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996253724511e-05,
+      "loss": 1.3371,
+      "step": 651
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996240114079e-05,
+      "loss": 1.3247,
+      "step": 652
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996226478968e-05,
+      "loss": 1.2976,
+      "step": 653
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999962128191774e-05,
+      "loss": 1.1112,
+      "step": 654
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996199134709e-05,
+      "loss": 1.1392,
+      "step": 655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999961854255606e-05,
+      "loss": 0.9464,
+      "step": 656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999961716917345e-05,
+      "loss": 1.1398,
+      "step": 657
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996157933228e-05,
+      "loss": 1.2702,
+      "step": 658
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999961441500436e-05,
+      "loss": 1.046,
+      "step": 659
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996130342181e-05,
+      "loss": 1.203,
+      "step": 660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996116509637e-05,
+      "loss": 1.1421,
+      "step": 661
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996102652417e-05,
+      "loss": 0.8796,
+      "step": 662
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999960887705165e-05,
+      "loss": 1.0829,
+      "step": 663
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996074863938e-05,
+      "loss": 1.2028,
+      "step": 664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996060932679e-05,
+      "loss": 1.2499,
+      "step": 665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999960469767427e-05,
+      "loss": 1.5727,
+      "step": 666
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999960329961264e-05,
+      "loss": 1.4672,
+      "step": 667
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996018990832e-05,
+      "loss": 1.3084,
+      "step": 668
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996004960858e-05,
+      "loss": 1.1889,
+      "step": 669
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999959909062054e-05,
+      "loss": 1.1393,
+      "step": 670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995976826874e-05,
+      "loss": 1.2297,
+      "step": 671
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995962722864e-05,
+      "loss": 1.2911,
+      "step": 672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995948594174e-05,
+      "loss": 1.0032,
+      "step": 673
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999959344408066e-05,
+      "loss": 0.9387,
+      "step": 674
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995920262759e-05,
+      "loss": 0.9681,
+      "step": 675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995906060033e-05,
+      "loss": 0.9953,
+      "step": 676
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995891832628e-05,
+      "loss": 1.0853,
+      "step": 677
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999958775805446e-05,
+      "loss": 0.9147,
+      "step": 678
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995863303781e-05,
+      "loss": 1.2149,
+      "step": 679
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999584900234e-05,
+      "loss": 0.9156,
+      "step": 680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999583467622e-05,
+      "loss": 1.2145,
+      "step": 681
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999582032542e-05,
+      "loss": 1.2697,
+      "step": 682
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995805949942e-05,
+      "loss": 1.5116,
+      "step": 683
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995791549784e-05,
+      "loss": 1.0923,
+      "step": 684
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999957771249483e-05,
+      "loss": 1.0256,
+      "step": 685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999957626754336e-05,
+      "loss": 1.2579,
+      "step": 686
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995748201239e-05,
+      "loss": 1.8896,
+      "step": 687
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995733702366e-05,
+      "loss": 1.9751,
+      "step": 688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995719178815e-05,
+      "loss": 2.1633,
+      "step": 689
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995704630584e-05,
+      "loss": 1.6439,
+      "step": 690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995690057674e-05,
+      "loss": 0.7501,
+      "step": 691
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999956754600865e-05,
+      "loss": 0.9558,
+      "step": 692
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999956608378184e-05,
+      "loss": 1.0196,
+      "step": 693
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999956461908725e-05,
+      "loss": 1.3408,
+      "step": 694
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999956315192475e-05,
+      "loss": 1.1365,
+      "step": 695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995616822943e-05,
+      "loss": 1.2901,
+      "step": 696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999560210196e-05,
+      "loss": 1.348,
+      "step": 697
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995587356298e-05,
+      "loss": 1.3301,
+      "step": 698
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995572585957e-05,
+      "loss": 1.2014,
+      "step": 699
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999955577909376e-05,
+      "loss": 1.1904,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995542971238e-05,
+      "loss": 1.4346,
+      "step": 701
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995528126862e-05,
+      "loss": 1.1291,
+      "step": 702
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995513257805e-05,
+      "loss": 1.0225,
+      "step": 703
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999954983640696e-05,
+      "loss": 1.902,
+      "step": 704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995483445655e-05,
+      "loss": 1.111,
+      "step": 705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995468502562e-05,
+      "loss": 1.0966,
+      "step": 706
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999954535347906e-05,
+      "loss": 0.9121,
+      "step": 707
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999543854234e-05,
+      "loss": 0.9493,
+      "step": 708
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999542352521e-05,
+      "loss": 0.9592,
+      "step": 709
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995408483401e-05,
+      "loss": 0.8201,
+      "step": 710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995393416913e-05,
+      "loss": 1.1763,
+      "step": 711
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995378325747e-05,
+      "loss": 1.1213,
+      "step": 712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999953632099015e-05,
+      "loss": 1.0797,
+      "step": 713
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995348069377e-05,
+      "loss": 0.9418,
+      "step": 714
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995332904174e-05,
+      "loss": 1.2825,
+      "step": 715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999953177142915e-05,
+      "loss": 1.2253,
+      "step": 716
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999953024997305e-05,
+      "loss": 1.0951,
+      "step": 717
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999952872604903e-05,
+      "loss": 1.2137,
+      "step": 718
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995271996572e-05,
+      "loss": 1.1174,
+      "step": 719
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995256707975e-05,
+      "loss": 1.2376,
+      "step": 720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995241394698e-05,
+      "loss": 1.2504,
+      "step": 721
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995226056742e-05,
+      "loss": 1.1348,
+      "step": 722
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995210694107e-05,
+      "loss": 1.1022,
+      "step": 723
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995195306794e-05,
+      "loss": 1.1405,
+      "step": 724
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995179894802e-05,
+      "loss": 1.237,
+      "step": 725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999951644581314e-05,
+      "loss": 1.009,
+      "step": 726
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995148996781e-05,
+      "loss": 1.4083,
+      "step": 727
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995133510752e-05,
+      "loss": 1.8452,
+      "step": 728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995118000044e-05,
+      "loss": 1.0214,
+      "step": 729
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995102464657e-05,
+      "loss": 1.0336,
+      "step": 730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999950869045916e-05,
+      "loss": 1.1379,
+      "step": 731
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999950713198474e-05,
+      "loss": 1.0476,
+      "step": 732
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999950557104234e-05,
+      "loss": 1.3398,
+      "step": 733
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999950400763216e-05,
+      "loss": 1.0744,
+      "step": 734
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999950244175406e-05,
+      "loss": 1.0007,
+      "step": 735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999950087340805e-05,
+      "loss": 1.3092,
+      "step": 736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994993025941e-05,
+      "loss": 1.1256,
+      "step": 737
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999949772931235e-05,
+      "loss": 1.6756,
+      "step": 738
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999949615356266e-05,
+      "loss": 1.9275,
+      "step": 739
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994945753451e-05,
+      "loss": 0.9195,
+      "step": 740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994929946597e-05,
+      "loss": 1.3474,
+      "step": 741
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999949141150624e-05,
+      "loss": 1.4102,
+      "step": 742
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994898258851e-05,
+      "loss": 1.3585,
+      "step": 743
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994882377959e-05,
+      "loss": 1.2699,
+      "step": 744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994866472389e-05,
+      "loss": 1.0139,
+      "step": 745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999485054214e-05,
+      "loss": 0.8573,
+      "step": 746
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999948345872114e-05,
+      "loss": 0.7978,
+      "step": 747
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994818607605e-05,
+      "loss": 0.9753,
+      "step": 748
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994802603319e-05,
+      "loss": 1.3202,
+      "step": 749
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994786574354e-05,
+      "loss": 0.3981,
+      "step": 750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999477052071e-05,
+      "loss": 0.3395,
+      "step": 751
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999947544423885e-05,
+      "loss": 0.3216,
+      "step": 752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999947383393876e-05,
+      "loss": 0.215,
+      "step": 753
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999947222117074e-05,
+      "loss": 0.2558,
+      "step": 754
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999947060593475e-05,
+      "loss": 0.1257,
+      "step": 755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999468988231e-05,
+      "loss": 0.096,
+      "step": 756
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994673680592e-05,
+      "loss": 0.0708,
+      "step": 757
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994657454197e-05,
+      "loss": 0.0575,
+      "step": 758
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994641203122e-05,
+      "loss": 0.0431,
+      "step": 759
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999946249273687e-05,
+      "loss": 0.1335,
+      "step": 760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999946086269365e-05,
+      "loss": 0.1145,
+      "step": 761
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999945923018245e-05,
+      "loss": 0.3766,
+      "step": 762
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994575952035e-05,
+      "loss": 1.2983,
+      "step": 763
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994559577565e-05,
+      "loss": 1.3596,
+      "step": 764
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994543178417e-05,
+      "loss": 1.3108,
+      "step": 765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999452675459e-05,
+      "loss": 1.0224,
+      "step": 766
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994510306085e-05,
+      "loss": 1.2363,
+      "step": 767
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999944938329e-05,
+      "loss": 1.2409,
+      "step": 768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994477335036e-05,
+      "loss": 1.1756,
+      "step": 769
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999944608124935e-05,
+      "loss": 0.8086,
+      "step": 770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999944442652726e-05,
+      "loss": 1.2566,
+      "step": 771
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994427693372e-05,
+      "loss": 1.3554,
+      "step": 772
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999944110967925e-05,
+      "loss": 1.1814,
+      "step": 773
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994394475535e-05,
+      "loss": 1.1509,
+      "step": 774
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994377829598e-05,
+      "loss": 0.5566,
+      "step": 775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994361158982e-05,
+      "loss": 1.0299,
+      "step": 776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999943444636874e-05,
+      "loss": 1.093,
+      "step": 777
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994327743713e-05,
+      "loss": 1.3153,
+      "step": 778
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994310999061e-05,
+      "loss": 1.3283,
+      "step": 779
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999942942297296e-05,
+      "loss": 1.1775,
+      "step": 780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994277435719e-05,
+      "loss": 1.3292,
+      "step": 781
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999942606170296e-05,
+      "loss": 1.416,
+      "step": 782
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999942437736616e-05,
+      "loss": 1.2101,
+      "step": 783
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999942269056143e-05,
+      "loss": 1.0418,
+      "step": 784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999942100128887e-05,
+      "loss": 1.1471,
+      "step": 785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994193095484e-05,
+      "loss": 1.1467,
+      "step": 786
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999941761534005e-05,
+      "loss": 1.0736,
+      "step": 787
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994159186638e-05,
+      "loss": 1.2616,
+      "step": 788
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994142195197e-05,
+      "loss": 1.2858,
+      "step": 789
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994125179076e-05,
+      "loss": 1.3641,
+      "step": 790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994108138277e-05,
+      "loss": 1.2877,
+      "step": 791
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994091072799e-05,
+      "loss": 0.8191,
+      "step": 792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994073982642e-05,
+      "loss": 0.6352,
+      "step": 793
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994056867806e-05,
+      "loss": 1.0221,
+      "step": 794
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999940397282906e-05,
+      "loss": 0.8679,
+      "step": 795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999940225640976e-05,
+      "loss": 1.6629,
+      "step": 796
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999994005375225e-05,
+      "loss": 1.2877,
+      "step": 797
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993988161673e-05,
+      "loss": 1.117,
+      "step": 798
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993970923443e-05,
+      "loss": 1.016,
+      "step": 799
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993953660534e-05,
+      "loss": 1.0511,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993936372946e-05,
+      "loss": 1.0768,
+      "step": 801
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993919060679e-05,
+      "loss": 1.0474,
+      "step": 802
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993901723733e-05,
+      "loss": 0.9152,
+      "step": 803
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993884362108e-05,
+      "loss": 1.1368,
+      "step": 804
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999938669758045e-05,
+      "loss": 1.0106,
+      "step": 805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993849564822e-05,
+      "loss": 1.1477,
+      "step": 806
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999938321291604e-05,
+      "loss": 1.2261,
+      "step": 807
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999381466882e-05,
+      "loss": 0.8828,
+      "step": 808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999937971838004e-05,
+      "loss": 0.9267,
+      "step": 809
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999937796741033e-05,
+      "loss": 0.9857,
+      "step": 810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999937621397265e-05,
+      "loss": 0.7243,
+      "step": 811
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999374458067e-05,
+      "loss": 0.2533,
+      "step": 812
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993726996935e-05,
+      "loss": 0.1383,
+      "step": 813
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999937093885217e-05,
+      "loss": 0.1169,
+      "step": 814
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993691755429e-05,
+      "loss": 0.5441,
+      "step": 815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993674097658e-05,
+      "loss": 1.1295,
+      "step": 816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993656415208e-05,
+      "loss": 1.2738,
+      "step": 817
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993638708079e-05,
+      "loss": 1.1248,
+      "step": 818
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999362097627e-05,
+      "loss": 1.1047,
+      "step": 819
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993603219784e-05,
+      "loss": 1.0343,
+      "step": 820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993585438618e-05,
+      "loss": 1.3274,
+      "step": 821
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993567632773e-05,
+      "loss": 1.233,
+      "step": 822
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999354980225e-05,
+      "loss": 0.9655,
+      "step": 823
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993531947047e-05,
+      "loss": 1.2446,
+      "step": 824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993514067166e-05,
+      "loss": 1.204,
+      "step": 825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993496162606e-05,
+      "loss": 1.334,
+      "step": 826
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999934782333666e-05,
+      "loss": 1.5279,
+      "step": 827
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993460279448e-05,
+      "loss": 0.6742,
+      "step": 828
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993442300851e-05,
+      "loss": 0.4579,
+      "step": 829
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993424297576e-05,
+      "loss": 0.55,
+      "step": 830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999934062696215e-05,
+      "loss": 0.8013,
+      "step": 831
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993388216988e-05,
+      "loss": 0.9383,
+      "step": 832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993370139675e-05,
+      "loss": 1.1955,
+      "step": 833
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993352037684e-05,
+      "loss": 1.0137,
+      "step": 834
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993333911014e-05,
+      "loss": 0.9991,
+      "step": 835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999933157596645e-05,
+      "loss": 0.9199,
+      "step": 836
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993297583637e-05,
+      "loss": 0.4115,
+      "step": 837
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999327938293e-05,
+      "loss": 0.3562,
+      "step": 838
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993261157544e-05,
+      "loss": 1.0957,
+      "step": 839
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999324290748e-05,
+      "loss": 0.9675,
+      "step": 840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993224632736e-05,
+      "loss": 0.9914,
+      "step": 841
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999932063333144e-05,
+      "loss": 0.9066,
+      "step": 842
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999931880092135e-05,
+      "loss": 1.0886,
+      "step": 843
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993169660433e-05,
+      "loss": 0.969,
+      "step": 844
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993151286974e-05,
+      "loss": 1.0533,
+      "step": 845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993132888836e-05,
+      "loss": 1.0183,
+      "step": 846
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999311446602e-05,
+      "loss": 1.1548,
+      "step": 847
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993096018524e-05,
+      "loss": 0.987,
+      "step": 848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999307754635e-05,
+      "loss": 1.4268,
+      "step": 849
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999930590494956e-05,
+      "loss": 1.5397,
+      "step": 850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999930405279635e-05,
+      "loss": 0.9337,
+      "step": 851
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993021981752e-05,
+      "loss": 1.3076,
+      "step": 852
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993003410863e-05,
+      "loss": 1.1003,
+      "step": 853
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999929848152936e-05,
+      "loss": 1.0851,
+      "step": 854
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992966195046e-05,
+      "loss": 1.2169,
+      "step": 855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999294755012e-05,
+      "loss": 1.2069,
+      "step": 856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992928880514e-05,
+      "loss": 1.184,
+      "step": 857
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999291018623e-05,
+      "loss": 1.4656,
+      "step": 858
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992891467267e-05,
+      "loss": 1.2775,
+      "step": 859
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992872723624e-05,
+      "loss": 1.6209,
+      "step": 860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999928539553034e-05,
+      "loss": 1.8363,
+      "step": 861
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992835162304e-05,
+      "loss": 1.6992,
+      "step": 862
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999928163446254e-05,
+      "loss": 2.3404,
+      "step": 863
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992797502268e-05,
+      "loss": 1.5144,
+      "step": 864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992778635231e-05,
+      "loss": 1.5439,
+      "step": 865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999927597435156e-05,
+      "loss": 1.3937,
+      "step": 866
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992740827121e-05,
+      "loss": 1.6777,
+      "step": 867
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992721886048e-05,
+      "loss": 1.1788,
+      "step": 868
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992702920296e-05,
+      "loss": 1.2119,
+      "step": 869
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992683929865e-05,
+      "loss": 1.3642,
+      "step": 870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992664914755e-05,
+      "loss": 0.9385,
+      "step": 871
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992645874967e-05,
+      "loss": 1.0575,
+      "step": 872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992626810499e-05,
+      "loss": 1.091,
+      "step": 873
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992607721353e-05,
+      "loss": 1.205,
+      "step": 874
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999925886075275e-05,
+      "loss": 0.9686,
+      "step": 875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999925694690233e-05,
+      "loss": 1.0848,
+      "step": 876
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992550305841e-05,
+      "loss": 1.1814,
+      "step": 877
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992531117978e-05,
+      "loss": 0.8181,
+      "step": 878
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992511905438e-05,
+      "loss": 1.0436,
+      "step": 879
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992492668219e-05,
+      "loss": 1.1664,
+      "step": 880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999924734063195e-05,
+      "loss": 1.5344,
+      "step": 881
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992454119743e-05,
+      "loss": 1.1005,
+      "step": 882
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999924348084864e-05,
+      "loss": 1.3829,
+      "step": 883
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992415472551e-05,
+      "loss": 1.0473,
+      "step": 884
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992396111938e-05,
+      "loss": 0.9251,
+      "step": 885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999923767266444e-05,
+      "loss": 1.0935,
+      "step": 886
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992357316673e-05,
+      "loss": 1.082,
+      "step": 887
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999923378820224e-05,
+      "loss": 0.3298,
+      "step": 888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992318422693e-05,
+      "loss": 0.5366,
+      "step": 889
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992298938685e-05,
+      "loss": 1.022,
+      "step": 890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992279429997e-05,
+      "loss": 0.8881,
+      "step": 891
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999922598966313e-05,
+      "loss": 0.8658,
+      "step": 892
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999922403385864e-05,
+      "loss": 1.2105,
+      "step": 893
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992220755862e-05,
+      "loss": 1.1332,
+      "step": 894
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999922011484604e-05,
+      "loss": 5.3122,
+      "step": 895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999921815163786e-05,
+      "loss": 3.7951,
+      "step": 896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992161859618e-05,
+      "loss": 5.5103,
+      "step": 897
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999921421781784e-05,
+      "loss": 3.4476,
+      "step": 898
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999921224720605e-05,
+      "loss": 2.2574,
+      "step": 899
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999921027412635e-05,
+      "loss": 1.302,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999920829857874e-05,
+      "loss": 1.1378,
+      "step": 901
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999992063205633e-05,
+      "loss": 1.1153,
+      "step": 902
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999920434008e-05,
+      "loss": 1.1502,
+      "step": 903
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999920235712874e-05,
+      "loss": 1.266,
+      "step": 904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999920037170953e-05,
+      "loss": 1.4186,
+      "step": 905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999919838382255e-05,
+      "loss": 0.9976,
+      "step": 906
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991963934676e-05,
+      "loss": 1.2569,
+      "step": 907
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991944006448e-05,
+      "loss": 1.2904,
+      "step": 908
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999919240535416e-05,
+      "loss": 0.6151,
+      "step": 909
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991904075956e-05,
+      "loss": 1.2937,
+      "step": 910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999918840736915e-05,
+      "loss": 1.1849,
+      "step": 911
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991864046748e-05,
+      "loss": 1.1805,
+      "step": 912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999918439951255e-05,
+      "loss": 1.2076,
+      "step": 913
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999918239188245e-05,
+      "loss": 1.1705,
+      "step": 914
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999918038178436e-05,
+      "loss": 0.8461,
+      "step": 915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991783692185e-05,
+      "loss": 1.127,
+      "step": 916
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991763541848e-05,
+      "loss": 1.2225,
+      "step": 917
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991743366831e-05,
+      "loss": 1.0974,
+      "step": 918
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991723167135e-05,
+      "loss": 0.53,
+      "step": 919
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999917029427614e-05,
+      "loss": 0.7431,
+      "step": 920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999916826937076e-05,
+      "loss": 1.4294,
+      "step": 921
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991662419976e-05,
+      "loss": 1.2934,
+      "step": 922
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991642121565e-05,
+      "loss": 0.9691,
+      "step": 923
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999916217984755e-05,
+      "loss": 0.9543,
+      "step": 924
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999916014507065e-05,
+      "loss": 1.2091,
+      "step": 925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991581078259e-05,
+      "loss": 1.0221,
+      "step": 926
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999915606811323e-05,
+      "loss": 1.0449,
+      "step": 927
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991540259327e-05,
+      "loss": 0.9845,
+      "step": 928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999915198128436e-05,
+      "loss": 1.1451,
+      "step": 929
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999149934168e-05,
+      "loss": 1.0454,
+      "step": 930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999914788458376e-05,
+      "loss": 1.2146,
+      "step": 931
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991458325318e-05,
+      "loss": 1.3721,
+      "step": 932
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999914377801184e-05,
+      "loss": 1.1491,
+      "step": 933
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999141721024e-05,
+      "loss": 0.996,
+      "step": 934
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991396615682e-05,
+      "loss": 1.0118,
+      "step": 935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991375996446e-05,
+      "loss": 0.6254,
+      "step": 936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999913553525315e-05,
+      "loss": 0.995,
+      "step": 937
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991334683937e-05,
+      "loss": 1.1517,
+      "step": 938
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999913139906645e-05,
+      "loss": 1.1732,
+      "step": 939
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991293272713e-05,
+      "loss": 0.7338,
+      "step": 940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991272530082e-05,
+      "loss": 1.2936,
+      "step": 941
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991251762773e-05,
+      "loss": 1.0981,
+      "step": 942
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999912309707855e-05,
+      "loss": 1.2161,
+      "step": 943
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991210154118e-05,
+      "loss": 1.5329,
+      "step": 944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991189312772e-05,
+      "loss": 1.1148,
+      "step": 945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991168446747e-05,
+      "loss": 1.5379,
+      "step": 946
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991147556043e-05,
+      "loss": 1.166,
+      "step": 947
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999911266406606e-05,
+      "loss": 1.0611,
+      "step": 948
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999911057006e-05,
+      "loss": 1.2562,
+      "step": 949
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999108473586e-05,
+      "loss": 1.1501,
+      "step": 950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991063746441e-05,
+      "loss": 1.2442,
+      "step": 951
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991042732343e-05,
+      "loss": 0.8504,
+      "step": 952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999991021693566e-05,
+      "loss": 0.6655,
+      "step": 953
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999100063011e-05,
+      "loss": 0.8098,
+      "step": 954
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990979541976e-05,
+      "loss": 0.6979,
+      "step": 955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990958429163e-05,
+      "loss": 0.6262,
+      "step": 956
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999093729167e-05,
+      "loss": 0.9136,
+      "step": 957
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990916129499e-05,
+      "loss": 0.6292,
+      "step": 958
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990894942649e-05,
+      "loss": 0.7384,
+      "step": 959
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990873731121e-05,
+      "loss": 0.6144,
+      "step": 960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990852494913e-05,
+      "loss": 0.8134,
+      "step": 961
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990831234026e-05,
+      "loss": 0.6764,
+      "step": 962
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990809948461e-05,
+      "loss": 1.2664,
+      "step": 963
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990788638217e-05,
+      "loss": 1.2815,
+      "step": 964
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990767303293e-05,
+      "loss": 1.1905,
+      "step": 965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990745943692e-05,
+      "loss": 1.2767,
+      "step": 966
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999907245594114e-05,
+      "loss": 1.1545,
+      "step": 967
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990703150451e-05,
+      "loss": 1.0269,
+      "step": 968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990681716813e-05,
+      "loss": 1.0386,
+      "step": 969
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999906602584956e-05,
+      "loss": 1.2325,
+      "step": 970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990638775499e-05,
+      "loss": 1.1714,
+      "step": 971
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990617267824e-05,
+      "loss": 1.1213,
+      "step": 972
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999059573547e-05,
+      "loss": 1.087,
+      "step": 973
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990574178438e-05,
+      "loss": 0.8446,
+      "step": 974
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990552596726e-05,
+      "loss": 1.1204,
+      "step": 975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990530990336e-05,
+      "loss": 1.1727,
+      "step": 976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999905093592664e-05,
+      "loss": 1.0612,
+      "step": 977
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990487703518e-05,
+      "loss": 1.2293,
+      "step": 978
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990466023091e-05,
+      "loss": 1.2555,
+      "step": 979
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999904443179855e-05,
+      "loss": 1.0958,
+      "step": 980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999904225882e-05,
+      "loss": 1.0161,
+      "step": 981
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990400833737e-05,
+      "loss": 1.1351,
+      "step": 982
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990379054594e-05,
+      "loss": 1.3182,
+      "step": 983
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999903572507736e-05,
+      "loss": 1.3075,
+      "step": 984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990335422273e-05,
+      "loss": 1.2306,
+      "step": 985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990313569094e-05,
+      "loss": 1.2427,
+      "step": 986
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990291691237e-05,
+      "loss": 0.4983,
+      "step": 987
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999902697886994e-05,
+      "loss": 0.4873,
+      "step": 988
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990247861484e-05,
+      "loss": 0.4938,
+      "step": 989
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999022590959e-05,
+      "loss": 0.9492,
+      "step": 990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999902039330167e-05,
+      "loss": 1.1465,
+      "step": 991
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990181931765e-05,
+      "loss": 1.27,
+      "step": 992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990159905834e-05,
+      "loss": 0.4938,
+      "step": 993
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999901378552236e-05,
+      "loss": 1.0486,
+      "step": 994
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999901157799356e-05,
+      "loss": 1.1557,
+      "step": 995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990093679968e-05,
+      "loss": 1.3443,
+      "step": 996
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999900715553216e-05,
+      "loss": 1.1414,
+      "step": 997
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990049405996e-05,
+      "loss": 1.1585,
+      "step": 998
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990027231992e-05,
+      "loss": 1.1172,
+      "step": 999
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999990005033309e-05,
+      "loss": 1.2645,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.1082977056503296,
+      "eval_runtime": 84.9058,
+      "eval_samples_per_second": 16.312,
+      "eval_steps_per_second": 4.087,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989982809948e-05,
+      "loss": 1.1184,
+      "step": 1001
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989960561907e-05,
+      "loss": 1.1318,
+      "step": 1002
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989938289188e-05,
+      "loss": 1.0316,
+      "step": 1003
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999899159917895e-05,
+      "loss": 1.1357,
+      "step": 1004
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989893669712e-05,
+      "loss": 1.1337,
+      "step": 1005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989871322956e-05,
+      "loss": 0.9274,
+      "step": 1006
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989848951522e-05,
+      "loss": 0.9013,
+      "step": 1007
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989826555408e-05,
+      "loss": 0.8428,
+      "step": 1008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999898041346155e-05,
+      "loss": 0.8037,
+      "step": 1009
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999897816891436e-05,
+      "loss": 0.7398,
+      "step": 1010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989759218994e-05,
+      "loss": 0.6844,
+      "step": 1011
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989736724165e-05,
+      "loss": 0.577,
+      "step": 1012
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989714204657e-05,
+      "loss": 0.6148,
+      "step": 1013
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989691660471e-05,
+      "loss": 0.8637,
+      "step": 1014
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999896690916045e-05,
+      "loss": 1.3844,
+      "step": 1015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998964649806e-05,
+      "loss": 1.4286,
+      "step": 1016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989623879837e-05,
+      "loss": 0.8812,
+      "step": 1017
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989601236935e-05,
+      "loss": 0.8998,
+      "step": 1018
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989578569354e-05,
+      "loss": 1.0472,
+      "step": 1019
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989555877094e-05,
+      "loss": 1.1919,
+      "step": 1020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999895331601556e-05,
+      "loss": 1.1963,
+      "step": 1021
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989510418538e-05,
+      "loss": 1.198,
+      "step": 1022
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989487652242e-05,
+      "loss": 1.0706,
+      "step": 1023
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999894648612666e-05,
+      "loss": 1.2298,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989442045613e-05,
+      "loss": 1.3413,
+      "step": 1025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998941920528e-05,
+      "loss": 0.837,
+      "step": 1026
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989396340268e-05,
+      "loss": 1.1898,
+      "step": 1027
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999893734505775e-05,
+      "loss": 1.1632,
+      "step": 1028
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999893505362086e-05,
+      "loss": 1.014,
+      "step": 1029
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998932759716e-05,
+      "loss": 0.9097,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989304633433e-05,
+      "loss": 1.3816,
+      "step": 1031
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999892816450274e-05,
+      "loss": 1.1644,
+      "step": 1032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999892586319425e-05,
+      "loss": 1.205,
+      "step": 1033
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989235594179e-05,
+      "loss": 1.3464,
+      "step": 1034
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999892125317366e-05,
+      "loss": 1.2883,
+      "step": 1035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989189444615e-05,
+      "loss": 1.085,
+      "step": 1036
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999891663328155e-05,
+      "loss": 1.1615,
+      "step": 1037
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989143196336e-05,
+      "loss": 1.2407,
+      "step": 1038
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999891200351784e-05,
+      "loss": 1.1057,
+      "step": 1039
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989096849342e-05,
+      "loss": 1.2035,
+      "step": 1040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989073638827e-05,
+      "loss": 1.2299,
+      "step": 1041
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999890504036315e-05,
+      "loss": 1.3304,
+      "step": 1042
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989027143759e-05,
+      "loss": 1.0313,
+      "step": 1043
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999989003859207e-05,
+      "loss": 1.3824,
+      "step": 1044
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988980549976e-05,
+      "loss": 1.0822,
+      "step": 1045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999889572160665e-05,
+      "loss": 1.9536,
+      "step": 1046
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988933857478e-05,
+      "loss": 1.9319,
+      "step": 1047
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998891047421e-05,
+      "loss": 1.1924,
+      "step": 1048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988887066265e-05,
+      "loss": 0.978,
+      "step": 1049
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998886363364e-05,
+      "loss": 1.021,
+      "step": 1050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988840176336e-05,
+      "loss": 1.1141,
+      "step": 1051
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988816694353e-05,
+      "loss": 1.2393,
+      "step": 1052
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988793187692e-05,
+      "loss": 1.2058,
+      "step": 1053
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999887696563514e-05,
+      "loss": 1.3326,
+      "step": 1054
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999887461003326e-05,
+      "loss": 1.24,
+      "step": 1055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988722519634e-05,
+      "loss": 1.3852,
+      "step": 1056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999886989142576e-05,
+      "loss": 0.7316,
+      "step": 1057
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988675284202e-05,
+      "loss": 1.0525,
+      "step": 1058
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999886516294666e-05,
+      "loss": 1.064,
+      "step": 1059
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988627950054e-05,
+      "loss": 1.0082,
+      "step": 1060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988604245962e-05,
+      "loss": 0.9874,
+      "step": 1061
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988580517191e-05,
+      "loss": 0.9579,
+      "step": 1062
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998855676374e-05,
+      "loss": 1.0129,
+      "step": 1063
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999885329856125e-05,
+      "loss": 1.2541,
+      "step": 1064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988509182805e-05,
+      "loss": 1.1398,
+      "step": 1065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988485355318e-05,
+      "loss": 0.6827,
+      "step": 1066
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999884615031536e-05,
+      "loss": 0.9613,
+      "step": 1067
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988437626309e-05,
+      "loss": 1.0136,
+      "step": 1068
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999884137247864e-05,
+      "loss": 0.9416,
+      "step": 1069
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999883897985844e-05,
+      "loss": 1.2144,
+      "step": 1070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999883658477046e-05,
+      "loss": 1.235,
+      "step": 1071
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988341872145e-05,
+      "loss": 0.8769,
+      "step": 1072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988317871906e-05,
+      "loss": 0.8905,
+      "step": 1073
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999882938469896e-05,
+      "loss": 1.2377,
+      "step": 1074
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988269797394e-05,
+      "loss": 1.2858,
+      "step": 1075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988245723119e-05,
+      "loss": 1.1109,
+      "step": 1076
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999882216241656e-05,
+      "loss": 1.2624,
+      "step": 1077
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988197500533e-05,
+      "loss": 0.8667,
+      "step": 1078
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988173352222e-05,
+      "loss": 1.0818,
+      "step": 1079
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988149179232e-05,
+      "loss": 0.8727,
+      "step": 1080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999881249815634e-05,
+      "loss": 1.1173,
+      "step": 1081
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999881007592156e-05,
+      "loss": 1.2649,
+      "step": 1082
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988076512189e-05,
+      "loss": 1.3832,
+      "step": 1083
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988052240484e-05,
+      "loss": 1.2634,
+      "step": 1084
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999880279440995e-05,
+      "loss": 1.2364,
+      "step": 1085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999880036230365e-05,
+      "loss": 1.3662,
+      "step": 1086
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987979277295e-05,
+      "loss": 1.0146,
+      "step": 1087
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999879549068737e-05,
+      "loss": 1.1297,
+      "step": 1088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999879305117745e-05,
+      "loss": 1.1345,
+      "step": 1089
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987906091996e-05,
+      "loss": 0.7416,
+      "step": 1090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999878816475395e-05,
+      "loss": 0.8146,
+      "step": 1091
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999878571784037e-05,
+      "loss": 0.7677,
+      "step": 1092
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999878326845886e-05,
+      "loss": 0.7257,
+      "step": 1093
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987808166096e-05,
+      "loss": 0.7448,
+      "step": 1094
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999877836229224e-05,
+      "loss": 0.7,
+      "step": 1095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987759055071e-05,
+      "loss": 0.6121,
+      "step": 1096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987734462541e-05,
+      "loss": 0.7789,
+      "step": 1097
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987709845333e-05,
+      "loss": 0.649,
+      "step": 1098
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987685203445e-05,
+      "loss": 0.8089,
+      "step": 1099
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999876605368786e-05,
+      "loss": 0.7264,
+      "step": 1100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987635845633e-05,
+      "loss": 0.6354,
+      "step": 1101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987611129709e-05,
+      "loss": 0.7173,
+      "step": 1102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987586389106e-05,
+      "loss": 0.7339,
+      "step": 1103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987561623824e-05,
+      "loss": 0.7132,
+      "step": 1104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999875368338634e-05,
+      "loss": 1.0778,
+      "step": 1105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987512019224e-05,
+      "loss": 1.1789,
+      "step": 1106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987487179906e-05,
+      "loss": 1.0053,
+      "step": 1107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987462315908e-05,
+      "loss": 1.1175,
+      "step": 1108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987437427233e-05,
+      "loss": 1.2642,
+      "step": 1109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987412513878e-05,
+      "loss": 1.1858,
+      "step": 1110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987387575844e-05,
+      "loss": 1.1409,
+      "step": 1111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999873626131323e-05,
+      "loss": 1.4153,
+      "step": 1112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999873376257404e-05,
+      "loss": 1.3314,
+      "step": 1113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987312613671e-05,
+      "loss": 1.2541,
+      "step": 1114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999872875769216e-05,
+      "loss": 1.2764,
+      "step": 1115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987262515494e-05,
+      "loss": 1.5257,
+      "step": 1116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987237429388e-05,
+      "loss": 1.2683,
+      "step": 1117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987212318602e-05,
+      "loss": 0.6983,
+      "step": 1118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999871871831385e-05,
+      "loss": 1.0439,
+      "step": 1119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987162022995e-05,
+      "loss": 1.1931,
+      "step": 1120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987136838173e-05,
+      "loss": 0.7782,
+      "step": 1121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999871116286724e-05,
+      "loss": 1.1135,
+      "step": 1122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987086394493e-05,
+      "loss": 1.1923,
+      "step": 1123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987061135635e-05,
+      "loss": 1.0822,
+      "step": 1124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987035852098e-05,
+      "loss": 1.1745,
+      "step": 1125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999987010543882e-05,
+      "loss": 0.9733,
+      "step": 1126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986985210987e-05,
+      "loss": 0.8489,
+      "step": 1127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986959853414e-05,
+      "loss": 1.1524,
+      "step": 1128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999869344711615e-05,
+      "loss": 1.1779,
+      "step": 1129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998690906423e-05,
+      "loss": 1.19,
+      "step": 1130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998688363262e-05,
+      "loss": 1.0589,
+      "step": 1131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986858176332e-05,
+      "loss": 1.1153,
+      "step": 1132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999868326953635e-05,
+      "loss": 1.199,
+      "step": 1133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999868071897174e-05,
+      "loss": 0.846,
+      "step": 1134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986781659392e-05,
+      "loss": 0.9576,
+      "step": 1135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986756104388e-05,
+      "loss": 0.5289,
+      "step": 1136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986730524705e-05,
+      "loss": 0.4011,
+      "step": 1137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986704920343e-05,
+      "loss": 0.2992,
+      "step": 1138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986679291303e-05,
+      "loss": 0.2196,
+      "step": 1139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986653637584e-05,
+      "loss": 0.1587,
+      "step": 1140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999866279591856e-05,
+      "loss": 0.2478,
+      "step": 1141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999866022561084e-05,
+      "loss": 0.2585,
+      "step": 1142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986576528353e-05,
+      "loss": 0.2091,
+      "step": 1143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986550775918e-05,
+      "loss": 0.3503,
+      "step": 1144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999865249988045e-05,
+      "loss": 0.1784,
+      "step": 1145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986499197013e-05,
+      "loss": 0.2116,
+      "step": 1146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986473370542e-05,
+      "loss": 0.1828,
+      "step": 1147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986447519392e-05,
+      "loss": 0.2068,
+      "step": 1148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986421643563e-05,
+      "loss": 0.2104,
+      "step": 1149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986395743056e-05,
+      "loss": 1.1287,
+      "step": 1150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986369817869e-05,
+      "loss": 1.2888,
+      "step": 1151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986343868004e-05,
+      "loss": 1.1025,
+      "step": 1152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998631789346e-05,
+      "loss": 1.352,
+      "step": 1153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986291894238e-05,
+      "loss": 1.4697,
+      "step": 1154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986265870336e-05,
+      "loss": 1.2039,
+      "step": 1155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999862398217556e-05,
+      "loss": 1.298,
+      "step": 1156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999862137484965e-05,
+      "loss": 1.1438,
+      "step": 1157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986187650558e-05,
+      "loss": 1.1965,
+      "step": 1158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986161527942e-05,
+      "loss": 1.2826,
+      "step": 1159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999861353806464e-05,
+      "loss": 0.9909,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999986109208672e-05,
+      "loss": 1.1301,
+      "step": 1161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999860830120186e-05,
+      "loss": 1.2068,
+      "step": 1162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999860567906866e-05,
+      "loss": 1.205,
+      "step": 1163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999860305446755e-05,
+      "loss": 0.9465,
+      "step": 1164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999860042739866e-05,
+      "loss": 1.0319,
+      "step": 1165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985977978618e-05,
+      "loss": 1.2249,
+      "step": 1166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999859516585706e-05,
+      "loss": 1.1404,
+      "step": 1167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985925313845e-05,
+      "loss": 1.1021,
+      "step": 1168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998589894444e-05,
+      "loss": 0.7697,
+      "step": 1169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985872550356e-05,
+      "loss": 1.2475,
+      "step": 1170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999858461315936e-05,
+      "loss": 1.346,
+      "step": 1171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985819688152e-05,
+      "loss": 1.2817,
+      "step": 1172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999857932200326e-05,
+      "loss": 0.905,
+      "step": 1173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985766727233e-05,
+      "loss": 0.8015,
+      "step": 1174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985740209756e-05,
+      "loss": 0.9413,
+      "step": 1175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999857136675994e-05,
+      "loss": 1.1069,
+      "step": 1176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985687100764e-05,
+      "loss": 1.0106,
+      "step": 1177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999856605092496e-05,
+      "loss": 1.1865,
+      "step": 1178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999856338930566e-05,
+      "loss": 1.0042,
+      "step": 1179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985607252186e-05,
+      "loss": 1.2633,
+      "step": 1180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999855805866345e-05,
+      "loss": 1.1735,
+      "step": 1181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999855538964055e-05,
+      "loss": 1.2769,
+      "step": 1182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985527181497e-05,
+      "loss": 1.1381,
+      "step": 1183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999855004419105e-05,
+      "loss": 1.0372,
+      "step": 1184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985473677644e-05,
+      "loss": 1.5588,
+      "step": 1185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999854468887e-05,
+      "loss": 1.1077,
+      "step": 1186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985420075077e-05,
+      "loss": 1.1749,
+      "step": 1187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985393236775e-05,
+      "loss": 0.679,
+      "step": 1188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985366373794e-05,
+      "loss": 1.0461,
+      "step": 1189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985339486134e-05,
+      "loss": 1.2824,
+      "step": 1190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985312573795e-05,
+      "loss": 1.0176,
+      "step": 1191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999852856367776e-05,
+      "loss": 1.2261,
+      "step": 1192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985258675082e-05,
+      "loss": 1.4413,
+      "step": 1193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999852316887064e-05,
+      "loss": 1.1552,
+      "step": 1194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985204677653e-05,
+      "loss": 1.1012,
+      "step": 1195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985177641921e-05,
+      "loss": 1.3743,
+      "step": 1196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985150581509e-05,
+      "loss": 1.2877,
+      "step": 1197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985123496419e-05,
+      "loss": 1.0601,
+      "step": 1198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999850963866504e-05,
+      "loss": 1.0229,
+      "step": 1199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985069252202e-05,
+      "loss": 0.5861,
+      "step": 1200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985042093076e-05,
+      "loss": 1.0952,
+      "step": 1201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999850149092705e-05,
+      "loss": 0.8716,
+      "step": 1202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984987700787e-05,
+      "loss": 1.3299,
+      "step": 1203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984960467623e-05,
+      "loss": 1.3952,
+      "step": 1204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999849332097816e-05,
+      "loss": 1.201,
+      "step": 1205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984905927261e-05,
+      "loss": 0.9686,
+      "step": 1206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984878620061e-05,
+      "loss": 1.1564,
+      "step": 1207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984851288183e-05,
+      "loss": 0.7878,
+      "step": 1208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999848239316264e-05,
+      "loss": 1.445,
+      "step": 1209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984796550391e-05,
+      "loss": 1.3462,
+      "step": 1210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984769144476e-05,
+      "loss": 1.3696,
+      "step": 1211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984741713883e-05,
+      "loss": 1.0053,
+      "step": 1212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984714258611e-05,
+      "loss": 1.0261,
+      "step": 1213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998468677866e-05,
+      "loss": 1.2284,
+      "step": 1214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998465927403e-05,
+      "loss": 1.2134,
+      "step": 1215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984631744722e-05,
+      "loss": 1.3078,
+      "step": 1216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999846041907336e-05,
+      "loss": 1.2526,
+      "step": 1217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984576612068e-05,
+      "loss": 1.0028,
+      "step": 1218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984549008723e-05,
+      "loss": 0.9113,
+      "step": 1219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999845213807e-05,
+      "loss": 0.8816,
+      "step": 1220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984493727997e-05,
+      "loss": 0.8277,
+      "step": 1221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984466050616e-05,
+      "loss": 0.8075,
+      "step": 1222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984438348556e-05,
+      "loss": 0.7724,
+      "step": 1223
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984410621817e-05,
+      "loss": 0.9919,
+      "step": 1224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999843828703994e-05,
+      "loss": 1.1758,
+      "step": 1225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999843550943026e-05,
+      "loss": 1.1239,
+      "step": 1226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984327293528e-05,
+      "loss": 1.1041,
+      "step": 1227
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999842994680734e-05,
+      "loss": 1.1451,
+      "step": 1228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999842716179405e-05,
+      "loss": 1.1521,
+      "step": 1229
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984243743129e-05,
+      "loss": 1.0867,
+      "step": 1230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999842158436385e-05,
+      "loss": 1.0515,
+      "step": 1231
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998418791947e-05,
+      "loss": 1.0314,
+      "step": 1232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999984159970621e-05,
+      "loss": 0.9818,
+      "step": 1233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999841319970945e-05,
+      "loss": 1.0535,
+      "step": 1234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999841039988894e-05,
+      "loss": 1.0195,
+      "step": 1235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999840759760044e-05,
+      "loss": 1.1967,
+      "step": 1236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999840479284416e-05,
+      "loss": 1.082,
+      "step": 1237
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999840198562e-05,
+      "loss": 1.3885,
+      "step": 1238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999839917592786e-05,
+      "loss": 0.8606,
+      "step": 1239
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998396363768e-05,
+      "loss": 1.3506,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983935491401e-05,
+      "loss": 1.1122,
+      "step": 1241
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983907320444e-05,
+      "loss": 1.3343,
+      "step": 1242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983879124808e-05,
+      "loss": 1.3739,
+      "step": 1243
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983850904494e-05,
+      "loss": 1.3176,
+      "step": 1244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999838226594994e-05,
+      "loss": 0.8282,
+      "step": 1245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999837943898276e-05,
+      "loss": 0.6075,
+      "step": 1246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983766095477e-05,
+      "loss": 0.7719,
+      "step": 1247
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999837377764467e-05,
+      "loss": 0.5532,
+      "step": 1248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983709432738e-05,
+      "loss": 0.5228,
+      "step": 1249
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983681064351e-05,
+      "loss": 0.6887,
+      "step": 1250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983652671285e-05,
+      "loss": 1.1635,
+      "step": 1251
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998362425354e-05,
+      "loss": 1.174,
+      "step": 1252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999835958111165e-05,
+      "loss": 1.3225,
+      "step": 1253
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999835673440136e-05,
+      "loss": 1.2051,
+      "step": 1254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983538852232e-05,
+      "loss": 1.1052,
+      "step": 1255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983510335772e-05,
+      "loss": 1.0838,
+      "step": 1256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983481794633e-05,
+      "loss": 0.8987,
+      "step": 1257
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983453228816e-05,
+      "loss": 0.5425,
+      "step": 1258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983424638319e-05,
+      "loss": 0.5464,
+      "step": 1259
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983396023144e-05,
+      "loss": 0.5096,
+      "step": 1260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998336738329e-05,
+      "loss": 1.0058,
+      "step": 1261
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999833387187575e-05,
+      "loss": 1.2524,
+      "step": 1262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999833100295456e-05,
+      "loss": 1.0138,
+      "step": 1263
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983281315656e-05,
+      "loss": 1.1427,
+      "step": 1264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983252577086e-05,
+      "loss": 1.548,
+      "step": 1265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983223813838e-05,
+      "loss": 1.1719,
+      "step": 1266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983195025911e-05,
+      "loss": 1.4455,
+      "step": 1267
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983166213306e-05,
+      "loss": 1.4792,
+      "step": 1268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983137376022e-05,
+      "loss": 1.1389,
+      "step": 1269
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999831085140594e-05,
+      "loss": 1.0004,
+      "step": 1270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983079627417e-05,
+      "loss": 0.4547,
+      "step": 1271
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983050716096e-05,
+      "loss": 0.8922,
+      "step": 1272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999983021780097e-05,
+      "loss": 0.8002,
+      "step": 1273
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982992819419e-05,
+      "loss": 1.1497,
+      "step": 1274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999829638340625e-05,
+      "loss": 0.9905,
+      "step": 1275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999829348240264e-05,
+      "loss": 1.1683,
+      "step": 1276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999829057893124e-05,
+      "loss": 1.0706,
+      "step": 1277
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999828767299187e-05,
+      "loss": 1.1382,
+      "step": 1278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999828476458464e-05,
+      "loss": 1.2152,
+      "step": 1279
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982818537096e-05,
+      "loss": 1.0906,
+      "step": 1280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982789403666e-05,
+      "loss": 1.0177,
+      "step": 1281
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999827602455574e-05,
+      "loss": 1.2553,
+      "step": 1282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999827310627706e-05,
+      "loss": 0.9395,
+      "step": 1283
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999827018553046e-05,
+      "loss": 1.0483,
+      "step": 1284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998267262316e-05,
+      "loss": 1.2686,
+      "step": 1285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999826433663366e-05,
+      "loss": 1.2582,
+      "step": 1286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982614084834e-05,
+      "loss": 1.1988,
+      "step": 1287
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982584778653e-05,
+      "loss": 1.1987,
+      "step": 1288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982555447793e-05,
+      "loss": 0.9949,
+      "step": 1289
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999825260922554e-05,
+      "loss": 1.3115,
+      "step": 1290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999824967120373e-05,
+      "loss": 1.182,
+      "step": 1291
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999824673071415e-05,
+      "loss": 1.107,
+      "step": 1292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982437877566e-05,
+      "loss": 1.0447,
+      "step": 1293
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999824084233125e-05,
+      "loss": 0.838,
+      "step": 1294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998237894438e-05,
+      "loss": 1.0267,
+      "step": 1295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999823494407695e-05,
+      "loss": 0.9289,
+      "step": 1296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982319912479e-05,
+      "loss": 1.1514,
+      "step": 1297
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999822903595106e-05,
+      "loss": 1.3189,
+      "step": 1298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982260781863e-05,
+      "loss": 1.1155,
+      "step": 1299
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999822311795364e-05,
+      "loss": 1.1044,
+      "step": 1300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982201552531e-05,
+      "loss": 1.2472,
+      "step": 1301
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982171900848e-05,
+      "loss": 0.9781,
+      "step": 1302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982142224485e-05,
+      "loss": 0.716,
+      "step": 1303
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982112523444e-05,
+      "loss": 0.5798,
+      "step": 1304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982082797724e-05,
+      "loss": 1.3287,
+      "step": 1305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982053047324e-05,
+      "loss": 1.0971,
+      "step": 1306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999982023272247e-05,
+      "loss": 1.2485,
+      "step": 1307
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998199347249e-05,
+      "loss": 0.9496,
+      "step": 1308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981963648055e-05,
+      "loss": 1.0331,
+      "step": 1309
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981933798941e-05,
+      "loss": 1.045,
+      "step": 1310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981903925148e-05,
+      "loss": 0.9116,
+      "step": 1311
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999818740266766e-05,
+      "loss": 1.256,
+      "step": 1312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981844103526e-05,
+      "loss": 1.1436,
+      "step": 1313
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981814155698e-05,
+      "loss": 1.1256,
+      "step": 1314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999817841831896e-05,
+      "loss": 1.0737,
+      "step": 1315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981754186003e-05,
+      "loss": 1.2136,
+      "step": 1316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981724164138e-05,
+      "loss": 1.008,
+      "step": 1317
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999816941175935e-05,
+      "loss": 0.9908,
+      "step": 1318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998166404637e-05,
+      "loss": 1.1998,
+      "step": 1319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981633950469e-05,
+      "loss": 1.0553,
+      "step": 1320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999816038298885e-05,
+      "loss": 1.2117,
+      "step": 1321
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981573684629e-05,
+      "loss": 0.8267,
+      "step": 1322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981543514691e-05,
+      "loss": 1.5073,
+      "step": 1323
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999815133200745e-05,
+      "loss": 1.2936,
+      "step": 1324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981483100778e-05,
+      "loss": 1.1269,
+      "step": 1325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981452856805e-05,
+      "loss": 1.1422,
+      "step": 1326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999814225881515e-05,
+      "loss": 1.0733,
+      "step": 1327
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999813922948205e-05,
+      "loss": 1.1965,
+      "step": 1328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999813619768096e-05,
+      "loss": 1.5187,
+      "step": 1329
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999813316341196e-05,
+      "loss": 1.0381,
+      "step": 1330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981301266752e-05,
+      "loss": 1.0388,
+      "step": 1331
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999812708747054e-05,
+      "loss": 1.3295,
+      "step": 1332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981240457979e-05,
+      "loss": 1.1589,
+      "step": 1333
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999812100165754e-05,
+      "loss": 1.2435,
+      "step": 1334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999811795504916e-05,
+      "loss": 1.0928,
+      "step": 1335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998114905973e-05,
+      "loss": 1.1308,
+      "step": 1336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999811185442894e-05,
+      "loss": 1.06,
+      "step": 1337
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999810880041695e-05,
+      "loss": 1.6996,
+      "step": 1338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981057439372e-05,
+      "loss": 2.2645,
+      "step": 1339
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999981026849895e-05,
+      "loss": 2.2996,
+      "step": 1340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980996235739e-05,
+      "loss": 1.1277,
+      "step": 1341
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999809655969047e-05,
+      "loss": 1.1166,
+      "step": 1342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980934933392e-05,
+      "loss": 1.1923,
+      "step": 1343
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999809042452e-05,
+      "loss": 1.2633,
+      "step": 1344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980873532329e-05,
+      "loss": 0.8925,
+      "step": 1345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999808427947795e-05,
+      "loss": 0.9157,
+      "step": 1346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980812032551e-05,
+      "loss": 1.1102,
+      "step": 1347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980781245645e-05,
+      "loss": 1.0015,
+      "step": 1348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980750434058e-05,
+      "loss": 1.0523,
+      "step": 1349
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999807195977946e-05,
+      "loss": 1.1134,
+      "step": 1350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999806887368505e-05,
+      "loss": 0.9844,
+      "step": 1351
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999806578512287e-05,
+      "loss": 1.3819,
+      "step": 1352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999806269409276e-05,
+      "loss": 1.3509,
+      "step": 1353
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980596005949e-05,
+      "loss": 0.937,
+      "step": 1354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980565046291e-05,
+      "loss": 1.0986,
+      "step": 1355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980534061953e-05,
+      "loss": 0.9307,
+      "step": 1356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999805030529374e-05,
+      "loss": 1.0425,
+      "step": 1357
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999804720192426e-05,
+      "loss": 1.0921,
+      "step": 1358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998044096087e-05,
+      "loss": 1.4279,
+      "step": 1359
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999804098778176e-05,
+      "loss": 1.0246,
+      "step": 1360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980378770087e-05,
+      "loss": 1.2811,
+      "step": 1361
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999803476376774e-05,
+      "loss": 1.2623,
+      "step": 1362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980316480589e-05,
+      "loss": 1.1786,
+      "step": 1363
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980285298822e-05,
+      "loss": 1.7484,
+      "step": 1364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980254092376e-05,
+      "loss": 1.5065,
+      "step": 1365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980222861251e-05,
+      "loss": 1.063,
+      "step": 1366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999801916054475e-05,
+      "loss": 1.1653,
+      "step": 1367
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980160324966e-05,
+      "loss": 1.2328,
+      "step": 1368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980129019805e-05,
+      "loss": 1.1489,
+      "step": 1369
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999800976899654e-05,
+      "loss": 1.5534,
+      "step": 1370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980066335447e-05,
+      "loss": 2.7763,
+      "step": 1371
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99998003495625e-05,
+      "loss": 2.337,
+      "step": 1372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999980003552375e-05,
+      "loss": 2.4659,
+      "step": 1373
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997997212382e-05,
+      "loss": 2.3945,
+      "step": 1374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999799406705864e-05,
+      "loss": 2.1941,
+      "step": 1375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999799091926743e-05,
+      "loss": 1.6053,
+      "step": 1376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979877690084e-05,
+      "loss": 1.0788,
+      "step": 1377
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979846162814e-05,
+      "loss": 1.3246,
+      "step": 1378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979814610865e-05,
+      "loss": 1.4813,
+      "step": 1379
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979783034239e-05,
+      "loss": 1.3126,
+      "step": 1380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979751432933e-05,
+      "loss": 1.178,
+      "step": 1381
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979719806949e-05,
+      "loss": 1.5922,
+      "step": 1382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979688156285e-05,
+      "loss": 1.3641,
+      "step": 1383
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999796564809434e-05,
+      "loss": 1.3358,
+      "step": 1384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979624780923e-05,
+      "loss": 0.9263,
+      "step": 1385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979593056223e-05,
+      "loss": 1.2803,
+      "step": 1386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979561306844e-05,
+      "loss": 0.8805,
+      "step": 1387
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979529532788e-05,
+      "loss": 1.5963,
+      "step": 1388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979497734052e-05,
+      "loss": 1.3765,
+      "step": 1389
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979465910637e-05,
+      "loss": 1.1817,
+      "step": 1390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979434062544e-05,
+      "loss": 1.3105,
+      "step": 1391
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979402189772e-05,
+      "loss": 1.043,
+      "step": 1392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999793702923216e-05,
+      "loss": 1.0101,
+      "step": 1393
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979338370192e-05,
+      "loss": 1.3187,
+      "step": 1394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979306423384e-05,
+      "loss": 1.1142,
+      "step": 1395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979274451897e-05,
+      "loss": 0.9806,
+      "step": 1396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999792424557315e-05,
+      "loss": 1.1195,
+      "step": 1397
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979210434886e-05,
+      "loss": 1.2506,
+      "step": 1398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979178389363e-05,
+      "loss": 1.13,
+      "step": 1399
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999791463191616e-05,
+      "loss": 1.2353,
+      "step": 1400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997911422428e-05,
+      "loss": 1.1788,
+      "step": 1401
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979082104721e-05,
+      "loss": 1.0935,
+      "step": 1402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999790499604835e-05,
+      "loss": 1.0114,
+      "step": 1403
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999979017791566e-05,
+      "loss": 1.1398,
+      "step": 1404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978985597971e-05,
+      "loss": 1.2195,
+      "step": 1405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978953379696e-05,
+      "loss": 1.1132,
+      "step": 1406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999789211367435e-05,
+      "loss": 1.2819,
+      "step": 1407
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999788888691114e-05,
+      "loss": 0.621,
+      "step": 1408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999788565768e-05,
+      "loss": 0.3403,
+      "step": 1409
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978824259811e-05,
+      "loss": 0.3945,
+      "step": 1410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978791918143e-05,
+      "loss": 1.0303,
+      "step": 1411
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999787595517965e-05,
+      "loss": 1.1808,
+      "step": 1412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999787271607714e-05,
+      "loss": 1.3936,
+      "step": 1413
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999786947450665e-05,
+      "loss": 2.0301,
+      "step": 1414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978662304684e-05,
+      "loss": 1.1428,
+      "step": 1415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978629839622e-05,
+      "loss": 0.9583,
+      "step": 1416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999785973498816e-05,
+      "loss": 1.1777,
+      "step": 1417
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978564835462e-05,
+      "loss": 1.103,
+      "step": 1418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978532296364e-05,
+      "loss": 1.017,
+      "step": 1419
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978499732587e-05,
+      "loss": 1.0472,
+      "step": 1420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978467144132e-05,
+      "loss": 0.9334,
+      "step": 1421
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978434530998e-05,
+      "loss": 1.0641,
+      "step": 1422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978401893185e-05,
+      "loss": 1.1676,
+      "step": 1423
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978369230693e-05,
+      "loss": 0.9384,
+      "step": 1424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978336543523e-05,
+      "loss": 0.9575,
+      "step": 1425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978303831674e-05,
+      "loss": 1.1896,
+      "step": 1426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978271095146e-05,
+      "loss": 1.1317,
+      "step": 1427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999782383339396e-05,
+      "loss": 1.0979,
+      "step": 1428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978205548054e-05,
+      "loss": 1.0988,
+      "step": 1429
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997817273749e-05,
+      "loss": 1.1114,
+      "step": 1430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978139902247e-05,
+      "loss": 2.2097,
+      "step": 1431
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978107042326e-05,
+      "loss": 1.0582,
+      "step": 1432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999780741577265e-05,
+      "loss": 0.8381,
+      "step": 1433
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978041248447e-05,
+      "loss": 0.8595,
+      "step": 1434
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999978008314489e-05,
+      "loss": 0.8939,
+      "step": 1435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977975355853e-05,
+      "loss": 1.2282,
+      "step": 1436
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977942372538e-05,
+      "loss": 1.3273,
+      "step": 1437
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977909364544e-05,
+      "loss": 1.1084,
+      "step": 1438
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977876331872e-05,
+      "loss": 1.0852,
+      "step": 1439
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999778432745204e-05,
+      "loss": 1.3899,
+      "step": 1440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997781019249e-05,
+      "loss": 1.2291,
+      "step": 1441
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999777770857816e-05,
+      "loss": 1.1377,
+      "step": 1442
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999777439543935e-05,
+      "loss": 1.0589,
+      "step": 1443
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999777107983276e-05,
+      "loss": 1.0669,
+      "step": 1444
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977677617582e-05,
+      "loss": 0.9086,
+      "step": 1445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977644412159e-05,
+      "loss": 1.2046,
+      "step": 1446
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977611182056e-05,
+      "loss": 1.195,
+      "step": 1447
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977577927275e-05,
+      "loss": 1.3642,
+      "step": 1448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977544647816e-05,
+      "loss": 1.078,
+      "step": 1449
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977511343677e-05,
+      "loss": 1.2531,
+      "step": 1450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997747801486e-05,
+      "loss": 1.2787,
+      "step": 1451
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999774446613637e-05,
+      "loss": 0.9946,
+      "step": 1452
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977411283189e-05,
+      "loss": 1.0959,
+      "step": 1453
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999773778803354e-05,
+      "loss": 1.214,
+      "step": 1454
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977344452803e-05,
+      "loss": 1.1483,
+      "step": 1455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977311000592e-05,
+      "loss": 1.084,
+      "step": 1456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977277523703e-05,
+      "loss": 0.9704,
+      "step": 1457
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999772440221344e-05,
+      "loss": 0.7158,
+      "step": 1458
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977210495887e-05,
+      "loss": 1.0292,
+      "step": 1459
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977176944961e-05,
+      "loss": 1.1296,
+      "step": 1460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999771433693564e-05,
+      "loss": 1.0778,
+      "step": 1461
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999977109769074e-05,
+      "loss": 3.3233,
+      "step": 1462
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999770761441115e-05,
+      "loss": 3.9099,
+      "step": 1463
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999770424944707e-05,
+      "loss": 1.3,
+      "step": 1464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999770088201514e-05,
+      "loss": 0.973,
+      "step": 1465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999769751211536e-05,
+      "loss": 1.1086,
+      "step": 1466
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976941397476e-05,
+      "loss": 0.8508,
+      "step": 1467
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976907649121e-05,
+      "loss": 1.0619,
+      "step": 1468
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976873876087e-05,
+      "loss": 1.0714,
+      "step": 1469
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976840078374e-05,
+      "loss": 2.7562,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976806255982e-05,
+      "loss": 0.9887,
+      "step": 1471
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976772408912e-05,
+      "loss": 0.974,
+      "step": 1472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999767385371624e-05,
+      "loss": 1.245,
+      "step": 1473
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976704640734e-05,
+      "loss": 1.5144,
+      "step": 1474
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976670719628e-05,
+      "loss": 1.0975,
+      "step": 1475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976636773843e-05,
+      "loss": 1.8456,
+      "step": 1476
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976602803379e-05,
+      "loss": 1.646,
+      "step": 1477
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999765688082356e-05,
+      "loss": 1.3266,
+      "step": 1478
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976534788414e-05,
+      "loss": 1.3988,
+      "step": 1479
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999765007439144e-05,
+      "loss": 1.3974,
+      "step": 1480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976466674735e-05,
+      "loss": 1.1976,
+      "step": 1481
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976432580878e-05,
+      "loss": 1.5675,
+      "step": 1482
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999763984623415e-05,
+      "loss": 1.0349,
+      "step": 1483
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999763643191264e-05,
+      "loss": 1.3132,
+      "step": 1484
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976330151233e-05,
+      "loss": 1.1919,
+      "step": 1485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997629595866e-05,
+      "loss": 0.9964,
+      "step": 1486
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976261741409e-05,
+      "loss": 1.2105,
+      "step": 1487
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999762274994794e-05,
+      "loss": 0.8005,
+      "step": 1488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999761932328706e-05,
+      "loss": 1.0002,
+      "step": 1489
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999761589415834e-05,
+      "loss": 1.3317,
+      "step": 1490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976124625617e-05,
+      "loss": 1.5958,
+      "step": 1491
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976090284972e-05,
+      "loss": 0.9418,
+      "step": 1492
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999760559196494e-05,
+      "loss": 1.1749,
+      "step": 1493
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976021529647e-05,
+      "loss": 1.041,
+      "step": 1494
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999759871149666e-05,
+      "loss": 1.3281,
+      "step": 1495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999759526756064e-05,
+      "loss": 0.9078,
+      "step": 1496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999759182115685e-05,
+      "loss": 1.1638,
+      "step": 1497
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999758837228514e-05,
+      "loss": 1.2198,
+      "step": 1498
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975849209456e-05,
+      "loss": 1.4662,
+      "step": 1499
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975814671382e-05,
+      "loss": 1.384,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.1025360822677612,
+      "eval_runtime": 85.1844,
+      "eval_samples_per_second": 16.259,
+      "eval_steps_per_second": 4.074,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999757801086286e-05,
+      "loss": 1.0703,
+      "step": 1501
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975745521196e-05,
+      "loss": 1.1068,
+      "step": 1502
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975710909086e-05,
+      "loss": 1.5193,
+      "step": 1503
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975676272297e-05,
+      "loss": 1.0456,
+      "step": 1504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975641610829e-05,
+      "loss": 1.1337,
+      "step": 1505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975606924682e-05,
+      "loss": 1.0031,
+      "step": 1506
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975572213856e-05,
+      "loss": 1.2249,
+      "step": 1507
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975537478353e-05,
+      "loss": 1.0518,
+      "step": 1508
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997550271817e-05,
+      "loss": 1.1407,
+      "step": 1509
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975467933309e-05,
+      "loss": 1.0663,
+      "step": 1510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975433123768e-05,
+      "loss": 1.0735,
+      "step": 1511
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999753982895494e-05,
+      "loss": 1.2828,
+      "step": 1512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999753634306525e-05,
+      "loss": 1.3558,
+      "step": 1513
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975328547076e-05,
+      "loss": 1.0236,
+      "step": 1514
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999752936388205e-05,
+      "loss": 0.9801,
+      "step": 1515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975258705887e-05,
+      "loss": 1.0956,
+      "step": 1516
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975223748275e-05,
+      "loss": 1.2429,
+      "step": 1517
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999751887659833e-05,
+      "loss": 1.0751,
+      "step": 1518
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999751537590136e-05,
+      "loss": 1.5672,
+      "step": 1519
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975118727365e-05,
+      "loss": 1.2634,
+      "step": 1520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975083671038e-05,
+      "loss": 1.2086,
+      "step": 1521
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999975048590032e-05,
+      "loss": 1.0235,
+      "step": 1522
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999750134843476e-05,
+      "loss": 1.1494,
+      "step": 1523
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974978353984e-05,
+      "loss": 1.0878,
+      "step": 1524
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974943198942e-05,
+      "loss": 1.1574,
+      "step": 1525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999749080192217e-05,
+      "loss": 1.267,
+      "step": 1526
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974872814822e-05,
+      "loss": 1.0234,
+      "step": 1527
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974837585744e-05,
+      "loss": 1.2157,
+      "step": 1528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974802331987e-05,
+      "loss": 1.3503,
+      "step": 1529
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974767053552e-05,
+      "loss": 1.2035,
+      "step": 1530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999747317504375e-05,
+      "loss": 1.1835,
+      "step": 1531
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974696422645e-05,
+      "loss": 1.3205,
+      "step": 1532
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974661070174e-05,
+      "loss": 1.2915,
+      "step": 1533
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974625693023e-05,
+      "loss": 1.0228,
+      "step": 1534
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999745902911946e-05,
+      "loss": 1.0057,
+      "step": 1535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974554864687e-05,
+      "loss": 0.9307,
+      "step": 1536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999745194135e-05,
+      "loss": 1.2788,
+      "step": 1537
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974483937635e-05,
+      "loss": 1.3145,
+      "step": 1538
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999744484370914e-05,
+      "loss": 1.4141,
+      "step": 1539
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974412911869e-05,
+      "loss": 0.8175,
+      "step": 1540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974377361968e-05,
+      "loss": 1.0975,
+      "step": 1541
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974341787388e-05,
+      "loss": 1.3513,
+      "step": 1542
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974306188129e-05,
+      "loss": 1.0626,
+      "step": 1543
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974270564192e-05,
+      "loss": 1.0888,
+      "step": 1544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999742349155766e-05,
+      "loss": 1.0065,
+      "step": 1545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974199242282e-05,
+      "loss": 1.0015,
+      "step": 1546
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974163544309e-05,
+      "loss": 1.1223,
+      "step": 1547
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974127821656e-05,
+      "loss": 1.1169,
+      "step": 1548
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999740920743256e-05,
+      "loss": 1.3053,
+      "step": 1549
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974056302317e-05,
+      "loss": 1.1333,
+      "step": 1550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999740205056286e-05,
+      "loss": 0.7307,
+      "step": 1551
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973984684262e-05,
+      "loss": 1.3703,
+      "step": 1552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973948838216e-05,
+      "loss": 1.4943,
+      "step": 1553
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973912967492e-05,
+      "loss": 1.1022,
+      "step": 1554
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999738770720895e-05,
+      "loss": 0.9436,
+      "step": 1555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973841152008e-05,
+      "loss": 1.1686,
+      "step": 1556
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999738052072474e-05,
+      "loss": 1.0699,
+      "step": 1557
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973769237809e-05,
+      "loss": 1.023,
+      "step": 1558
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999737332436914e-05,
+      "loss": 1.1993,
+      "step": 1559
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973697224895e-05,
+      "loss": 0.9739,
+      "step": 1560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999736611814195e-05,
+      "loss": 1.2015,
+      "step": 1561
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973625113266e-05,
+      "loss": 1.3037,
+      "step": 1562
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999735890204344e-05,
+      "loss": 1.1989,
+      "step": 1563
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973552902923e-05,
+      "loss": 0.9803,
+      "step": 1564
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973516760733e-05,
+      "loss": 1.1173,
+      "step": 1565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973480593865e-05,
+      "loss": 0.8269,
+      "step": 1566
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973444402318e-05,
+      "loss": 1.2442,
+      "step": 1567
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999734081860925e-05,
+      "loss": 1.1248,
+      "step": 1568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973371945188e-05,
+      "loss": 1.2366,
+      "step": 1569
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999733356796046e-05,
+      "loss": 1.3797,
+      "step": 1570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999732993893426e-05,
+      "loss": 1.4179,
+      "step": 1571
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973263074402e-05,
+      "loss": 1.1726,
+      "step": 1572
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973226734783e-05,
+      "loss": 0.4636,
+      "step": 1573
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973190370485e-05,
+      "loss": 0.8854,
+      "step": 1574
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999731539815086e-05,
+      "loss": 1.2157,
+      "step": 1575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999731175678536e-05,
+      "loss": 1.6862,
+      "step": 1576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997308112952e-05,
+      "loss": 0.8838,
+      "step": 1577
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999730446665075e-05,
+      "loss": 1.1561,
+      "step": 1578
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999973008178816e-05,
+      "loss": 0.9012,
+      "step": 1579
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972971666446e-05,
+      "loss": 1.02,
+      "step": 1580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972935129398e-05,
+      "loss": 1.1196,
+      "step": 1581
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997289856767e-05,
+      "loss": 1.2262,
+      "step": 1582
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999728619812644e-05,
+      "loss": 1.259,
+      "step": 1583
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999728253701796e-05,
+      "loss": 1.1746,
+      "step": 1584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972788734416e-05,
+      "loss": 1.0513,
+      "step": 1585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999727520739744e-05,
+      "loss": 1.0383,
+      "step": 1586
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999727153888535e-05,
+      "loss": 1.0011,
+      "step": 1587
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972678679055e-05,
+      "loss": 1.0647,
+      "step": 1588
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972641944576e-05,
+      "loss": 0.817,
+      "step": 1589
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997260518542e-05,
+      "loss": 1.0495,
+      "step": 1590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972568401584e-05,
+      "loss": 1.1049,
+      "step": 1591
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997253159307e-05,
+      "loss": 0.9831,
+      "step": 1592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972494759878e-05,
+      "loss": 0.9951,
+      "step": 1593
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999724579020055e-05,
+      "loss": 0.8695,
+      "step": 1594
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972421019456e-05,
+      "loss": 1.1208,
+      "step": 1595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972384112227e-05,
+      "loss": 1.3532,
+      "step": 1596
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997234718032e-05,
+      "loss": 1.1476,
+      "step": 1597
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972310223734e-05,
+      "loss": 1.2644,
+      "step": 1598
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972273242469e-05,
+      "loss": 1.1528,
+      "step": 1599
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972236236525e-05,
+      "loss": 1.3417,
+      "step": 1600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999721992059034e-05,
+      "loss": 1.1045,
+      "step": 1601
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999721621506026e-05,
+      "loss": 1.084,
+      "step": 1602
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999721250706234e-05,
+      "loss": 1.1412,
+      "step": 1603
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972087965965e-05,
+      "loss": 0.6463,
+      "step": 1604
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972050836628e-05,
+      "loss": 0.5842,
+      "step": 1605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999972013682612e-05,
+      "loss": 0.4859,
+      "step": 1606
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971976503919e-05,
+      "loss": 1.1645,
+      "step": 1607
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971939300546e-05,
+      "loss": 0.9786,
+      "step": 1608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999719020724945e-05,
+      "loss": 1.294,
+      "step": 1609
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971864819764e-05,
+      "loss": 1.2126,
+      "step": 1610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999718275423555e-05,
+      "loss": 1.1962,
+      "step": 1611
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971790240268e-05,
+      "loss": 1.073,
+      "step": 1612
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971752913501e-05,
+      "loss": 1.1562,
+      "step": 1613
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971715562057e-05,
+      "loss": 1.1068,
+      "step": 1614
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971678185933e-05,
+      "loss": 1.1081,
+      "step": 1615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971640785131e-05,
+      "loss": 1.0775,
+      "step": 1616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997160335965e-05,
+      "loss": 1.083,
+      "step": 1617
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999715659094906e-05,
+      "loss": 1.0268,
+      "step": 1618
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999715284346524e-05,
+      "loss": 0.9439,
+      "step": 1619
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971490935136e-05,
+      "loss": 0.7094,
+      "step": 1620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999714534109405e-05,
+      "loss": 1.4053,
+      "step": 1621
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971415862066e-05,
+      "loss": 0.924,
+      "step": 1622
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999713782885134e-05,
+      "loss": 1.1412,
+      "step": 1623
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999713406902814e-05,
+      "loss": 1.0323,
+      "step": 1624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999713030673717e-05,
+      "loss": 1.2048,
+      "step": 1625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971265419783e-05,
+      "loss": 0.7828,
+      "step": 1626
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999712277475154e-05,
+      "loss": 0.9369,
+      "step": 1627
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999711900505695e-05,
+      "loss": 1.1362,
+      "step": 1628
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999711523289445e-05,
+      "loss": 1.134,
+      "step": 1629
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971114582641e-05,
+      "loss": 1.2234,
+      "step": 1630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971076811659e-05,
+      "loss": 1.2499,
+      "step": 1631
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971039015998e-05,
+      "loss": 1.2591,
+      "step": 1632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999971001195659e-05,
+      "loss": 1.4315,
+      "step": 1633
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970963350641e-05,
+      "loss": 1.0759,
+      "step": 1634
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970925480944e-05,
+      "loss": 1.169,
+      "step": 1635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970887586568e-05,
+      "loss": 1.393,
+      "step": 1636
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970849667515e-05,
+      "loss": 1.3423,
+      "step": 1637
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999708117237814e-05,
+      "loss": 1.0546,
+      "step": 1638
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997077375537e-05,
+      "loss": 0.8396,
+      "step": 1639
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970735762281e-05,
+      "loss": 1.1582,
+      "step": 1640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970697744511e-05,
+      "loss": 1.021,
+      "step": 1641
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970659702064e-05,
+      "loss": 1.0423,
+      "step": 1642
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970621634938e-05,
+      "loss": 0.8419,
+      "step": 1643
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970583543133e-05,
+      "loss": 1.0889,
+      "step": 1644
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999705454266496e-05,
+      "loss": 1.2187,
+      "step": 1645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970507285488e-05,
+      "loss": 1.3234,
+      "step": 1646
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970469119647e-05,
+      "loss": 1.1936,
+      "step": 1647
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970430929128e-05,
+      "loss": 1.1664,
+      "step": 1648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999703927139296e-05,
+      "loss": 0.8755,
+      "step": 1649
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970354474053e-05,
+      "loss": 0.4793,
+      "step": 1650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970316209498e-05,
+      "loss": 0.4663,
+      "step": 1651
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970277920264e-05,
+      "loss": 1.0177,
+      "step": 1652
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970239606352e-05,
+      "loss": 0.8776,
+      "step": 1653
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970201267761e-05,
+      "loss": 1.1567,
+      "step": 1654
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99997016290449e-05,
+      "loss": 1.1212,
+      "step": 1655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970124516542e-05,
+      "loss": 1.0755,
+      "step": 1656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970086103915e-05,
+      "loss": 1.2144,
+      "step": 1657
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999970047666609e-05,
+      "loss": 1.1866,
+      "step": 1658
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999700092046245e-05,
+      "loss": 0.9839,
+      "step": 1659
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999699707179614e-05,
+      "loss": 1.3086,
+      "step": 1660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996993220662e-05,
+      "loss": 1.105,
+      "step": 1661
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969893670599e-05,
+      "loss": 1.253,
+      "step": 1662
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999698551099e-05,
+      "loss": 1.546,
+      "step": 1663
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969816524522e-05,
+      "loss": 1.0515,
+      "step": 1664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969777914466e-05,
+      "loss": 1.2819,
+      "step": 1665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999697392797306e-05,
+      "loss": 0.9853,
+      "step": 1666
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999697006203175e-05,
+      "loss": 1.0811,
+      "step": 1667
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999696619362245e-05,
+      "loss": 1.0727,
+      "step": 1668
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969623227454e-05,
+      "loss": 1.1194,
+      "step": 1669
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999695844940045e-05,
+      "loss": 0.961,
+      "step": 1670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969545735876e-05,
+      "loss": 1.0931,
+      "step": 1671
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969506953069e-05,
+      "loss": 1.2467,
+      "step": 1672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969468145583e-05,
+      "loss": 1.2632,
+      "step": 1673
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999694293134195e-05,
+      "loss": 1.3062,
+      "step": 1674
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999693904565765e-05,
+      "loss": 1.0255,
+      "step": 1675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969351575055e-05,
+      "loss": 0.4736,
+      "step": 1676
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969312668855e-05,
+      "loss": 0.4632,
+      "step": 1677
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969273737976e-05,
+      "loss": 1.1215,
+      "step": 1678
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999692347824186e-05,
+      "loss": 1.2863,
+      "step": 1679
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999691958021826e-05,
+      "loss": 1.2319,
+      "step": 1680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969156797268e-05,
+      "loss": 1.1239,
+      "step": 1681
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999691177676745e-05,
+      "loss": 1.1725,
+      "step": 1682
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999690787134024e-05,
+      "loss": 1.3194,
+      "step": 1683
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999690396344525e-05,
+      "loss": 1.767,
+      "step": 1684
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969000530823e-05,
+      "loss": 1.1727,
+      "step": 1685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999689614025146e-05,
+      "loss": 0.942,
+      "step": 1686
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999689222495286e-05,
+      "loss": 0.8949,
+      "step": 1687
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968883071863e-05,
+      "loss": 1.0356,
+      "step": 1688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968843869519e-05,
+      "loss": 1.1545,
+      "step": 1689
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968804642496e-05,
+      "loss": 1.0806,
+      "step": 1690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968765390796e-05,
+      "loss": 1.6976,
+      "step": 1691
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968726114416e-05,
+      "loss": 1.4155,
+      "step": 1692
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968686813358e-05,
+      "loss": 1.2434,
+      "step": 1693
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999686474876205e-05,
+      "loss": 1.0439,
+      "step": 1694
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999686081372046e-05,
+      "loss": 0.9256,
+      "step": 1695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996856876211e-05,
+      "loss": 1.2005,
+      "step": 1696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999685293623376e-05,
+      "loss": 1.0603,
+      "step": 1697
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968489937886e-05,
+      "loss": 1.3092,
+      "step": 1698
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968450488756e-05,
+      "loss": 1.3336,
+      "step": 1699
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968411014947e-05,
+      "loss": 1.0654,
+      "step": 1700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968371516459e-05,
+      "loss": 1.1481,
+      "step": 1701
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968331993293e-05,
+      "loss": 0.9474,
+      "step": 1702
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968292445448e-05,
+      "loss": 1.5143,
+      "step": 1703
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999682528729254e-05,
+      "loss": 2.0976,
+      "step": 1704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968213275723e-05,
+      "loss": 1.2388,
+      "step": 1705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968173653843e-05,
+      "loss": 1.3075,
+      "step": 1706
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999681340072835e-05,
+      "loss": 1.5414,
+      "step": 1707
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968094336045e-05,
+      "loss": 0.8223,
+      "step": 1708
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968054640128e-05,
+      "loss": 1.1105,
+      "step": 1709
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999968014919534e-05,
+      "loss": 1.3593,
+      "step": 1710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996797517426e-05,
+      "loss": 1.2691,
+      "step": 1711
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967935404307e-05,
+      "loss": 0.9297,
+      "step": 1712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967895609676e-05,
+      "loss": 1.17,
+      "step": 1713
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967855790366e-05,
+      "loss": 0.9769,
+      "step": 1714
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967815946378e-05,
+      "loss": 0.8034,
+      "step": 1715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999677760777114e-05,
+      "loss": 0.9674,
+      "step": 1716
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967736184366e-05,
+      "loss": 0.9551,
+      "step": 1717
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967696266341e-05,
+      "loss": 1.3977,
+      "step": 1718
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999676563236384e-05,
+      "loss": 0.9968,
+      "step": 1719
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967616356257e-05,
+      "loss": 0.7371,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967576364197e-05,
+      "loss": 0.9805,
+      "step": 1721
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999675363474584e-05,
+      "loss": 1.1072,
+      "step": 1722
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967496306041e-05,
+      "loss": 1.0521,
+      "step": 1723
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967456239945e-05,
+      "loss": 0.7374,
+      "step": 1724
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967416149171e-05,
+      "loss": 1.0237,
+      "step": 1725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999673760337175e-05,
+      "loss": 1.0984,
+      "step": 1726
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967335893585e-05,
+      "loss": 1.1678,
+      "step": 1727
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967295728775e-05,
+      "loss": 1.3822,
+      "step": 1728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967255539286e-05,
+      "loss": 1.2259,
+      "step": 1729
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967215325118e-05,
+      "loss": 1.2165,
+      "step": 1730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967175086272e-05,
+      "loss": 1.2102,
+      "step": 1731
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999671348227465e-05,
+      "loss": 1.3913,
+      "step": 1732
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999967094534543e-05,
+      "loss": 0.938,
+      "step": 1733
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999670542216614e-05,
+      "loss": 1.6322,
+      "step": 1734
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999670138841005e-05,
+      "loss": 1.2323,
+      "step": 1735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999669735218605e-05,
+      "loss": 1.0992,
+      "step": 1736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966933134943e-05,
+      "loss": 1.1555,
+      "step": 1737
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999668927233456e-05,
+      "loss": 1.0509,
+      "step": 1738
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966852287071e-05,
+      "loss": 1.0388,
+      "step": 1739
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966811826117e-05,
+      "loss": 1.2675,
+      "step": 1740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966771340484e-05,
+      "loss": 1.161,
+      "step": 1741
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999667308301736e-05,
+      "loss": 1.75,
+      "step": 1742
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999666902951835e-05,
+      "loss": 0.7144,
+      "step": 1743
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966649735515e-05,
+      "loss": 0.7856,
+      "step": 1744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966609151168e-05,
+      "loss": 1.1215,
+      "step": 1745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999665685421425e-05,
+      "loss": 1.0481,
+      "step": 1746
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966527908438e-05,
+      "loss": 1.1639,
+      "step": 1747
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966487250055e-05,
+      "loss": 0.5334,
+      "step": 1748
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966446566994e-05,
+      "loss": 1.1725,
+      "step": 1749
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966405859253e-05,
+      "loss": 1.1146,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999663651268346e-05,
+      "loss": 1.1789,
+      "step": 1751
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999663243697376e-05,
+      "loss": 1.1226,
+      "step": 1752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999662835879615e-05,
+      "loss": 1.1025,
+      "step": 1753
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966242781507e-05,
+      "loss": 1.0359,
+      "step": 1754
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966201950374e-05,
+      "loss": 1.102,
+      "step": 1755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966161094562e-05,
+      "loss": 1.283,
+      "step": 1756
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999661202140715e-05,
+      "loss": 1.1886,
+      "step": 1757
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999966079308902e-05,
+      "loss": 1.2364,
+      "step": 1758
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999660383790546e-05,
+      "loss": 1.0793,
+      "step": 1759
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965997424529e-05,
+      "loss": 0.7617,
+      "step": 1760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965956445324e-05,
+      "loss": 1.2649,
+      "step": 1761
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996591544144e-05,
+      "loss": 1.3203,
+      "step": 1762
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999658744128785e-05,
+      "loss": 0.9644,
+      "step": 1763
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965833359638e-05,
+      "loss": 1.1444,
+      "step": 1764
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999657922817186e-05,
+      "loss": 1.0832,
+      "step": 1765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965751179121e-05,
+      "loss": 1.0513,
+      "step": 1766
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965710051844e-05,
+      "loss": 0.9997,
+      "step": 1767
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999656688998895e-05,
+      "loss": 1.2584,
+      "step": 1768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965627723256e-05,
+      "loss": 1.4774,
+      "step": 1769
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965586521943e-05,
+      "loss": 1.3373,
+      "step": 1770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965545295952e-05,
+      "loss": 0.9803,
+      "step": 1771
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965504045283e-05,
+      "loss": 0.8081,
+      "step": 1772
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999654627699345e-05,
+      "loss": 1.2502,
+      "step": 1773
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965421469908e-05,
+      "loss": 0.9243,
+      "step": 1774
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999653801452024e-05,
+      "loss": 0.9536,
+      "step": 1775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999653387958186e-05,
+      "loss": 1.033,
+      "step": 1776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965297421756e-05,
+      "loss": 1.0354,
+      "step": 1777
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965256023015e-05,
+      "loss": 1.0212,
+      "step": 1778
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999652145995944e-05,
+      "loss": 1.3128,
+      "step": 1779
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965173151496e-05,
+      "loss": 0.8684,
+      "step": 1780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996513167872e-05,
+      "loss": 1.0906,
+      "step": 1781
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965090181264e-05,
+      "loss": 1.2967,
+      "step": 1782
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999650486591295e-05,
+      "loss": 1.1864,
+      "step": 1783
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965007112317e-05,
+      "loss": 1.2853,
+      "step": 1784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964965540825e-05,
+      "loss": 1.3813,
+      "step": 1785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999649239446553e-05,
+      "loss": 1.1605,
+      "step": 1786
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964882323807e-05,
+      "loss": 1.12,
+      "step": 1787
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999648406782795e-05,
+      "loss": 1.1585,
+      "step": 1788
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964799008074e-05,
+      "loss": 0.8351,
+      "step": 1789
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996475731319e-05,
+      "loss": 1.0039,
+      "step": 1790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964715593627e-05,
+      "loss": 0.9833,
+      "step": 1791
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999646738493855e-05,
+      "loss": 1.0115,
+      "step": 1792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964632080465e-05,
+      "loss": 0.8991,
+      "step": 1793
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999645902868666e-05,
+      "loss": 0.9605,
+      "step": 1794
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999645484685884e-05,
+      "loss": 0.9477,
+      "step": 1795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964506625633e-05,
+      "loss": 0.8547,
+      "step": 1796
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964464757998e-05,
+      "loss": 0.9197,
+      "step": 1797
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964422865685e-05,
+      "loss": 1.047,
+      "step": 1798
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964380948694e-05,
+      "loss": 0.7114,
+      "step": 1799
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999643390070225e-05,
+      "loss": 1.0012,
+      "step": 1800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964297040674e-05,
+      "loss": 1.1069,
+      "step": 1801
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999642550496467e-05,
+      "loss": 1.0559,
+      "step": 1802
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964213033941e-05,
+      "loss": 1.1771,
+      "step": 1803
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999641709935556e-05,
+      "loss": 0.9095,
+      "step": 1804
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964128928493e-05,
+      "loss": 1.3731,
+      "step": 1805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964086838751e-05,
+      "loss": 1.2707,
+      "step": 1806
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996404472433e-05,
+      "loss": 1.3383,
+      "step": 1807
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999964002585231e-05,
+      "loss": 1.1379,
+      "step": 1808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963960421453e-05,
+      "loss": 1.5423,
+      "step": 1809
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963918232997e-05,
+      "loss": 1.1511,
+      "step": 1810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963876019862e-05,
+      "loss": 1.3019,
+      "step": 1811
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963833782049e-05,
+      "loss": 1.1668,
+      "step": 1812
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963791519556e-05,
+      "loss": 0.9874,
+      "step": 1813
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999637492323856e-05,
+      "loss": 1.431,
+      "step": 1814
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999637069205366e-05,
+      "loss": 1.8638,
+      "step": 1815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999636645840084e-05,
+      "loss": 1.1395,
+      "step": 1816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963622222803e-05,
+      "loss": 1.125,
+      "step": 1817
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999635798369174e-05,
+      "loss": 1.0607,
+      "step": 1818
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999635374263545e-05,
+      "loss": 1.1098,
+      "step": 1819
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999634949911124e-05,
+      "loss": 0.6851,
+      "step": 1820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963452531191e-05,
+      "loss": 1.1234,
+      "step": 1821
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963410046592e-05,
+      "loss": 1.3048,
+      "step": 1822
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963367537313e-05,
+      "loss": 1.1247,
+      "step": 1823
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999633250033574e-05,
+      "loss": 1.0778,
+      "step": 1824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963282444722e-05,
+      "loss": 0.8994,
+      "step": 1825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963239861409e-05,
+      "loss": 1.1417,
+      "step": 1826
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999631972534166e-05,
+      "loss": 1.0574,
+      "step": 1827
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963154620746e-05,
+      "loss": 1.1599,
+      "step": 1828
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963111963396e-05,
+      "loss": 1.238,
+      "step": 1829
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999963069281368e-05,
+      "loss": 1.6336,
+      "step": 1830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999630265746616e-05,
+      "loss": 2.0859,
+      "step": 1831
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999629838432765e-05,
+      "loss": 1.2952,
+      "step": 1832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962941087213e-05,
+      "loss": 1.384,
+      "step": 1833
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962898306471e-05,
+      "loss": 0.9765,
+      "step": 1834
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962855501049e-05,
+      "loss": 1.0935,
+      "step": 1835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996281267095e-05,
+      "loss": 1.1919,
+      "step": 1836
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999627698161725e-05,
+      "loss": 0.9998,
+      "step": 1837
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962726936715e-05,
+      "loss": 1.155,
+      "step": 1838
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999626840325794e-05,
+      "loss": 1.1796,
+      "step": 1839
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962641103766e-05,
+      "loss": 1.3332,
+      "step": 1840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962598150274e-05,
+      "loss": 1.0477,
+      "step": 1841
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962555172103e-05,
+      "loss": 1.112,
+      "step": 1842
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999625121692536e-05,
+      "loss": 0.9866,
+      "step": 1843
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999624691417254e-05,
+      "loss": 0.7711,
+      "step": 1844
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962426089518e-05,
+      "loss": 0.9799,
+      "step": 1845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962383012633e-05,
+      "loss": 0.6033,
+      "step": 1846
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999623399110694e-05,
+      "loss": 0.7475,
+      "step": 1847
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962296784827e-05,
+      "loss": 0.7878,
+      "step": 1848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962253633906e-05,
+      "loss": 1.272,
+      "step": 1849
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999622104583065e-05,
+      "loss": 1.2431,
+      "step": 1850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962167258029e-05,
+      "loss": 1.1266,
+      "step": 1851
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962124033072e-05,
+      "loss": 1.2129,
+      "step": 1852
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999620807834366e-05,
+      "loss": 1.1271,
+      "step": 1853
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962037509123e-05,
+      "loss": 1.1641,
+      "step": 1854
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996199421013e-05,
+      "loss": 1.5963,
+      "step": 1855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961950886459e-05,
+      "loss": 1.5356,
+      "step": 1856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996190753811e-05,
+      "loss": 1.118,
+      "step": 1857
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961864165081e-05,
+      "loss": 1.4149,
+      "step": 1858
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999618207673746e-05,
+      "loss": 0.9634,
+      "step": 1859
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961777344989e-05,
+      "loss": 1.0974,
+      "step": 1860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961733897925e-05,
+      "loss": 1.1586,
+      "step": 1861
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961690426183e-05,
+      "loss": 1.1899,
+      "step": 1862
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961646929762e-05,
+      "loss": 1.4185,
+      "step": 1863
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961603408663e-05,
+      "loss": 0.9067,
+      "step": 1864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961559862885e-05,
+      "loss": 1.2734,
+      "step": 1865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999615162924276e-05,
+      "loss": 1.0768,
+      "step": 1866
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999614726972925e-05,
+      "loss": 1.469,
+      "step": 1867
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961429077479e-05,
+      "loss": 1.3085,
+      "step": 1868
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961385432986e-05,
+      "loss": 1.064,
+      "step": 1869
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961341763816e-05,
+      "loss": 1.2152,
+      "step": 1870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961298069966e-05,
+      "loss": 0.6752,
+      "step": 1871
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961254351438e-05,
+      "loss": 0.4043,
+      "step": 1872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961210608232e-05,
+      "loss": 0.7948,
+      "step": 1873
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999611668403455e-05,
+      "loss": 1.1583,
+      "step": 1874
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961123047782e-05,
+      "loss": 1.4685,
+      "step": 1875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996107923054e-05,
+      "loss": 1.2818,
+      "step": 1876
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999961035388619e-05,
+      "loss": 1.0971,
+      "step": 1877
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996099152202e-05,
+      "loss": 1.4049,
+      "step": 1878
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999609476307415e-05,
+      "loss": 1.1807,
+      "step": 1879
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960903714785e-05,
+      "loss": 1.435,
+      "step": 1880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99996085977415e-05,
+      "loss": 1.14,
+      "step": 1881
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960815808836e-05,
+      "loss": 1.0288,
+      "step": 1882
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999607718188435e-05,
+      "loss": 0.9891,
+      "step": 1883
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960727804173e-05,
+      "loss": 1.1035,
+      "step": 1884
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960683764824e-05,
+      "loss": 1.2492,
+      "step": 1885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960639700796e-05,
+      "loss": 1.1905,
+      "step": 1886
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960595612089e-05,
+      "loss": 1.0795,
+      "step": 1887
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960551498704e-05,
+      "loss": 1.0844,
+      "step": 1888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999605073606406e-05,
+      "loss": 1.2944,
+      "step": 1889
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999604631978986e-05,
+      "loss": 1.1076,
+      "step": 1890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999604190104774e-05,
+      "loss": 1.2012,
+      "step": 1891
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999603747983784e-05,
+      "loss": 1.2852,
+      "step": 1892
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999603305616e-05,
+      "loss": 1.0921,
+      "step": 1893
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960286300144e-05,
+      "loss": 1.2454,
+      "step": 1894
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960242014009e-05,
+      "loss": 1.0453,
+      "step": 1895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999601977031957e-05,
+      "loss": 1.1153,
+      "step": 1896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999601533677036e-05,
+      "loss": 1.0698,
+      "step": 1897
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960109007533e-05,
+      "loss": 1.2448,
+      "step": 1898
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999960064622684e-05,
+      "loss": 1.2356,
+      "step": 1899
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999600202131567e-05,
+      "loss": 0.9474,
+      "step": 1900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995997577895e-05,
+      "loss": 1.1506,
+      "step": 1901
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959931320065e-05,
+      "loss": 1.3618,
+      "step": 1902
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999598868365014e-05,
+      "loss": 1.0169,
+      "step": 1903
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995984232826e-05,
+      "loss": 1.2114,
+      "step": 1904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999597977953395e-05,
+      "loss": 1.1541,
+      "step": 1905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999597532377405e-05,
+      "loss": 0.8533,
+      "step": 1906
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959708655463e-05,
+      "loss": 1.1301,
+      "step": 1907
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959664048507e-05,
+      "loss": 0.9514,
+      "step": 1908
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959619416872e-05,
+      "loss": 1.0316,
+      "step": 1909
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999595747605585e-05,
+      "loss": 1.1069,
+      "step": 1910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959530079567e-05,
+      "loss": 0.9771,
+      "step": 1911
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999594853738966e-05,
+      "loss": 0.8837,
+      "step": 1912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959440643548e-05,
+      "loss": 0.7769,
+      "step": 1913
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995939588852e-05,
+      "loss": 0.7168,
+      "step": 1914
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959351108815e-05,
+      "loss": 0.6889,
+      "step": 1915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999593063044306e-05,
+      "loss": 0.5915,
+      "step": 1916
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959261475367e-05,
+      "loss": 0.6089,
+      "step": 1917
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999592166216256e-05,
+      "loss": 0.6052,
+      "step": 1918
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959171743205e-05,
+      "loss": 0.5808,
+      "step": 1919
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999959126840107e-05,
+      "loss": 0.5909,
+      "step": 1920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995908191233e-05,
+      "loss": 0.5403,
+      "step": 1921
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999590369598734e-05,
+      "loss": 0.6327,
+      "step": 1922
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995899198274e-05,
+      "loss": 0.5225,
+      "step": 1923
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958946980927e-05,
+      "loss": 1.0703,
+      "step": 1924
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999589019544355e-05,
+      "loss": 1.2725,
+      "step": 1925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958856903265e-05,
+      "loss": 1.1882,
+      "step": 1926
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999588118274174e-05,
+      "loss": 1.03,
+      "step": 1927
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995876672689e-05,
+      "loss": 1.3146,
+      "step": 1928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958721601684e-05,
+      "loss": 1.1833,
+      "step": 1929
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999586764518e-05,
+      "loss": 1.0596,
+      "step": 1930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958631277238e-05,
+      "loss": 1.2622,
+      "step": 1931
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958586077996e-05,
+      "loss": 0.862,
+      "step": 1932
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958540854077e-05,
+      "loss": 1.0546,
+      "step": 1933
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958495605479e-05,
+      "loss": 0.5355,
+      "step": 1934
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958450332202e-05,
+      "loss": 1.4974,
+      "step": 1935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999584050342466e-05,
+      "loss": 1.2706,
+      "step": 1936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958359711613e-05,
+      "loss": 1.0658,
+      "step": 1937
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999583143643e-05,
+      "loss": 1.0819,
+      "step": 1938
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995826899231e-05,
+      "loss": 1.0302,
+      "step": 1939
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995822359564e-05,
+      "loss": 1.0842,
+      "step": 1940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958178174292e-05,
+      "loss": 1.1691,
+      "step": 1941
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999958132728266e-05,
+      "loss": 1.2302,
+      "step": 1942
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999580872575605e-05,
+      "loss": 0.9427,
+      "step": 1943
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999580417621775e-05,
+      "loss": 1.3759,
+      "step": 1944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957996242115e-05,
+      "loss": 1.1939,
+      "step": 1945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957950697374e-05,
+      "loss": 0.8473,
+      "step": 1946
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999579051279555e-05,
+      "loss": 1.031,
+      "step": 1947
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957859533857e-05,
+      "loss": 0.941,
+      "step": 1948
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957813915082e-05,
+      "loss": 0.9295,
+      "step": 1949
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999577682716265e-05,
+      "loss": 1.6521,
+      "step": 1950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999577226034935e-05,
+      "loss": 1.2613,
+      "step": 1951
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957676910681e-05,
+      "loss": 1.4755,
+      "step": 1952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957631193191e-05,
+      "loss": 1.5328,
+      "step": 1953
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957585451022e-05,
+      "loss": 1.3601,
+      "step": 1954
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999575396841746e-05,
+      "loss": 1.2172,
+      "step": 1955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999574938926485e-05,
+      "loss": 1.0954,
+      "step": 1956
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999574480764446e-05,
+      "loss": 1.0489,
+      "step": 1957
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957402235561e-05,
+      "loss": 1.1921,
+      "step": 1958
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995735637e-05,
+      "loss": 1.1143,
+      "step": 1959
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995731047976e-05,
+      "loss": 1.2846,
+      "step": 1960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957264564841e-05,
+      "loss": 1.1944,
+      "step": 1961
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957218625244e-05,
+      "loss": 1.2594,
+      "step": 1962
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999571726609686e-05,
+      "loss": 0.9978,
+      "step": 1963
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957126672015e-05,
+      "loss": 0.971,
+      "step": 1964
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957080658382e-05,
+      "loss": 1.1946,
+      "step": 1965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999957034620071e-05,
+      "loss": 1.0838,
+      "step": 1966
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956988557081e-05,
+      "loss": 1.233,
+      "step": 1967
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999569424694124e-05,
+      "loss": 1.2316,
+      "step": 1968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956896357066e-05,
+      "loss": 1.1695,
+      "step": 1969
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956850220041e-05,
+      "loss": 1.1943,
+      "step": 1970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956804058337e-05,
+      "loss": 1.1467,
+      "step": 1971
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999567578719545e-05,
+      "loss": 1.1771,
+      "step": 1972
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956711660894e-05,
+      "loss": 1.071,
+      "step": 1973
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999566654251544e-05,
+      "loss": 1.0812,
+      "step": 1974
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956619164736e-05,
+      "loss": 0.6495,
+      "step": 1975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995657287964e-05,
+      "loss": 0.8987,
+      "step": 1976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999565265698656e-05,
+      "loss": 1.7146,
+      "step": 1977
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956480235412e-05,
+      "loss": 0.9659,
+      "step": 1978
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995643387628e-05,
+      "loss": 0.9835,
+      "step": 1979
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956387492469e-05,
+      "loss": 1.2594,
+      "step": 1980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999563410839806e-05,
+      "loss": 1.1015,
+      "step": 1981
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956294650813e-05,
+      "loss": 1.5359,
+      "step": 1982
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956248192967e-05,
+      "loss": 1.1858,
+      "step": 1983
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956201710443e-05,
+      "loss": 2.232,
+      "step": 1984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956155203239e-05,
+      "loss": 1.0845,
+      "step": 1985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999561086713576e-05,
+      "loss": 1.1301,
+      "step": 1986
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999560621147975e-05,
+      "loss": 3.6122,
+      "step": 1987
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956015533559e-05,
+      "loss": 1.1042,
+      "step": 1988
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955968927642e-05,
+      "loss": 1.5292,
+      "step": 1989
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955922297047e-05,
+      "loss": 1.2125,
+      "step": 1990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955875641772e-05,
+      "loss": 1.2061,
+      "step": 1991
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999558289618196e-05,
+      "loss": 1.3218,
+      "step": 1992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999557822571887e-05,
+      "loss": 1.2492,
+      "step": 1993
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955735527879e-05,
+      "loss": 1.6787,
+      "step": 1994
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999556887738906e-05,
+      "loss": 1.187,
+      "step": 1995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955641995224e-05,
+      "loss": 1.8543,
+      "step": 1996
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955595191879e-05,
+      "loss": 0.6718,
+      "step": 1997
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955548363855e-05,
+      "loss": 0.9185,
+      "step": 1998
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999555015111534e-05,
+      "loss": 1.3068,
+      "step": 1999
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999554546337724e-05,
+      "loss": 1.372,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.09769606590271,
+      "eval_runtime": 85.4918,
+      "eval_samples_per_second": 16.2,
+      "eval_steps_per_second": 4.059,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999554077317136e-05,
+      "loss": 1.1012,
+      "step": 2001
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999553608049757e-05,
+      "loss": 1.2555,
+      "step": 2002
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955313853559e-05,
+      "loss": 1.185,
+      "step": 2003
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955266877465e-05,
+      "loss": 1.1243,
+      "step": 2004
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955219876692e-05,
+      "loss": 1.296,
+      "step": 2005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995517285124e-05,
+      "loss": 1.1022,
+      "step": 2006
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995512580111e-05,
+      "loss": 1.2041,
+      "step": 2007
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999550787263014e-05,
+      "loss": 1.0095,
+      "step": 2008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999955031626814e-05,
+      "loss": 1.1557,
+      "step": 2009
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999549845026484e-05,
+      "loss": 1.3879,
+      "step": 2010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999549373538035e-05,
+      "loss": 0.9205,
+      "step": 2011
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999548901802815e-05,
+      "loss": 0.1468,
+      "step": 2012
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995484298208e-05,
+      "loss": 0.268,
+      "step": 2013
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999547957592e-05,
+      "loss": 0.6679,
+      "step": 2014
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954748511641e-05,
+      "loss": 0.8559,
+      "step": 2015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999547012394054e-05,
+      "loss": 1.0634,
+      "step": 2016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995465394249e-05,
+      "loss": 1.2137,
+      "step": 2017
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954606620896e-05,
+      "loss": 1.2434,
+      "step": 2018
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954559274624e-05,
+      "loss": 1.0491,
+      "step": 2019
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954511903673e-05,
+      "loss": 1.0955,
+      "step": 2020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954464508044e-05,
+      "loss": 1.3627,
+      "step": 2021
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999544170877366e-05,
+      "loss": 0.8931,
+      "step": 2022
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995436964275e-05,
+      "loss": 1.1627,
+      "step": 2023
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954322173085e-05,
+      "loss": 1.2295,
+      "step": 2024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954274678742e-05,
+      "loss": 1.1959,
+      "step": 2025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995422715972e-05,
+      "loss": 1.1635,
+      "step": 2026
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999541796160196e-05,
+      "loss": 0.9611,
+      "step": 2027
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954132047642e-05,
+      "loss": 0.7987,
+      "step": 2028
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999540844545846e-05,
+      "loss": 1.4583,
+      "step": 2029
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954036836849e-05,
+      "loss": 1.2441,
+      "step": 2030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953989194435e-05,
+      "loss": 1.5323,
+      "step": 2031
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999539415273415e-05,
+      "loss": 1.1934,
+      "step": 2032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953893835571e-05,
+      "loss": 1.0853,
+      "step": 2033
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953846119121e-05,
+      "loss": 1.1559,
+      "step": 2034
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999537983779934e-05,
+      "loss": 1.0736,
+      "step": 2035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953750612186e-05,
+      "loss": 1.0305,
+      "step": 2036
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999537028217015e-05,
+      "loss": 1.2131,
+      "step": 2037
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953655006537e-05,
+      "loss": 1.1931,
+      "step": 2038
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953607166695e-05,
+      "loss": 1.179,
+      "step": 2039
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953559302174e-05,
+      "loss": 1.0621,
+      "step": 2040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953511412976e-05,
+      "loss": 1.3458,
+      "step": 2041
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953463499098e-05,
+      "loss": 1.1783,
+      "step": 2042
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953415560542e-05,
+      "loss": 1.159,
+      "step": 2043
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999533675973076e-05,
+      "loss": 1.3375,
+      "step": 2044
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999533196093946e-05,
+      "loss": 0.6313,
+      "step": 2045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953271596803e-05,
+      "loss": 0.9884,
+      "step": 2046
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999532235595325e-05,
+      "loss": 1.2119,
+      "step": 2047
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953175497585e-05,
+      "loss": 1.2758,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953127410957e-05,
+      "loss": 0.9148,
+      "step": 2049
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953079299652e-05,
+      "loss": 1.0894,
+      "step": 2050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953031163668e-05,
+      "loss": 0.7798,
+      "step": 2051
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999529830030056e-05,
+      "loss": 0.7435,
+      "step": 2052
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952934817664e-05,
+      "loss": 0.8528,
+      "step": 2053
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952886607645e-05,
+      "loss": 1.2994,
+      "step": 2054
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952838372948e-05,
+      "loss": 1.0807,
+      "step": 2055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999527901135715e-05,
+      "loss": 1.2836,
+      "step": 2056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952741829517e-05,
+      "loss": 1.291,
+      "step": 2057
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952693520783e-05,
+      "loss": 1.1041,
+      "step": 2058
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952645187371e-05,
+      "loss": 1.0205,
+      "step": 2059
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952596829281e-05,
+      "loss": 0.9736,
+      "step": 2060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952548446512e-05,
+      "loss": 1.2984,
+      "step": 2061
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952500039065e-05,
+      "loss": 1.2229,
+      "step": 2062
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995245160694e-05,
+      "loss": 1.6288,
+      "step": 2063
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952403150135e-05,
+      "loss": 1.4125,
+      "step": 2064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952354668652e-05,
+      "loss": 1.3746,
+      "step": 2065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952306162492e-05,
+      "loss": 1.1113,
+      "step": 2066
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999522576316516e-05,
+      "loss": 1.2301,
+      "step": 2067
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952209076134e-05,
+      "loss": 1.1858,
+      "step": 2068
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952160495937e-05,
+      "loss": 1.0311,
+      "step": 2069
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999952111891062e-05,
+      "loss": 1.2018,
+      "step": 2070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999520632615084e-05,
+      "loss": 1.0876,
+      "step": 2071
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999520146072766e-05,
+      "loss": 1.0011,
+      "step": 2072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951965928366e-05,
+      "loss": 1.1342,
+      "step": 2073
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999519172247775e-05,
+      "loss": 0.8343,
+      "step": 2074
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999518684965096e-05,
+      "loss": 1.3023,
+      "step": 2075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951819743563e-05,
+      "loss": 1.0484,
+      "step": 2076
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995177096594e-05,
+      "loss": 0.8818,
+      "step": 2077
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951722163637e-05,
+      "loss": 0.5781,
+      "step": 2078
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951673336656e-05,
+      "loss": 1.1657,
+      "step": 2079
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999516244849956e-05,
+      "loss": 1.2215,
+      "step": 2080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951575608658e-05,
+      "loss": 1.0539,
+      "step": 2081
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951526707641e-05,
+      "loss": 1.2303,
+      "step": 2082
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951477781946e-05,
+      "loss": 1.5738,
+      "step": 2083
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951428831572e-05,
+      "loss": 1.1878,
+      "step": 2084
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999513798565204e-05,
+      "loss": 1.1655,
+      "step": 2085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951330856789e-05,
+      "loss": 1.0752,
+      "step": 2086
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995128183238e-05,
+      "loss": 0.9155,
+      "step": 2087
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951232783293e-05,
+      "loss": 1.0812,
+      "step": 2088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951183709527e-05,
+      "loss": 1.0981,
+      "step": 2089
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999511346110827e-05,
+      "loss": 1.0767,
+      "step": 2090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951085487959e-05,
+      "loss": 0.9912,
+      "step": 2091
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999951036340158e-05,
+      "loss": 1.015,
+      "step": 2092
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950987167679e-05,
+      "loss": 1.6259,
+      "step": 2093
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995093797052e-05,
+      "loss": 1.4019,
+      "step": 2094
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950888748684e-05,
+      "loss": 1.5632,
+      "step": 2095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999508395021685e-05,
+      "loss": 1.2167,
+      "step": 2096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950790230975e-05,
+      "loss": 1.2337,
+      "step": 2097
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950740935103e-05,
+      "loss": 1.116,
+      "step": 2098
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950691614552e-05,
+      "loss": 1.1899,
+      "step": 2099
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950642269323e-05,
+      "loss": 1.1664,
+      "step": 2100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999505928994154e-05,
+      "loss": 0.9768,
+      "step": 2101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99995054350483e-05,
+      "loss": 1.3383,
+      "step": 2102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950494085565e-05,
+      "loss": 1.09,
+      "step": 2103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999504446416225e-05,
+      "loss": 1.6185,
+      "step": 2104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950395173001e-05,
+      "loss": 1.2097,
+      "step": 2105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950345679702e-05,
+      "loss": 0.9709,
+      "step": 2106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950296161723e-05,
+      "loss": 1.4045,
+      "step": 2107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999502466190664e-05,
+      "loss": 1.187,
+      "step": 2108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950197051731e-05,
+      "loss": 1.2063,
+      "step": 2109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950147459718e-05,
+      "loss": 1.1091,
+      "step": 2110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999500978430256e-05,
+      "loss": 1.2056,
+      "step": 2111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999950048201655e-05,
+      "loss": 0.812,
+      "step": 2112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999499985356055e-05,
+      "loss": 0.5464,
+      "step": 2113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999499488448785e-05,
+      "loss": 1.2549,
+      "step": 2114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949899129472e-05,
+      "loss": 1.3707,
+      "step": 2115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949849389388e-05,
+      "loss": 1.2616,
+      "step": 2116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949799624626e-05,
+      "loss": 1.1822,
+      "step": 2117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949749835184e-05,
+      "loss": 1.0703,
+      "step": 2118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949700021065e-05,
+      "loss": 1.0636,
+      "step": 2119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949650182266e-05,
+      "loss": 1.1332,
+      "step": 2120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994960031879e-05,
+      "loss": 1.0944,
+      "step": 2121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949550430635e-05,
+      "loss": 1.0199,
+      "step": 2122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999495005178015e-05,
+      "loss": 0.8863,
+      "step": 2123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949450580289e-05,
+      "loss": 1.181,
+      "step": 2124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999494006180994e-05,
+      "loss": 1.0959,
+      "step": 2125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994935063123e-05,
+      "loss": 1.0288,
+      "step": 2126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949300619683e-05,
+      "loss": 1.1203,
+      "step": 2127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949250583457e-05,
+      "loss": 1.0955,
+      "step": 2128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999492005225535e-05,
+      "loss": 1.0168,
+      "step": 2129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994915043697e-05,
+      "loss": 0.8773,
+      "step": 2130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994910032671e-05,
+      "loss": 0.4928,
+      "step": 2131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994905019177e-05,
+      "loss": 0.8741,
+      "step": 2132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999949000032153e-05,
+      "loss": 1.1476,
+      "step": 2133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999489498478555e-05,
+      "loss": 1.126,
+      "step": 2134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948899638881e-05,
+      "loss": 0.8158,
+      "step": 2135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948849405228e-05,
+      "loss": 0.7589,
+      "step": 2136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999487991468964e-05,
+      "loss": 0.7096,
+      "step": 2137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948748863886e-05,
+      "loss": 0.9201,
+      "step": 2138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948698556197e-05,
+      "loss": 0.8196,
+      "step": 2139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994864822383e-05,
+      "loss": 1.1322,
+      "step": 2140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948597866785e-05,
+      "loss": 1.1709,
+      "step": 2141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948547485061e-05,
+      "loss": 1.1259,
+      "step": 2142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948497078658e-05,
+      "loss": 1.9433,
+      "step": 2143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948446647578e-05,
+      "loss": 1.3849,
+      "step": 2144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948396191818e-05,
+      "loss": 1.5158,
+      "step": 2145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999483457113807e-05,
+      "loss": 1.5195,
+      "step": 2146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999482952062646e-05,
+      "loss": 1.4063,
+      "step": 2147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994824467647e-05,
+      "loss": 2.0924,
+      "step": 2148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948194121997e-05,
+      "loss": 1.1986,
+      "step": 2149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999481435428456e-05,
+      "loss": 0.603,
+      "step": 2150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999480929390164e-05,
+      "loss": 0.5798,
+      "step": 2151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948042310508e-05,
+      "loss": 0.2908,
+      "step": 2152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947991657321e-05,
+      "loss": 0.2309,
+      "step": 2153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947940979456e-05,
+      "loss": 1.0838,
+      "step": 2154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999478902769126e-05,
+      "loss": 1.4886,
+      "step": 2155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994783954969e-05,
+      "loss": 1.1981,
+      "step": 2156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999477887977895e-05,
+      "loss": 1.1871,
+      "step": 2157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947738021211e-05,
+      "loss": 1.2981,
+      "step": 2158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947687219954e-05,
+      "loss": 0.7443,
+      "step": 2159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999476363940176e-05,
+      "loss": 1.2174,
+      "step": 2160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947585543404e-05,
+      "loss": 0.9883,
+      "step": 2161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947534668111e-05,
+      "loss": 0.7254,
+      "step": 2162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994748376814e-05,
+      "loss": 1.322,
+      "step": 2163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947432843491e-05,
+      "loss": 1.4698,
+      "step": 2164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999473818941636e-05,
+      "loss": 1.3622,
+      "step": 2165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999473309201565e-05,
+      "loss": 0.9578,
+      "step": 2166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947279921472e-05,
+      "loss": 1.0192,
+      "step": 2167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999472288981083e-05,
+      "loss": 0.8398,
+      "step": 2168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947177850067e-05,
+      "loss": 0.6333,
+      "step": 2169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999471267773476e-05,
+      "loss": 0.5143,
+      "step": 2170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999947075679949e-05,
+      "loss": 0.8075,
+      "step": 2171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999470245578717e-05,
+      "loss": 0.9884,
+      "step": 2172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946973411117e-05,
+      "loss": 1.4855,
+      "step": 2173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946922239683e-05,
+      "loss": 1.0709,
+      "step": 2174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946871043571e-05,
+      "loss": 1.4966,
+      "step": 2175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946819822781e-05,
+      "loss": 1.3703,
+      "step": 2176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946768577312e-05,
+      "loss": 0.8544,
+      "step": 2177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946717307164e-05,
+      "loss": 0.8872,
+      "step": 2178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999466660123386e-05,
+      "loss": 1.1566,
+      "step": 2179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946614692834e-05,
+      "loss": 1.0904,
+      "step": 2180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999465633486514e-05,
+      "loss": 0.8584,
+      "step": 2181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946511979791e-05,
+      "loss": 1.3786,
+      "step": 2182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946460586251e-05,
+      "loss": 1.7855,
+      "step": 2183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946409168033e-05,
+      "loss": 1.0695,
+      "step": 2184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946357725137e-05,
+      "loss": 1.0416,
+      "step": 2185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999463062575616e-05,
+      "loss": 0.7917,
+      "step": 2186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999462547653086e-05,
+      "loss": 4.7444,
+      "step": 2187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946203248377e-05,
+      "loss": 6.0795,
+      "step": 2188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946151706767e-05,
+      "loss": 5.9444,
+      "step": 2189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946100140479e-05,
+      "loss": 5.8905,
+      "step": 2190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999946048549512e-05,
+      "loss": 5.7887,
+      "step": 2191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999459969338666e-05,
+      "loss": 1.2573,
+      "step": 2192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999459452935435e-05,
+      "loss": 0.9757,
+      "step": 2193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945893628541e-05,
+      "loss": 0.6567,
+      "step": 2194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999458419388604e-05,
+      "loss": 0.8967,
+      "step": 2195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945790224502e-05,
+      "loss": 1.2589,
+      "step": 2196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945738485464e-05,
+      "loss": 1.1499,
+      "step": 2197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999456867217486e-05,
+      "loss": 0.9498,
+      "step": 2198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945634933355e-05,
+      "loss": 1.3002,
+      "step": 2199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945583120282e-05,
+      "loss": 1.4875,
+      "step": 2200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945531282532e-05,
+      "loss": 1.1735,
+      "step": 2201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945479420102e-05,
+      "loss": 0.207,
+      "step": 2202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999454275329945e-05,
+      "loss": 0.1521,
+      "step": 2203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999453756212075e-05,
+      "loss": 0.1431,
+      "step": 2204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945323684744e-05,
+      "loss": 0.0981,
+      "step": 2205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945271723601e-05,
+      "loss": 0.082,
+      "step": 2206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999452197377796e-05,
+      "loss": 0.0651,
+      "step": 2207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945167727279e-05,
+      "loss": 0.0447,
+      "step": 2208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945115692101e-05,
+      "loss": 0.0372,
+      "step": 2209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999945063632245e-05,
+      "loss": 0.0263,
+      "step": 2210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994501154771e-05,
+      "loss": 0.0216,
+      "step": 2211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944959438496e-05,
+      "loss": 0.0183,
+      "step": 2212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944907304604e-05,
+      "loss": 0.0182,
+      "step": 2213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999448551460345e-05,
+      "loss": 0.0174,
+      "step": 2214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944802962786e-05,
+      "loss": 0.0185,
+      "step": 2215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944750754858e-05,
+      "loss": 0.0171,
+      "step": 2216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944698522253e-05,
+      "loss": 0.0146,
+      "step": 2217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999446462649694e-05,
+      "loss": 0.0241,
+      "step": 2218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944593983007e-05,
+      "loss": 0.0217,
+      "step": 2219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944541676367e-05,
+      "loss": 0.0212,
+      "step": 2220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999444893450476e-05,
+      "loss": 0.02,
+      "step": 2221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994443698905e-05,
+      "loss": 0.019,
+      "step": 2222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944384608375e-05,
+      "loss": 0.019,
+      "step": 2223
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994433220302e-05,
+      "loss": 0.017,
+      "step": 2224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944279772988e-05,
+      "loss": 0.0168,
+      "step": 2225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999442273182765e-05,
+      "loss": 0.0184,
+      "step": 2226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944174838887e-05,
+      "loss": 0.0185,
+      "step": 2227
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999441223348195e-05,
+      "loss": 0.0151,
+      "step": 2228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999944069806073e-05,
+      "loss": 0.0121,
+      "step": 2229
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999440172526494e-05,
+      "loss": 0.0186,
+      "step": 2230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943964674546e-05,
+      "loss": 0.0138,
+      "step": 2231
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999439120717646e-05,
+      "loss": 0.1738,
+      "step": 2232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943859444305e-05,
+      "loss": 0.8234,
+      "step": 2233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999438067921666e-05,
+      "loss": 0.9426,
+      "step": 2234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994375411535e-05,
+      "loss": 1.2646,
+      "step": 2235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943701413855e-05,
+      "loss": 1.3297,
+      "step": 2236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999436486876814e-05,
+      "loss": 1.3769,
+      "step": 2237
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999435959368303e-05,
+      "loss": 1.1826,
+      "step": 2238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999435431613e-05,
+      "loss": 1.0608,
+      "step": 2239
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999434903610914e-05,
+      "loss": 1.1587,
+      "step": 2240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943437536205e-05,
+      "loss": 1.1704,
+      "step": 2241
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943384686639e-05,
+      "loss": 1.0112,
+      "step": 2242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943331812396e-05,
+      "loss": 1.4902,
+      "step": 2243
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943278913474e-05,
+      "loss": 0.7887,
+      "step": 2244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943225989873e-05,
+      "loss": 1.4997,
+      "step": 2245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999431730415944e-05,
+      "loss": 1.2394,
+      "step": 2246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943120068638e-05,
+      "loss": 1.1772,
+      "step": 2247
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943067071002e-05,
+      "loss": 1.0988,
+      "step": 2248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999943014048688e-05,
+      "loss": 1.0802,
+      "step": 2249
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999429610016954e-05,
+      "loss": 1.0132,
+      "step": 2250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942907930025e-05,
+      "loss": 0.8749,
+      "step": 2251
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942854833676e-05,
+      "loss": 1.2088,
+      "step": 2252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942801712648e-05,
+      "loss": 1.4583,
+      "step": 2253
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942748566942e-05,
+      "loss": 1.1602,
+      "step": 2254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942695396558e-05,
+      "loss": 1.3084,
+      "step": 2255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942642201495e-05,
+      "loss": 1.084,
+      "step": 2256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942588981754e-05,
+      "loss": 1.3865,
+      "step": 2257
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942535737335e-05,
+      "loss": 1.3727,
+      "step": 2258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942482468237e-05,
+      "loss": 0.8791,
+      "step": 2259
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999424291744615e-05,
+      "loss": 1.2743,
+      "step": 2260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942375856006e-05,
+      "loss": 1.3113,
+      "step": 2261
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942322512873e-05,
+      "loss": 1.4118,
+      "step": 2262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942269145062e-05,
+      "loss": 0.969,
+      "step": 2263
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942215752572e-05,
+      "loss": 0.9726,
+      "step": 2264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942162335404e-05,
+      "loss": 1.5553,
+      "step": 2265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999421088935585e-05,
+      "loss": 1.2159,
+      "step": 2266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942055427033e-05,
+      "loss": 1.1292,
+      "step": 2267
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999420019358304e-05,
+      "loss": 0.9931,
+      "step": 2268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999419484199486e-05,
+      "loss": 1.0924,
+      "step": 2269
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941894879388e-05,
+      "loss": 1.0195,
+      "step": 2270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994184131415e-05,
+      "loss": 1.1812,
+      "step": 2271
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941787724234e-05,
+      "loss": 1.09,
+      "step": 2272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941734109639e-05,
+      "loss": 1.1697,
+      "step": 2273
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999416804703646e-05,
+      "loss": 0.9411,
+      "step": 2274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999416268064133e-05,
+      "loss": 1.2827,
+      "step": 2275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999415731177836e-05,
+      "loss": 1.2085,
+      "step": 2276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941519404475e-05,
+      "loss": 1.2183,
+      "step": 2277
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941465666488e-05,
+      "loss": 0.8818,
+      "step": 2278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941411903822e-05,
+      "loss": 0.7588,
+      "step": 2279
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941358116479e-05,
+      "loss": 1.3433,
+      "step": 2280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941304304457e-05,
+      "loss": 1.3804,
+      "step": 2281
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999412504677565e-05,
+      "loss": 1.2872,
+      "step": 2282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941196606378e-05,
+      "loss": 1.1368,
+      "step": 2283
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999411427203206e-05,
+      "loss": 1.0265,
+      "step": 2284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941088809585e-05,
+      "loss": 1.0054,
+      "step": 2285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999410348741715e-05,
+      "loss": 1.2381,
+      "step": 2286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994098091408e-05,
+      "loss": 0.956,
+      "step": 2287
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999409269293084e-05,
+      "loss": 1.1399,
+      "step": 2288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994087291986e-05,
+      "loss": 1.4057,
+      "step": 2289
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940818885732e-05,
+      "loss": 1.1755,
+      "step": 2290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940764826927e-05,
+      "loss": 1.1603,
+      "step": 2291
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940710743443e-05,
+      "loss": 1.2242,
+      "step": 2292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994065663528e-05,
+      "loss": 0.7225,
+      "step": 2293
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994060250244e-05,
+      "loss": 0.2577,
+      "step": 2294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999405483449206e-05,
+      "loss": 0.1954,
+      "step": 2295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999404941627234e-05,
+      "loss": 0.1391,
+      "step": 2296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940439955848e-05,
+      "loss": 0.1212,
+      "step": 2297
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940385724293e-05,
+      "loss": 0.0721,
+      "step": 2298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994033146806e-05,
+      "loss": 0.0758,
+      "step": 2299
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940277187149e-05,
+      "loss": 0.0615,
+      "step": 2300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994022288156e-05,
+      "loss": 0.0825,
+      "step": 2301
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940168551293e-05,
+      "loss": 0.0567,
+      "step": 2302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940114196347e-05,
+      "loss": 0.0523,
+      "step": 2303
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999940059816723e-05,
+      "loss": 0.0547,
+      "step": 2304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99994000541242e-05,
+      "loss": 0.4968,
+      "step": 2305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939950983439e-05,
+      "loss": 1.3247,
+      "step": 2306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993989652978e-05,
+      "loss": 1.3734,
+      "step": 2307
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939842051442e-05,
+      "loss": 1.1558,
+      "step": 2308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939787548426e-05,
+      "loss": 1.2981,
+      "step": 2309
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939733020732e-05,
+      "loss": 1.5171,
+      "step": 2310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939678468359e-05,
+      "loss": 1.6943,
+      "step": 2311
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939623891308e-05,
+      "loss": 1.5977,
+      "step": 2312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939569289578e-05,
+      "loss": 1.1567,
+      "step": 2313
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993951466317e-05,
+      "loss": 1.3617,
+      "step": 2314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939460012084e-05,
+      "loss": 1.2513,
+      "step": 2315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999394053363196e-05,
+      "loss": 1.3635,
+      "step": 2316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939350635877e-05,
+      "loss": 1.0933,
+      "step": 2317
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999392959107555e-05,
+      "loss": 1.3159,
+      "step": 2318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939241160956e-05,
+      "loss": 1.1969,
+      "step": 2319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939186386478e-05,
+      "loss": 1.2352,
+      "step": 2320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939131587322e-05,
+      "loss": 1.3023,
+      "step": 2321
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999939076763487e-05,
+      "loss": 1.5029,
+      "step": 2322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999390219149745e-05,
+      "loss": 1.002,
+      "step": 2323
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938967041783e-05,
+      "loss": 0.6181,
+      "step": 2324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938912143913e-05,
+      "loss": 0.8718,
+      "step": 2325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938857221366e-05,
+      "loss": 0.7674,
+      "step": 2326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999388022741386e-05,
+      "loss": 1.4088,
+      "step": 2327
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938747302234e-05,
+      "loss": 1.239,
+      "step": 2328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999386923056515e-05,
+      "loss": 0.9995,
+      "step": 2329
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993863728439e-05,
+      "loss": 1.0201,
+      "step": 2330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999385822384505e-05,
+      "loss": 1.0667,
+      "step": 2331
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938527167832e-05,
+      "loss": 1.2822,
+      "step": 2332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938472072536e-05,
+      "loss": 0.7923,
+      "step": 2333
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938416952561e-05,
+      "loss": 1.1033,
+      "step": 2334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938361807908e-05,
+      "loss": 1.1808,
+      "step": 2335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938306638576e-05,
+      "loss": 1.1299,
+      "step": 2336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938251444567e-05,
+      "loss": 1.1362,
+      "step": 2337
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938196225879e-05,
+      "loss": 1.1539,
+      "step": 2338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999381409825124e-05,
+      "loss": 0.6446,
+      "step": 2339
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999938085714468e-05,
+      "loss": 1.214,
+      "step": 2340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999380304217446e-05,
+      "loss": 1.4017,
+      "step": 2341
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937975104343e-05,
+      "loss": 1.9942,
+      "step": 2342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999379197622636e-05,
+      "loss": 1.5023,
+      "step": 2343
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937864395505e-05,
+      "loss": 1.3173,
+      "step": 2344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999378090040694e-05,
+      "loss": 1.2286,
+      "step": 2345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999377535879546e-05,
+      "loss": 1.4841,
+      "step": 2346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937698147161e-05,
+      "loss": 0.9893,
+      "step": 2347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993764268169e-05,
+      "loss": 1.1911,
+      "step": 2348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993758719154e-05,
+      "loss": 1.119,
+      "step": 2349
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937531676712e-05,
+      "loss": 1.0328,
+      "step": 2350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999374761372055e-05,
+      "loss": 1.1877,
+      "step": 2351
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937420573021e-05,
+      "loss": 1.46,
+      "step": 2352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937364984158e-05,
+      "loss": 0.8874,
+      "step": 2353
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999373093706165e-05,
+      "loss": 1.2046,
+      "step": 2354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937253732397e-05,
+      "loss": 0.9197,
+      "step": 2355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999371980694985e-05,
+      "loss": 1.5218,
+      "step": 2356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999371423819225e-05,
+      "loss": 0.7522,
+      "step": 2357
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999370866696674e-05,
+      "loss": 1.3164,
+      "step": 2358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999937030932735e-05,
+      "loss": 1.1041,
+      "step": 2359
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936975171124e-05,
+      "loss": 1.2685,
+      "step": 2360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936919384833e-05,
+      "loss": 1.3134,
+      "step": 2361
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936863573866e-05,
+      "loss": 1.016,
+      "step": 2362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936807738219e-05,
+      "loss": 0.9835,
+      "step": 2363
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936751877895e-05,
+      "loss": 0.7403,
+      "step": 2364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999366959928915e-05,
+      "loss": 0.5867,
+      "step": 2365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999366400832106e-05,
+      "loss": 0.8516,
+      "step": 2366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936584148851e-05,
+      "loss": 1.3296,
+      "step": 2367
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999365281898134e-05,
+      "loss": 1.0289,
+      "step": 2368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936472206097e-05,
+      "loss": 0.4003,
+      "step": 2369
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936416197702e-05,
+      "loss": 1.0497,
+      "step": 2370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999363601646296e-05,
+      "loss": 1.4006,
+      "step": 2371
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936304106878e-05,
+      "loss": 0.778,
+      "step": 2372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936248024448e-05,
+      "loss": 0.7985,
+      "step": 2373
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993619191734e-05,
+      "loss": 1.0902,
+      "step": 2374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999361357855544e-05,
+      "loss": 1.1798,
+      "step": 2375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999936079629091e-05,
+      "loss": 1.3368,
+      "step": 2376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999360234479474e-05,
+      "loss": 1.3867,
+      "step": 2377
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935967242126e-05,
+      "loss": 1.1832,
+      "step": 2378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935911011627e-05,
+      "loss": 1.3546,
+      "step": 2379
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935854756449e-05,
+      "loss": 1.3045,
+      "step": 2380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935798476593e-05,
+      "loss": 1.2272,
+      "step": 2381
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935742172059e-05,
+      "loss": 1.219,
+      "step": 2382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935685842846e-05,
+      "loss": 1.1794,
+      "step": 2383
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999356294889554e-05,
+      "loss": 0.8872,
+      "step": 2384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999355731103855e-05,
+      "loss": 1.0011,
+      "step": 2385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999355167071385e-05,
+      "loss": 1.196,
+      "step": 2386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999354602792124e-05,
+      "loss": 1.1468,
+      "step": 2387
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999354038266084e-05,
+      "loss": 0.9644,
+      "step": 2388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935347349326e-05,
+      "loss": 0.9631,
+      "step": 2389
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935290847365e-05,
+      "loss": 1.2356,
+      "step": 2390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935234320726e-05,
+      "loss": 1.1133,
+      "step": 2391
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935177769408e-05,
+      "loss": 1.3155,
+      "step": 2392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935121193413e-05,
+      "loss": 1.4621,
+      "step": 2393
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999935064592739e-05,
+      "loss": 0.9106,
+      "step": 2394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999350079673855e-05,
+      "loss": 1.099,
+      "step": 2395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934951317356e-05,
+      "loss": 1.2296,
+      "step": 2396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934894642646e-05,
+      "loss": 0.8653,
+      "step": 2397
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999348379432595e-05,
+      "loss": 1.285,
+      "step": 2398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934781219193e-05,
+      "loss": 1.5817,
+      "step": 2399
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999347244704494e-05,
+      "loss": 1.3223,
+      "step": 2400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934667697027e-05,
+      "loss": 1.2201,
+      "step": 2401
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934610898927e-05,
+      "loss": 1.035,
+      "step": 2402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999345540761484e-05,
+      "loss": 1.4577,
+      "step": 2403
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934497228691e-05,
+      "loss": 0.9728,
+      "step": 2404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999344403565556e-05,
+      "loss": 0.9041,
+      "step": 2405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999343834597425e-05,
+      "loss": 0.8214,
+      "step": 2406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993432653825e-05,
+      "loss": 0.936,
+      "step": 2407
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999342695920795e-05,
+      "loss": 0.9635,
+      "step": 2408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934212621231e-05,
+      "loss": 1.1556,
+      "step": 2409
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934155625704e-05,
+      "loss": 1.254,
+      "step": 2410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999340986054985e-05,
+      "loss": 1.209,
+      "step": 2411
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999934041560615e-05,
+      "loss": 1.0671,
+      "step": 2412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933984491054e-05,
+      "loss": 0.8076,
+      "step": 2413
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933927396814e-05,
+      "loss": 1.0531,
+      "step": 2414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933870277895e-05,
+      "loss": 0.9312,
+      "step": 2415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933813134299e-05,
+      "loss": 1.3744,
+      "step": 2416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999337559660244e-05,
+      "loss": 1.0359,
+      "step": 2417
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993369877307e-05,
+      "loss": 0.8227,
+      "step": 2418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933641555439e-05,
+      "loss": 1.1006,
+      "step": 2419
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999335843131286e-05,
+      "loss": 0.8073,
+      "step": 2420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933527046141e-05,
+      "loss": 1.2168,
+      "step": 2421
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999334697544745e-05,
+      "loss": 0.9359,
+      "step": 2422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99993341243813e-05,
+      "loss": 1.2773,
+      "step": 2423
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933355097107e-05,
+      "loss": 1.1156,
+      "step": 2424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999332977314063e-05,
+      "loss": 1.1494,
+      "step": 2425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933240341026e-05,
+      "loss": 0.8976,
+      "step": 2426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933182925969e-05,
+      "loss": 1.162,
+      "step": 2427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999331254862326e-05,
+      "loss": 0.8506,
+      "step": 2428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933068021818e-05,
+      "loss": 0.6927,
+      "step": 2429
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999933010532725e-05,
+      "loss": 1.1664,
+      "step": 2430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932953018954e-05,
+      "loss": 0.9944,
+      "step": 2431
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999328954805046e-05,
+      "loss": 1.021,
+      "step": 2432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932837917377e-05,
+      "loss": 0.7567,
+      "step": 2433
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932780329572e-05,
+      "loss": 1.3012,
+      "step": 2434
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932722717088e-05,
+      "loss": 1.0829,
+      "step": 2435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932665079925e-05,
+      "loss": 1.6098,
+      "step": 2436
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932607418084e-05,
+      "loss": 1.2366,
+      "step": 2437
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932549731565e-05,
+      "loss": 1.191,
+      "step": 2438
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999324920203685e-05,
+      "loss": 1.2385,
+      "step": 2439
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932434284492e-05,
+      "loss": 0.7776,
+      "step": 2440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932376523939e-05,
+      "loss": 0.5032,
+      "step": 2441
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932318738707e-05,
+      "loss": 0.4156,
+      "step": 2442
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999322609287964e-05,
+      "loss": 0.4252,
+      "step": 2443
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999322030942074e-05,
+      "loss": 0.3237,
+      "step": 2444
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999321452349406e-05,
+      "loss": 0.452,
+      "step": 2445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932087350996e-05,
+      "loss": 0.4207,
+      "step": 2446
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999932029442372e-05,
+      "loss": 0.2601,
+      "step": 2447
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999319715090706e-05,
+      "loss": 0.4093,
+      "step": 2448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999319135510905e-05,
+      "loss": 1.6132,
+      "step": 2449
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931855568432e-05,
+      "loss": 1.3601,
+      "step": 2450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931797561096e-05,
+      "loss": 1.1905,
+      "step": 2451
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931739529081e-05,
+      "loss": 1.187,
+      "step": 2452
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999316814723876e-05,
+      "loss": 1.2807,
+      "step": 2453
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931623391016e-05,
+      "loss": 1.5622,
+      "step": 2454
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999315652849663e-05,
+      "loss": 1.0686,
+      "step": 2455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931507154238e-05,
+      "loss": 1.002,
+      "step": 2456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999314489988325e-05,
+      "loss": 1.3571,
+      "step": 2457
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999313908187476e-05,
+      "loss": 1.0784,
+      "step": 2458
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999313326139855e-05,
+      "loss": 1.0493,
+      "step": 2459
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999312743845436e-05,
+      "loss": 1.3248,
+      "step": 2460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999312161304245e-05,
+      "loss": 1.2138,
+      "step": 2461
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931157851627e-05,
+      "loss": 1.3235,
+      "step": 2462
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931099548151e-05,
+      "loss": 1.1581,
+      "step": 2463
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999931041219997e-05,
+      "loss": 1.0672,
+      "step": 2464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930982867165e-05,
+      "loss": 1.2655,
+      "step": 2465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930924489654e-05,
+      "loss": 1.191,
+      "step": 2466
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930866087466e-05,
+      "loss": 1.1727,
+      "step": 2467
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930807660598e-05,
+      "loss": 1.3226,
+      "step": 2468
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930749209053e-05,
+      "loss": 1.5273,
+      "step": 2469
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999306907328294e-05,
+      "loss": 1.1624,
+      "step": 2470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930632231927e-05,
+      "loss": 0.857,
+      "step": 2471
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999305737063474e-05,
+      "loss": 1.1461,
+      "step": 2472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930515156089e-05,
+      "loss": 1.1581,
+      "step": 2473
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930456581152e-05,
+      "loss": 1.0932,
+      "step": 2474
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930397981537e-05,
+      "loss": 1.0128,
+      "step": 2475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930339357244e-05,
+      "loss": 1.2064,
+      "step": 2476
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999302807082726e-05,
+      "loss": 1.0275,
+      "step": 2477
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930222034622e-05,
+      "loss": 0.8178,
+      "step": 2478
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930163336295e-05,
+      "loss": 1.0451,
+      "step": 2479
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930104613289e-05,
+      "loss": 1.4416,
+      "step": 2480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999930045865604e-05,
+      "loss": 0.8916,
+      "step": 2481
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929987093241e-05,
+      "loss": 1.195,
+      "step": 2482
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999299282962e-05,
+      "loss": 1.3613,
+      "step": 2483
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929869474481e-05,
+      "loss": 1.3498,
+      "step": 2484
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929810628083e-05,
+      "loss": 1.0009,
+      "step": 2485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999297517570076e-05,
+      "loss": 1.3047,
+      "step": 2486
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929692861253e-05,
+      "loss": 0.9789,
+      "step": 2487
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999296339408206e-05,
+      "loss": 1.2014,
+      "step": 2488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999295749957104e-05,
+      "loss": 1.2294,
+      "step": 2489
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929516025922e-05,
+      "loss": 1.0622,
+      "step": 2490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999294570314545e-05,
+      "loss": 1.1546,
+      "step": 2491
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929398012309e-05,
+      "loss": 1.2043,
+      "step": 2492
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999293389684855e-05,
+      "loss": 1.0917,
+      "step": 2493
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929279899984e-05,
+      "loss": 1.0515,
+      "step": 2494
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929220806804e-05,
+      "loss": 1.1921,
+      "step": 2495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929161688946e-05,
+      "loss": 1.1185,
+      "step": 2496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999929102546409e-05,
+      "loss": 1.4151,
+      "step": 2497
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999290433791946e-05,
+      "loss": 1.1235,
+      "step": 2498
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928984187301e-05,
+      "loss": 0.8122,
+      "step": 2499
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999289249707296e-05,
+      "loss": 1.1508,
+      "step": 2500
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.0854716300964355,
+      "eval_runtime": 83.847,
+      "eval_samples_per_second": 16.518,
+      "eval_steps_per_second": 4.138,
+      "step": 2500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999288657294804e-05,
+      "loss": 1.0851,
+      "step": 2501
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928806463553e-05,
+      "loss": 0.8544,
+      "step": 2502
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928747172946e-05,
+      "loss": 0.7983,
+      "step": 2503
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999286878576626e-05,
+      "loss": 1.1961,
+      "step": 2504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999286285176995e-05,
+      "loss": 1.0891,
+      "step": 2505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999285691530586e-05,
+      "loss": 1.0302,
+      "step": 2506
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99992850976374e-05,
+      "loss": 1.1088,
+      "step": 2507
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928450349743e-05,
+      "loss": 1.0297,
+      "step": 2508
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928390911067e-05,
+      "loss": 1.2901,
+      "step": 2509
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928331447714e-05,
+      "loss": 1.0585,
+      "step": 2510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928271959682e-05,
+      "loss": 1.1662,
+      "step": 2511
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999282124469714e-05,
+      "loss": 0.6689,
+      "step": 2512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928152909583e-05,
+      "loss": 0.6319,
+      "step": 2513
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999928093347516e-05,
+      "loss": 0.9722,
+      "step": 2514
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999280337607715e-05,
+      "loss": 0.8079,
+      "step": 2515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927974149348e-05,
+      "loss": 1.1531,
+      "step": 2516
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927914513247e-05,
+      "loss": 0.886,
+      "step": 2517
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927854852467e-05,
+      "loss": 0.9439,
+      "step": 2518
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999277951670096e-05,
+      "loss": 1.2437,
+      "step": 2519
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999277354568735e-05,
+      "loss": 1.2504,
+      "step": 2520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927675722059e-05,
+      "loss": 1.299,
+      "step": 2521
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999276159625665e-05,
+      "loss": 0.8641,
+      "step": 2522
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999275561783956e-05,
+      "loss": 0.6387,
+      "step": 2523
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927496369547e-05,
+      "loss": 1.1639,
+      "step": 2524
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927436536019e-05,
+      "loss": 1.0768,
+      "step": 2525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927376677814e-05,
+      "loss": 1.2307,
+      "step": 2526
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99992731679493e-05,
+      "loss": 1.1291,
+      "step": 2527
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999927256887368e-05,
+      "loss": 1.1474,
+      "step": 2528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999271969551284e-05,
+      "loss": 1.1732,
+      "step": 2529
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99992713699821e-05,
+      "loss": 1.2119,
+      "step": 2530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999270770166136e-05,
+      "loss": 1.0065,
+      "step": 2531
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999270170103385e-05,
+      "loss": 1.0581,
+      "step": 2532
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999269569793855e-05,
+      "loss": 1.0729,
+      "step": 2533
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926896923755e-05,
+      "loss": 1.1211,
+      "step": 2534
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926836843445e-05,
+      "loss": 1.17,
+      "step": 2535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926776738457e-05,
+      "loss": 1.1372,
+      "step": 2536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926716608791e-05,
+      "loss": 1.7751,
+      "step": 2537
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999266564544465e-05,
+      "loss": 1.0604,
+      "step": 2538
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926596275424e-05,
+      "loss": 1.3684,
+      "step": 2539
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926536071724e-05,
+      "loss": 1.2431,
+      "step": 2540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926475843345e-05,
+      "loss": 1.0332,
+      "step": 2541
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926415590288e-05,
+      "loss": 1.3016,
+      "step": 2542
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999263553125525e-05,
+      "loss": 1.0958,
+      "step": 2543
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926295010139e-05,
+      "loss": 1.142,
+      "step": 2544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926234683047e-05,
+      "loss": 1.189,
+      "step": 2545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999261743312774e-05,
+      "loss": 1.0052,
+      "step": 2546
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926113954829e-05,
+      "loss": 1.0415,
+      "step": 2547
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999260535537026e-05,
+      "loss": 1.2109,
+      "step": 2548
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925993127899e-05,
+      "loss": 1.2191,
+      "step": 2549
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925932677415e-05,
+      "loss": 1.1117,
+      "step": 2550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999258722022545e-05,
+      "loss": 1.0124,
+      "step": 2551
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999258117024154e-05,
+      "loss": 1.0574,
+      "step": 2552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925751177898e-05,
+      "loss": 1.0316,
+      "step": 2553
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999256906287016e-05,
+      "loss": 0.5798,
+      "step": 2554
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925630054828e-05,
+      "loss": 0.5607,
+      "step": 2555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999255694562766e-05,
+      "loss": 0.5936,
+      "step": 2556
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925508833046e-05,
+      "loss": 1.1024,
+      "step": 2557
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925448185138e-05,
+      "loss": 1.2206,
+      "step": 2558
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999253875125505e-05,
+      "loss": 0.9331,
+      "step": 2559
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999253268152856e-05,
+      "loss": 0.9955,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925266093343e-05,
+      "loss": 0.8683,
+      "step": 2561
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925205346721e-05,
+      "loss": 0.6031,
+      "step": 2562
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925144575421e-05,
+      "loss": 0.9559,
+      "step": 2563
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925083779444e-05,
+      "loss": 0.6049,
+      "step": 2564
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999925022958788e-05,
+      "loss": 1.2506,
+      "step": 2565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999249621134534e-05,
+      "loss": 1.506,
+      "step": 2566
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924901243441e-05,
+      "loss": 1.3293,
+      "step": 2567
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924840348751e-05,
+      "loss": 1.9336,
+      "step": 2568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924779429382e-05,
+      "loss": 0.9746,
+      "step": 2569
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924718485335e-05,
+      "loss": 0.761,
+      "step": 2570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999246575166095e-05,
+      "loss": 0.308,
+      "step": 2571
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924596523206e-05,
+      "loss": 1.4471,
+      "step": 2572
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999245355051245e-05,
+      "loss": 1.4313,
+      "step": 2573
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924474462365e-05,
+      "loss": 0.8333,
+      "step": 2574
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924413394926e-05,
+      "loss": 0.7514,
+      "step": 2575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999243523028105e-05,
+      "loss": 1.001,
+      "step": 2576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924291186016e-05,
+      "loss": 1.1125,
+      "step": 2577
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999242300445435e-05,
+      "loss": 1.2343,
+      "step": 2578
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924168878393e-05,
+      "loss": 1.1152,
+      "step": 2579
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999924107687564e-05,
+      "loss": 0.926,
+      "step": 2580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999240464720564e-05,
+      "loss": 1.0037,
+      "step": 2581
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923985231871e-05,
+      "loss": 1.2988,
+      "step": 2582
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999239239670074e-05,
+      "loss": 0.8706,
+      "step": 2583
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923862677466e-05,
+      "loss": 0.9814,
+      "step": 2584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923801363246e-05,
+      "loss": 0.9953,
+      "step": 2585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923740024347e-05,
+      "loss": 0.7982,
+      "step": 2586
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923678660771e-05,
+      "loss": 0.2924,
+      "step": 2587
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923617272516e-05,
+      "loss": 0.2321,
+      "step": 2588
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999235558595836e-05,
+      "loss": 0.2196,
+      "step": 2589
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923494421973e-05,
+      "loss": 0.6108,
+      "step": 2590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923432959683e-05,
+      "loss": 1.3284,
+      "step": 2591
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999233714727164e-05,
+      "loss": 0.884,
+      "step": 2592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999233099610706e-05,
+      "loss": 0.7834,
+      "step": 2593
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923248424747e-05,
+      "loss": 0.7138,
+      "step": 2594
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923186863745e-05,
+      "loss": 1.0278,
+      "step": 2595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923125278065e-05,
+      "loss": 1.1555,
+      "step": 2596
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999923063667706e-05,
+      "loss": 1.1282,
+      "step": 2597
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999230020326706e-05,
+      "loss": 1.1194,
+      "step": 2598
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999229403729554e-05,
+      "loss": 1.0122,
+      "step": 2599
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922878688563e-05,
+      "loss": 1.1518,
+      "step": 2600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922816979492e-05,
+      "loss": 1.2748,
+      "step": 2601
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922755245743e-05,
+      "loss": 0.9259,
+      "step": 2602
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922693487315e-05,
+      "loss": 0.8864,
+      "step": 2603
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922631704209e-05,
+      "loss": 1.0629,
+      "step": 2604
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922569896426e-05,
+      "loss": 1.0549,
+      "step": 2605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922508063964e-05,
+      "loss": 1.0743,
+      "step": 2606
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999224462068236e-05,
+      "loss": 1.0465,
+      "step": 2607
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999223843250054e-05,
+      "loss": 1.3165,
+      "step": 2608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922322418509e-05,
+      "loss": 1.1523,
+      "step": 2609
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999222604873344e-05,
+      "loss": 1.0081,
+      "step": 2610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999221985314814e-05,
+      "loss": 0.9723,
+      "step": 2611
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99992213655095e-05,
+      "loss": 1.3478,
+      "step": 2612
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922074545741e-05,
+      "loss": 1.2355,
+      "step": 2613
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999922012515854e-05,
+      "loss": 1.1177,
+      "step": 2614
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999219504612884e-05,
+      "loss": 0.9973,
+      "step": 2615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999218883820445e-05,
+      "loss": 1.0677,
+      "step": 2616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921826278122e-05,
+      "loss": 0.9187,
+      "step": 2617
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921764149522e-05,
+      "loss": 1.0991,
+      "step": 2618
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921701996244e-05,
+      "loss": 1.2343,
+      "step": 2619
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921639818288e-05,
+      "loss": 1.2575,
+      "step": 2620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921577615653e-05,
+      "loss": 1.0961,
+      "step": 2621
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999215153883405e-05,
+      "loss": 1.0279,
+      "step": 2622
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921453136349e-05,
+      "loss": 1.2034,
+      "step": 2623
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99992139085968e-05,
+      "loss": 4.8598,
+      "step": 2624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921328558333e-05,
+      "loss": 3.16,
+      "step": 2625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999212662323075e-05,
+      "loss": 1.0854,
+      "step": 2626
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921203881604e-05,
+      "loss": 1.1011,
+      "step": 2627
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921141506222e-05,
+      "loss": 1.0632,
+      "step": 2628
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999921079106162e-05,
+      "loss": 1.1298,
+      "step": 2629
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999210166814236e-05,
+      "loss": 1.1682,
+      "step": 2630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920954232008e-05,
+      "loss": 1.1885,
+      "step": 2631
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920891757913e-05,
+      "loss": 1.7747,
+      "step": 2632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920829259141e-05,
+      "loss": 1.326,
+      "step": 2633
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999207667356896e-05,
+      "loss": 1.2759,
+      "step": 2634
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920704187561e-05,
+      "loss": 0.9534,
+      "step": 2635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999206416147535e-05,
+      "loss": 1.0857,
+      "step": 2636
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920579017269e-05,
+      "loss": 0.8371,
+      "step": 2637
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920516395105e-05,
+      "loss": 1.1331,
+      "step": 2638
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999204537482634e-05,
+      "loss": 0.9473,
+      "step": 2639
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999203910767435e-05,
+      "loss": 1.1575,
+      "step": 2640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920328380546e-05,
+      "loss": 1.06,
+      "step": 2641
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99992026565967e-05,
+      "loss": 0.4564,
+      "step": 2642
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920202914116e-05,
+      "loss": 1.0863,
+      "step": 2643
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920140143883e-05,
+      "loss": 1.8009,
+      "step": 2644
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999200773489726e-05,
+      "loss": 1.1431,
+      "step": 2645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999920014529383e-05,
+      "loss": 0.7587,
+      "step": 2646
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919951685117e-05,
+      "loss": 0.8565,
+      "step": 2647
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919888816171e-05,
+      "loss": 1.1586,
+      "step": 2648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919825922549e-05,
+      "loss": 0.9109,
+      "step": 2649
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919763004247e-05,
+      "loss": 1.2484,
+      "step": 2650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999197000612675e-05,
+      "loss": 1.5153,
+      "step": 2651
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919637093609e-05,
+      "loss": 1.4671,
+      "step": 2652
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919574101274e-05,
+      "loss": 1.179,
+      "step": 2653
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991951108426e-05,
+      "loss": 1.284,
+      "step": 2654
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919448042568e-05,
+      "loss": 0.8949,
+      "step": 2655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919384976198e-05,
+      "loss": 1.2341,
+      "step": 2656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999193218851496e-05,
+      "loss": 1.2149,
+      "step": 2657
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999192587694226e-05,
+      "loss": 0.8877,
+      "step": 2658
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919195629018e-05,
+      "loss": 0.9975,
+      "step": 2659
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919132463935e-05,
+      "loss": 1.095,
+      "step": 2660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999190692741735e-05,
+      "loss": 0.7553,
+      "step": 2661
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999919006059735e-05,
+      "loss": 0.8629,
+      "step": 2662
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999189428206174e-05,
+      "loss": 0.904,
+      "step": 2663
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999188795568216e-05,
+      "loss": 1.3531,
+      "step": 2664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918816268349e-05,
+      "loss": 1.2525,
+      "step": 2665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918752955197e-05,
+      "loss": 0.8583,
+      "step": 2666
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918689617367e-05,
+      "loss": 1.2082,
+      "step": 2667
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918626254859e-05,
+      "loss": 1.3743,
+      "step": 2668
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918562867672e-05,
+      "loss": 1.4376,
+      "step": 2669
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918499455808e-05,
+      "loss": 1.577,
+      "step": 2670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918436019266e-05,
+      "loss": 1.177,
+      "step": 2671
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918372558046e-05,
+      "loss": 1.1328,
+      "step": 2672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918309072147e-05,
+      "loss": 1.1072,
+      "step": 2673
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918245561569e-05,
+      "loss": 1.086,
+      "step": 2674
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999181820263144e-05,
+      "loss": 1.1498,
+      "step": 2675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999918118466381e-05,
+      "loss": 1.0412,
+      "step": 2676
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999180548817704e-05,
+      "loss": 1.4574,
+      "step": 2677
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991799127248e-05,
+      "loss": 0.992,
+      "step": 2678
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917927638513e-05,
+      "loss": 0.3997,
+      "step": 2679
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917863979867e-05,
+      "loss": 0.8342,
+      "step": 2680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999178002965436e-05,
+      "loss": 1.1407,
+      "step": 2681
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917736588542e-05,
+      "loss": 1.3863,
+      "step": 2682
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917672855861e-05,
+      "loss": 1.0188,
+      "step": 2683
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999176090985035e-05,
+      "loss": 1.1638,
+      "step": 2684
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917545316467e-05,
+      "loss": 1.0254,
+      "step": 2685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999174815097524e-05,
+      "loss": 1.1862,
+      "step": 2686
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917417678359e-05,
+      "loss": 1.2064,
+      "step": 2687
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917353822289e-05,
+      "loss": 0.6489,
+      "step": 2688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991728994154e-05,
+      "loss": 0.8799,
+      "step": 2689
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917226036113e-05,
+      "loss": 1.1395,
+      "step": 2690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999171621060076e-05,
+      "loss": 1.3707,
+      "step": 2691
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999917098151225e-05,
+      "loss": 1.0209,
+      "step": 2692
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999170341717626e-05,
+      "loss": 1.2067,
+      "step": 2693
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916970167623e-05,
+      "loss": 1.3174,
+      "step": 2694
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916906138806e-05,
+      "loss": 1.1166,
+      "step": 2695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991684208531e-05,
+      "loss": 1.2999,
+      "step": 2696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916778007136e-05,
+      "loss": 1.196,
+      "step": 2697
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999167139042834e-05,
+      "loss": 1.3003,
+      "step": 2698
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916649776754e-05,
+      "loss": 1.1829,
+      "step": 2699
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916585624546e-05,
+      "loss": 1.0895,
+      "step": 2700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916521447659e-05,
+      "loss": 0.8053,
+      "step": 2701
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999164572460945e-05,
+      "loss": 1.4009,
+      "step": 2702
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916393019852e-05,
+      "loss": 1.1801,
+      "step": 2703
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916328768931e-05,
+      "loss": 1.2571,
+      "step": 2704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999162644933315e-05,
+      "loss": 1.1818,
+      "step": 2705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916200193055e-05,
+      "loss": 1.4611,
+      "step": 2706
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999161358680995e-05,
+      "loss": 1.211,
+      "step": 2707
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999916071518467e-05,
+      "loss": 1.0912,
+      "step": 2708
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999160071441555e-05,
+      "loss": 1.0762,
+      "step": 2709
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999159427451655e-05,
+      "loss": 1.2525,
+      "step": 2710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915878321498e-05,
+      "loss": 1.1884,
+      "step": 2711
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915813873152e-05,
+      "loss": 1.2574,
+      "step": 2712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915749400128e-05,
+      "loss": 1.4895,
+      "step": 2713
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915684902427e-05,
+      "loss": 0.8249,
+      "step": 2714
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999156203800464e-05,
+      "loss": 0.9618,
+      "step": 2715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915555832988e-05,
+      "loss": 1.293,
+      "step": 2716
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999154912612516e-05,
+      "loss": 1.3095,
+      "step": 2717
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915426664837e-05,
+      "loss": 1.0918,
+      "step": 2718
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915362043744e-05,
+      "loss": 1.1069,
+      "step": 2719
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915297397974e-05,
+      "loss": 1.0256,
+      "step": 2720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915232727526e-05,
+      "loss": 1.2754,
+      "step": 2721
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999915168032399e-05,
+      "loss": 1.1649,
+      "step": 2722
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999151033125934e-05,
+      "loss": 1.2403,
+      "step": 2723
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991503856811e-05,
+      "loss": 0.715,
+      "step": 2724
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914973798949e-05,
+      "loss": 1.0529,
+      "step": 2725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991490900511e-05,
+      "loss": 1.1878,
+      "step": 2726
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914844186592e-05,
+      "loss": 0.9524,
+      "step": 2727
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999147793433965e-05,
+      "loss": 0.9358,
+      "step": 2728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914714475523e-05,
+      "loss": 0.5873,
+      "step": 2729
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999146495829716e-05,
+      "loss": 0.9812,
+      "step": 2730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914584665741e-05,
+      "loss": 1.1639,
+      "step": 2731
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914519723834e-05,
+      "loss": 1.1829,
+      "step": 2732
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914454757247e-05,
+      "loss": 1.0936,
+      "step": 2733
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914389765984e-05,
+      "loss": 1.0867,
+      "step": 2734
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914324750041e-05,
+      "loss": 0.9176,
+      "step": 2735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914259709421e-05,
+      "loss": 1.2482,
+      "step": 2736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914194644123e-05,
+      "loss": 1.1445,
+      "step": 2737
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999914129554146e-05,
+      "loss": 1.02,
+      "step": 2738
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999140644394914e-05,
+      "loss": 1.0444,
+      "step": 2739
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913999300159e-05,
+      "loss": 2.032,
+      "step": 2740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913934136148e-05,
+      "loss": 1.6777,
+      "step": 2741
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913868947459e-05,
+      "loss": 1.1249,
+      "step": 2742
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913803734092e-05,
+      "loss": 1.2135,
+      "step": 2743
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913738496047e-05,
+      "loss": 1.1091,
+      "step": 2744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913673233323e-05,
+      "loss": 0.9304,
+      "step": 2745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913607945922e-05,
+      "loss": 1.0457,
+      "step": 2746
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913542633843e-05,
+      "loss": 1.1601,
+      "step": 2747
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913477297085e-05,
+      "loss": 1.213,
+      "step": 2748
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991341193565e-05,
+      "loss": 1.1778,
+      "step": 2749
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913346549536e-05,
+      "loss": 1.1547,
+      "step": 2750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913281138744e-05,
+      "loss": 1.2153,
+      "step": 2751
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913215703275e-05,
+      "loss": 1.0718,
+      "step": 2752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999131502431266e-05,
+      "loss": 1.0972,
+      "step": 2753
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999130847583006e-05,
+      "loss": 1.3415,
+      "step": 2754
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999913019248796e-05,
+      "loss": 1.0715,
+      "step": 2755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912953714614e-05,
+      "loss": 0.9399,
+      "step": 2756
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912888155753e-05,
+      "loss": 1.0309,
+      "step": 2757
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999128225722155e-05,
+      "loss": 1.2394,
+      "step": 2758
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999127569639985e-05,
+      "loss": 1.0969,
+      "step": 2759
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999126913311044e-05,
+      "loss": 1.1535,
+      "step": 2760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912625673532e-05,
+      "loss": 1.1519,
+      "step": 2761
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912559991281e-05,
+      "loss": 1.7157,
+      "step": 2762
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912494284352e-05,
+      "loss": 1.1947,
+      "step": 2763
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912428552745e-05,
+      "loss": 0.9904,
+      "step": 2764
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991236279646e-05,
+      "loss": 0.857,
+      "step": 2765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912297015497e-05,
+      "loss": 1.1261,
+      "step": 2766
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912231209855e-05,
+      "loss": 1.2156,
+      "step": 2767
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912165379537e-05,
+      "loss": 1.2385,
+      "step": 2768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991209952454e-05,
+      "loss": 1.002,
+      "step": 2769
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999912033644864e-05,
+      "loss": 1.1895,
+      "step": 2770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999119677405105e-05,
+      "loss": 1.2197,
+      "step": 2771
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911901811479e-05,
+      "loss": 1.1403,
+      "step": 2772
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911835857769e-05,
+      "loss": 1.2452,
+      "step": 2773
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911769879382e-05,
+      "loss": 0.8898,
+      "step": 2774
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999117038763155e-05,
+      "loss": 0.8207,
+      "step": 2775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911637848572e-05,
+      "loss": 0.6392,
+      "step": 2776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991157179615e-05,
+      "loss": 0.7881,
+      "step": 2777
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991150571905e-05,
+      "loss": 1.3896,
+      "step": 2778
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911439617272e-05,
+      "loss": 1.2835,
+      "step": 2779
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999113734908156e-05,
+      "loss": 1.1221,
+      "step": 2780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911307339682e-05,
+      "loss": 0.4653,
+      "step": 2781
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911241163869e-05,
+      "loss": 0.9905,
+      "step": 2782
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999111749633785e-05,
+      "loss": 1.1892,
+      "step": 2783
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999111087382105e-05,
+      "loss": 1.2269,
+      "step": 2784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999911042488365e-05,
+      "loss": 1.7996,
+      "step": 2785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991097621384e-05,
+      "loss": 1.5628,
+      "step": 2786
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999910909914637e-05,
+      "loss": 0.7233,
+      "step": 2787
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999910843590756e-05,
+      "loss": 0.6555,
+      "step": 2788
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999910777242197e-05,
+      "loss": 1.121,
+      "step": 2789
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99991071086896e-05,
+      "loss": 1.4148,
+      "step": 2790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999106444710454e-05,
+      "loss": 1.2223,
+      "step": 2791
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999105780484524e-05,
+      "loss": 1.1463,
+      "step": 2792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999105116011815e-05,
+      "loss": 1.0044,
+      "step": 2793
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999910445129232e-05,
+      "loss": 1.2547,
+      "step": 2794
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999910378632605e-05,
+      "loss": 1.0682,
+      "step": 2795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999103121113e-05,
+      "loss": 1.0591,
+      "step": 2796
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999910245565317e-05,
+      "loss": 1.3869,
+      "step": 2797
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999101789946555e-05,
+      "loss": 1.2375,
+      "step": 2798
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999101123993165e-05,
+      "loss": 1.1,
+      "step": 2799
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999910045779299e-05,
+      "loss": 0.9864,
+      "step": 2800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909979134603e-05,
+      "loss": 1.2769,
+      "step": 2801
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999099124652294e-05,
+      "loss": 1.2261,
+      "step": 2802
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909845771178e-05,
+      "loss": 1.41,
+      "step": 2803
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999097790524485e-05,
+      "loss": 1.2601,
+      "step": 2804
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999097123090413e-05,
+      "loss": 1.0879,
+      "step": 2805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909645540955e-05,
+      "loss": 1.2381,
+      "step": 2806
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909578748191e-05,
+      "loss": 1.0534,
+      "step": 2807
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909511930749e-05,
+      "loss": 1.2444,
+      "step": 2808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999094450886294e-05,
+      "loss": 1.0326,
+      "step": 2809
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999093782218306e-05,
+      "loss": 1.0756,
+      "step": 2810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909311330355e-05,
+      "loss": 0.9788,
+      "step": 2811
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909244414201e-05,
+      "loss": 1.0639,
+      "step": 2812
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909177473369e-05,
+      "loss": 1.063,
+      "step": 2813
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999909110507859e-05,
+      "loss": 1.05,
+      "step": 2814
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990904351767e-05,
+      "loss": 0.9695,
+      "step": 2815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908976502804e-05,
+      "loss": 0.9567,
+      "step": 2816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990890946326e-05,
+      "loss": 1.1835,
+      "step": 2817
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908842399037e-05,
+      "loss": 1.1377,
+      "step": 2818
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908775310137e-05,
+      "loss": 1.3342,
+      "step": 2819
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908708196559e-05,
+      "loss": 0.7403,
+      "step": 2820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908641058302e-05,
+      "loss": 1.149,
+      "step": 2821
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908573895367e-05,
+      "loss": 1.3792,
+      "step": 2822
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908506707754e-05,
+      "loss": 1.1136,
+      "step": 2823
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908439495464e-05,
+      "loss": 1.1038,
+      "step": 2824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908372258495e-05,
+      "loss": 1.0534,
+      "step": 2825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908304996849e-05,
+      "loss": 1.4179,
+      "step": 2826
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908237710523e-05,
+      "loss": 0.6761,
+      "step": 2827
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908170399521e-05,
+      "loss": 1.2538,
+      "step": 2828
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990810306384e-05,
+      "loss": 0.8459,
+      "step": 2829
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999908035703481e-05,
+      "loss": 1.1891,
+      "step": 2830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907968318444e-05,
+      "loss": 1.0343,
+      "step": 2831
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907900908729e-05,
+      "loss": 0.8952,
+      "step": 2832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907833474336e-05,
+      "loss": 1.2053,
+      "step": 2833
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907766015265e-05,
+      "loss": 0.9051,
+      "step": 2834
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907698531516e-05,
+      "loss": 1.2077,
+      "step": 2835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907631023089e-05,
+      "loss": 1.2348,
+      "step": 2836
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907563489984e-05,
+      "loss": 0.9451,
+      "step": 2837
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999074959322e-05,
+      "loss": 1.019,
+      "step": 2838
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990742834974e-05,
+      "loss": 1.1164,
+      "step": 2839
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999073607425995e-05,
+      "loss": 0.9338,
+      "step": 2840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907293110783e-05,
+      "loss": 1.0339,
+      "step": 2841
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907225454287e-05,
+      "loss": 0.8994,
+      "step": 2842
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999071577731135e-05,
+      "loss": 0.8476,
+      "step": 2843
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907090067262e-05,
+      "loss": 0.8179,
+      "step": 2844
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907022336733e-05,
+      "loss": 0.8631,
+      "step": 2845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906954581526e-05,
+      "loss": 1.1398,
+      "step": 2846
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990688680164e-05,
+      "loss": 0.8859,
+      "step": 2847
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906818997077e-05,
+      "loss": 1.5999,
+      "step": 2848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906751167835e-05,
+      "loss": 1.3751,
+      "step": 2849
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999066833139155e-05,
+      "loss": 1.2111,
+      "step": 2850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906615435318e-05,
+      "loss": 1.1603,
+      "step": 2851
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999065475320424e-05,
+      "loss": 1.0992,
+      "step": 2852
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906479604088e-05,
+      "loss": 1.5599,
+      "step": 2853
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999064116514574e-05,
+      "loss": 1.2482,
+      "step": 2854
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906343674147e-05,
+      "loss": 0.7927,
+      "step": 2855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906275672159e-05,
+      "loss": 1.3003,
+      "step": 2856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906207645494e-05,
+      "loss": 0.9895,
+      "step": 2857
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990613959415e-05,
+      "loss": 1.076,
+      "step": 2858
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999060715181276e-05,
+      "loss": 1.2395,
+      "step": 2859
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906003417428e-05,
+      "loss": 1.2853,
+      "step": 2860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905935292051e-05,
+      "loss": 0.9141,
+      "step": 2861
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905867141995e-05,
+      "loss": 1.4543,
+      "step": 2862
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905798967261e-05,
+      "loss": 0.7374,
+      "step": 2863
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905730767849e-05,
+      "loss": 1.238,
+      "step": 2864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905662543759e-05,
+      "loss": 1.0451,
+      "step": 2865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905594294991e-05,
+      "loss": 1.1267,
+      "step": 2866
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905526021546e-05,
+      "loss": 1.1472,
+      "step": 2867
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905457723422e-05,
+      "loss": 1.1766,
+      "step": 2868
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990538940062e-05,
+      "loss": 1.3368,
+      "step": 2869
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990532105314e-05,
+      "loss": 1.1428,
+      "step": 2870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905252680982e-05,
+      "loss": 1.3337,
+      "step": 2871
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905184284146e-05,
+      "loss": 1.0391,
+      "step": 2872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999905115862632e-05,
+      "loss": 1.5314,
+      "step": 2873
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990504741644e-05,
+      "loss": 1.2044,
+      "step": 2874
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990497894557e-05,
+      "loss": 1.0157,
+      "step": 2875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904910450022e-05,
+      "loss": 1.0111,
+      "step": 2876
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999048419297956e-05,
+      "loss": 0.7298,
+      "step": 2877
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904773384892e-05,
+      "loss": 0.9956,
+      "step": 2878
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990470481531e-05,
+      "loss": 0.5304,
+      "step": 2879
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990463622105e-05,
+      "loss": 0.5684,
+      "step": 2880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904567602112e-05,
+      "loss": 1.1679,
+      "step": 2881
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999044989584954e-05,
+      "loss": 0.807,
+      "step": 2882
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904430290202e-05,
+      "loss": 0.5727,
+      "step": 2883
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990436159723e-05,
+      "loss": 0.6624,
+      "step": 2884
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904292879579e-05,
+      "loss": 1.0362,
+      "step": 2885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904224137251e-05,
+      "loss": 1.2741,
+      "step": 2886
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999041553702455e-05,
+      "loss": 1.2492,
+      "step": 2887
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904086578561e-05,
+      "loss": 1.405,
+      "step": 2888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999904017762199e-05,
+      "loss": 0.9898,
+      "step": 2889
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903948921159e-05,
+      "loss": 1.2159,
+      "step": 2890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903880055441e-05,
+      "loss": 0.9097,
+      "step": 2891
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903811165045e-05,
+      "loss": 1.3631,
+      "step": 2892
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999037422499703e-05,
+      "loss": 6.0654,
+      "step": 2893
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903673310219e-05,
+      "loss": 5.9507,
+      "step": 2894
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903604345789e-05,
+      "loss": 5.8381,
+      "step": 2895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903535356681e-05,
+      "loss": 5.6796,
+      "step": 2896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903466342895e-05,
+      "loss": 5.9238,
+      "step": 2897
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903397304431e-05,
+      "loss": 4.1682,
+      "step": 2898
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903328241289e-05,
+      "loss": 1.2678,
+      "step": 2899
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903259153469e-05,
+      "loss": 1.1465,
+      "step": 2900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903190040971e-05,
+      "loss": 1.0097,
+      "step": 2901
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903120903795e-05,
+      "loss": 1.0424,
+      "step": 2902
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999903051741941e-05,
+      "loss": 1.1693,
+      "step": 2903
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902982555409e-05,
+      "loss": 0.7356,
+      "step": 2904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902913344199e-05,
+      "loss": 1.0111,
+      "step": 2905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902844108311e-05,
+      "loss": 1.2294,
+      "step": 2906
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902774847745e-05,
+      "loss": 0.8252,
+      "step": 2907
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999027055625014e-05,
+      "loss": 0.9026,
+      "step": 2908
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902636252579e-05,
+      "loss": 1.0901,
+      "step": 2909
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902566917979e-05,
+      "loss": 1.0077,
+      "step": 2910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999024975587015e-05,
+      "loss": 1.1404,
+      "step": 2911
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902428174745e-05,
+      "loss": 0.9697,
+      "step": 2912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902358766112e-05,
+      "loss": 1.0877,
+      "step": 2913
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999022893327994e-05,
+      "loss": 1.2615,
+      "step": 2914
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990221987481e-05,
+      "loss": 1.2332,
+      "step": 2915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999902150392142e-05,
+      "loss": 1.2203,
+      "step": 2916
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999020808847965e-05,
+      "loss": 1.2718,
+      "step": 2917
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999020113527726e-05,
+      "loss": 1.2264,
+      "step": 2918
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999019417960716e-05,
+      "loss": 1.016,
+      "step": 2919
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999018722146914e-05,
+      "loss": 0.9891,
+      "step": 2920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999018026086334e-05,
+      "loss": 1.0894,
+      "step": 2921
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901732977898e-05,
+      "loss": 0.9991,
+      "step": 2922
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901663322484e-05,
+      "loss": 1.234,
+      "step": 2923
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901593642393e-05,
+      "loss": 1.3263,
+      "step": 2924
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999015239376235e-05,
+      "loss": 0.9724,
+      "step": 2925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901454208176e-05,
+      "loss": 1.0791,
+      "step": 2926
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990138445405e-05,
+      "loss": 1.2334,
+      "step": 2927
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901314675246e-05,
+      "loss": 0.8321,
+      "step": 2928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999012448717655e-05,
+      "loss": 1.2932,
+      "step": 2929
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901175043605e-05,
+      "loss": 1.3547,
+      "step": 2930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901105190768e-05,
+      "loss": 0.7176,
+      "step": 2931
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999901035313252e-05,
+      "loss": 0.7461,
+      "step": 2932
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999009654110594e-05,
+      "loss": 1.4159,
+      "step": 2933
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900895484188e-05,
+      "loss": 1.0933,
+      "step": 2934
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900825532639e-05,
+      "loss": 1.0124,
+      "step": 2935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900755556411e-05,
+      "loss": 1.2458,
+      "step": 2936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999006855555064e-05,
+      "loss": 1.2437,
+      "step": 2937
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900615529923e-05,
+      "loss": 0.7774,
+      "step": 2938
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900545479662e-05,
+      "loss": 1.2332,
+      "step": 2939
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900475404723e-05,
+      "loss": 1.1314,
+      "step": 2940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900405305106e-05,
+      "loss": 0.9918,
+      "step": 2941
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990033518081e-05,
+      "loss": 0.5156,
+      "step": 2942
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999002650318374e-05,
+      "loss": 0.386,
+      "step": 2943
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999900194858187e-05,
+      "loss": 0.3806,
+      "step": 2944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999001246598575e-05,
+      "loss": 0.2731,
+      "step": 2945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99990005443685e-05,
+      "loss": 0.311,
+      "step": 2946
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899984189166e-05,
+      "loss": 0.1946,
+      "step": 2947
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998999139168026e-05,
+      "loss": 0.1996,
+      "step": 2948
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899843619762e-05,
+      "loss": 0.1449,
+      "step": 2949
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998997732980434e-05,
+      "loss": 0.1897,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899702951647e-05,
+      "loss": 0.0818,
+      "step": 2951
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899632580572e-05,
+      "loss": 0.153,
+      "step": 2952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998995621848195e-05,
+      "loss": 0.1676,
+      "step": 2953
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899491764389e-05,
+      "loss": 0.0627,
+      "step": 2954
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989942131928e-05,
+      "loss": 0.1163,
+      "step": 2955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899350849494e-05,
+      "loss": 0.1337,
+      "step": 2956
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899280355029e-05,
+      "loss": 0.139,
+      "step": 2957
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899209835887e-05,
+      "loss": 0.0755,
+      "step": 2958
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899139292067e-05,
+      "loss": 0.0569,
+      "step": 2959
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999899068723568e-05,
+      "loss": 0.0567,
+      "step": 2960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898998130392e-05,
+      "loss": 0.2035,
+      "step": 2961
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898927512538e-05,
+      "loss": 0.2997,
+      "step": 2962
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998988568700056e-05,
+      "loss": 0.3259,
+      "step": 2963
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998987862027956e-05,
+      "loss": 0.1894,
+      "step": 2964
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898715510908e-05,
+      "loss": 0.0832,
+      "step": 2965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898644794342e-05,
+      "loss": 1.3086,
+      "step": 2966
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898574053098e-05,
+      "loss": 1.2232,
+      "step": 2967
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898503287176e-05,
+      "loss": 0.9243,
+      "step": 2968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898432496576e-05,
+      "loss": 1.2614,
+      "step": 2969
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998983616812984e-05,
+      "loss": 1.134,
+      "step": 2970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998982908413425e-05,
+      "loss": 1.37,
+      "step": 2971
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898219976709e-05,
+      "loss": 1.1254,
+      "step": 2972
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898149087397e-05,
+      "loss": 1.1534,
+      "step": 2973
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999898078173408e-05,
+      "loss": 1.2074,
+      "step": 2974
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989800723474e-05,
+      "loss": 0.8483,
+      "step": 2975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897936271395e-05,
+      "loss": 1.3476,
+      "step": 2976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897865283372e-05,
+      "loss": 1.29,
+      "step": 2977
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998977942706705e-05,
+      "loss": 1.2638,
+      "step": 2978
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897723233291e-05,
+      "loss": 0.9091,
+      "step": 2979
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897652171234e-05,
+      "loss": 1.1177,
+      "step": 2980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897581084499e-05,
+      "loss": 1.0874,
+      "step": 2981
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897509973086e-05,
+      "loss": 1.0801,
+      "step": 2982
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897438836995e-05,
+      "loss": 1.0494,
+      "step": 2983
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897367676226e-05,
+      "loss": 1.1651,
+      "step": 2984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897296490779e-05,
+      "loss": 0.8595,
+      "step": 2985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897225280654e-05,
+      "loss": 0.8869,
+      "step": 2986
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998971540458514e-05,
+      "loss": 1.2285,
+      "step": 2987
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897082786371e-05,
+      "loss": 1.128,
+      "step": 2988
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998970115022126e-05,
+      "loss": 1.3749,
+      "step": 2989
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999896940193376e-05,
+      "loss": 1.3221,
+      "step": 2990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999896868859861e-05,
+      "loss": 1.254,
+      "step": 2991
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998967975016695e-05,
+      "loss": 1.1455,
+      "step": 2992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999896726118799e-05,
+      "loss": 1.0821,
+      "step": 2993
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998966547112506e-05,
+      "loss": 1.1394,
+      "step": 2994
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999896583279025e-05,
+      "loss": 0.8994,
+      "step": 2995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998965118221206e-05,
+      "loss": 1.2445,
+      "step": 2996
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998964403405386e-05,
+      "loss": 1.1261,
+      "step": 2997
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999896368834279e-05,
+      "loss": 1.1851,
+      "step": 2998
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999896297303341e-05,
+      "loss": 1.5682,
+      "step": 2999
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998962257477256e-05,
+      "loss": 0.9643,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.0737282037734985,
+      "eval_runtime": 85.3847,
+      "eval_samples_per_second": 16.221,
+      "eval_steps_per_second": 4.064,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999896154167432e-05,
+      "loss": 1.2074,
+      "step": 3001
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998960825624593e-05,
+      "loss": 1.3158,
+      "step": 3002
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998960109328105e-05,
+      "loss": 1.0965,
+      "step": 3003
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998959392784825e-05,
+      "loss": 1.0788,
+      "step": 3004
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895867599478e-05,
+      "loss": 0.8845,
+      "step": 3005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998957958957946e-05,
+      "loss": 0.793,
+      "step": 3006
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895724167433e-05,
+      "loss": 0.9951,
+      "step": 3007
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895652414393e-05,
+      "loss": 1.0859,
+      "step": 3008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895580636677e-05,
+      "loss": 0.8108,
+      "step": 3009
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998955088342816e-05,
+      "loss": 0.5787,
+      "step": 3010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895437007209e-05,
+      "loss": 0.5608,
+      "step": 3011
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895365155458e-05,
+      "loss": 0.5967,
+      "step": 3012
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989529327903e-05,
+      "loss": 0.4456,
+      "step": 3013
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895221377923e-05,
+      "loss": 0.5504,
+      "step": 3014
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895149452138e-05,
+      "loss": 0.8343,
+      "step": 3015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895077501676e-05,
+      "loss": 1.6915,
+      "step": 3016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999895005526536e-05,
+      "loss": 1.2349,
+      "step": 3017
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998949335267174e-05,
+      "loss": 1.139,
+      "step": 3018
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999894861502221e-05,
+      "loss": 1.1852,
+      "step": 3019
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999894789453047e-05,
+      "loss": 1.2533,
+      "step": 3020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999894717379195e-05,
+      "loss": 1.2747,
+      "step": 3021
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998946452806655e-05,
+      "loss": 1.1085,
+      "step": 3022
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998945731574574e-05,
+      "loss": 1.2659,
+      "step": 3023
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999894501009572e-05,
+      "loss": 0.7887,
+      "step": 3024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998944288370084e-05,
+      "loss": 1.1709,
+      "step": 3025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999894356639767e-05,
+      "loss": 1.1869,
+      "step": 3026
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998942844178475e-05,
+      "loss": 1.1321,
+      "step": 3027
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989421217125e-05,
+      "loss": 1.2509,
+      "step": 3028
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998941398999754e-05,
+      "loss": 1.3264,
+      "step": 3029
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999894067604022e-05,
+      "loss": 1.0859,
+      "step": 3030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998939952833915e-05,
+      "loss": 0.9069,
+      "step": 3031
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998939229380825e-05,
+      "loss": 1.1984,
+      "step": 3032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999893850568095e-05,
+      "loss": 0.9934,
+      "step": 3033
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998937781734304e-05,
+      "loss": 0.997,
+      "step": 3034
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999893705754089e-05,
+      "loss": 1.177,
+      "step": 3035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999893633310068e-05,
+      "loss": 1.0475,
+      "step": 3036
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998935608413685e-05,
+      "loss": 0.9387,
+      "step": 3037
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999893488347993e-05,
+      "loss": 1.0551,
+      "step": 3038
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998934158299385e-05,
+      "loss": 1.1771,
+      "step": 3039
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998933432872065e-05,
+      "loss": 1.2724,
+      "step": 3040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998932707197966e-05,
+      "loss": 1.2089,
+      "step": 3041
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999893198127708e-05,
+      "loss": 1.0284,
+      "step": 3042
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999893125510943e-05,
+      "loss": 0.9842,
+      "step": 3043
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999893052869499e-05,
+      "loss": 1.0936,
+      "step": 3044
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892980203378e-05,
+      "loss": 1.0716,
+      "step": 3045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998929075125784e-05,
+      "loss": 1.3936,
+      "step": 3046
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892834797101e-05,
+      "loss": 1.2452,
+      "step": 3047
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892762056946e-05,
+      "loss": 1.0022,
+      "step": 3048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998926892921124e-05,
+      "loss": 1.0688,
+      "step": 3049
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892616502602e-05,
+      "loss": 1.0423,
+      "step": 3050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998925436884126e-05,
+      "loss": 1.1952,
+      "step": 3051
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892470849546e-05,
+      "loss": 1.157,
+      "step": 3052
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998923979860015e-05,
+      "loss": 1.149,
+      "step": 3053
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892325097779e-05,
+      "loss": 0.8852,
+      "step": 3054
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998922521848786e-05,
+      "loss": 1.442,
+      "step": 3055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998921792473005e-05,
+      "loss": 1.1477,
+      "step": 3056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999892106285044e-05,
+      "loss": 1.0485,
+      "step": 3057
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989203329811e-05,
+      "loss": 0.9521,
+      "step": 3058
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891960286498e-05,
+      "loss": 1.3072,
+      "step": 3059
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891887250208e-05,
+      "loss": 1.0286,
+      "step": 3060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989181418924e-05,
+      "loss": 1.2539,
+      "step": 3061
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891741103595e-05,
+      "loss": 1.1245,
+      "step": 3062
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891667993272e-05,
+      "loss": 0.932,
+      "step": 3063
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998915948582705e-05,
+      "loss": 1.4209,
+      "step": 3064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891521698591e-05,
+      "loss": 1.5105,
+      "step": 3065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891448514234e-05,
+      "loss": 1.2014,
+      "step": 3066
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891375305199e-05,
+      "loss": 1.3503,
+      "step": 3067
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891302071486e-05,
+      "loss": 0.9712,
+      "step": 3068
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891228813095e-05,
+      "loss": 1.1086,
+      "step": 3069
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891155530026e-05,
+      "loss": 1.3851,
+      "step": 3070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998910822222804e-05,
+      "loss": 1.1959,
+      "step": 3071
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999891008889856e-05,
+      "loss": 0.9626,
+      "step": 3072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999890935532754e-05,
+      "loss": 1.0227,
+      "step": 3073
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998908621509736e-05,
+      "loss": 1.2477,
+      "step": 3074
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999890788744516e-05,
+      "loss": 1.1234,
+      "step": 3075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989071531338e-05,
+      "loss": 1.2972,
+      "step": 3076
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998906418575666e-05,
+      "loss": 1.0932,
+      "step": 3077
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999890568377075e-05,
+      "loss": 1.2099,
+      "step": 3078
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998904948719054e-05,
+      "loss": 1.0578,
+      "step": 3079
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999890421342058e-05,
+      "loss": 1.0771,
+      "step": 3080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999890347787533e-05,
+      "loss": 1.2784,
+      "step": 3081
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989027420833e-05,
+      "loss": 1.0203,
+      "step": 3082
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999890200604449e-05,
+      "loss": 1.3236,
+      "step": 3083
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99989012697589e-05,
+      "loss": 0.9522,
+      "step": 3084
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999890053322654e-05,
+      "loss": 0.9891,
+      "step": 3085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988997964474e-05,
+      "loss": 1.0306,
+      "step": 3086
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889905942147e-05,
+      "loss": 1.2939,
+      "step": 3087
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889832214877e-05,
+      "loss": 1.1802,
+      "step": 3088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889758462929e-05,
+      "loss": 1.0999,
+      "step": 3089
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889684686303e-05,
+      "loss": 1.2702,
+      "step": 3090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889610884999e-05,
+      "loss": 1.2539,
+      "step": 3091
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889537059018e-05,
+      "loss": 0.94,
+      "step": 3092
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889463208358e-05,
+      "loss": 1.1902,
+      "step": 3093
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889389333021e-05,
+      "loss": 0.987,
+      "step": 3094
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889315433006e-05,
+      "loss": 1.141,
+      "step": 3095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998892415083134e-05,
+      "loss": 1.2143,
+      "step": 3096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889167558942e-05,
+      "loss": 0.8637,
+      "step": 3097
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889093584893e-05,
+      "loss": 1.0951,
+      "step": 3098
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999889019586167e-05,
+      "loss": 1.2597,
+      "step": 3099
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998889455627625e-05,
+      "loss": 0.9368,
+      "step": 3100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998888715146794e-05,
+      "loss": 1.5598,
+      "step": 3101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888797441919e-05,
+      "loss": 1.1202,
+      "step": 3102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888723344482e-05,
+      "loss": 1.0475,
+      "step": 3103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888649222366e-05,
+      "loss": 1.0778,
+      "step": 3104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998885750755726e-05,
+      "loss": 1.1118,
+      "step": 3105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998885009041005e-05,
+      "loss": 0.9362,
+      "step": 3106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888426707951e-05,
+      "loss": 1.0847,
+      "step": 3107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998883524871244e-05,
+      "loss": 1.0924,
+      "step": 3108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888278241619e-05,
+      "loss": 0.9675,
+      "step": 3109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888203971437e-05,
+      "loss": 1.0317,
+      "step": 3110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888129676575e-05,
+      "loss": 1.1201,
+      "step": 3111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888055357037e-05,
+      "loss": 1.1086,
+      "step": 3112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998879810128205e-05,
+      "loss": 1.2691,
+      "step": 3113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999887906643926e-05,
+      "loss": 1.0789,
+      "step": 3114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999887832250354e-05,
+      "loss": 0.8669,
+      "step": 3115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998877578321046e-05,
+      "loss": 0.9003,
+      "step": 3116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999887683389177e-05,
+      "loss": 0.9656,
+      "step": 3117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998876089215705e-05,
+      "loss": 0.7094,
+      "step": 3118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999887534429287e-05,
+      "loss": 0.9274,
+      "step": 3119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999887459912326e-05,
+      "loss": 1.2837,
+      "step": 3120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998873853706875e-05,
+      "loss": 0.894,
+      "step": 3121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988731080437e-05,
+      "loss": 0.8587,
+      "step": 3122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998872362133754e-05,
+      "loss": 1.2183,
+      "step": 3123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999887161597703e-05,
+      "loss": 1.1431,
+      "step": 3124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999887086957352e-05,
+      "loss": 1.3707,
+      "step": 3125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998870122923235e-05,
+      "loss": 1.0031,
+      "step": 3126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999886937602618e-05,
+      "loss": 0.9755,
+      "step": 3127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999886862888234e-05,
+      "loss": 1.0886,
+      "step": 3128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999886788149172e-05,
+      "loss": 0.8562,
+      "step": 3129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999886713385432e-05,
+      "loss": 0.464,
+      "step": 3130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999886638597015e-05,
+      "loss": 0.6532,
+      "step": 3131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988656378392e-05,
+      "loss": 1.1652,
+      "step": 3132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998864889461466e-05,
+      "loss": 1.2203,
+      "step": 3133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998864140836956e-05,
+      "loss": 1.025,
+      "step": 3134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999886339196567e-05,
+      "loss": 1.2002,
+      "step": 3135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999886264284761e-05,
+      "loss": 1.3056,
+      "step": 3136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998861893482764e-05,
+      "loss": 1.0095,
+      "step": 3137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998861143871135e-05,
+      "loss": 1.3763,
+      "step": 3138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998860394012734e-05,
+      "loss": 1.0908,
+      "step": 3139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998859643907556e-05,
+      "loss": 1.0629,
+      "step": 3140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988588935556e-05,
+      "loss": 1.3534,
+      "step": 3141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998858142956866e-05,
+      "loss": 0.9142,
+      "step": 3142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998857392111354e-05,
+      "loss": 0.9733,
+      "step": 3143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998856641019064e-05,
+      "loss": 1.0786,
+      "step": 3144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998855889679996e-05,
+      "loss": 0.9734,
+      "step": 3145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999885513809415e-05,
+      "loss": 1.4096,
+      "step": 3146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999885438626152e-05,
+      "loss": 1.1417,
+      "step": 3147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999885363418212e-05,
+      "loss": 1.0368,
+      "step": 3148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999885288185593e-05,
+      "loss": 1.1928,
+      "step": 3149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998852129282966e-05,
+      "loss": 0.9801,
+      "step": 3150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999885137646324e-05,
+      "loss": 0.9239,
+      "step": 3151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999885062339672e-05,
+      "loss": 0.866,
+      "step": 3152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884987008342e-05,
+      "loss": 0.8846,
+      "step": 3153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884911652335e-05,
+      "loss": 0.8873,
+      "step": 3154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988483627165e-05,
+      "loss": 1.2745,
+      "step": 3155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884760866287e-05,
+      "loss": 1.1213,
+      "step": 3156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998846854362465e-05,
+      "loss": 0.9286,
+      "step": 3157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998846099815276e-05,
+      "loss": 0.4891,
+      "step": 3158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884534502132e-05,
+      "loss": 1.1473,
+      "step": 3159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884458998057e-05,
+      "loss": 0.6707,
+      "step": 3160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884383469306e-05,
+      "loss": 1.0229,
+      "step": 3161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884307915875e-05,
+      "loss": 1.3011,
+      "step": 3162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884232337768e-05,
+      "loss": 0.8993,
+      "step": 3163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884156734983e-05,
+      "loss": 1.1166,
+      "step": 3164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998840811075195e-05,
+      "loss": 1.1444,
+      "step": 3165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999884005455378e-05,
+      "loss": 1.1135,
+      "step": 3166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999883929778559e-05,
+      "loss": 0.9868,
+      "step": 3167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999883854077063e-05,
+      "loss": 1.1363,
+      "step": 3168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999883778350889e-05,
+      "loss": 1.2838,
+      "step": 3169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999883702600037e-05,
+      "loss": 1.428,
+      "step": 3170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998836268245067e-05,
+      "loss": 1.1863,
+      "step": 3171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998835510242993e-05,
+      "loss": 1.1418,
+      "step": 3172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998834751994136e-05,
+      "loss": 0.9341,
+      "step": 3173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988339934985e-05,
+      "loss": 1.0903,
+      "step": 3174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998833234756086e-05,
+      "loss": 1.192,
+      "step": 3175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988324757669e-05,
+      "loss": 1.1706,
+      "step": 3176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999883171653093e-05,
+      "loss": 1.1696,
+      "step": 3177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999883095704819e-05,
+      "loss": 1.2754,
+      "step": 3178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998830197318665e-05,
+      "loss": 0.8601,
+      "step": 3179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882943734237e-05,
+      "loss": 1.2475,
+      "step": 3180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882867711928e-05,
+      "loss": 1.006,
+      "step": 3181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882791664943e-05,
+      "loss": 1.1577,
+      "step": 3182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882715593279e-05,
+      "loss": 0.969,
+      "step": 3183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882639496938e-05,
+      "loss": 1.0363,
+      "step": 3184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882563375919e-05,
+      "loss": 1.0267,
+      "step": 3185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998824872302216e-05,
+      "loss": 1.2042,
+      "step": 3186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882411059847e-05,
+      "loss": 1.5013,
+      "step": 3187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882334864795e-05,
+      "loss": 1.2098,
+      "step": 3188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882258645065e-05,
+      "loss": 1.028,
+      "step": 3189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882182400657e-05,
+      "loss": 0.9882,
+      "step": 3190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882106131571e-05,
+      "loss": 1.0897,
+      "step": 3191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999882029837808e-05,
+      "loss": 1.2585,
+      "step": 3192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998819535193666e-05,
+      "loss": 1.1727,
+      "step": 3193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998818771762465e-05,
+      "loss": 1.0733,
+      "step": 3194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998818008084505e-05,
+      "loss": 1.1633,
+      "step": 3195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998817244159754e-05,
+      "loss": 1.2635,
+      "step": 3196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999881647998823e-05,
+      "loss": 0.9241,
+      "step": 3197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998815715569925e-05,
+      "loss": 1.1062,
+      "step": 3198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998814950904846e-05,
+      "loss": 1.1206,
+      "step": 3199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999881418599299e-05,
+      "loss": 1.0057,
+      "step": 3200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998813420834356e-05,
+      "loss": 1.1484,
+      "step": 3201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999881265542894e-05,
+      "loss": 1.0997,
+      "step": 3202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999881188977675e-05,
+      "loss": 1.0839,
+      "step": 3203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999881112387778e-05,
+      "loss": 1.5233,
+      "step": 3204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998810357732026e-05,
+      "loss": 1.1323,
+      "step": 3205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880959133951e-05,
+      "loss": 1.2861,
+      "step": 3206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880882470021e-05,
+      "loss": 0.8339,
+      "step": 3207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880805781413e-05,
+      "loss": 1.2662,
+      "step": 3208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880729068127e-05,
+      "loss": 1.0926,
+      "step": 3209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880652330163e-05,
+      "loss": 0.9723,
+      "step": 3210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880575567522e-05,
+      "loss": 0.9267,
+      "step": 3211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880498780203e-05,
+      "loss": 1.014,
+      "step": 3212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880421968206e-05,
+      "loss": 1.7897,
+      "step": 3213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880345131531e-05,
+      "loss": 0.9724,
+      "step": 3214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99988026827018e-05,
+      "loss": 1.209,
+      "step": 3215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998801913841494e-05,
+      "loss": 0.9055,
+      "step": 3216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998801144734406e-05,
+      "loss": 1.1476,
+      "step": 3217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880037538055e-05,
+      "loss": 1.0604,
+      "step": 3218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879960577992e-05,
+      "loss": 1.1971,
+      "step": 3219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879883593251e-05,
+      "loss": 2.9813,
+      "step": 3220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879806583832e-05,
+      "loss": 6.0157,
+      "step": 3221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998797295497356e-05,
+      "loss": 5.9594,
+      "step": 3222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998796524909606e-05,
+      "loss": 2.482,
+      "step": 3223
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879575407509e-05,
+      "loss": 1.1952,
+      "step": 3224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998794982993794e-05,
+      "loss": 1.3123,
+      "step": 3225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879421166571e-05,
+      "loss": 1.3332,
+      "step": 3226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879344009085e-05,
+      "loss": 1.144,
+      "step": 3227
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879266826922e-05,
+      "loss": 1.2806,
+      "step": 3228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879189620081e-05,
+      "loss": 1.1283,
+      "step": 3229
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998791123885624e-05,
+      "loss": 1.2873,
+      "step": 3230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999879035132366e-05,
+      "loss": 1.1551,
+      "step": 3231
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998789578514914e-05,
+      "loss": 1.0223,
+      "step": 3232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987888054594e-05,
+      "loss": 1.0714,
+      "step": 3233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987880321571e-05,
+      "loss": 1.1004,
+      "step": 3234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998787258608026e-05,
+      "loss": 1.0357,
+      "step": 3235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999878648481218e-05,
+      "loss": 0.8907,
+      "step": 3236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999878571076955e-05,
+      "loss": 0.9581,
+      "step": 3237
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999878493648014e-05,
+      "loss": 1.0673,
+      "step": 3238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998784161943954e-05,
+      "loss": 1.2363,
+      "step": 3239
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998783387160986e-05,
+      "loss": 1.3043,
+      "step": 3240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999878261213125e-05,
+      "loss": 0.9974,
+      "step": 3241
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999878183685473e-05,
+      "loss": 0.7341,
+      "step": 3242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998781061331434e-05,
+      "loss": 0.7755,
+      "step": 3243
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999878028556136e-05,
+      "loss": 0.951,
+      "step": 3244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877950954451e-05,
+      "loss": 0.7519,
+      "step": 3245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877873328089e-05,
+      "loss": 1.0954,
+      "step": 3246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877795677048e-05,
+      "loss": 1.0842,
+      "step": 3247
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987771800133e-05,
+      "loss": 1.1799,
+      "step": 3248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877640300934e-05,
+      "loss": 1.0409,
+      "step": 3249
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877562575861e-05,
+      "loss": 1.2298,
+      "step": 3250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877484826109e-05,
+      "loss": 1.1191,
+      "step": 3251
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998774070516796e-05,
+      "loss": 0.9785,
+      "step": 3252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877329252573e-05,
+      "loss": 1.2481,
+      "step": 3253
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877251428788e-05,
+      "loss": 1.4826,
+      "step": 3254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998771735803264e-05,
+      "loss": 1.1167,
+      "step": 3255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999877095707186e-05,
+      "loss": 1.1116,
+      "step": 3256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998770178093676e-05,
+      "loss": 0.9439,
+      "step": 3257
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998769398868725e-05,
+      "loss": 1.2351,
+      "step": 3258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998768619396995e-05,
+      "loss": 1.0686,
+      "step": 3259
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999876783967848e-05,
+      "loss": 1.5349,
+      "step": 3260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998767059713196e-05,
+      "loss": 1.1874,
+      "step": 3261
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998766279501133e-05,
+      "loss": 1.3539,
+      "step": 3262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998765499042286e-05,
+      "loss": 1.2347,
+      "step": 3263
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998764718336673e-05,
+      "loss": 1.2608,
+      "step": 3264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998763937384277e-05,
+      "loss": 1.4032,
+      "step": 3265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987631561851e-05,
+      "loss": 1.3431,
+      "step": 3266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999876237473915e-05,
+      "loss": 1.027,
+      "step": 3267
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998761593046425e-05,
+      "loss": 1.5965,
+      "step": 3268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998760811106916e-05,
+      "loss": 1.0208,
+      "step": 3269
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998760028920636e-05,
+      "loss": 1.1091,
+      "step": 3270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875924648758e-05,
+      "loss": 1.2286,
+      "step": 3271
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998758463807735e-05,
+      "loss": 0.8837,
+      "step": 3272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875768088113e-05,
+      "loss": 0.7764,
+      "step": 3273
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875689770773e-05,
+      "loss": 0.7657,
+      "step": 3274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875611428756e-05,
+      "loss": 0.9198,
+      "step": 3275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875533062062e-05,
+      "loss": 1.1006,
+      "step": 3276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875454670689e-05,
+      "loss": 1.0076,
+      "step": 3277
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875376254639e-05,
+      "loss": 1.3636,
+      "step": 3278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875297813912e-05,
+      "loss": 1.1701,
+      "step": 3279
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875219348506e-05,
+      "loss": 1.1995,
+      "step": 3280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875140858423e-05,
+      "loss": 0.9329,
+      "step": 3281
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999875062343662e-05,
+      "loss": 1.2552,
+      "step": 3282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998749838042235e-05,
+      "loss": 1.1106,
+      "step": 3283
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999874905240107e-05,
+      "loss": 1.3376,
+      "step": 3284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999874826651313e-05,
+      "loss": 1.1603,
+      "step": 3285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998747480378413e-05,
+      "loss": 1.4663,
+      "step": 3286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998746693996915e-05,
+      "loss": 1.3069,
+      "step": 3287
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999874590736865e-05,
+      "loss": 1.1898,
+      "step": 3288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987451204936e-05,
+      "loss": 1.2227,
+      "step": 3289
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999874433337177e-05,
+      "loss": 1.0331,
+      "step": 3290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999874354600317e-05,
+      "loss": 1.115,
+      "step": 3291
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999874275838779e-05,
+      "loss": 1.0944,
+      "step": 3292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999874197052563e-05,
+      "loss": 1.2408,
+      "step": 3293
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998741182416695e-05,
+      "loss": 1.0281,
+      "step": 3294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998740394060987e-05,
+      "loss": 1.1958,
+      "step": 3295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987396054585e-05,
+      "loss": 0.6619,
+      "step": 3296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998738816609235e-05,
+      "loss": 1.2319,
+      "step": 3297
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999873802751319e-05,
+      "loss": 1.0441,
+      "step": 3298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999873723817037e-05,
+      "loss": 1.2682,
+      "step": 3299
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999873644858077e-05,
+      "loss": 1.1649,
+      "step": 3300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987356587444e-05,
+      "loss": 0.9821,
+      "step": 3301
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998734868661255e-05,
+      "loss": 0.9071,
+      "step": 3302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999873407833133e-05,
+      "loss": 0.9467,
+      "step": 3303
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998733287754626e-05,
+      "loss": 1.3175,
+      "step": 3304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998732496931144e-05,
+      "loss": 1.2642,
+      "step": 3305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998731705860884e-05,
+      "loss": 1.2276,
+      "step": 3306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998730914543846e-05,
+      "loss": 1.0944,
+      "step": 3307
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999873012298004e-05,
+      "loss": 0.7492,
+      "step": 3308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998729331169444e-05,
+      "loss": 1.5178,
+      "step": 3309
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998728539112086e-05,
+      "loss": 1.1938,
+      "step": 3310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998727746807936e-05,
+      "loss": 1.2197,
+      "step": 3311
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999872695425702e-05,
+      "loss": 1.0732,
+      "step": 3312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998726161459324e-05,
+      "loss": 0.9876,
+      "step": 3313
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999872536841485e-05,
+      "loss": 0.8108,
+      "step": 3314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987245751236e-05,
+      "loss": 1.0706,
+      "step": 3315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999872378158558e-05,
+      "loss": 1.2819,
+      "step": 3316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999872298780077e-05,
+      "loss": 1.1775,
+      "step": 3317
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999872219376919e-05,
+      "loss": 1.0579,
+      "step": 3318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998721399490835e-05,
+      "loss": 1.0045,
+      "step": 3319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987206049657e-05,
+      "loss": 1.0541,
+      "step": 3320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871981019378e-05,
+      "loss": 1.5234,
+      "step": 3321
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987190151751e-05,
+      "loss": 1.2823,
+      "step": 3322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871821990964e-05,
+      "loss": 1.5049,
+      "step": 3323
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871742439739e-05,
+      "loss": 1.2312,
+      "step": 3324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871662863837e-05,
+      "loss": 1.2007,
+      "step": 3325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871583263258e-05,
+      "loss": 1.0416,
+      "step": 3326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998715036380005e-05,
+      "loss": 1.4341,
+      "step": 3327
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871423988066e-05,
+      "loss": 0.9972,
+      "step": 3328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871344313453e-05,
+      "loss": 1.1302,
+      "step": 3329
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871264614163e-05,
+      "loss": 1.2175,
+      "step": 3330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998711848901945e-05,
+      "loss": 0.804,
+      "step": 3331
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871105141549e-05,
+      "loss": 0.8429,
+      "step": 3332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998710253682255e-05,
+      "loss": 0.966,
+      "step": 3333
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999870945570225e-05,
+      "loss": 1.0244,
+      "step": 3334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998708657475466e-05,
+      "loss": 1.2352,
+      "step": 3335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99987078590019e-05,
+      "loss": 1.2682,
+      "step": 3336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999870706028156e-05,
+      "loss": 1.2687,
+      "step": 3337
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998706261314444e-05,
+      "loss": 1.1056,
+      "step": 3338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999870546210055e-05,
+      "loss": 0.7855,
+      "step": 3339
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999870466263988e-05,
+      "loss": 1.506,
+      "step": 3340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999870386293244e-05,
+      "loss": 1.3185,
+      "step": 3341
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998703062978215e-05,
+      "loss": 1.0603,
+      "step": 3342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999870226277721e-05,
+      "loss": 0.8978,
+      "step": 3343
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998701462329437e-05,
+      "loss": 0.7992,
+      "step": 3344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998700661634884e-05,
+      "loss": 1.1016,
+      "step": 3345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869986069355e-05,
+      "loss": 1.0676,
+      "step": 3346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869905950545e-05,
+      "loss": 1.3322,
+      "step": 3347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998698258070564e-05,
+      "loss": 0.8448,
+      "step": 3348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869745638891e-05,
+      "loss": 1.2603,
+      "step": 3349
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869665446047e-05,
+      "loss": 2.0377,
+      "step": 3350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869585228526e-05,
+      "loss": 1.1663,
+      "step": 3351
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998695049863263e-05,
+      "loss": 1.3606,
+      "step": 3352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99986942471945e-05,
+      "loss": 1.2747,
+      "step": 3353
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869344427896e-05,
+      "loss": 1.1279,
+      "step": 3354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869264111664e-05,
+      "loss": 1.0564,
+      "step": 3355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869183770754e-05,
+      "loss": 0.5569,
+      "step": 3356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869103405167e-05,
+      "loss": 0.9706,
+      "step": 3357
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999869023014903e-05,
+      "loss": 0.992,
+      "step": 3358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99986894259996e-05,
+      "loss": 0.9301,
+      "step": 3359
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998688621603395e-05,
+      "loss": 1.0552,
+      "step": 3360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999868781696042e-05,
+      "loss": 1.2832,
+      "step": 3361
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998687012070664e-05,
+      "loss": 1.0131,
+      "step": 3362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999868620693413e-05,
+      "loss": 1.0057,
+      "step": 3363
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999868540155083e-05,
+      "loss": 0.7254,
+      "step": 3364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999868459592074e-05,
+      "loss": 0.9847,
+      "step": 3365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999868379004388e-05,
+      "loss": 1.0289,
+      "step": 3366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998682983920245e-05,
+      "loss": 1.1107,
+      "step": 3367
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998682177549826e-05,
+      "loss": 0.4372,
+      "step": 3368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998681370932635e-05,
+      "loss": 0.9341,
+      "step": 3369
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998680564068674e-05,
+      "loss": 1.035,
+      "step": 3370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999867975695793e-05,
+      "loss": 1.0862,
+      "step": 3371
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999867894960041e-05,
+      "loss": 1.0916,
+      "step": 3372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998678141996115e-05,
+      "loss": 1.4372,
+      "step": 3373
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999867733414504e-05,
+      "loss": 1.4361,
+      "step": 3374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99986765260472e-05,
+      "loss": 1.2444,
+      "step": 3375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999867571770257e-05,
+      "loss": 1.3493,
+      "step": 3376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999867490911117e-05,
+      "loss": 1.1362,
+      "step": 3377
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999867410027299e-05,
+      "loss": 1.1176,
+      "step": 3378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999867329118804e-05,
+      "loss": 1.4117,
+      "step": 3379
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998672481856305e-05,
+      "loss": 1.6114,
+      "step": 3380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99986716722778e-05,
+      "loss": 1.5723,
+      "step": 3381
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998670862452515e-05,
+      "loss": 1.3304,
+      "step": 3382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998670052380454e-05,
+      "loss": 1.0998,
+      "step": 3383
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998669242061614e-05,
+      "loss": 1.1452,
+      "step": 3384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998668431496004e-05,
+      "loss": 1.1085,
+      "step": 3385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998667620683615e-05,
+      "loss": 1.2093,
+      "step": 3386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999866680962445e-05,
+      "loss": 1.1872,
+      "step": 3387
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998665998318504e-05,
+      "loss": 1.1485,
+      "step": 3388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998665186765795e-05,
+      "loss": 0.9688,
+      "step": 3389
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998664374966294e-05,
+      "loss": 1.3017,
+      "step": 3390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999866356292003e-05,
+      "loss": 1.1592,
+      "step": 3391
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999866275062698e-05,
+      "loss": 1.1275,
+      "step": 3392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999866193808716e-05,
+      "loss": 0.9936,
+      "step": 3393
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999866112530056e-05,
+      "loss": 1.0442,
+      "step": 3394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999866031226719e-05,
+      "loss": 0.9005,
+      "step": 3395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998659498987035e-05,
+      "loss": 0.7956,
+      "step": 3396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999865868546011e-05,
+      "loss": 0.771,
+      "step": 3397
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998657871686406e-05,
+      "loss": 0.7461,
+      "step": 3398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998657057665924e-05,
+      "loss": 0.6217,
+      "step": 3399
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998656243398664e-05,
+      "loss": 0.5203,
+      "step": 3400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999865542888463e-05,
+      "loss": 0.596,
+      "step": 3401
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998654614123824e-05,
+      "loss": 0.6091,
+      "step": 3402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999865379911624e-05,
+      "loss": 0.6239,
+      "step": 3403
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999865298386188e-05,
+      "loss": 0.5065,
+      "step": 3404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998652168360736e-05,
+      "loss": 0.6384,
+      "step": 3405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999865135261282e-05,
+      "loss": 0.645,
+      "step": 3406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999865053661814e-05,
+      "loss": 1.1136,
+      "step": 3407
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864972037667e-05,
+      "loss": 0.9981,
+      "step": 3408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864890388843e-05,
+      "loss": 1.2167,
+      "step": 3409
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864808715341e-05,
+      "loss": 1.2866,
+      "step": 3410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998647270171616e-05,
+      "loss": 1.0734,
+      "step": 3411
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864645294305e-05,
+      "loss": 1.1872,
+      "step": 3412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99986456354677e-05,
+      "loss": 1.9395,
+      "step": 3413
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864481774558e-05,
+      "loss": 1.3458,
+      "step": 3414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864399977668e-05,
+      "loss": 1.2487,
+      "step": 3415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864318156101e-05,
+      "loss": 1.4208,
+      "step": 3416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998642363098556e-05,
+      "loss": 1.1311,
+      "step": 3417
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999864154438933e-05,
+      "loss": 1.2627,
+      "step": 3418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998640725433326e-05,
+      "loss": 1.3189,
+      "step": 3419
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999863990623055e-05,
+      "loss": 1.2433,
+      "step": 3420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998639086781e-05,
+      "loss": 1.1867,
+      "step": 3421
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998638267084664e-05,
+      "loss": 0.8289,
+      "step": 3422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999863744714156e-05,
+      "loss": 1.4112,
+      "step": 3423
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998636626951674e-05,
+      "loss": 1.2278,
+      "step": 3424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999863580651501e-05,
+      "loss": 0.9466,
+      "step": 3425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999863498583158e-05,
+      "loss": 1.1715,
+      "step": 3426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999863416490137e-05,
+      "loss": 1.4013,
+      "step": 3427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998633343724374e-05,
+      "loss": 1.0322,
+      "step": 3428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998632522300614e-05,
+      "loss": 1.1111,
+      "step": 3429
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998631700630076e-05,
+      "loss": 0.8467,
+      "step": 3430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999863087871276e-05,
+      "loss": 1.0402,
+      "step": 3431
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999863005654867e-05,
+      "loss": 0.9011,
+      "step": 3432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999862923413781e-05,
+      "loss": 1.1107,
+      "step": 3433
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998628411480165e-05,
+      "loss": 0.7305,
+      "step": 3434
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998627588575744e-05,
+      "loss": 0.5344,
+      "step": 3435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999862676542455e-05,
+      "loss": 0.5887,
+      "step": 3436
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998625942026575e-05,
+      "loss": 0.53,
+      "step": 3437
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998625118381834e-05,
+      "loss": 0.6023,
+      "step": 3438
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998624294490315e-05,
+      "loss": 0.5898,
+      "step": 3439
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999862347035201e-05,
+      "loss": 0.5584,
+      "step": 3440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999862264596694e-05,
+      "loss": 0.5397,
+      "step": 3441
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999862182133509e-05,
+      "loss": 0.732,
+      "step": 3442
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998620996456465e-05,
+      "loss": 0.588,
+      "step": 3443
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999862017133106e-05,
+      "loss": 0.5927,
+      "step": 3444
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861934595889e-05,
+      "loss": 0.5751,
+      "step": 3445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861852033994e-05,
+      "loss": 0.5968,
+      "step": 3446
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99986176944742e-05,
+      "loss": 0.6897,
+      "step": 3447
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998616868361694e-05,
+      "loss": 0.4709,
+      "step": 3448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998616042002415e-05,
+      "loss": 0.5125,
+      "step": 3449
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861521539636e-05,
+      "loss": 0.4163,
+      "step": 3450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861438854353e-05,
+      "loss": 0.4144,
+      "step": 3451
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861356144392e-05,
+      "loss": 0.4661,
+      "step": 3452
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861273409754e-05,
+      "loss": 0.3836,
+      "step": 3453
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861190650438e-05,
+      "loss": 0.3678,
+      "step": 3454
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861107866444e-05,
+      "loss": 0.4695,
+      "step": 3455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999861025057773e-05,
+      "loss": 0.3993,
+      "step": 3456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860942224425e-05,
+      "loss": 0.3604,
+      "step": 3457
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860859366398e-05,
+      "loss": 0.4166,
+      "step": 3458
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860776483695e-05,
+      "loss": 0.4779,
+      "step": 3459
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860693576313e-05,
+      "loss": 0.4345,
+      "step": 3460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860610644253e-05,
+      "loss": 0.4439,
+      "step": 3461
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998605276875175e-05,
+      "loss": 0.3738,
+      "step": 3462
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860444706103e-05,
+      "loss": 0.5384,
+      "step": 3463
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860361700012e-05,
+      "loss": 1.3531,
+      "step": 3464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998602786692426e-05,
+      "loss": 1.1958,
+      "step": 3465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998601956137955e-05,
+      "loss": 1.3425,
+      "step": 3466
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860112533671e-05,
+      "loss": 1.3543,
+      "step": 3467
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999860029428869e-05,
+      "loss": 0.9571,
+      "step": 3468
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99985994629939e-05,
+      "loss": 1.2027,
+      "step": 3469
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859863145233e-05,
+      "loss": 1.2444,
+      "step": 3470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859779966398e-05,
+      "loss": 1.1483,
+      "step": 3471
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859696762886e-05,
+      "loss": 1.2692,
+      "step": 3472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998596135346956e-05,
+      "loss": 1.1072,
+      "step": 3473
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859530281828e-05,
+      "loss": 1.0419,
+      "step": 3474
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859447004284e-05,
+      "loss": 1.103,
+      "step": 3475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998593637020615e-05,
+      "loss": 1.3038,
+      "step": 3476
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859280375161e-05,
+      "loss": 1.1934,
+      "step": 3477
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998591970235836e-05,
+      "loss": 1.0241,
+      "step": 3478
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998591136473286e-05,
+      "loss": 1.088,
+      "step": 3479
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859030246395e-05,
+      "loss": 1.1071,
+      "step": 3480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999858946820786e-05,
+      "loss": 1.1374,
+      "step": 3481
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999858863370498e-05,
+      "loss": 0.9614,
+      "step": 3482
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999858779895532e-05,
+      "loss": 1.1881,
+      "step": 3483
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998586963958894e-05,
+      "loss": 1.2886,
+      "step": 3484
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999858612871569e-05,
+      "loss": 0.7944,
+      "step": 3485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999858529322571e-05,
+      "loss": 1.3246,
+      "step": 3486
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998584457488955e-05,
+      "loss": 1.3053,
+      "step": 3487
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998583621505424e-05,
+      "loss": 1.3665,
+      "step": 3488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998582785275114e-05,
+      "loss": 1.2066,
+      "step": 3489
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999858194879804e-05,
+      "loss": 1.1096,
+      "step": 3490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999858111207418e-05,
+      "loss": 1.3171,
+      "step": 3491
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998580275103546e-05,
+      "loss": 0.9422,
+      "step": 3492
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999857943788614e-05,
+      "loss": 1.0387,
+      "step": 3493
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999857860042195e-05,
+      "loss": 1.2342,
+      "step": 3494
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998577762711e-05,
+      "loss": 1.0393,
+      "step": 3495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999857692475326e-05,
+      "loss": 0.9846,
+      "step": 3496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998576086548756e-05,
+      "loss": 1.7033,
+      "step": 3497
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998575248097466e-05,
+      "loss": 1.2261,
+      "step": 3498
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998574409399404e-05,
+      "loss": 0.8879,
+      "step": 3499
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998573570454564e-05,
+      "loss": 0.8896,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.0793774127960205,
+      "eval_runtime": 85.2881,
+      "eval_samples_per_second": 16.239,
+      "eval_steps_per_second": 4.069,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999857273126296e-05,
+      "loss": 0.9776,
+      "step": 3501
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999857189182457e-05,
+      "loss": 1.1962,
+      "step": 3502
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999857105213941e-05,
+      "loss": 0.9708,
+      "step": 3503
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999857021220747e-05,
+      "loss": 0.9686,
+      "step": 3504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856937202876e-05,
+      "loss": 1.0573,
+      "step": 3505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856853160327e-05,
+      "loss": 1.1712,
+      "step": 3506
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856769093101e-05,
+      "loss": 1.1592,
+      "step": 3507
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856685001196e-05,
+      "loss": 0.8742,
+      "step": 3508
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856600884615e-05,
+      "loss": 0.9963,
+      "step": 3509
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856516743356e-05,
+      "loss": 1.009,
+      "step": 3510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99985643257742e-05,
+      "loss": 1.2888,
+      "step": 3511
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856348386806e-05,
+      "loss": 1.0399,
+      "step": 3512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856264171514e-05,
+      "loss": 1.138,
+      "step": 3513
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999856179931545e-05,
+      "loss": 0.9903,
+      "step": 3514
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998560956668985e-05,
+      "loss": 0.8786,
+      "step": 3515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998560113775746e-05,
+      "loss": 1.2258,
+      "step": 3516
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999855927063573e-05,
+      "loss": 1.2219,
+      "step": 3517
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999855842724893e-05,
+      "loss": 1.0808,
+      "step": 3518
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998557583615365e-05,
+      "loss": 1.2123,
+      "step": 3519
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999855673973503e-05,
+      "loss": 1.3638,
+      "step": 3520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998555895607904e-05,
+      "loss": 1.071,
+      "step": 3521
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999855505123402e-05,
+      "loss": 0.5962,
+      "step": 3522
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998554206613345e-05,
+      "loss": 0.9059,
+      "step": 3523
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99985533617459e-05,
+      "loss": 0.9626,
+      "step": 3524
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999855251663169e-05,
+      "loss": 1.3906,
+      "step": 3525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999855167127069e-05,
+      "loss": 1.3815,
+      "step": 3526
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998550825662925e-05,
+      "loss": 1.3475,
+      "step": 3527
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999854997980838e-05,
+      "loss": 1.2545,
+      "step": 3528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998549133707064e-05,
+      "loss": 1.2959,
+      "step": 3529
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998548287358974e-05,
+      "loss": 0.9743,
+      "step": 3530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99985474407641e-05,
+      "loss": 1.2155,
+      "step": 3531
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999854659392245e-05,
+      "loss": 1.107,
+      "step": 3532
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998545746834034e-05,
+      "loss": 1.0236,
+      "step": 3533
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999854489949884e-05,
+      "loss": 1.0437,
+      "step": 3534
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998544051916865e-05,
+      "loss": 1.1194,
+      "step": 3535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999854320408812e-05,
+      "loss": 1.0564,
+      "step": 3536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998542356012604e-05,
+      "loss": 1.0349,
+      "step": 3537
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999854150769031e-05,
+      "loss": 1.2829,
+      "step": 3538
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999854065912124e-05,
+      "loss": 0.9777,
+      "step": 3539
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998539810305394e-05,
+      "loss": 1.2097,
+      "step": 3540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853896124277e-05,
+      "loss": 0.8308,
+      "step": 3541
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853811193338e-05,
+      "loss": 0.91,
+      "step": 3542
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853726237721e-05,
+      "loss": 0.9906,
+      "step": 3543
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853641257426e-05,
+      "loss": 1.122,
+      "step": 3544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853556252454e-05,
+      "loss": 0.9985,
+      "step": 3545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853471222805e-05,
+      "loss": 0.9434,
+      "step": 3546
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998533861684784e-05,
+      "loss": 1.1515,
+      "step": 3547
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853301089473e-05,
+      "loss": 1.1774,
+      "step": 3548
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853215985791e-05,
+      "loss": 0.9875,
+      "step": 3549
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853130857432e-05,
+      "loss": 1.1445,
+      "step": 3550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999853045704394e-05,
+      "loss": 1.3954,
+      "step": 3551
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998529605266805e-05,
+      "loss": 1.2126,
+      "step": 3552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852875324288e-05,
+      "loss": 0.8698,
+      "step": 3553
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852790097219e-05,
+      "loss": 1.1348,
+      "step": 3554
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998527048454716e-05,
+      "loss": 0.7995,
+      "step": 3555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852619569047e-05,
+      "loss": 1.3086,
+      "step": 3556
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852534267945e-05,
+      "loss": 1.1718,
+      "step": 3557
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852448942166e-05,
+      "loss": 1.0632,
+      "step": 3558
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852363591709e-05,
+      "loss": 1.0728,
+      "step": 3559
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852278216575e-05,
+      "loss": 1.2243,
+      "step": 3560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852192816762e-05,
+      "loss": 1.1648,
+      "step": 3561
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852107392273e-05,
+      "loss": 0.9032,
+      "step": 3562
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999852021943106e-05,
+      "loss": 1.2605,
+      "step": 3563
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999851936469262e-05,
+      "loss": 0.9793,
+      "step": 3564
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99985185097074e-05,
+      "loss": 1.275,
+      "step": 3565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999851765447541e-05,
+      "loss": 1.0971,
+      "step": 3566
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998516798996634e-05,
+      "loss": 1.2443,
+      "step": 3567
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998515943271096e-05,
+      "loss": 1.1871,
+      "step": 3568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999851508729877e-05,
+      "loss": 1.0477,
+      "step": 3569
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998514231079686e-05,
+      "loss": 1.0575,
+      "step": 3570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998513374613814e-05,
+      "loss": 0.9334,
+      "step": 3571
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999851251790118e-05,
+      "loss": 1.1136,
+      "step": 3572
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998511660941756e-05,
+      "loss": 1.2688,
+      "step": 3573
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998510803735564e-05,
+      "loss": 1.1204,
+      "step": 3574
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850994628259e-05,
+      "loss": 1.1439,
+      "step": 3575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850908858286e-05,
+      "loss": 1.1018,
+      "step": 3576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850823063634e-05,
+      "loss": 1.3292,
+      "step": 3577
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850737244305e-05,
+      "loss": 0.9875,
+      "step": 3578
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850651400298e-05,
+      "loss": 1.0082,
+      "step": 3579
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850565531614e-05,
+      "loss": 1.189,
+      "step": 3580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850479638252e-05,
+      "loss": 2.1198,
+      "step": 3581
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850393720213e-05,
+      "loss": 1.5383,
+      "step": 3582
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850307777497e-05,
+      "loss": 1.1514,
+      "step": 3583
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850221810103e-05,
+      "loss": 0.9158,
+      "step": 3584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999850135818032e-05,
+      "loss": 1.1748,
+      "step": 3585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998500498012824e-05,
+      "loss": 1.2605,
+      "step": 3586
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849963759856e-05,
+      "loss": 1.286,
+      "step": 3587
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849877693752e-05,
+      "loss": 1.0966,
+      "step": 3588
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998497916029714e-05,
+      "loss": 1.1221,
+      "step": 3589
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849705487513e-05,
+      "loss": 1.2224,
+      "step": 3590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998496193473765e-05,
+      "loss": 0.6984,
+      "step": 3591
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998495331825623e-05,
+      "loss": 1.261,
+      "step": 3592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849446993071e-05,
+      "loss": 1.4161,
+      "step": 3593
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849360778903e-05,
+      "loss": 1.2171,
+      "step": 3594
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849274540057e-05,
+      "loss": 0.8938,
+      "step": 3595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849188276534e-05,
+      "loss": 1.21,
+      "step": 3596
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849101988333e-05,
+      "loss": 1.4424,
+      "step": 3597
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999849015675454e-05,
+      "loss": 1.1366,
+      "step": 3598
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999848929337898e-05,
+      "loss": 0.9923,
+      "step": 3599
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999848842975665e-05,
+      "loss": 1.1938,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998487565887544e-05,
+      "loss": 1.095,
+      "step": 3601
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999848670177166e-05,
+      "loss": 1.1377,
+      "step": 3602
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998485837409004e-05,
+      "loss": 1.1227,
+      "step": 3603
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999848497279957e-05,
+      "loss": 1.1024,
+      "step": 3604
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999848410794337e-05,
+      "loss": 0.7742,
+      "step": 3605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999848324284039e-05,
+      "loss": 1.2553,
+      "step": 3606
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999848237749063e-05,
+      "loss": 0.9677,
+      "step": 3607
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984815118941e-05,
+      "loss": 1.0944,
+      "step": 3608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984806460508e-05,
+      "loss": 0.93,
+      "step": 3609
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999847977996072e-05,
+      "loss": 1.1312,
+      "step": 3610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999847891362387e-05,
+      "loss": 1.0116,
+      "step": 3611
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998478047040245e-05,
+      "loss": 0.9964,
+      "step": 3612
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998477180209834e-05,
+      "loss": 0.875,
+      "step": 3613
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999847631313266e-05,
+      "loss": 0.8657,
+      "step": 3614
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999847544580871e-05,
+      "loss": 1.2225,
+      "step": 3615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998474578237986e-05,
+      "loss": 0.9426,
+      "step": 3616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998473710420484e-05,
+      "loss": 1.1022,
+      "step": 3617
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999847284235622e-05,
+      "loss": 1.1475,
+      "step": 3618
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998471974045165e-05,
+      "loss": 1.1483,
+      "step": 3619
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999847110548734e-05,
+      "loss": 0.9323,
+      "step": 3620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999847023668275e-05,
+      "loss": 0.9708,
+      "step": 3621
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998469367631375e-05,
+      "loss": 0.8721,
+      "step": 3622
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999846849833323e-05,
+      "loss": 0.7354,
+      "step": 3623
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999846762878831e-05,
+      "loss": 0.7569,
+      "step": 3624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999846675899662e-05,
+      "loss": 0.6744,
+      "step": 3625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999846588895815e-05,
+      "loss": 0.6684,
+      "step": 3626
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984650186729e-05,
+      "loss": 0.7758,
+      "step": 3627
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998464148140886e-05,
+      "loss": 0.566,
+      "step": 3628
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998463277362095e-05,
+      "loss": 0.6658,
+      "step": 3629
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998462406336526e-05,
+      "loss": 0.5957,
+      "step": 3630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998461535064186e-05,
+      "loss": 0.5347,
+      "step": 3631
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998460663545075e-05,
+      "loss": 0.5478,
+      "step": 3632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998459791779186e-05,
+      "loss": 0.4344,
+      "step": 3633
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999845891976652e-05,
+      "loss": 0.8633,
+      "step": 3634
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999845804750709e-05,
+      "loss": 1.2405,
+      "step": 3635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999845717500087e-05,
+      "loss": 1.6678,
+      "step": 3636
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999845630224789e-05,
+      "loss": 1.3302,
+      "step": 3637
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998455429248125e-05,
+      "loss": 1.0957,
+      "step": 3638
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998454556001595e-05,
+      "loss": 0.8661,
+      "step": 3639
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999845368250829e-05,
+      "loss": 0.9106,
+      "step": 3640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984528087682e-05,
+      "loss": 0.9173,
+      "step": 3641
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998451934781344e-05,
+      "loss": 1.0613,
+      "step": 3642
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998451060547716e-05,
+      "loss": 0.9675,
+      "step": 3643
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998450186067317e-05,
+      "loss": 1.0819,
+      "step": 3644
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844931134013e-05,
+      "loss": 1.2227,
+      "step": 3645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844843636618e-05,
+      "loss": 1.172,
+      "step": 3646
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844756114545e-05,
+      "loss": 1.0692,
+      "step": 3647
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998446685677946e-05,
+      "loss": 0.88,
+      "step": 3648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844580996367e-05,
+      "loss": 1.0156,
+      "step": 3649
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844493400262e-05,
+      "loss": 1.1947,
+      "step": 3650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984440577948e-05,
+      "loss": 1.1163,
+      "step": 3651
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984431813402e-05,
+      "loss": 1.0156,
+      "step": 3652
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844230463883e-05,
+      "loss": 1.1083,
+      "step": 3653
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844142769068e-05,
+      "loss": 1.0469,
+      "step": 3654
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999844055049576e-05,
+      "loss": 0.8485,
+      "step": 3655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998439673054066e-05,
+      "loss": 0.9361,
+      "step": 3656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984387953656e-05,
+      "loss": 1.2245,
+      "step": 3657
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999843791743035e-05,
+      "loss": 1.2246,
+      "step": 3658
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998437039248336e-05,
+      "loss": 1.329,
+      "step": 3659
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999843616081955e-05,
+      "loss": 1.7445,
+      "step": 3660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998435282143986e-05,
+      "loss": 1.8381,
+      "step": 3661
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998434403221644e-05,
+      "loss": 1.6972,
+      "step": 3662
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999843352405254e-05,
+      "loss": 1.7102,
+      "step": 3663
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999843264463665e-05,
+      "loss": 1.7059,
+      "step": 3664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998431764973985e-05,
+      "loss": 1.6628,
+      "step": 3665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999843088506455e-05,
+      "loss": 1.7004,
+      "step": 3666
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999843000490834e-05,
+      "loss": 1.675,
+      "step": 3667
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842912450536e-05,
+      "loss": 1.6609,
+      "step": 3668
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984282438556e-05,
+      "loss": 1.6648,
+      "step": 3669
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842736295907e-05,
+      "loss": 1.6662,
+      "step": 3670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842648181576e-05,
+      "loss": 1.6543,
+      "step": 3671
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842560042569e-05,
+      "loss": 1.6691,
+      "step": 3672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998424718788837e-05,
+      "loss": 1.6507,
+      "step": 3673
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998423836905207e-05,
+      "loss": 1.6278,
+      "step": 3674
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842295477481e-05,
+      "loss": 1.6509,
+      "step": 3675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842207239763e-05,
+      "loss": 1.6284,
+      "step": 3676
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842118977369e-05,
+      "loss": 1.6695,
+      "step": 3677
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999842030690297e-05,
+      "loss": 1.6526,
+      "step": 3678
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841942378547e-05,
+      "loss": 1.6508,
+      "step": 3679
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984185404212e-05,
+      "loss": 1.6496,
+      "step": 3680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998417656810156e-05,
+      "loss": 1.6646,
+      "step": 3681
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998416772952336e-05,
+      "loss": 1.6578,
+      "step": 3682
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841588884775e-05,
+      "loss": 1.6691,
+      "step": 3683
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841500449638e-05,
+      "loss": 1.6716,
+      "step": 3684
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841411989824e-05,
+      "loss": 1.658,
+      "step": 3685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841323505333e-05,
+      "loss": 1.6474,
+      "step": 3686
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841234996164e-05,
+      "loss": 1.6518,
+      "step": 3687
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841146462319e-05,
+      "loss": 1.6826,
+      "step": 3688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999841057903795e-05,
+      "loss": 1.6797,
+      "step": 3689
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999840969320595e-05,
+      "loss": 1.6301,
+      "step": 3690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999840880712716e-05,
+      "loss": 1.6173,
+      "step": 3691
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999840792080161e-05,
+      "loss": 1.6426,
+      "step": 3692
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999840703422928e-05,
+      "loss": 1.6242,
+      "step": 3693
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998406147410176e-05,
+      "loss": 1.64,
+      "step": 3694
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99984052603443e-05,
+      "loss": 1.6469,
+      "step": 3695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998404373031646e-05,
+      "loss": 1.6164,
+      "step": 3696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998403485472225e-05,
+      "loss": 1.6305,
+      "step": 3697
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998402597666025e-05,
+      "loss": 1.6101,
+      "step": 3698
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998401709613054e-05,
+      "loss": 1.617,
+      "step": 3699
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999840082131331e-05,
+      "loss": 1.6174,
+      "step": 3700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839993276679e-05,
+      "loss": 1.6436,
+      "step": 3701
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983990439735e-05,
+      "loss": 1.6182,
+      "step": 3702
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839815493343e-05,
+      "loss": 1.5683,
+      "step": 3703
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839726564659e-05,
+      "loss": 1.559,
+      "step": 3704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839637611298e-05,
+      "loss": 1.5921,
+      "step": 3705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839548633259e-05,
+      "loss": 1.5961,
+      "step": 3706
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839459630543e-05,
+      "loss": 1.5753,
+      "step": 3707
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983937060315e-05,
+      "loss": 1.568,
+      "step": 3708
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839281551079e-05,
+      "loss": 1.5881,
+      "step": 3709
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839192474331e-05,
+      "loss": 1.5924,
+      "step": 3710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998391033729056e-05,
+      "loss": 1.5891,
+      "step": 3711
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998390142468026e-05,
+      "loss": 1.5866,
+      "step": 3712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998389250960225e-05,
+      "loss": 1.5519,
+      "step": 3713
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999838835920565e-05,
+      "loss": 1.5676,
+      "step": 3714
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983874672043e-05,
+      "loss": 1.5594,
+      "step": 3715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999838657495618e-05,
+      "loss": 1.5622,
+      "step": 3716
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999838568246128e-05,
+      "loss": 1.5621,
+      "step": 3717
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999838478971961e-05,
+      "loss": 1.5302,
+      "step": 3718
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999838389673117e-05,
+      "loss": 1.5501,
+      "step": 3719
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998383003495955e-05,
+      "loss": 1.5491,
+      "step": 3720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998382110013964e-05,
+      "loss": 1.5433,
+      "step": 3721
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998381216285195e-05,
+      "loss": 1.5156,
+      "step": 3722
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998380322309654e-05,
+      "loss": 1.5524,
+      "step": 3723
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999837942808735e-05,
+      "loss": 1.5503,
+      "step": 3724
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999837853361826e-05,
+      "loss": 1.5909,
+      "step": 3725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998377638902406e-05,
+      "loss": 1.5548,
+      "step": 3726
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998376743939774e-05,
+      "loss": 1.5458,
+      "step": 3727
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999837584873037e-05,
+      "loss": 1.5581,
+      "step": 3728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999837495327419e-05,
+      "loss": 1.5439,
+      "step": 3729
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999837405757124e-05,
+      "loss": 1.5403,
+      "step": 3730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998373161621515e-05,
+      "loss": 1.5434,
+      "step": 3731
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999837226542502e-05,
+      "loss": 1.5112,
+      "step": 3732
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999837136898175e-05,
+      "loss": 1.5424,
+      "step": 3733
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983704722917e-05,
+      "loss": 1.5404,
+      "step": 3734
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836957535488e-05,
+      "loss": 1.5589,
+      "step": 3735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836867817129e-05,
+      "loss": 1.5588,
+      "step": 3736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998367780740925e-05,
+      "loss": 1.5285,
+      "step": 3737
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836688306378e-05,
+      "loss": 1.5774,
+      "step": 3738
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836598513987e-05,
+      "loss": 1.5473,
+      "step": 3739
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836508696919e-05,
+      "loss": 1.5688,
+      "step": 3740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836418855173e-05,
+      "loss": 1.5648,
+      "step": 3741
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998363289887496e-05,
+      "loss": 1.4973,
+      "step": 3742
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836239097649e-05,
+      "loss": 1.5512,
+      "step": 3743
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998361491818716e-05,
+      "loss": 1.5802,
+      "step": 3744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999836059241416e-05,
+      "loss": 1.5288,
+      "step": 3745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999835969276284e-05,
+      "loss": 1.5351,
+      "step": 3746
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999835879286474e-05,
+      "loss": 1.5493,
+      "step": 3747
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998357892719866e-05,
+      "loss": 1.5846,
+      "step": 3748
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999835699232822e-05,
+      "loss": 1.6383,
+      "step": 3749
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998356091689804e-05,
+      "loss": 1.2697,
+      "step": 3750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998355190804616e-05,
+      "loss": 0.8748,
+      "step": 3751
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998354289672644e-05,
+      "loss": 0.9201,
+      "step": 3752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998353388293907e-05,
+      "loss": 0.8053,
+      "step": 3753
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983524866684e-05,
+      "loss": 0.4354,
+      "step": 3754
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999835158479611e-05,
+      "loss": 0.7464,
+      "step": 3755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998350682677055e-05,
+      "loss": 1.2396,
+      "step": 3756
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998349780311226e-05,
+      "loss": 1.0182,
+      "step": 3757
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999834887769862e-05,
+      "loss": 0.9524,
+      "step": 3758
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999834797483924e-05,
+      "loss": 0.3241,
+      "step": 3759
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983470717331e-05,
+      "loss": 0.3452,
+      "step": 3760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999834616838017e-05,
+      "loss": 1.3963,
+      "step": 3761
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998345264780475e-05,
+      "loss": 1.3835,
+      "step": 3762
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998344360934e-05,
+      "loss": 3.2567,
+      "step": 3763
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998343456840765e-05,
+      "loss": 2.4853,
+      "step": 3764
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999834255250075e-05,
+      "loss": 2.1915,
+      "step": 3765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999834164791396e-05,
+      "loss": 2.4211,
+      "step": 3766
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998340743080397e-05,
+      "loss": 2.2489,
+      "step": 3767
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998339838000065e-05,
+      "loss": 2.1991,
+      "step": 3768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999833893267295e-05,
+      "loss": 2.1998,
+      "step": 3769
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998338027099074e-05,
+      "loss": 2.2579,
+      "step": 3770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998337121278414e-05,
+      "loss": 2.0534,
+      "step": 3771
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998336215211e-05,
+      "loss": 2.1723,
+      "step": 3772
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998335308896796e-05,
+      "loss": 1.879,
+      "step": 3773
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999833440233582e-05,
+      "loss": 1.6274,
+      "step": 3774
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999833349552807e-05,
+      "loss": 1.8824,
+      "step": 3775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999833258847355e-05,
+      "loss": 1.5347,
+      "step": 3776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999833168117226e-05,
+      "loss": 1.5394,
+      "step": 3777
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998330773624194e-05,
+      "loss": 1.6869,
+      "step": 3778
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999832986582935e-05,
+      "loss": 1.8133,
+      "step": 3779
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998328957787745e-05,
+      "loss": 1.4076,
+      "step": 3780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999832804949936e-05,
+      "loss": 1.6906,
+      "step": 3781
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998327140964205e-05,
+      "loss": 1.7414,
+      "step": 3782
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999832623218227e-05,
+      "loss": 1.3289,
+      "step": 3783
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999832532315357e-05,
+      "loss": 1.5075,
+      "step": 3784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983244138781e-05,
+      "loss": 1.2666,
+      "step": 3785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999832350435584e-05,
+      "loss": 1.4103,
+      "step": 3786
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998322594586824e-05,
+      "loss": 1.4504,
+      "step": 3787
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999832168457103e-05,
+      "loss": 1.4091,
+      "step": 3788
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999832077430846e-05,
+      "loss": 1.3284,
+      "step": 3789
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998319863799115e-05,
+      "loss": 1.4356,
+      "step": 3790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998318953043e-05,
+      "loss": 1.2981,
+      "step": 3791
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999831804204012e-05,
+      "loss": 1.5094,
+      "step": 3792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998317130790464e-05,
+      "loss": 1.5671,
+      "step": 3793
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999831621929403e-05,
+      "loss": 1.4988,
+      "step": 3794
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998315307550826e-05,
+      "loss": 1.4861,
+      "step": 3795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999831439556084e-05,
+      "loss": 1.5498,
+      "step": 3796
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999831348332409e-05,
+      "loss": 1.2615,
+      "step": 3797
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999831257084057e-05,
+      "loss": 1.5546,
+      "step": 3798
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998311658110274e-05,
+      "loss": 1.5435,
+      "step": 3799
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983107451332e-05,
+      "loss": 1.6826,
+      "step": 3800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999830983190936e-05,
+      "loss": 1.4174,
+      "step": 3801
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999830891843875e-05,
+      "loss": 1.6398,
+      "step": 3802
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999830800472136e-05,
+      "loss": 1.4567,
+      "step": 3803
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99983070907572e-05,
+      "loss": 1.6013,
+      "step": 3804
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998306176546266e-05,
+      "loss": 1.7008,
+      "step": 3805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999830526208856e-05,
+      "loss": 1.6242,
+      "step": 3806
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998304347384085e-05,
+      "loss": 1.5235,
+      "step": 3807
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999830343243283e-05,
+      "loss": 1.3879,
+      "step": 3808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998302517234805e-05,
+      "loss": 1.4885,
+      "step": 3809
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999830160179001e-05,
+      "loss": 1.5374,
+      "step": 3810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999830068609844e-05,
+      "loss": 1.377,
+      "step": 3811
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998299770160095e-05,
+      "loss": 1.2827,
+      "step": 3812
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998298853974985e-05,
+      "loss": 1.3945,
+      "step": 3813
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982979375431e-05,
+      "loss": 1.3208,
+      "step": 3814
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999829702086444e-05,
+      "loss": 1.276,
+      "step": 3815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998296103939e-05,
+      "loss": 1.1584,
+      "step": 3816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982951867668e-05,
+      "loss": 1.002,
+      "step": 3817
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998294269347826e-05,
+      "loss": 0.9674,
+      "step": 3818
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999829335168207e-05,
+      "loss": 0.9041,
+      "step": 3819
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999829243376955e-05,
+      "loss": 0.6312,
+      "step": 3820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998291515610254e-05,
+      "loss": 0.6914,
+      "step": 3821
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999829059720418e-05,
+      "loss": 0.5184,
+      "step": 3822
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828967855134e-05,
+      "loss": 0.3013,
+      "step": 3823
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998288759651727e-05,
+      "loss": 0.2505,
+      "step": 3824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828784050534e-05,
+      "loss": 0.2985,
+      "step": 3825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828692111218e-05,
+      "loss": 0.899,
+      "step": 3826
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828600147225e-05,
+      "loss": 1.2663,
+      "step": 3827
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828508158555e-05,
+      "loss": 1.2636,
+      "step": 3828
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828416145207e-05,
+      "loss": 1.131,
+      "step": 3829
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828324107182e-05,
+      "loss": 1.1377,
+      "step": 3830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982823204448e-05,
+      "loss": 1.1853,
+      "step": 3831
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998281399571015e-05,
+      "loss": 1.2119,
+      "step": 3832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999828047845044e-05,
+      "loss": 1.1701,
+      "step": 3833
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827955708311e-05,
+      "loss": 1.2334,
+      "step": 3834
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827863546899e-05,
+      "loss": 0.8946,
+      "step": 3835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827771360811e-05,
+      "loss": 1.0699,
+      "step": 3836
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827679150045e-05,
+      "loss": 1.108,
+      "step": 3837
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827586914602e-05,
+      "loss": 1.0536,
+      "step": 3838
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827494654482e-05,
+      "loss": 1.1607,
+      "step": 3839
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998274023696845e-05,
+      "loss": 1.0009,
+      "step": 3840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982731006021e-05,
+      "loss": 1.0169,
+      "step": 3841
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998272177260586e-05,
+      "loss": 1.2903,
+      "step": 3842
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827125367229e-05,
+      "loss": 1.0584,
+      "step": 3843
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827032983723e-05,
+      "loss": 1.0524,
+      "step": 3844
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999826940575539e-05,
+      "loss": 0.9983,
+      "step": 3845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998268481426786e-05,
+      "loss": 0.9049,
+      "step": 3846
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999826755685141e-05,
+      "loss": 1.0455,
+      "step": 3847
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999826663202926e-05,
+      "loss": 1.2578,
+      "step": 3848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998265706960325e-05,
+      "loss": 1.1371,
+      "step": 3849
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999826478164463e-05,
+      "loss": 1.0095,
+      "step": 3850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998263856082164e-05,
+      "loss": 1.1335,
+      "step": 3851
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999826293027292e-05,
+      "loss": 1.1164,
+      "step": 3852
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982620042169e-05,
+      "loss": 0.9897,
+      "step": 3853
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998261077914114e-05,
+      "loss": 1.1324,
+      "step": 3854
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999826015136456e-05,
+      "loss": 0.9199,
+      "step": 3855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999825922456822e-05,
+      "loss": 0.6164,
+      "step": 3856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998258297525124e-05,
+      "loss": 1.0912,
+      "step": 3857
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999825737023525e-05,
+      "loss": 1.1302,
+      "step": 3858
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982564426986e-05,
+      "loss": 0.8905,
+      "step": 3859
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999825551491518e-05,
+      "loss": 0.9626,
+      "step": 3860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998254586884985e-05,
+      "loss": 0.9578,
+      "step": 3861
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999825365860802e-05,
+      "loss": 0.9679,
+      "step": 3862
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999825273008428e-05,
+      "loss": 1.0489,
+      "step": 3863
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999825180131377e-05,
+      "loss": 0.7817,
+      "step": 3864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999825087229649e-05,
+      "loss": 0.5133,
+      "step": 3865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998249943032435e-05,
+      "loss": 0.5381,
+      "step": 3866
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999824901352161e-05,
+      "loss": 0.5507,
+      "step": 3867
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999824808376401e-05,
+      "loss": 0.8889,
+      "step": 3868
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999824715375965e-05,
+      "loss": 0.5353,
+      "step": 3869
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998246223508506e-05,
+      "loss": 0.484,
+      "step": 3870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998245293010585e-05,
+      "loss": 0.3894,
+      "step": 3871
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982443622659e-05,
+      "loss": 0.3145,
+      "step": 3872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999824343127444e-05,
+      "loss": 1.4916,
+      "step": 3873
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999824250003621e-05,
+      "loss": 0.9999,
+      "step": 3874
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998241568551205e-05,
+      "loss": 0.8343,
+      "step": 3875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998240636819436e-05,
+      "loss": 1.1243,
+      "step": 3876
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823970484089e-05,
+      "loss": 1.0399,
+      "step": 3877
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823877261557e-05,
+      "loss": 0.8224,
+      "step": 3878
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823784014347e-05,
+      "loss": 1.0171,
+      "step": 3879
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823690742461e-05,
+      "loss": 1.9519,
+      "step": 3880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823597445897e-05,
+      "loss": 1.0492,
+      "step": 3881
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823504124657e-05,
+      "loss": 1.2882,
+      "step": 3882
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823410778739e-05,
+      "loss": 1.1063,
+      "step": 3883
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998233174081434e-05,
+      "loss": 0.9714,
+      "step": 3884
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823224012871e-05,
+      "loss": 0.9269,
+      "step": 3885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998231305929215e-05,
+      "loss": 1.3882,
+      "step": 3886
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999823037148295e-05,
+      "loss": 1.2153,
+      "step": 3887
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998229436789905e-05,
+      "loss": 0.9109,
+      "step": 3888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998228501850096e-05,
+      "loss": 0.9884,
+      "step": 3889
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998227566663516e-05,
+      "loss": 1.0662,
+      "step": 3890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999822663123016e-05,
+      "loss": 0.9719,
+      "step": 3891
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999822569555003e-05,
+      "loss": 1.3314,
+      "step": 3892
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999822475962313e-05,
+      "loss": 1.1694,
+      "step": 3893
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999822382344946e-05,
+      "loss": 1.3414,
+      "step": 3894
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999822288702901e-05,
+      "loss": 1.2034,
+      "step": 3895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982219503618e-05,
+      "loss": 1.2633,
+      "step": 3896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999822101344781e-05,
+      "loss": 1.0826,
+      "step": 3897
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999822007628705e-05,
+      "loss": 0.9299,
+      "step": 3898
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999821913887952e-05,
+      "loss": 1.712,
+      "step": 3899
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998218201225214e-05,
+      "loss": 1.2466,
+      "step": 3900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998217263324144e-05,
+      "loss": 1.0934,
+      "step": 3901
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998216325176296e-05,
+      "loss": 0.7847,
+      "step": 3902
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999821538678168e-05,
+      "loss": 1.2325,
+      "step": 3903
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999821444814029e-05,
+      "loss": 1.2659,
+      "step": 3904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998213509252126e-05,
+      "loss": 0.9979,
+      "step": 3905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999821257011719e-05,
+      "loss": 1.2732,
+      "step": 3906
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999821163073548e-05,
+      "loss": 0.6001,
+      "step": 3907
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999821069110701e-05,
+      "loss": 1.0395,
+      "step": 3908
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998209751231755e-05,
+      "loss": 0.9985,
+      "step": 3909
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820881110974e-05,
+      "loss": 0.8775,
+      "step": 3910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820787074094e-05,
+      "loss": 1.2523,
+      "step": 3911
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820693012538e-05,
+      "loss": 1.4449,
+      "step": 3912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820598926304e-05,
+      "loss": 1.343,
+      "step": 3913
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820504815394e-05,
+      "loss": 1.4019,
+      "step": 3914
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820410679805e-05,
+      "loss": 1.1844,
+      "step": 3915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99982031651954e-05,
+      "loss": 1.3271,
+      "step": 3916
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820222334598e-05,
+      "loss": 1.2497,
+      "step": 3917
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998201281249776e-05,
+      "loss": 0.9437,
+      "step": 3918
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999820033890681e-05,
+      "loss": 0.9859,
+      "step": 3919
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998199396317076e-05,
+      "loss": 1.2831,
+      "step": 3920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999819845348056e-05,
+      "loss": 1.0626,
+      "step": 3921
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998197510397284e-05,
+      "loss": 1.1052,
+      "step": 3922
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999819656706722e-05,
+      "loss": 0.8837,
+      "step": 3923
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99981956234904e-05,
+      "loss": 1.1141,
+      "step": 3924
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99981946796668e-05,
+      "loss": 0.9994,
+      "step": 3925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999819373559643e-05,
+      "loss": 1.1416,
+      "step": 3926
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99981927912793e-05,
+      "loss": 1.0897,
+      "step": 3927
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999819184671538e-05,
+      "loss": 1.2305,
+      "step": 3928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99981909019047e-05,
+      "loss": 1.2005,
+      "step": 3929
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818995684724e-05,
+      "loss": 0.9295,
+      "step": 3930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998189011543014e-05,
+      "loss": 1.2663,
+      "step": 3931
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818806599202e-05,
+      "loss": 1.0269,
+      "step": 3932
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818712019425e-05,
+      "loss": 1.0723,
+      "step": 3933
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818617414971e-05,
+      "loss": 1.3211,
+      "step": 3934
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99981852278584e-05,
+      "loss": 1.2296,
+      "step": 3935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818428132031e-05,
+      "loss": 1.2031,
+      "step": 3936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818333453545e-05,
+      "loss": 0.8096,
+      "step": 3937
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818238750382e-05,
+      "loss": 1.1207,
+      "step": 3938
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998181440225425e-05,
+      "loss": 1.106,
+      "step": 3939
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999818049270025e-05,
+      "loss": 1.1441,
+      "step": 3940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998179544928314e-05,
+      "loss": 1.1043,
+      "step": 3941
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99981785969096e-05,
+      "loss": 0.7397,
+      "step": 3942
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999817764864411e-05,
+      "loss": 0.4566,
+      "step": 3943
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999817670013186e-05,
+      "loss": 0.3368,
+      "step": 3944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999817575137283e-05,
+      "loss": 0.2477,
+      "step": 3945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998174802367024e-05,
+      "loss": 0.1339,
+      "step": 3946
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999817385311446e-05,
+      "loss": 0.1126,
+      "step": 3947
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998172903615116e-05,
+      "loss": 1.0654,
+      "step": 3948
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998171953869e-05,
+      "loss": 1.0436,
+      "step": 3949
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999817100387612e-05,
+      "loss": 1.3441,
+      "step": 3950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999817005363646e-05,
+      "loss": 1.4435,
+      "step": 3951
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998169103150026e-05,
+      "loss": 1.1704,
+      "step": 3952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999816815241683e-05,
+      "loss": 0.2951,
+      "step": 3953
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998167201436864e-05,
+      "loss": 1.0192,
+      "step": 3954
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998166250210116e-05,
+      "loss": 1.0236,
+      "step": 3955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998165298736604e-05,
+      "loss": 1.1952,
+      "step": 3956
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999816434701632e-05,
+      "loss": 1.1598,
+      "step": 3957
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999816339504926e-05,
+      "loss": 1.2215,
+      "step": 3958
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999816244283543e-05,
+      "loss": 1.4891,
+      "step": 3959
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998161490374836e-05,
+      "loss": 1.1688,
+      "step": 3960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999816053766746e-05,
+      "loss": 1.1998,
+      "step": 3961
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815958471332e-05,
+      "loss": 1.111,
+      "step": 3962
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815863151241e-05,
+      "loss": 1.1179,
+      "step": 3963
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815767806473e-05,
+      "loss": 1.0286,
+      "step": 3964
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815672437027e-05,
+      "loss": 1.1219,
+      "step": 3965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998155770429045e-05,
+      "loss": 1.2234,
+      "step": 3966
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815481624104e-05,
+      "loss": 1.1453,
+      "step": 3967
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998153861806276e-05,
+      "loss": 1.1588,
+      "step": 3968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815290712473e-05,
+      "loss": 0.6494,
+      "step": 3969
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815195219643e-05,
+      "loss": 0.6897,
+      "step": 3970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999815099702134e-05,
+      "loss": 1.1793,
+      "step": 3971
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998150041599484e-05,
+      "loss": 1.0342,
+      "step": 3972
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998149085930854e-05,
+      "loss": 1.1482,
+      "step": 3973
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814813001547e-05,
+      "loss": 1.2897,
+      "step": 3974
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998147173853296e-05,
+      "loss": 0.7381,
+      "step": 3975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814621744435e-05,
+      "loss": 1.1291,
+      "step": 3976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998145260788645e-05,
+      "loss": 1.1105,
+      "step": 3977
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814430388616e-05,
+      "loss": 1.2495,
+      "step": 3978
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814334673692e-05,
+      "loss": 1.0232,
+      "step": 3979
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814238934088e-05,
+      "loss": 1.4272,
+      "step": 3980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814143169809e-05,
+      "loss": 1.3486,
+      "step": 3981
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999814047380853e-05,
+      "loss": 1.285,
+      "step": 3982
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813951567219e-05,
+      "loss": 1.105,
+      "step": 3983
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813855728908e-05,
+      "loss": 0.9492,
+      "step": 3984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998137598659204e-05,
+      "loss": 1.1968,
+      "step": 3985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813663978255e-05,
+      "loss": 0.9647,
+      "step": 3986
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813568065913e-05,
+      "loss": 1.1586,
+      "step": 3987
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813472128894e-05,
+      "loss": 1.2225,
+      "step": 3988
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813376167197e-05,
+      "loss": 1.072,
+      "step": 3989
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813280180824e-05,
+      "loss": 1.1902,
+      "step": 3990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813184169774e-05,
+      "loss": 1.1988,
+      "step": 3991
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999813088134045e-05,
+      "loss": 1.206,
+      "step": 3992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998129920736405e-05,
+      "loss": 1.1791,
+      "step": 3993
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998128959885595e-05,
+      "loss": 1.5142,
+      "step": 3994
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998127998788006e-05,
+      "loss": 1.4691,
+      "step": 3995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998127037443646e-05,
+      "loss": 1.4362,
+      "step": 3996
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999812607585251e-05,
+      "loss": 1.2838,
+      "step": 3997
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998125114014606e-05,
+      "loss": 0.9268,
+      "step": 3998
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999812415192994e-05,
+      "loss": 0.9915,
+      "step": 3999
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999812318959849e-05,
+      "loss": 1.1468,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.09718918800354,
+      "eval_runtime": 106.6196,
+      "eval_samples_per_second": 12.99,
+      "eval_steps_per_second": 3.255,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999812222702027e-05,
+      "loss": 0.8135,
+      "step": 4001
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998121264195284e-05,
+      "loss": 1.2887,
+      "step": 4002
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999812030112353e-05,
+      "loss": 0.9391,
+      "step": 4003
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998119337805e-05,
+      "loss": 1.1028,
+      "step": 4004
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99981183742397e-05,
+      "loss": 1.8483,
+      "step": 4005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999811741042764e-05,
+      "loss": 0.5112,
+      "step": 4006
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998116446368794e-05,
+      "loss": 0.9238,
+      "step": 4007
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999811548206318e-05,
+      "loss": 1.0704,
+      "step": 4008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998114517510795e-05,
+      "loss": 0.8911,
+      "step": 4009
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998113552711645e-05,
+      "loss": 0.7326,
+      "step": 4010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999811258766572e-05,
+      "loss": 0.5943,
+      "step": 4011
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998111622373025e-05,
+      "loss": 0.9056,
+      "step": 4012
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998110656833555e-05,
+      "loss": 1.0,
+      "step": 4013
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810969104733e-05,
+      "loss": 1.0892,
+      "step": 4014
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998108725014314e-05,
+      "loss": 0.989,
+      "step": 4015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810775873454e-05,
+      "loss": 1.1427,
+      "step": 4016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810679220799e-05,
+      "loss": 1.3094,
+      "step": 4017
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810582543466e-05,
+      "loss": 1.5354,
+      "step": 4018
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810485841457e-05,
+      "loss": 1.2071,
+      "step": 4019
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810389114772e-05,
+      "loss": 1.0141,
+      "step": 4020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998102923634084e-05,
+      "loss": 1.1229,
+      "step": 4021
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810195587368e-05,
+      "loss": 1.4388,
+      "step": 4022
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998100987866505e-05,
+      "loss": 1.651,
+      "step": 4023
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999810001961256e-05,
+      "loss": 1.1134,
+      "step": 4024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999809905111185e-05,
+      "loss": 1.1017,
+      "step": 4025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998098082364365e-05,
+      "loss": 1.0174,
+      "step": 4026
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980971133701e-05,
+      "loss": 0.3147,
+      "step": 4027
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999809614412908e-05,
+      "loss": 1.336,
+      "step": 4028
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999809517464128e-05,
+      "loss": 0.9984,
+      "step": 4029
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999809420490671e-05,
+      "loss": 0.9577,
+      "step": 4030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998093234925366e-05,
+      "loss": 1.3655,
+      "step": 4031
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998092264697264e-05,
+      "loss": 0.8947,
+      "step": 4032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999809129422238e-05,
+      "loss": 1.1589,
+      "step": 4033
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999809032350073e-05,
+      "loss": 1.2562,
+      "step": 4034
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999808935253231e-05,
+      "loss": 0.9964,
+      "step": 4035
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999808838131712e-05,
+      "loss": 1.249,
+      "step": 4036
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998087409855156e-05,
+      "loss": 1.2319,
+      "step": 4037
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999808643814642e-05,
+      "loss": 1.2463,
+      "step": 4038
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999808546619091e-05,
+      "loss": 1.2636,
+      "step": 4039
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999808449398864e-05,
+      "loss": 1.3586,
+      "step": 4040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980835215396e-05,
+      "loss": 1.1718,
+      "step": 4041
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999808254884378e-05,
+      "loss": 1.187,
+      "step": 4042
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980815759012e-05,
+      "loss": 1.0817,
+      "step": 4043
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999808060271184e-05,
+      "loss": 1.162,
+      "step": 4044
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998079629275705e-05,
+      "loss": 1.1909,
+      "step": 4045
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807865559282e-05,
+      "loss": 1.1345,
+      "step": 4046
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807768166315e-05,
+      "loss": 0.9697,
+      "step": 4047
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807670748671e-05,
+      "loss": 0.9829,
+      "step": 4048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980757330635e-05,
+      "loss": 1.131,
+      "step": 4049
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807475839352e-05,
+      "loss": 1.06,
+      "step": 4050
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807378347677e-05,
+      "loss": 1.0493,
+      "step": 4051
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807280831324e-05,
+      "loss": 0.7613,
+      "step": 4052
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807183290296e-05,
+      "loss": 3.0747,
+      "step": 4053
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807085724589e-05,
+      "loss": 4.4983,
+      "step": 4054
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998069881342064e-05,
+      "loss": 4.1956,
+      "step": 4055
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999806890519146e-05,
+      "loss": 4.0042,
+      "step": 4056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999806792879409e-05,
+      "loss": 3.661,
+      "step": 4057
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998066952149945e-05,
+      "loss": 3.8345,
+      "step": 4058
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998065975259034e-05,
+      "loss": 3.6504,
+      "step": 4059
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998064998121344e-05,
+      "loss": 1.5604,
+      "step": 4060
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999806402073689e-05,
+      "loss": 0.8872,
+      "step": 4061
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998063043105666e-05,
+      "loss": 0.9651,
+      "step": 4062
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999806206522767e-05,
+      "loss": 1.0188,
+      "step": 4063
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999806108710291e-05,
+      "loss": 1.4743,
+      "step": 4064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999806010873137e-05,
+      "loss": 1.2101,
+      "step": 4065
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805913011307e-05,
+      "loss": 1.4814,
+      "step": 4066
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805815124799e-05,
+      "loss": 0.996,
+      "step": 4067
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998057172136144e-05,
+      "loss": 0.9714,
+      "step": 4068
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805619277753e-05,
+      "loss": 0.9029,
+      "step": 4069
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805521317214e-05,
+      "loss": 1.2617,
+      "step": 4070
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805423331998e-05,
+      "loss": 0.703,
+      "step": 4071
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805325322106e-05,
+      "loss": 1.2638,
+      "step": 4072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805227287536e-05,
+      "loss": 1.2354,
+      "step": 4073
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998051292282885e-05,
+      "loss": 1.0022,
+      "step": 4074
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999805031144365e-05,
+      "loss": 1.1542,
+      "step": 4075
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999804933035764e-05,
+      "loss": 0.9708,
+      "step": 4076
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999804834902486e-05,
+      "loss": 1.0911,
+      "step": 4077
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999804736744531e-05,
+      "loss": 1.4731,
+      "step": 4078
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998046385619e-05,
+      "loss": 1.1589,
+      "step": 4079
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998045403545904e-05,
+      "loss": 1.025,
+      "step": 4080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999804442122604e-05,
+      "loss": 0.9713,
+      "step": 4081
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999804343865941e-05,
+      "loss": 1.3273,
+      "step": 4082
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999804245584602e-05,
+      "loss": 1.2429,
+      "step": 4083
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999804147278585e-05,
+      "loss": 0.9776,
+      "step": 4084
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980404894789e-05,
+      "loss": 0.8755,
+      "step": 4085
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980395059252e-05,
+      "loss": 1.3534,
+      "step": 4086
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803852212471e-05,
+      "loss": 1.581,
+      "step": 4087
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803753807747e-05,
+      "loss": 2.4373,
+      "step": 4088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803655378344e-05,
+      "loss": 3.114,
+      "step": 4089
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803556924265e-05,
+      "loss": 1.136,
+      "step": 4090
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998034584455096e-05,
+      "loss": 1.281,
+      "step": 4091
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803359942076e-05,
+      "loss": 1.1828,
+      "step": 4092
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803261413966e-05,
+      "loss": 1.0111,
+      "step": 4093
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803162861179e-05,
+      "loss": 1.7322,
+      "step": 4094
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999803064283715e-05,
+      "loss": 1.4425,
+      "step": 4095
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999802965681574e-05,
+      "loss": 1.3707,
+      "step": 4096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998028670547556e-05,
+      "loss": 1.4249,
+      "step": 4097
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999802768403261e-05,
+      "loss": 1.0745,
+      "step": 4098
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999802669727089e-05,
+      "loss": 0.8969,
+      "step": 4099
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998025710262395e-05,
+      "loss": 1.0658,
+      "step": 4100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998024723007134e-05,
+      "loss": 0.9743,
+      "step": 4101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999802373550511e-05,
+      "loss": 1.1153,
+      "step": 4102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980227477563e-05,
+      "loss": 1.1346,
+      "step": 4103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999802175976073e-05,
+      "loss": 0.8344,
+      "step": 4104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998020771518394e-05,
+      "loss": 0.1926,
+      "step": 4105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998019783029284e-05,
+      "loss": 0.2064,
+      "step": 4106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99980187942934e-05,
+      "loss": 0.3738,
+      "step": 4107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801780531075e-05,
+      "loss": 0.3192,
+      "step": 4108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801681608133e-05,
+      "loss": 0.3141,
+      "step": 4109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801582660514e-05,
+      "loss": 0.302,
+      "step": 4110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801483688218e-05,
+      "loss": 0.3193,
+      "step": 4111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801384691245e-05,
+      "loss": 0.2444,
+      "step": 4112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801285669595e-05,
+      "loss": 0.2959,
+      "step": 4113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801186623268e-05,
+      "loss": 0.1291,
+      "step": 4114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999801087552264e-05,
+      "loss": 0.3326,
+      "step": 4115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998009884565834e-05,
+      "loss": 1.5303,
+      "step": 4116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998008893362254e-05,
+      "loss": 0.9613,
+      "step": 4117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998007901911904e-05,
+      "loss": 1.2479,
+      "step": 4118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999800691021479e-05,
+      "loss": 1.2053,
+      "step": 4119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998005918270897e-05,
+      "loss": 0.9582,
+      "step": 4120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999800492608024e-05,
+      "loss": 1.4681,
+      "step": 4121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999800393364281e-05,
+      "loss": 1.3266,
+      "step": 4122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999800294095861e-05,
+      "loss": 1.107,
+      "step": 4123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999800194802765e-05,
+      "loss": 1.2981,
+      "step": 4124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999800095484991e-05,
+      "loss": 1.0812,
+      "step": 4125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99979999614254e-05,
+      "loss": 0.9215,
+      "step": 4126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799896775412e-05,
+      "loss": 1.056,
+      "step": 4127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799797383607e-05,
+      "loss": 0.9254,
+      "step": 4128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799697967126e-05,
+      "loss": 1.2472,
+      "step": 4129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799598525967e-05,
+      "loss": 0.6347,
+      "step": 4130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799499060132e-05,
+      "loss": 0.6053,
+      "step": 4131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799399569619e-05,
+      "loss": 0.8091,
+      "step": 4132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997993000544295e-05,
+      "loss": 1.1355,
+      "step": 4133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999799200514563e-05,
+      "loss": 1.5105,
+      "step": 4134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99979910095002e-05,
+      "loss": 1.204,
+      "step": 4135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997990013607995e-05,
+      "loss": 1.1201,
+      "step": 4136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997989017469026e-05,
+      "loss": 1.3261,
+      "step": 4137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999798802108328e-05,
+      "loss": 1.1437,
+      "step": 4138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999798702445077e-05,
+      "loss": 1.2163,
+      "step": 4139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997986027571485e-05,
+      "loss": 1.0872,
+      "step": 4140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999798503044543e-05,
+      "loss": 1.0482,
+      "step": 4141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997984033072606e-05,
+      "loss": 1.1133,
+      "step": 4142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999798303545302e-05,
+      "loss": 2.1336,
+      "step": 4143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999798203758666e-05,
+      "loss": 1.3565,
+      "step": 4144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999798103947353e-05,
+      "loss": 0.8247,
+      "step": 4145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997980041113634e-05,
+      "loss": 1.0643,
+      "step": 4146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999797904250696e-05,
+      "loss": 1.3419,
+      "step": 4147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999797804365353e-05,
+      "loss": 1.2586,
+      "step": 4148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999797704455331e-05,
+      "loss": 2.8076,
+      "step": 4149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997976045206344e-05,
+      "loss": 1.7105,
+      "step": 4150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999797504561259e-05,
+      "loss": 1.0705,
+      "step": 4151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997974045772075e-05,
+      "loss": 1.1643,
+      "step": 4152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999797304568479e-05,
+      "loss": 1.1731,
+      "step": 4153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997972045350735e-05,
+      "loss": 1.028,
+      "step": 4154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999797104476991e-05,
+      "loss": 1.0386,
+      "step": 4155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999797004394231e-05,
+      "loss": 1.1987,
+      "step": 4156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999796904286795e-05,
+      "loss": 1.2542,
+      "step": 4157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997968041546814e-05,
+      "loss": 1.1764,
+      "step": 4158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999796703997891e-05,
+      "loss": 1.6205,
+      "step": 4159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999796603816424e-05,
+      "loss": 1.5423,
+      "step": 4160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99979650361028e-05,
+      "loss": 1.539,
+      "step": 4161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999796403379459e-05,
+      "loss": 1.297,
+      "step": 4162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999796303123961e-05,
+      "loss": 1.1458,
+      "step": 4163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999796202843786e-05,
+      "loss": 1.2349,
+      "step": 4164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997961025389336e-05,
+      "loss": 1.3741,
+      "step": 4165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999796002209405e-05,
+      "loss": 1.2188,
+      "step": 4166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999795901855199e-05,
+      "loss": 1.3562,
+      "step": 4167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999795801476317e-05,
+      "loss": 0.3673,
+      "step": 4168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999795701072757e-05,
+      "loss": 0.2597,
+      "step": 4169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99979560064452e-05,
+      "loss": 0.6811,
+      "step": 4170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999795500191607e-05,
+      "loss": 1.1625,
+      "step": 4171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997953997140165e-05,
+      "loss": 1.1949,
+      "step": 4172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997952992117494e-05,
+      "loss": 1.0916,
+      "step": 4173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997951986848045e-05,
+      "loss": 0.7343,
+      "step": 4174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999795098133184e-05,
+      "loss": 0.7105,
+      "step": 4175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999794997556886e-05,
+      "loss": 0.6363,
+      "step": 4176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997948969559105e-05,
+      "loss": 0.6748,
+      "step": 4177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997947963302585e-05,
+      "loss": 0.8273,
+      "step": 4178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999794695679929e-05,
+      "loss": 1.3885,
+      "step": 4179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999794595004924e-05,
+      "loss": 1.1066,
+      "step": 4180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999794494305241e-05,
+      "loss": 1.0197,
+      "step": 4181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999794393580882e-05,
+      "loss": 1.0707,
+      "step": 4182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999794292831845e-05,
+      "loss": 1.0189,
+      "step": 4183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997941920581316e-05,
+      "loss": 1.2503,
+      "step": 4184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999794091259741e-05,
+      "loss": 1.1458,
+      "step": 4185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999793990436674e-05,
+      "loss": 1.1863,
+      "step": 4186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997938895889294e-05,
+      "loss": 1.2241,
+      "step": 4187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999793788716508e-05,
+      "loss": 0.8661,
+      "step": 4188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997936878194107e-05,
+      "loss": 0.5257,
+      "step": 4189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999793586897635e-05,
+      "loss": 0.9842,
+      "step": 4190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999793485951184e-05,
+      "loss": 1.1282,
+      "step": 4191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999793384980055e-05,
+      "loss": 1.2302,
+      "step": 4192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999793283984249e-05,
+      "loss": 1.2022,
+      "step": 4193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997931829637665e-05,
+      "loss": 1.1716,
+      "step": 4194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997930819186075e-05,
+      "loss": 1.0513,
+      "step": 4195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999792980848771e-05,
+      "loss": 1.0939,
+      "step": 4196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997928797542575e-05,
+      "loss": 1.15,
+      "step": 4197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999792778635067e-05,
+      "loss": 1.2474,
+      "step": 4198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997926774912004e-05,
+      "loss": 1.2844,
+      "step": 4199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999792576322656e-05,
+      "loss": 1.1551,
+      "step": 4200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997924751294355e-05,
+      "loss": 1.2846,
+      "step": 4201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997923739115374e-05,
+      "loss": 0.9903,
+      "step": 4202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999792272668963e-05,
+      "loss": 1.1966,
+      "step": 4203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999792171401712e-05,
+      "loss": 1.275,
+      "step": 4204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997920701097836e-05,
+      "loss": 1.0189,
+      "step": 4205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999791968793178e-05,
+      "loss": 1.1909,
+      "step": 4206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997918674518953e-05,
+      "loss": 1.2318,
+      "step": 4207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999791766085936e-05,
+      "loss": 1.2186,
+      "step": 4208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997916646953e-05,
+      "loss": 0.8141,
+      "step": 4209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997915632799876e-05,
+      "loss": 1.029,
+      "step": 4210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997914618399974e-05,
+      "loss": 1.1777,
+      "step": 4211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999791360375331e-05,
+      "loss": 1.1809,
+      "step": 4212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999791258885988e-05,
+      "loss": 1.2198,
+      "step": 4213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999791157371967e-05,
+      "loss": 1.2193,
+      "step": 4214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997910558332696e-05,
+      "loss": 0.9905,
+      "step": 4215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999790954269895e-05,
+      "loss": 1.2994,
+      "step": 4216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997908526818436e-05,
+      "loss": 1.0537,
+      "step": 4217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999790751069116e-05,
+      "loss": 1.5454,
+      "step": 4218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999790649431711e-05,
+      "loss": 1.1544,
+      "step": 4219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99979054776963e-05,
+      "loss": 0.9599,
+      "step": 4220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997904460828704e-05,
+      "loss": 1.0451,
+      "step": 4221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999790344371435e-05,
+      "loss": 1.054,
+      "step": 4222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997902426353224e-05,
+      "loss": 1.161,
+      "step": 4223
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999790140874534e-05,
+      "loss": 0.8822,
+      "step": 4224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999790039089067e-05,
+      "loss": 1.0428,
+      "step": 4225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997899372789244e-05,
+      "loss": 1.132,
+      "step": 4226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997898354441044e-05,
+      "loss": 1.0825,
+      "step": 4227
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999789733584608e-05,
+      "loss": 1.1674,
+      "step": 4228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999789631700434e-05,
+      "loss": 1.1649,
+      "step": 4229
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999789529791583e-05,
+      "loss": 0.691,
+      "step": 4230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999789427858056e-05,
+      "loss": 1.1995,
+      "step": 4231
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997893258998517e-05,
+      "loss": 0.9215,
+      "step": 4232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999789223916971e-05,
+      "loss": 1.2847,
+      "step": 4233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999789121909413e-05,
+      "loss": 1.1242,
+      "step": 4234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999789019877178e-05,
+      "loss": 0.6223,
+      "step": 4235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999788917820266e-05,
+      "loss": 0.8613,
+      "step": 4236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999788815738678e-05,
+      "loss": 0.9802,
+      "step": 4237
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999788713632412e-05,
+      "loss": 1.3161,
+      "step": 4238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99978861150147e-05,
+      "loss": 1.0592,
+      "step": 4239
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999788509345851e-05,
+      "loss": 1.2151,
+      "step": 4240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997884071655546e-05,
+      "loss": 1.0687,
+      "step": 4241
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999788304960582e-05,
+      "loss": 1.2194,
+      "step": 4242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999788202730932e-05,
+      "loss": 1.1696,
+      "step": 4243
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999788100476606e-05,
+      "loss": 0.9879,
+      "step": 4244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997879981976016e-05,
+      "loss": 0.6206,
+      "step": 4245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999787895893922e-05,
+      "loss": 1.3536,
+      "step": 4246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999787793565565e-05,
+      "loss": 1.2582,
+      "step": 4247
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999787691212531e-05,
+      "loss": 1.3572,
+      "step": 4248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997875888348195e-05,
+      "loss": 1.0565,
+      "step": 4249
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999787486432432e-05,
+      "loss": 1.3061,
+      "step": 4250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999787384005367e-05,
+      "loss": 1.1021,
+      "step": 4251
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999787281553626e-05,
+      "loss": 0.7918,
+      "step": 4252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997871790772075e-05,
+      "loss": 1.0945,
+      "step": 4253
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999787076576112e-05,
+      "loss": 1.3962,
+      "step": 4254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786974050341e-05,
+      "loss": 1.0041,
+      "step": 4255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786871499892e-05,
+      "loss": 0.9295,
+      "step": 4256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997867689247664e-05,
+      "loss": 1.2329,
+      "step": 4257
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786666324964e-05,
+      "loss": 1.1603,
+      "step": 4258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786563700485e-05,
+      "loss": 1.1955,
+      "step": 4259
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786461051329e-05,
+      "loss": 1.642,
+      "step": 4260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786358377495e-05,
+      "loss": 1.9116,
+      "step": 4261
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786255678985e-05,
+      "loss": 1.9218,
+      "step": 4262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786152955799e-05,
+      "loss": 1.4183,
+      "step": 4263
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999786050207935e-05,
+      "loss": 1.0229,
+      "step": 4264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999785947435395e-05,
+      "loss": 0.7571,
+      "step": 4265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999785844638178e-05,
+      "loss": 0.6904,
+      "step": 4266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999785741816284e-05,
+      "loss": 1.4207,
+      "step": 4267
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999785638969713e-05,
+      "loss": 0.8324,
+      "step": 4268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999785536098466e-05,
+      "loss": 1.1555,
+      "step": 4269
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999785433202541e-05,
+      "loss": 1.3096,
+      "step": 4270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99978533028194e-05,
+      "loss": 0.7022,
+      "step": 4271
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997852273366616e-05,
+      "loss": 0.5015,
+      "step": 4272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999785124366706e-05,
+      "loss": 0.6592,
+      "step": 4273
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997850213720745e-05,
+      "loss": 0.8913,
+      "step": 4274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997849183527664e-05,
+      "loss": 1.3936,
+      "step": 4275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997848153087804e-05,
+      "loss": 1.2355,
+      "step": 4276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999784712240119e-05,
+      "loss": 0.8601,
+      "step": 4277
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999784609146779e-05,
+      "loss": 1.0973,
+      "step": 4278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999784506028763e-05,
+      "loss": 0.974,
+      "step": 4279
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999784402886071e-05,
+      "loss": 1.2342,
+      "step": 4280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999784299718701e-05,
+      "loss": 1.1617,
+      "step": 4281
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997841965266546e-05,
+      "loss": 1.1919,
+      "step": 4282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997840933099315e-05,
+      "loss": 1.281,
+      "step": 4283
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999783990068532e-05,
+      "loss": 1.119,
+      "step": 4284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997838868024546e-05,
+      "loss": 1.3177,
+      "step": 4285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999783783511701e-05,
+      "loss": 0.8571,
+      "step": 4286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997836801962706e-05,
+      "loss": 0.7544,
+      "step": 4287
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999783576856163e-05,
+      "loss": 0.5614,
+      "step": 4288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997834734913795e-05,
+      "loss": 1.0438,
+      "step": 4289
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997833701019186e-05,
+      "loss": 1.2406,
+      "step": 4290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997832666877806e-05,
+      "loss": 1.1987,
+      "step": 4291
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999783163248967e-05,
+      "loss": 1.4006,
+      "step": 4292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999783059785475e-05,
+      "loss": 0.9797,
+      "step": 4293
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782956297307e-05,
+      "loss": 1.0625,
+      "step": 4294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782852784463e-05,
+      "loss": 1.2627,
+      "step": 4295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997827492469406e-05,
+      "loss": 1.0369,
+      "step": 4296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782645684742e-05,
+      "loss": 1.4083,
+      "step": 4297
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782542097867e-05,
+      "loss": 1.428,
+      "step": 4298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997824384863145e-05,
+      "loss": 1.4295,
+      "step": 4299
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782334850086e-05,
+      "loss": 1.0226,
+      "step": 4300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99978223118918e-05,
+      "loss": 1.0865,
+      "step": 4301
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782127503597e-05,
+      "loss": 1.0492,
+      "step": 4302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782023793338e-05,
+      "loss": 1.5096,
+      "step": 4303
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999781920058402e-05,
+      "loss": 1.2298,
+      "step": 4304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999781816298789e-05,
+      "loss": 1.4012,
+      "step": 4305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997817125144995e-05,
+      "loss": 1.0163,
+      "step": 4306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999781608705533e-05,
+      "loss": 0.8868,
+      "step": 4307
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99978150487189e-05,
+      "loss": 1.0451,
+      "step": 4308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99978140101357e-05,
+      "loss": 1.1224,
+      "step": 4309
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999781297130573e-05,
+      "loss": 0.59,
+      "step": 4310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999781193222899e-05,
+      "loss": 0.5647,
+      "step": 4311
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999781089290549e-05,
+      "loss": 2.8877,
+      "step": 4312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780985333522e-05,
+      "loss": 2.3249,
+      "step": 4313
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780881351818e-05,
+      "loss": 0.5768,
+      "step": 4314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780777345438e-05,
+      "loss": 0.9765,
+      "step": 4315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780673314379e-05,
+      "loss": 1.2209,
+      "step": 4316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997805692586455e-05,
+      "loss": 1.1851,
+      "step": 4317
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780465178234e-05,
+      "loss": 1.0895,
+      "step": 4318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780361073146e-05,
+      "loss": 1.0949,
+      "step": 4319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780256943381e-05,
+      "loss": 1.3861,
+      "step": 4320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99978015278894e-05,
+      "loss": 1.0276,
+      "step": 4321
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999780048609822e-05,
+      "loss": 1.2675,
+      "step": 4322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999779944406027e-05,
+      "loss": 1.2956,
+      "step": 4323
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997798401775556e-05,
+      "loss": 1.1,
+      "step": 4324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999779735924407e-05,
+      "loss": 1.1209,
+      "step": 4325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997796316465814e-05,
+      "loss": 0.9514,
+      "step": 4326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999779527344079e-05,
+      "loss": 1.148,
+      "step": 4327
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997794230169e-05,
+      "loss": 0.991,
+      "step": 4328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999779318665045e-05,
+      "loss": 1.1714,
+      "step": 4329
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999779214288512e-05,
+      "loss": 1.024,
+      "step": 4330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999779109887304e-05,
+      "loss": 1.0824,
+      "step": 4331
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997790054614174e-05,
+      "loss": 1.2253,
+      "step": 4332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997789010108546e-05,
+      "loss": 0.9916,
+      "step": 4333
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997787965356154e-05,
+      "loss": 1.2658,
+      "step": 4334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999778692035699e-05,
+      "loss": 1.0458,
+      "step": 4335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997785875111056e-05,
+      "loss": 1.299,
+      "step": 4336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997784829618364e-05,
+      "loss": 0.933,
+      "step": 4337
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99977837838789e-05,
+      "loss": 1.2263,
+      "step": 4338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999778273789266e-05,
+      "loss": 1.5378,
+      "step": 4339
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999778169165966e-05,
+      "loss": 1.2621,
+      "step": 4340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997780645179896e-05,
+      "loss": 1.4907,
+      "step": 4341
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777959845336e-05,
+      "loss": 0.6645,
+      "step": 4342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997778551480056e-05,
+      "loss": 0.8306,
+      "step": 4343
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777750425999e-05,
+      "loss": 0.6357,
+      "step": 4344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777645679316e-05,
+      "loss": 0.7529,
+      "step": 4345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777540907955e-05,
+      "loss": 0.9319,
+      "step": 4346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997774361119175e-05,
+      "loss": 0.7715,
+      "step": 4347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777331291203e-05,
+      "loss": 0.8799,
+      "step": 4348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777226445813e-05,
+      "loss": 0.8824,
+      "step": 4349
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777121575745e-05,
+      "loss": 1.3207,
+      "step": 4350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999777016681001e-05,
+      "loss": 1.0959,
+      "step": 4351
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99977691176158e-05,
+      "loss": 0.6392,
+      "step": 4352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999776806817482e-05,
+      "loss": 1.5281,
+      "step": 4353
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999776701848707e-05,
+      "loss": 1.1428,
+      "step": 4354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997765968552565e-05,
+      "loss": 1.2032,
+      "step": 4355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999776491837128e-05,
+      "loss": 1.1605,
+      "step": 4356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999776386794324e-05,
+      "loss": 1.1478,
+      "step": 4357
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999776281726842e-05,
+      "loss": 1.1389,
+      "step": 4358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999776176634684e-05,
+      "loss": 1.0784,
+      "step": 4359
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999776071517849e-05,
+      "loss": 1.0957,
+      "step": 4360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997759663763375e-05,
+      "loss": 1.1626,
+      "step": 4361
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997758612101484e-05,
+      "loss": 1.1153,
+      "step": 4362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997757560192836e-05,
+      "loss": 1.3685,
+      "step": 4363
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997756508037416e-05,
+      "loss": 1.077,
+      "step": 4364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999775545563523e-05,
+      "loss": 1.1683,
+      "step": 4365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999775440298628e-05,
+      "loss": 1.2724,
+      "step": 4366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999775335009056e-05,
+      "loss": 1.0369,
+      "step": 4367
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999775229694807e-05,
+      "loss": 0.9634,
+      "step": 4368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999775124355881e-05,
+      "loss": 1.1885,
+      "step": 4369
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997750189922785e-05,
+      "loss": 1.1612,
+      "step": 4370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774913604e-05,
+      "loss": 1.0395,
+      "step": 4371
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997748081910446e-05,
+      "loss": 0.9069,
+      "step": 4372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774702753412e-05,
+      "loss": 1.0056,
+      "step": 4373
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774597291102e-05,
+      "loss": 1.2466,
+      "step": 4374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774491804117e-05,
+      "loss": 0.7215,
+      "step": 4375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774386292454e-05,
+      "loss": 1.1564,
+      "step": 4376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774280756116e-05,
+      "loss": 1.313,
+      "step": 4377
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774175195099e-05,
+      "loss": 1.1211,
+      "step": 4378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999774069609406e-05,
+      "loss": 0.9678,
+      "step": 4379
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999773963999037e-05,
+      "loss": 1.1849,
+      "step": 4380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999773858363991e-05,
+      "loss": 0.7452,
+      "step": 4381
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997737527042675e-05,
+      "loss": 1.5299,
+      "step": 4382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999773647019868e-05,
+      "loss": 0.9208,
+      "step": 4383
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997735413107916e-05,
+      "loss": 0.8891,
+      "step": 4384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999773435577039e-05,
+      "loss": 1.143,
+      "step": 4385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999773329818609e-05,
+      "loss": 1.2671,
+      "step": 4386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999773224035502e-05,
+      "loss": 1.1009,
+      "step": 4387
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997731182277196e-05,
+      "loss": 1.114,
+      "step": 4388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999773012395259e-05,
+      "loss": 1.0124,
+      "step": 4389
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999772906538123e-05,
+      "loss": 1.0004,
+      "step": 4390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997728006563097e-05,
+      "loss": 1.1445,
+      "step": 4391
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99977269474982e-05,
+      "loss": 1.0378,
+      "step": 4392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999772588818653e-05,
+      "loss": 1.1042,
+      "step": 4393
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99977248286281e-05,
+      "loss": 1.2522,
+      "step": 4394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99977237688229e-05,
+      "loss": 1.1237,
+      "step": 4395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999772270877093e-05,
+      "loss": 1.1444,
+      "step": 4396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99977216484722e-05,
+      "loss": 1.9499,
+      "step": 4397
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997720587926686e-05,
+      "loss": 3.5417,
+      "step": 4398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997719527134423e-05,
+      "loss": 1.2548,
+      "step": 4399
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999771846609539e-05,
+      "loss": 1.3411,
+      "step": 4400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997717404809586e-05,
+      "loss": 1.3086,
+      "step": 4401
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999771634327701e-05,
+      "loss": 1.2577,
+      "step": 4402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997715281497676e-05,
+      "loss": 1.2029,
+      "step": 4403
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999771421947157e-05,
+      "loss": 1.2334,
+      "step": 4404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99977131571987e-05,
+      "loss": 1.2065,
+      "step": 4405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999771209467907e-05,
+      "loss": 1.1416,
+      "step": 4406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997711031912665e-05,
+      "loss": 1.1523,
+      "step": 4407
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770996889949e-05,
+      "loss": 1.1319,
+      "step": 4408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770890563955e-05,
+      "loss": 1.4375,
+      "step": 4409
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997707842132845e-05,
+      "loss": 1.2708,
+      "step": 4410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997706778379376e-05,
+      "loss": 1.4632,
+      "step": 4411
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770571437914e-05,
+      "loss": 1.3789,
+      "step": 4412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770465013213e-05,
+      "loss": 1.0032,
+      "step": 4413
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770358563836e-05,
+      "loss": 0.7754,
+      "step": 4414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770252089782e-05,
+      "loss": 1.0801,
+      "step": 4415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770145591052e-05,
+      "loss": 1.0647,
+      "step": 4416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999770039067644e-05,
+      "loss": 1.2998,
+      "step": 4417
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99976993251956e-05,
+      "loss": 1.0556,
+      "step": 4418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997698259467996e-05,
+      "loss": 1.1464,
+      "step": 4419
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999769719349363e-05,
+      "loss": 1.1842,
+      "step": 4420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999769612727249e-05,
+      "loss": 1.059,
+      "step": 4421
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999769506080458e-05,
+      "loss": 1.0721,
+      "step": 4422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999769399408991e-05,
+      "loss": 1.0131,
+      "step": 4423
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999769292712847e-05,
+      "loss": 1.7383,
+      "step": 4424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999769185992026e-05,
+      "loss": 1.2293,
+      "step": 4425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999769079246529e-05,
+      "loss": 0.9657,
+      "step": 4426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999768972476355e-05,
+      "loss": 1.0584,
+      "step": 4427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999768865681505e-05,
+      "loss": 1.2185,
+      "step": 4428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997687588619765e-05,
+      "loss": 1.156,
+      "step": 4429
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999768652017773e-05,
+      "loss": 1.4052,
+      "step": 4430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999768545148893e-05,
+      "loss": 1.2739,
+      "step": 4431
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999768438255336e-05,
+      "loss": 1.1099,
+      "step": 4432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999768331337101e-05,
+      "loss": 1.3071,
+      "step": 4433
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997682243941906e-05,
+      "loss": 0.9026,
+      "step": 4434
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999768117426603e-05,
+      "loss": 1.1763,
+      "step": 4435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997680104343396e-05,
+      "loss": 0.6924,
+      "step": 4436
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999767903417399e-05,
+      "loss": 1.1063,
+      "step": 4437
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997677963757815e-05,
+      "loss": 6.2614,
+      "step": 4438
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999767689309488e-05,
+      "loss": 6.0042,
+      "step": 4439
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999767582218518e-05,
+      "loss": 5.9976,
+      "step": 4440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99976747510287e-05,
+      "loss": 5.9647,
+      "step": 4441
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999767367962547e-05,
+      "loss": 5.9497,
+      "step": 4442
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999767260797546e-05,
+      "loss": 5.9563,
+      "step": 4443
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999767153607869e-05,
+      "loss": 5.8703,
+      "step": 4444
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999767046393515e-05,
+      "loss": 5.8784,
+      "step": 4445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999766939154484e-05,
+      "loss": 5.8673,
+      "step": 4446
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999766831890777e-05,
+      "loss": 5.9005,
+      "step": 4447
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999766724602394e-05,
+      "loss": 4.7112,
+      "step": 4448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997666172893334e-05,
+      "loss": 1.3249,
+      "step": 4449
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997665099515965e-05,
+      "loss": 1.5641,
+      "step": 4450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997664025891824e-05,
+      "loss": 1.8465,
+      "step": 4451
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997662952020926e-05,
+      "loss": 1.9022,
+      "step": 4452
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999766187790326e-05,
+      "loss": 1.2422,
+      "step": 4453
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999766080353882e-05,
+      "loss": 1.1355,
+      "step": 4454
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999765972892762e-05,
+      "loss": 1.0728,
+      "step": 4455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999765865406965e-05,
+      "loss": 0.7308,
+      "step": 4456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997657578964915e-05,
+      "loss": 1.221,
+      "step": 4457
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999765650361341e-05,
+      "loss": 1.0683,
+      "step": 4458
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999765542801514e-05,
+      "loss": 1.0572,
+      "step": 4459
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999765435217011e-05,
+      "loss": 1.3279,
+      "step": 4460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997653276078316e-05,
+      "loss": 1.1065,
+      "step": 4461
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999765219973975e-05,
+      "loss": 1.2609,
+      "step": 4462
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999765112315441e-05,
+      "loss": 1.0018,
+      "step": 4463
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997650046322315e-05,
+      "loss": 1.2095,
+      "step": 4464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999764896924345e-05,
+      "loss": 1.0187,
+      "step": 4465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999764789191782e-05,
+      "loss": 1.28,
+      "step": 4466
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999764681434542e-05,
+      "loss": 1.14,
+      "step": 4467
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999764573652626e-05,
+      "loss": 0.8774,
+      "step": 4468
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999764465846033e-05,
+      "loss": 1.1061,
+      "step": 4469
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999764358014763e-05,
+      "loss": 1.3555,
+      "step": 4470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999764250158817e-05,
+      "loss": 1.042,
+      "step": 4471
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997641422781936e-05,
+      "loss": 1.2689,
+      "step": 4472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997640343728945e-05,
+      "loss": 1.1008,
+      "step": 4473
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999763926442919e-05,
+      "loss": 1.4017,
+      "step": 4474
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997638184882655e-05,
+      "loss": 0.9575,
+      "step": 4475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997637105089364e-05,
+      "loss": 1.3155,
+      "step": 4476
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99976360250493e-05,
+      "loss": 1.0595,
+      "step": 4477
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999763494476248e-05,
+      "loss": 1.228,
+      "step": 4478
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999763386422889e-05,
+      "loss": 0.9576,
+      "step": 4479
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997632783448527e-05,
+      "loss": 1.1748,
+      "step": 4480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997631702421406e-05,
+      "loss": 1.2856,
+      "step": 4481
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997630621147515e-05,
+      "loss": 1.1432,
+      "step": 4482
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999762953962686e-05,
+      "loss": 1.0681,
+      "step": 4483
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999762845785944e-05,
+      "loss": 1.021,
+      "step": 4484
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999762737584525e-05,
+      "loss": 1.1034,
+      "step": 4485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999762629358429e-05,
+      "loss": 0.9276,
+      "step": 4486
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999762521107657e-05,
+      "loss": 0.9006,
+      "step": 4487
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997624128322086e-05,
+      "loss": 0.8601,
+      "step": 4488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999762304532083e-05,
+      "loss": 1.2345,
+      "step": 4489
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997621962072816e-05,
+      "loss": 1.0603,
+      "step": 4490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999762087857803e-05,
+      "loss": 1.2632,
+      "step": 4491
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999761979483648e-05,
+      "loss": 1.0961,
+      "step": 4492
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999761871084816e-05,
+      "loss": 1.4085,
+      "step": 4493
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999761762661308e-05,
+      "loss": 0.9822,
+      "step": 4494
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997616542131235e-05,
+      "loss": 0.8745,
+      "step": 4495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999761545740262e-05,
+      "loss": 1.2333,
+      "step": 4496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997614372427236e-05,
+      "loss": 1.2191,
+      "step": 4497
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997613287205094e-05,
+      "loss": 1.3622,
+      "step": 4498
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999761220173618e-05,
+      "loss": 1.6796,
+      "step": 4499
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99976111160205e-05,
+      "loss": 1.0056,
+      "step": 4500
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.0784711837768555,
+      "eval_runtime": 85.0826,
+      "eval_samples_per_second": 16.278,
+      "eval_steps_per_second": 4.078,
+      "step": 4500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999761003005806e-05,
+      "loss": 1.3483,
+      "step": 4501
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999760894384885e-05,
+      "loss": 1.1076,
+      "step": 4502
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997607857392875e-05,
+      "loss": 0.3563,
+      "step": 4503
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999760677069013e-05,
+      "loss": 0.8917,
+      "step": 4504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997605683740625e-05,
+      "loss": 1.2865,
+      "step": 4505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997604596544354e-05,
+      "loss": 1.0649,
+      "step": 4506
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999760350910132e-05,
+      "loss": 0.8918,
+      "step": 4507
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999760242141151e-05,
+      "loss": 1.1132,
+      "step": 4508
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999760133347494e-05,
+      "loss": 1.0425,
+      "step": 4509
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997600245291604e-05,
+      "loss": 1.1745,
+      "step": 4510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997599156861504e-05,
+      "loss": 0.932,
+      "step": 4511
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999759806818464e-05,
+      "loss": 1.0716,
+      "step": 4512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997596979261e-05,
+      "loss": 0.7812,
+      "step": 4513
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99975958900906e-05,
+      "loss": 0.8982,
+      "step": 4514
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999759480067344e-05,
+      "loss": 1.2918,
+      "step": 4515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99975937110095e-05,
+      "loss": 0.9576,
+      "step": 4516
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999759262109881e-05,
+      "loss": 1.1978,
+      "step": 4517
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997591530941345e-05,
+      "loss": 1.0162,
+      "step": 4518
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999759044053712e-05,
+      "loss": 1.1299,
+      "step": 4519
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999758934988612e-05,
+      "loss": 1.3023,
+      "step": 4520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999758825898837e-05,
+      "loss": 1.0354,
+      "step": 4521
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999758716784384e-05,
+      "loss": 1.2752,
+      "step": 4522
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999758607645255e-05,
+      "loss": 0.9105,
+      "step": 4523
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999758498481449e-05,
+      "loss": 1.1427,
+      "step": 4524
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997583892929676e-05,
+      "loss": 1.123,
+      "step": 4525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999758280079809e-05,
+      "loss": 1.0269,
+      "step": 4526
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999758170841973e-05,
+      "loss": 1.4752,
+      "step": 4527
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997580615794615e-05,
+      "loss": 1.9687,
+      "step": 4528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997579522922735e-05,
+      "loss": 1.4191,
+      "step": 4529
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997578429804084e-05,
+      "loss": 1.8628,
+      "step": 4530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999757733643867e-05,
+      "loss": 2.0473,
+      "step": 4531
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999757624282648e-05,
+      "loss": 1.7376,
+      "step": 4532
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999757514896754e-05,
+      "loss": 1.7128,
+      "step": 4533
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999757405486183e-05,
+      "loss": 1.2395,
+      "step": 4534
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997572960509355e-05,
+      "loss": 0.7994,
+      "step": 4535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999757186591011e-05,
+      "loss": 1.5496,
+      "step": 4536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997570771064095e-05,
+      "loss": 1.3805,
+      "step": 4537
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999756967597132e-05,
+      "loss": 1.1317,
+      "step": 4538
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997568580631785e-05,
+      "loss": 1.0239,
+      "step": 4539
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999756748504548e-05,
+      "loss": 0.9977,
+      "step": 4540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997566389212416e-05,
+      "loss": 0.7617,
+      "step": 4541
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999756529313258e-05,
+      "loss": 0.7574,
+      "step": 4542
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997564196805976e-05,
+      "loss": 0.4952,
+      "step": 4543
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999756310023261e-05,
+      "loss": 0.4071,
+      "step": 4544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999756200341248e-05,
+      "loss": 0.3267,
+      "step": 4545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999756090634558e-05,
+      "loss": 0.2456,
+      "step": 4546
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999755980903191e-05,
+      "loss": 0.1915,
+      "step": 4547
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997558711471486e-05,
+      "loss": 0.2422,
+      "step": 4548
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999755761366429e-05,
+      "loss": 0.1811,
+      "step": 4549
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999755651561033e-05,
+      "loss": 0.1229,
+      "step": 4550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997555417309613e-05,
+      "loss": 0.1014,
+      "step": 4551
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999755431876212e-05,
+      "loss": 0.5357,
+      "step": 4552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997553219967865e-05,
+      "loss": 1.2375,
+      "step": 4553
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999755212092685e-05,
+      "loss": 1.3645,
+      "step": 4554
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999755102163906e-05,
+      "loss": 1.3012,
+      "step": 4555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999754992210451e-05,
+      "loss": 1.2112,
+      "step": 4556
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997548822323195e-05,
+      "loss": 1.3208,
+      "step": 4557
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997547722295113e-05,
+      "loss": 1.0551,
+      "step": 4558
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999754662202026e-05,
+      "loss": 1.1901,
+      "step": 4559
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997545521498656e-05,
+      "loss": 0.8277,
+      "step": 4560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999754442073028e-05,
+      "loss": 0.3503,
+      "step": 4561
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997543319715135e-05,
+      "loss": 1.1548,
+      "step": 4562
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999754221845323e-05,
+      "loss": 1.1193,
+      "step": 4563
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997541116944556e-05,
+      "loss": 1.0134,
+      "step": 4564
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997540015189116e-05,
+      "loss": 1.3742,
+      "step": 4565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753891318691e-05,
+      "loss": 1.2755,
+      "step": 4566
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997537810937944e-05,
+      "loss": 1.4311,
+      "step": 4567
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753670844222e-05,
+      "loss": 1.3954,
+      "step": 4568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753560569972e-05,
+      "loss": 1.3045,
+      "step": 4569
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753450271045e-05,
+      "loss": 2.1417,
+      "step": 4570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753339947442e-05,
+      "loss": 1.2134,
+      "step": 4571
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753229599163e-05,
+      "loss": 1.1868,
+      "step": 4572
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753119226207e-05,
+      "loss": 1.072,
+      "step": 4573
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999753008828575e-05,
+      "loss": 0.9794,
+      "step": 4574
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999752898406266e-05,
+      "loss": 1.0879,
+      "step": 4575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99975278795928e-05,
+      "loss": 1.0484,
+      "step": 4576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999752677487618e-05,
+      "loss": 0.909,
+      "step": 4577
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99975256699128e-05,
+      "loss": 0.8647,
+      "step": 4578
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997524564702645e-05,
+      "loss": 1.1287,
+      "step": 4579
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999752345924573e-05,
+      "loss": 1.1584,
+      "step": 4580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997522353542056e-05,
+      "loss": 1.4009,
+      "step": 4581
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997521247591614e-05,
+      "loss": 0.9556,
+      "step": 4582
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99975201413944e-05,
+      "loss": 0.6594,
+      "step": 4583
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997519034950425e-05,
+      "loss": 1.0026,
+      "step": 4584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997517928259683e-05,
+      "loss": 1.068,
+      "step": 4585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997516821322184e-05,
+      "loss": 1.2799,
+      "step": 4586
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999751571413791e-05,
+      "loss": 1.2804,
+      "step": 4587
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999751460670688e-05,
+      "loss": 1.1973,
+      "step": 4588
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999751349902908e-05,
+      "loss": 1.1968,
+      "step": 4589
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999751239110452e-05,
+      "loss": 1.2127,
+      "step": 4590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999751128293318e-05,
+      "loss": 1.3084,
+      "step": 4591
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99975101745151e-05,
+      "loss": 1.0746,
+      "step": 4592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997509065850234e-05,
+      "loss": 1.1344,
+      "step": 4593
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750795693861e-05,
+      "loss": 1.1177,
+      "step": 4594
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750684778022e-05,
+      "loss": 1.4068,
+      "step": 4595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750573837507e-05,
+      "loss": 1.1157,
+      "step": 4596
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750462872315e-05,
+      "loss": 1.0543,
+      "step": 4597
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750351882447e-05,
+      "loss": 0.8104,
+      "step": 4598
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750240867902e-05,
+      "loss": 1.3103,
+      "step": 4599
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997501298286806e-05,
+      "loss": 1.2957,
+      "step": 4600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999750018764783e-05,
+      "loss": 1.1328,
+      "step": 4601
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999749907676209e-05,
+      "loss": 0.9242,
+      "step": 4602
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999749796562958e-05,
+      "loss": 1.1808,
+      "step": 4603
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999749685425031e-05,
+      "loss": 1.2292,
+      "step": 4604
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997495742624275e-05,
+      "loss": 0.935,
+      "step": 4605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997494630751474e-05,
+      "loss": 1.0585,
+      "step": 4606
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974935186319e-05,
+      "loss": 0.9551,
+      "step": 4607
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999749240626558e-05,
+      "loss": 0.9226,
+      "step": 4608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999749129365248e-05,
+      "loss": 1.1205,
+      "step": 4609
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999749018079262e-05,
+      "loss": 1.2519,
+      "step": 4610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999748906768599e-05,
+      "loss": 1.063,
+      "step": 4611
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974879543326e-05,
+      "loss": 1.1115,
+      "step": 4612
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999748684073245e-05,
+      "loss": 1.1365,
+      "step": 4613
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999748572688553e-05,
+      "loss": 1.1795,
+      "step": 4614
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999748461279185e-05,
+      "loss": 0.8403,
+      "step": 4615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974834984514e-05,
+      "loss": 0.9619,
+      "step": 4616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997482383864186e-05,
+      "loss": 1.5095,
+      "step": 4617
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999748126903021e-05,
+      "loss": 1.1162,
+      "step": 4618
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999748015394947e-05,
+      "loss": 0.8096,
+      "step": 4619
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997479038621964e-05,
+      "loss": 1.1393,
+      "step": 4620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997477923047684e-05,
+      "loss": 1.0588,
+      "step": 4621
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999747680722665e-05,
+      "loss": 1.0214,
+      "step": 4622
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997475691158856e-05,
+      "loss": 1.0753,
+      "step": 4623
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999747457484429e-05,
+      "loss": 1.0879,
+      "step": 4624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997473458282964e-05,
+      "loss": 1.1705,
+      "step": 4625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999747234147487e-05,
+      "loss": 1.4716,
+      "step": 4626
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999747122442001e-05,
+      "loss": 0.9731,
+      "step": 4627
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997470107118383e-05,
+      "loss": 0.9341,
+      "step": 4628
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997468989569994e-05,
+      "loss": 1.1177,
+      "step": 4629
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999746787177485e-05,
+      "loss": 1.4303,
+      "step": 4630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999746675373292e-05,
+      "loss": 1.0908,
+      "step": 4631
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999746563544425e-05,
+      "loss": 1.4172,
+      "step": 4632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974645169088e-05,
+      "loss": 0.9805,
+      "step": 4633
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997463398126596e-05,
+      "loss": 1.1843,
+      "step": 4634
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999746227909762e-05,
+      "loss": 1.1419,
+      "step": 4635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999746115982188e-05,
+      "loss": 0.9037,
+      "step": 4636
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997460040299376e-05,
+      "loss": 0.9535,
+      "step": 4637
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997458920530113e-05,
+      "loss": 1.1616,
+      "step": 4638
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999745780051408e-05,
+      "loss": 0.9401,
+      "step": 4639
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999745668025128e-05,
+      "loss": 0.9385,
+      "step": 4640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999745555974173e-05,
+      "loss": 1.5485,
+      "step": 4641
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974544389854e-05,
+      "loss": 1.2252,
+      "step": 4642
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999745331798231e-05,
+      "loss": 1.1477,
+      "step": 4643
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997452196732454e-05,
+      "loss": 1.497,
+      "step": 4644
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997451075235834e-05,
+      "loss": 0.9156,
+      "step": 4645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999744995349245e-05,
+      "loss": 0.6902,
+      "step": 4646
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999744883150231e-05,
+      "loss": 0.6409,
+      "step": 4647
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974477092654e-05,
+      "loss": 0.8875,
+      "step": 4648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999744658678172e-05,
+      "loss": 1.017,
+      "step": 4649
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999744546405128e-05,
+      "loss": 1.4372,
+      "step": 4650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997444341074076e-05,
+      "loss": 1.0174,
+      "step": 4651
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999744321785011e-05,
+      "loss": 1.1617,
+      "step": 4652
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999744209437938e-05,
+      "loss": 1.3952,
+      "step": 4653
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999744097066188e-05,
+      "loss": 1.1098,
+      "step": 4654
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999743984669763e-05,
+      "loss": 1.1052,
+      "step": 4655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974387224866e-05,
+      "loss": 0.9504,
+      "step": 4656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997437598028806e-05,
+      "loss": 0.9464,
+      "step": 4657
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997436473324256e-05,
+      "loss": 1.1921,
+      "step": 4658
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999743534837294e-05,
+      "loss": 1.0369,
+      "step": 4659
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999743422317486e-05,
+      "loss": 1.0909,
+      "step": 4660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999743309773002e-05,
+      "loss": 0.9977,
+      "step": 4661
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974319720384e-05,
+      "loss": 1.1727,
+      "step": 4662
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999743084610003e-05,
+      "loss": 0.9632,
+      "step": 4663
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999742971991489e-05,
+      "loss": 1.3036,
+      "step": 4664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999742859348299e-05,
+      "loss": 1.1178,
+      "step": 4665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999742746680433e-05,
+      "loss": 1.66,
+      "step": 4666
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99974263398789e-05,
+      "loss": 0.2656,
+      "step": 4667
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999742521270671e-05,
+      "loss": 0.4481,
+      "step": 4668
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999742408528775e-05,
+      "loss": 0.273,
+      "step": 4669
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999742295762202e-05,
+      "loss": 1.0934,
+      "step": 4670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999742182970953e-05,
+      "loss": 0.6311,
+      "step": 4671
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997420701550286e-05,
+      "loss": 0.6401,
+      "step": 4672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999741957314427e-05,
+      "loss": 1.102,
+      "step": 4673
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997418444491496e-05,
+      "loss": 0.8887,
+      "step": 4674
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999741731559196e-05,
+      "loss": 1.336,
+      "step": 4675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999741618644565e-05,
+      "loss": 1.7969,
+      "step": 4676
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999741505705258e-05,
+      "loss": 1.2247,
+      "step": 4677
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999741392741274e-05,
+      "loss": 1.0536,
+      "step": 4678
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997412797526144e-05,
+      "loss": 1.0859,
+      "step": 4679
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999741166739278e-05,
+      "loss": 1.4554,
+      "step": 4680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999741053701266e-05,
+      "loss": 1.3994,
+      "step": 4681
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999740940638576e-05,
+      "loss": 1.5389,
+      "step": 4682
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997408275512115e-05,
+      "loss": 1.4944,
+      "step": 4683
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997407144391697e-05,
+      "loss": 1.5548,
+      "step": 4684
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997406013024514e-05,
+      "loss": 1.2535,
+      "step": 4685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999740488141057e-05,
+      "loss": 1.2727,
+      "step": 4686
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999740374954986e-05,
+      "loss": 0.9823,
+      "step": 4687
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999740261744239e-05,
+      "loss": 1.006,
+      "step": 4688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999740148508815e-05,
+      "loss": 1.1687,
+      "step": 4689
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997400352487154e-05,
+      "loss": 0.9116,
+      "step": 4690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997399219639384e-05,
+      "loss": 1.3216,
+      "step": 4691
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999739808654486e-05,
+      "loss": 1.1733,
+      "step": 4692
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997396953203566e-05,
+      "loss": 1.2683,
+      "step": 4693
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999739581961551e-05,
+      "loss": 1.9192,
+      "step": 4694
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999739468578069e-05,
+      "loss": 1.497,
+      "step": 4695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999739355169911e-05,
+      "loss": 1.1082,
+      "step": 4696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997392417370756e-05,
+      "loss": 1.5696,
+      "step": 4697
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999739128279565e-05,
+      "loss": 0.9002,
+      "step": 4698
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999739014797378e-05,
+      "loss": 0.6441,
+      "step": 4699
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997389012905136e-05,
+      "loss": 1.2714,
+      "step": 4700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999738787758974e-05,
+      "loss": 1.1863,
+      "step": 4701
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999738674202757e-05,
+      "loss": 1.1173,
+      "step": 4702
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997385606218644e-05,
+      "loss": 1.4007,
+      "step": 4703
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999738447016295e-05,
+      "loss": 1.0387,
+      "step": 4704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999738333386049e-05,
+      "loss": 1.1106,
+      "step": 4705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999738219731127e-05,
+      "loss": 1.189,
+      "step": 4706
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999738106051529e-05,
+      "loss": 1.2896,
+      "step": 4707
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999737992347254e-05,
+      "loss": 1.2141,
+      "step": 4708
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999737878618304e-05,
+      "loss": 1.0269,
+      "step": 4709
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999737764864676e-05,
+      "loss": 1.1955,
+      "step": 4710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999737651086372e-05,
+      "loss": 1.3306,
+      "step": 4711
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999737537283392e-05,
+      "loss": 0.8874,
+      "step": 4712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997374234557354e-05,
+      "loss": 1.1114,
+      "step": 4713
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999737309603403e-05,
+      "loss": 1.0591,
+      "step": 4714
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997371957263935e-05,
+      "loss": 1.1689,
+      "step": 4715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999737081824708e-05,
+      "loss": 1.0724,
+      "step": 4716
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736967898346e-05,
+      "loss": 1.063,
+      "step": 4717
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736853947308e-05,
+      "loss": 1.1839,
+      "step": 4718
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736739971593e-05,
+      "loss": 1.0832,
+      "step": 4719
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736625971202e-05,
+      "loss": 1.253,
+      "step": 4720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736511946135e-05,
+      "loss": 0.9533,
+      "step": 4721
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736397896391e-05,
+      "loss": 0.6984,
+      "step": 4722
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736283821971e-05,
+      "loss": 1.1763,
+      "step": 4723
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736169722874e-05,
+      "loss": 1.289,
+      "step": 4724
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999736055599102e-05,
+      "loss": 1.1467,
+      "step": 4725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999735941450653e-05,
+      "loss": 0.9113,
+      "step": 4726
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997358272775273e-05,
+      "loss": 1.0437,
+      "step": 4727
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999735713079726e-05,
+      "loss": 1.0288,
+      "step": 4728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999735598857248e-05,
+      "loss": 1.0476,
+      "step": 4729
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999735484610094e-05,
+      "loss": 0.4842,
+      "step": 4730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997353703382625e-05,
+      "loss": 1.1656,
+      "step": 4731
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997352560417556e-05,
+      "loss": 1.1774,
+      "step": 4732
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999735141720572e-05,
+      "loss": 1.4356,
+      "step": 4733
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999735027374712e-05,
+      "loss": 0.9103,
+      "step": 4734
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999734913004177e-05,
+      "loss": 1.6251,
+      "step": 4735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999734798608964e-05,
+      "loss": 1.2213,
+      "step": 4736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997346841890755e-05,
+      "loss": 1.0874,
+      "step": 4737
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997345697445106e-05,
+      "loss": 1.3342,
+      "step": 4738
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999734455275269e-05,
+      "loss": 1.1486,
+      "step": 4739
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997343407813514e-05,
+      "loss": 1.2113,
+      "step": 4740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999734226262757e-05,
+      "loss": 1.1141,
+      "step": 4741
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999734111719487e-05,
+      "loss": 1.1153,
+      "step": 4742
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99973399715154e-05,
+      "loss": 1.1899,
+      "step": 4743
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999733882558918e-05,
+      "loss": 1.1211,
+      "step": 4744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999733767941618e-05,
+      "loss": 1.0558,
+      "step": 4745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997336532996426e-05,
+      "loss": 1.1537,
+      "step": 4746
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99973353863299e-05,
+      "loss": 1.1311,
+      "step": 4747
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997334239416623e-05,
+      "loss": 0.7318,
+      "step": 4748
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999733309225658e-05,
+      "loss": 0.8849,
+      "step": 4749
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999733194484977e-05,
+      "loss": 1.0484,
+      "step": 4750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99973307971962e-05,
+      "loss": 1.1667,
+      "step": 4751
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997329649295866e-05,
+      "loss": 1.5542,
+      "step": 4752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997328501148764e-05,
+      "loss": 1.0831,
+      "step": 4753
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997327352754905e-05,
+      "loss": 1.4354,
+      "step": 4754
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999732620411428e-05,
+      "loss": 1.2352,
+      "step": 4755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999732505522689e-05,
+      "loss": 1.4719,
+      "step": 4756
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999732390609274e-05,
+      "loss": 1.057,
+      "step": 4757
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999732275671183e-05,
+      "loss": 1.0866,
+      "step": 4758
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997321607084154e-05,
+      "loss": 1.104,
+      "step": 4759
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997320457209714e-05,
+      "loss": 1.0629,
+      "step": 4760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999731930708851e-05,
+      "loss": 1.6223,
+      "step": 4761
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999731815672054e-05,
+      "loss": 1.3682,
+      "step": 4762
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997317006105816e-05,
+      "loss": 1.2582,
+      "step": 4763
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999731585524432e-05,
+      "loss": 0.9874,
+      "step": 4764
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997314704136065e-05,
+      "loss": 0.8081,
+      "step": 4765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999731355278105e-05,
+      "loss": 1.0638,
+      "step": 4766
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997312401179276e-05,
+      "loss": 1.5015,
+      "step": 4767
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999731124933073e-05,
+      "loss": 1.5185,
+      "step": 4768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999731009723542e-05,
+      "loss": 1.2712,
+      "step": 4769
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999730894489335e-05,
+      "loss": 1.0055,
+      "step": 4770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999730779230452e-05,
+      "loss": 1.085,
+      "step": 4771
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997306639468926e-05,
+      "loss": 1.2079,
+      "step": 4772
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999730548638657e-05,
+      "loss": 1.1686,
+      "step": 4773
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999730433305745e-05,
+      "loss": 1.0924,
+      "step": 4774
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999730317948156e-05,
+      "loss": 1.3168,
+      "step": 4775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999730202565892e-05,
+      "loss": 1.1525,
+      "step": 4776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997300871589506e-05,
+      "loss": 1.1351,
+      "step": 4777
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999729971727334e-05,
+      "loss": 1.0702,
+      "step": 4778
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997298562710405e-05,
+      "loss": 1.0449,
+      "step": 4779
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99972974079007e-05,
+      "loss": 1.6182,
+      "step": 4780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997296252844246e-05,
+      "loss": 1.2316,
+      "step": 4781
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999729509754102e-05,
+      "loss": 1.1996,
+      "step": 4782
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999729394199103e-05,
+      "loss": 1.1948,
+      "step": 4783
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997292786194285e-05,
+      "loss": 0.9327,
+      "step": 4784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997291630150776e-05,
+      "loss": 1.0551,
+      "step": 4785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99972904738605e-05,
+      "loss": 1.0891,
+      "step": 4786
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999728931732346e-05,
+      "loss": 1.0666,
+      "step": 4787
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999728816053966e-05,
+      "loss": 1.1281,
+      "step": 4788
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99972870035091e-05,
+      "loss": 1.5918,
+      "step": 4789
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999728584623178e-05,
+      "loss": 1.0612,
+      "step": 4790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999728468870769e-05,
+      "loss": 1.0924,
+      "step": 4791
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997283530936836e-05,
+      "loss": 1.1297,
+      "step": 4792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997282372919225e-05,
+      "loss": 1.176,
+      "step": 4793
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999728121465485e-05,
+      "loss": 1.1923,
+      "step": 4794
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999728005614371e-05,
+      "loss": 1.0567,
+      "step": 4795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999727889738581e-05,
+      "loss": 1.1293,
+      "step": 4796
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999727773838115e-05,
+      "loss": 0.6926,
+      "step": 4797
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999727657912972e-05,
+      "loss": 1.2504,
+      "step": 4798
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997275419631535e-05,
+      "loss": 0.7057,
+      "step": 4799
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999727425988659e-05,
+      "loss": 0.9938,
+      "step": 4800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999727309989487e-05,
+      "loss": 1.1662,
+      "step": 4801
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999727193965639e-05,
+      "loss": 1.3126,
+      "step": 4802
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999727077917115e-05,
+      "loss": 1.1409,
+      "step": 4803
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999726961843915e-05,
+      "loss": 1.003,
+      "step": 4804
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997268457460386e-05,
+      "loss": 1.1619,
+      "step": 4805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997267296234865e-05,
+      "loss": 1.1168,
+      "step": 4806
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999726613476257e-05,
+      "loss": 1.2055,
+      "step": 4807
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999726497304352e-05,
+      "loss": 1.1727,
+      "step": 4808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999726381107771e-05,
+      "loss": 1.1429,
+      "step": 4809
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999726264886513e-05,
+      "loss": 1.2134,
+      "step": 4810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997261486405787e-05,
+      "loss": 0.9725,
+      "step": 4811
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999726032369969e-05,
+      "loss": 1.3631,
+      "step": 4812
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999725916074683e-05,
+      "loss": 1.3103,
+      "step": 4813
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99972579975472e-05,
+      "loss": 1.0344,
+      "step": 4814
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999725683410082e-05,
+      "loss": 1.7591,
+      "step": 4815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999725567040766e-05,
+      "loss": 1.6107,
+      "step": 4816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999725450646775e-05,
+      "loss": 1.2663,
+      "step": 4817
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997253342281075e-05,
+      "loss": 1.1365,
+      "step": 4818
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999725217784764e-05,
+      "loss": 0.8303,
+      "step": 4819
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997251013167435e-05,
+      "loss": 0.4453,
+      "step": 4820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997249848240474e-05,
+      "loss": 1.1691,
+      "step": 4821
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999724868306675e-05,
+      "loss": 1.5815,
+      "step": 4822
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999724751764626e-05,
+      "loss": 1.109,
+      "step": 4823
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999724635197901e-05,
+      "loss": 0.8534,
+      "step": 4824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997245186065e-05,
+      "loss": 1.1309,
+      "step": 4825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999724401990423e-05,
+      "loss": 1.1172,
+      "step": 4826
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999724285349669e-05,
+      "loss": 1.7788,
+      "step": 4827
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999724168684239e-05,
+      "loss": 1.3092,
+      "step": 4828
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997240519941335e-05,
+      "loss": 1.0972,
+      "step": 4829
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997239352793515e-05,
+      "loss": 0.6636,
+      "step": 4830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999723818539892e-05,
+      "loss": 0.448,
+      "step": 4831
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997237017757576e-05,
+      "loss": 0.4685,
+      "step": 4832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999723584986947e-05,
+      "loss": 0.3573,
+      "step": 4833
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999723468173459e-05,
+      "loss": 0.8482,
+      "step": 4834
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999723351335296e-05,
+      "loss": 1.3927,
+      "step": 4835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997232344724565e-05,
+      "loss": 1.117,
+      "step": 4836
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999723117584941e-05,
+      "loss": 1.0969,
+      "step": 4837
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997230006727486e-05,
+      "loss": 1.0169,
+      "step": 4838
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99972288373588e-05,
+      "loss": 0.8773,
+      "step": 4839
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997227667743364e-05,
+      "loss": 1.0116,
+      "step": 4840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997226497881156e-05,
+      "loss": 1.153,
+      "step": 4841
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997225327772184e-05,
+      "loss": 0.7223,
+      "step": 4842
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997224157416454e-05,
+      "loss": 0.5854,
+      "step": 4843
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997222986813966e-05,
+      "loss": 1.1327,
+      "step": 4844
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99972218159647e-05,
+      "loss": 1.3386,
+      "step": 4845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999722064486869e-05,
+      "loss": 0.6762,
+      "step": 4846
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999721947352591e-05,
+      "loss": 1.5002,
+      "step": 4847
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997218301936365e-05,
+      "loss": 1.6148,
+      "step": 4848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999721713010007e-05,
+      "loss": 1.7661,
+      "step": 4849
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997215958016995e-05,
+      "loss": 1.9748,
+      "step": 4850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999721478568718e-05,
+      "loss": 1.3449,
+      "step": 4851
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999721361311058e-05,
+      "loss": 1.1963,
+      "step": 4852
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997212440287234e-05,
+      "loss": 1.0218,
+      "step": 4853
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999721126721712e-05,
+      "loss": 1.3188,
+      "step": 4854
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999721009390025e-05,
+      "loss": 1.0898,
+      "step": 4855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997208920336606e-05,
+      "loss": 1.1505,
+      "step": 4856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999720774652621e-05,
+      "loss": 1.1612,
+      "step": 4857
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997206572469046e-05,
+      "loss": 1.2213,
+      "step": 4858
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997205398165126e-05,
+      "loss": 1.2131,
+      "step": 4859
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999720422361445e-05,
+      "loss": 1.0219,
+      "step": 4860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997203048817e-05,
+      "loss": 1.2768,
+      "step": 4861
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997201873772794e-05,
+      "loss": 1.0464,
+      "step": 4862
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997200698481816e-05,
+      "loss": 1.0345,
+      "step": 4863
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999719952294409e-05,
+      "loss": 1.2539,
+      "step": 4864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999719834715959e-05,
+      "loss": 1.0661,
+      "step": 4865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999719717112834e-05,
+      "loss": 1.0236,
+      "step": 4866
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999719599485032e-05,
+      "loss": 1.1617,
+      "step": 4867
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997194818325544e-05,
+      "loss": 0.8693,
+      "step": 4868
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997193641554e-05,
+      "loss": 1.0996,
+      "step": 4869
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997192464535705e-05,
+      "loss": 1.2102,
+      "step": 4870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999719128727063e-05,
+      "loss": 0.6804,
+      "step": 4871
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999719010975881e-05,
+      "loss": 0.9678,
+      "step": 4872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999718893200022e-05,
+      "loss": 0.9744,
+      "step": 4873
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997187753994876e-05,
+      "loss": 1.253,
+      "step": 4874
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999718657574276e-05,
+      "loss": 1.1902,
+      "step": 4875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997185397243885e-05,
+      "loss": 0.7328,
+      "step": 4876
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997184218498247e-05,
+      "loss": 0.7258,
+      "step": 4877
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999718303950586e-05,
+      "loss": 0.5045,
+      "step": 4878
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99971818602667e-05,
+      "loss": 0.601,
+      "step": 4879
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999718068078078e-05,
+      "loss": 0.4128,
+      "step": 4880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997179501048095e-05,
+      "loss": 0.7225,
+      "step": 4881
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997178321068655e-05,
+      "loss": 1.0695,
+      "step": 4882
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997177140842457e-05,
+      "loss": 1.2155,
+      "step": 4883
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999717596036949e-05,
+      "loss": 1.5828,
+      "step": 4884
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999717477964976e-05,
+      "loss": 0.9714,
+      "step": 4885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999717359868327e-05,
+      "loss": 1.0561,
+      "step": 4886
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999717241747002e-05,
+      "loss": 1.0278,
+      "step": 4887
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999717123601001e-05,
+      "loss": 0.9739,
+      "step": 4888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997170054303236e-05,
+      "loss": 1.0747,
+      "step": 4889
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99971688723497e-05,
+      "loss": 1.3003,
+      "step": 4890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999716769014941e-05,
+      "loss": 1.0654,
+      "step": 4891
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999716650770234e-05,
+      "loss": 1.1268,
+      "step": 4892
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999716532500852e-05,
+      "loss": 1.0623,
+      "step": 4893
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999716414206794e-05,
+      "loss": 1.1484,
+      "step": 4894
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99971629588806e-05,
+      "loss": 1.3828,
+      "step": 4895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997161775446496e-05,
+      "loss": 1.2322,
+      "step": 4896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999716059176563e-05,
+      "loss": 1.1974,
+      "step": 4897
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997159407838e-05,
+      "loss": 1.1084,
+      "step": 4898
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999715822366361e-05,
+      "loss": 1.2828,
+      "step": 4899
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997157039242466e-05,
+      "loss": 1.0407,
+      "step": 4900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999715585457455e-05,
+      "loss": 1.1489,
+      "step": 4901
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999715466965988e-05,
+      "loss": 1.4901,
+      "step": 4902
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999715348449845e-05,
+      "loss": 1.0606,
+      "step": 4903
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999715229909025e-05,
+      "loss": 1.0927,
+      "step": 4904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999715111343529e-05,
+      "loss": 1.3185,
+      "step": 4905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999714992753358e-05,
+      "loss": 1.3352,
+      "step": 4906
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99971487413851e-05,
+      "loss": 1.4179,
+      "step": 4907
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997147554989856e-05,
+      "loss": 1.1385,
+      "step": 4908
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999714636834785e-05,
+      "loss": 1.1543,
+      "step": 4909
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997145181459095e-05,
+      "loss": 1.3222,
+      "step": 4910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999714399432356e-05,
+      "loss": 1.5961,
+      "step": 4911
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997142806941276e-05,
+      "loss": 1.1299,
+      "step": 4912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997141619312234e-05,
+      "loss": 1.1121,
+      "step": 4913
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999714043143642e-05,
+      "loss": 1.0932,
+      "step": 4914
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999713924331385e-05,
+      "loss": 1.041,
+      "step": 4915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999713805494453e-05,
+      "loss": 1.1886,
+      "step": 4916
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997136866328434e-05,
+      "loss": 0.8678,
+      "step": 4917
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997135677465576e-05,
+      "loss": 0.9387,
+      "step": 4918
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999713448835596e-05,
+      "loss": 1.246,
+      "step": 4919
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999713329899959e-05,
+      "loss": 1.2206,
+      "step": 4920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999713210939645e-05,
+      "loss": 0.9171,
+      "step": 4921
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997130919546555e-05,
+      "loss": 1.0763,
+      "step": 4922
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997129729449896e-05,
+      "loss": 1.3132,
+      "step": 4923
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999712853910647e-05,
+      "loss": 0.8297,
+      "step": 4924
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999712734851629e-05,
+      "loss": 1.1485,
+      "step": 4925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997126157679344e-05,
+      "loss": 0.9347,
+      "step": 4926
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999712496659564e-05,
+      "loss": 1.1448,
+      "step": 4927
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999712377526518e-05,
+      "loss": 1.4814,
+      "step": 4928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997122583687954e-05,
+      "loss": 0.9763,
+      "step": 4929
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997121391863964e-05,
+      "loss": 1.3074,
+      "step": 4930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999712019979322e-05,
+      "loss": 1.1433,
+      "step": 4931
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999711900747571e-05,
+      "loss": 1.2763,
+      "step": 4932
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999711781491144e-05,
+      "loss": 0.9959,
+      "step": 4933
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999711662210041e-05,
+      "loss": 0.9883,
+      "step": 4934
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997115429042615e-05,
+      "loss": 1.2426,
+      "step": 4935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997114235738066e-05,
+      "loss": 1.2025,
+      "step": 4936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999711304218675e-05,
+      "loss": 1.0928,
+      "step": 4937
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999711184838868e-05,
+      "loss": 0.9726,
+      "step": 4938
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999711065434384e-05,
+      "loss": 1.0069,
+      "step": 4939
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997109460052244e-05,
+      "loss": 1.2129,
+      "step": 4940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997108265513886e-05,
+      "loss": 0.8876,
+      "step": 4941
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997107070728764e-05,
+      "loss": 1.1348,
+      "step": 4942
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997105875696884e-05,
+      "loss": 1.8116,
+      "step": 4943
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997104680418247e-05,
+      "loss": 1.475,
+      "step": 4944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999710348489285e-05,
+      "loss": 0.8913,
+      "step": 4945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997102289120685e-05,
+      "loss": 1.3088,
+      "step": 4946
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999710109310176e-05,
+      "loss": 1.3785,
+      "step": 4947
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999709989683607e-05,
+      "loss": 0.842,
+      "step": 4948
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999709870032363e-05,
+      "loss": 1.0386,
+      "step": 4949
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999709750356443e-05,
+      "loss": 1.2069,
+      "step": 4950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999709630655846e-05,
+      "loss": 1.0357,
+      "step": 4951
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997095109305735e-05,
+      "loss": 0.5687,
+      "step": 4952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997093911806245e-05,
+      "loss": 0.929,
+      "step": 4953
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999709271406e-05,
+      "loss": 1.0482,
+      "step": 4954
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997091516066985e-05,
+      "loss": 1.1471,
+      "step": 4955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997090317827215e-05,
+      "loss": 1.1224,
+      "step": 4956
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999708911934069e-05,
+      "loss": 1.1038,
+      "step": 4957
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997087920607395e-05,
+      "loss": 1.1545,
+      "step": 4958
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999708672162734e-05,
+      "loss": 1.0817,
+      "step": 4959
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997085522400525e-05,
+      "loss": 0.8527,
+      "step": 4960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999708432292696e-05,
+      "loss": 1.2293,
+      "step": 4961
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997083123206624e-05,
+      "loss": 1.2644,
+      "step": 4962
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999708192323952e-05,
+      "loss": 1.1254,
+      "step": 4963
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999708072302567e-05,
+      "loss": 1.0951,
+      "step": 4964
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999707952256505e-05,
+      "loss": 0.8272,
+      "step": 4965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999707832185767e-05,
+      "loss": 1.4679,
+      "step": 4966
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999707712090354e-05,
+      "loss": 1.8612,
+      "step": 4967
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999707591970264e-05,
+      "loss": 1.3245,
+      "step": 4968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997074718254976e-05,
+      "loss": 1.3916,
+      "step": 4969
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999707351656056e-05,
+      "loss": 1.0122,
+      "step": 4970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997072314619376e-05,
+      "loss": 0.8691,
+      "step": 4971
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999707111243144e-05,
+      "loss": 1.1927,
+      "step": 4972
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706990999674e-05,
+      "loss": 0.959,
+      "step": 4973
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706870731528e-05,
+      "loss": 1.1453,
+      "step": 4974
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706750438705e-05,
+      "loss": 1.2579,
+      "step": 4975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706630121207e-05,
+      "loss": 1.1374,
+      "step": 4976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706509779033e-05,
+      "loss": 1.4763,
+      "step": 4977
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706389412182e-05,
+      "loss": 1.2916,
+      "step": 4978
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706269020656e-05,
+      "loss": 1.1468,
+      "step": 4979
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706148604453e-05,
+      "loss": 0.9316,
+      "step": 4980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999706028163575e-05,
+      "loss": 0.7751,
+      "step": 4981
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99970590769802e-05,
+      "loss": 1.495,
+      "step": 4982
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997057872077894e-05,
+      "loss": 1.0799,
+      "step": 4983
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999705666692883e-05,
+      "loss": 1.0632,
+      "step": 4984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997055461533006e-05,
+      "loss": 1.1453,
+      "step": 4985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999705425589042e-05,
+      "loss": 0.639,
+      "step": 4986
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999705305000106e-05,
+      "loss": 0.8285,
+      "step": 4987
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997051843864965e-05,
+      "loss": 1.1244,
+      "step": 4988
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999705063748209e-05,
+      "loss": 0.9547,
+      "step": 4989
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999704943085246e-05,
+      "loss": 0.4331,
+      "step": 4990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999704822397607e-05,
+      "loss": 0.1439,
+      "step": 4991
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997047016852924e-05,
+      "loss": 1.0579,
+      "step": 4992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997045809483013e-05,
+      "loss": 1.2167,
+      "step": 4993
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997044601866345e-05,
+      "loss": 1.0966,
+      "step": 4994
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999704339400291e-05,
+      "loss": 1.6641,
+      "step": 4995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999704218589272e-05,
+      "loss": 1.1495,
+      "step": 4996
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999704097753577e-05,
+      "loss": 1.1242,
+      "step": 4997
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999703976893207e-05,
+      "loss": 1.1382,
+      "step": 4998
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999703856008159e-05,
+      "loss": 1.2479,
+      "step": 4999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997037350984355e-05,
+      "loss": 1.1091,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0793377161026,
+      "eval_runtime": 84.714,
+      "eval_samples_per_second": 16.349,
+      "eval_steps_per_second": 4.096,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999703614164036e-05,
+      "loss": 1.0244,
+      "step": 5001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999703493204961e-05,
+      "loss": 1.2538,
+      "step": 5002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99970337222121e-05,
+      "loss": 0.9095,
+      "step": 5003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997032512127826e-05,
+      "loss": 1.175,
+      "step": 5004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999703130179679e-05,
+      "loss": 0.9529,
+      "step": 5005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997030091219e-05,
+      "loss": 1.0424,
+      "step": 5006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999702888039445e-05,
+      "loss": 0.7228,
+      "step": 5007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999702766932314e-05,
+      "loss": 1.1532,
+      "step": 5008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999702645800506e-05,
+      "loss": 0.8094,
+      "step": 5009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997025246440236e-05,
+      "loss": 1.148,
+      "step": 5010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999702403462864e-05,
+      "loss": 1.2554,
+      "step": 5011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997022822570286e-05,
+      "loss": 1.2795,
+      "step": 5012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999702161026517e-05,
+      "loss": 1.2422,
+      "step": 5013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999702039771329e-05,
+      "loss": 1.2922,
+      "step": 5014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999701918491466e-05,
+      "loss": 1.2931,
+      "step": 5015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999701797186926e-05,
+      "loss": 1.2758,
+      "step": 5016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999701675857711e-05,
+      "loss": 1.284,
+      "step": 5017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99970155450382e-05,
+      "loss": 1.0614,
+      "step": 5018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999701433125252e-05,
+      "loss": 1.303,
+      "step": 5019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997013117220095e-05,
+      "loss": 1.4622,
+      "step": 5020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99970119029409e-05,
+      "loss": 1.2645,
+      "step": 5021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997010688414945e-05,
+      "loss": 1.2847,
+      "step": 5022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999700947364223e-05,
+      "loss": 1.1669,
+      "step": 5023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999700825862276e-05,
+      "loss": 0.8389,
+      "step": 5024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999700704335653e-05,
+      "loss": 1.2053,
+      "step": 5025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999700582784353e-05,
+      "loss": 0.8173,
+      "step": 5026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997004612083775e-05,
+      "loss": 0.9234,
+      "step": 5027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997003396077266e-05,
+      "loss": 1.0486,
+      "step": 5028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999700217982399e-05,
+      "loss": 1.0777,
+      "step": 5029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999700096332396e-05,
+      "loss": 1.1065,
+      "step": 5030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996999746577164e-05,
+      "loss": 0.9859,
+      "step": 5031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999699852958361e-05,
+      "loss": 1.3586,
+      "step": 5032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969973123433e-05,
+      "loss": 1.192,
+      "step": 5033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999699609485622e-05,
+      "loss": 0.9577,
+      "step": 5034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996994877122396e-05,
+      "loss": 0.8894,
+      "step": 5035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969936591418e-05,
+      "loss": 0.8399,
+      "step": 5036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999699244091445e-05,
+      "loss": 0.7582,
+      "step": 5037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999699122244034e-05,
+      "loss": 1.0065,
+      "step": 5038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996990003719466e-05,
+      "loss": 0.5715,
+      "step": 5039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999698878475184e-05,
+      "loss": 0.5834,
+      "step": 5040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996987565537445e-05,
+      "loss": 0.8216,
+      "step": 5041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996986346076294e-05,
+      "loss": 1.2597,
+      "step": 5042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996985126368386e-05,
+      "loss": 1.2998,
+      "step": 5043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996983906413714e-05,
+      "loss": 1.0993,
+      "step": 5044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996982686212284e-05,
+      "loss": 0.9802,
+      "step": 5045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999698146576409e-05,
+      "loss": 1.2207,
+      "step": 5046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999698024506915e-05,
+      "loss": 0.8932,
+      "step": 5047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996979024127434e-05,
+      "loss": 0.8309,
+      "step": 5048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999697780293897e-05,
+      "loss": 0.8186,
+      "step": 5049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999697658150374e-05,
+      "loss": 1.1159,
+      "step": 5050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999697535982175e-05,
+      "loss": 1.0577,
+      "step": 5051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996974137893005e-05,
+      "loss": 0.9394,
+      "step": 5052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969729157175e-05,
+      "loss": 1.1254,
+      "step": 5053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999697169329523e-05,
+      "loss": 1.2276,
+      "step": 5054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996970470626204e-05,
+      "loss": 0.9111,
+      "step": 5055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996969247710426e-05,
+      "loss": 0.6828,
+      "step": 5056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999696802454788e-05,
+      "loss": 0.4971,
+      "step": 5057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999696680113857e-05,
+      "loss": 0.3282,
+      "step": 5058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996965577482506e-05,
+      "loss": 0.7744,
+      "step": 5059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999696435357968e-05,
+      "loss": 1.1189,
+      "step": 5060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969631294301e-05,
+      "loss": 0.795,
+      "step": 5061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996961905033755e-05,
+      "loss": 1.2361,
+      "step": 5062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999696068039065e-05,
+      "loss": 1.1867,
+      "step": 5063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996959455500794e-05,
+      "loss": 1.2989,
+      "step": 5064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999695823036417e-05,
+      "loss": 1.2148,
+      "step": 5065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996957004980796e-05,
+      "loss": 0.9647,
+      "step": 5066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999695577935065e-05,
+      "loss": 1.4714,
+      "step": 5067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996954553473754e-05,
+      "loss": 1.3342,
+      "step": 5068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999695332735009e-05,
+      "loss": 1.1808,
+      "step": 5069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996952100979674e-05,
+      "loss": 1.1745,
+      "step": 5070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969508743625e-05,
+      "loss": 1.0786,
+      "step": 5071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999694964749856e-05,
+      "loss": 1.0861,
+      "step": 5072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996948420387866e-05,
+      "loss": 1.1887,
+      "step": 5073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999694719303041e-05,
+      "loss": 1.0008,
+      "step": 5074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969459654262e-05,
+      "loss": 1.1709,
+      "step": 5075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999694473757522e-05,
+      "loss": 1.1251,
+      "step": 5076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999694350947749e-05,
+      "loss": 1.036,
+      "step": 5077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996942281133e-05,
+      "loss": 1.4192,
+      "step": 5078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999694105254174e-05,
+      "loss": 0.9987,
+      "step": 5079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999693982370373e-05,
+      "loss": 1.015,
+      "step": 5080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999693859461896e-05,
+      "loss": 0.9053,
+      "step": 5081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996937365287434e-05,
+      "loss": 1.1885,
+      "step": 5082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999693613570914e-05,
+      "loss": 1.2264,
+      "step": 5083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996934905884094e-05,
+      "loss": 1.0292,
+      "step": 5084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996933675812284e-05,
+      "loss": 0.6453,
+      "step": 5085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996932445493716e-05,
+      "loss": 1.0079,
+      "step": 5086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999693121492839e-05,
+      "loss": 1.0922,
+      "step": 5087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969299841163e-05,
+      "loss": 1.1088,
+      "step": 5088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999692875305746e-05,
+      "loss": 1.046,
+      "step": 5089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996927521751854e-05,
+      "loss": 1.7119,
+      "step": 5090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969262901995e-05,
+      "loss": 1.4077,
+      "step": 5091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999692505840038e-05,
+      "loss": 1.9513,
+      "step": 5092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999692382635449e-05,
+      "loss": 1.406,
+      "step": 5093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999692259406186e-05,
+      "loss": 1.2516,
+      "step": 5094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999692136152245e-05,
+      "loss": 1.1783,
+      "step": 5095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99969201287363e-05,
+      "loss": 0.7014,
+      "step": 5096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996918895703384e-05,
+      "loss": 0.8977,
+      "step": 5097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996917662423704e-05,
+      "loss": 0.8868,
+      "step": 5098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999691642889727e-05,
+      "loss": 1.018,
+      "step": 5099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999691519512407e-05,
+      "loss": 1.0971,
+      "step": 5100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999691396110412e-05,
+      "loss": 1.1603,
+      "step": 5101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999691272683741e-05,
+      "loss": 0.1859,
+      "step": 5102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999691149232394e-05,
+      "loss": 0.1624,
+      "step": 5103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996910257563715e-05,
+      "loss": 0.1466,
+      "step": 5104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999690902255672e-05,
+      "loss": 0.1275,
+      "step": 5105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999690778730297e-05,
+      "loss": 0.1011,
+      "step": 5106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996906551802465e-05,
+      "loss": 0.0959,
+      "step": 5107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996905316055195e-05,
+      "loss": 0.2996,
+      "step": 5108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996904080061175e-05,
+      "loss": 1.068,
+      "step": 5109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999690284382039e-05,
+      "loss": 1.4908,
+      "step": 5110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999690160733285e-05,
+      "loss": 1.1142,
+      "step": 5111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999690037059854e-05,
+      "loss": 1.3284,
+      "step": 5112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999689913361749e-05,
+      "loss": 1.0132,
+      "step": 5113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996897896389666e-05,
+      "loss": 1.1871,
+      "step": 5114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996896658915086e-05,
+      "loss": 0.8212,
+      "step": 5115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999689542119375e-05,
+      "loss": 0.6875,
+      "step": 5116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996894183225653e-05,
+      "loss": 1.264,
+      "step": 5117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99968929450108e-05,
+      "loss": 1.0435,
+      "step": 5118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999689170654919e-05,
+      "loss": 1.2933,
+      "step": 5119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996890467840815e-05,
+      "loss": 1.3859,
+      "step": 5120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999688922888568e-05,
+      "loss": 0.7856,
+      "step": 5121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999688798968379e-05,
+      "loss": 0.5347,
+      "step": 5122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999688675023515e-05,
+      "loss": 0.4216,
+      "step": 5123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999688551053974e-05,
+      "loss": 1.0338,
+      "step": 5124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996884270597576e-05,
+      "loss": 1.2116,
+      "step": 5125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999688303040865e-05,
+      "loss": 1.1681,
+      "step": 5126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999688178997296e-05,
+      "loss": 1.3969,
+      "step": 5127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999688054929053e-05,
+      "loss": 1.1493,
+      "step": 5128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999687930836132e-05,
+      "loss": 1.4235,
+      "step": 5129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999687806718536e-05,
+      "loss": 0.9103,
+      "step": 5130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996876825762646e-05,
+      "loss": 0.8201,
+      "step": 5131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999687558409317e-05,
+      "loss": 1.1494,
+      "step": 5132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996874342176934e-05,
+      "loss": 1.1671,
+      "step": 5133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999687310001394e-05,
+      "loss": 1.1026,
+      "step": 5134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996871857604186e-05,
+      "loss": 1.0921,
+      "step": 5135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996870614947675e-05,
+      "loss": 1.3104,
+      "step": 5136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999686937204441e-05,
+      "loss": 1.1806,
+      "step": 5137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999686812889438e-05,
+      "loss": 1.0932,
+      "step": 5138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99968668854976e-05,
+      "loss": 0.892,
+      "step": 5139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999686564185405e-05,
+      "loss": 0.6406,
+      "step": 5140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999686439796374e-05,
+      "loss": 0.4112,
+      "step": 5141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999686315382668e-05,
+      "loss": 0.5258,
+      "step": 5142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999686190944286e-05,
+      "loss": 0.7285,
+      "step": 5143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999686066481229e-05,
+      "loss": 1.1035,
+      "step": 5144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996859419934944e-05,
+      "loss": 0.8942,
+      "step": 5145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996858174810843e-05,
+      "loss": 1.0368,
+      "step": 5146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999685692943999e-05,
+      "loss": 0.9548,
+      "step": 5147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999685568382238e-05,
+      "loss": 1.049,
+      "step": 5148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999685443795801e-05,
+      "loss": 0.7284,
+      "step": 5149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996853191846885e-05,
+      "loss": 1.198,
+      "step": 5150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999685194548899e-05,
+      "loss": 1.3568,
+      "step": 5151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999685069888434e-05,
+      "loss": 1.1284,
+      "step": 5152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999684945203293e-05,
+      "loss": 0.92,
+      "step": 5153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999684820493478e-05,
+      "loss": 1.5527,
+      "step": 5154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999684695758986e-05,
+      "loss": 1.4233,
+      "step": 5155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996845709998173e-05,
+      "loss": 1.3858,
+      "step": 5156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999684446215973e-05,
+      "loss": 0.9554,
+      "step": 5157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999684321407454e-05,
+      "loss": 1.1443,
+      "step": 5158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999684196574258e-05,
+      "loss": 1.0987,
+      "step": 5159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996840717163875e-05,
+      "loss": 0.961,
+      "step": 5160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996839468338396e-05,
+      "loss": 0.8515,
+      "step": 5161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999683821926617e-05,
+      "loss": 1.2304,
+      "step": 5162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999683696994718e-05,
+      "loss": 1.3475,
+      "step": 5163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999683572038143e-05,
+      "loss": 1.1656,
+      "step": 5164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999683447056893e-05,
+      "loss": 1.0646,
+      "step": 5165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996833220509665e-05,
+      "loss": 1.1677,
+      "step": 5166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999683197020365e-05,
+      "loss": 1.1873,
+      "step": 5167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996830719650865e-05,
+      "loss": 1.361,
+      "step": 5168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996829468851325e-05,
+      "loss": 1.0357,
+      "step": 5169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996828217805034e-05,
+      "loss": 1.2511,
+      "step": 5170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999682696651198e-05,
+      "loss": 1.0471,
+      "step": 5171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996825714972166e-05,
+      "loss": 1.0212,
+      "step": 5172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996824463185596e-05,
+      "loss": 0.9291,
+      "step": 5173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999682321115227e-05,
+      "loss": 0.6947,
+      "step": 5174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999682195887219e-05,
+      "loss": 0.5708,
+      "step": 5175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999682070634534e-05,
+      "loss": 0.3236,
+      "step": 5176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999681945357174e-05,
+      "loss": 0.3532,
+      "step": 5177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999681820055138e-05,
+      "loss": 0.3641,
+      "step": 5178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996816947284256e-05,
+      "loss": 0.749,
+      "step": 5179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999681569377038e-05,
+      "loss": 1.5034,
+      "step": 5180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996814440009744e-05,
+      "loss": 1.1136,
+      "step": 5181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996813186002355e-05,
+      "loss": 0.9345,
+      "step": 5182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99968119317482e-05,
+      "loss": 1.3294,
+      "step": 5183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996810677247296e-05,
+      "loss": 1.0999,
+      "step": 5184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999680942249962e-05,
+      "loss": 0.8845,
+      "step": 5185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999680816750521e-05,
+      "loss": 1.214,
+      "step": 5186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996806912264024e-05,
+      "loss": 1.164,
+      "step": 5187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999680565677608e-05,
+      "loss": 0.9096,
+      "step": 5188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999680440104138e-05,
+      "loss": 0.6272,
+      "step": 5189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996803145059925e-05,
+      "loss": 1.0698,
+      "step": 5190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999680188883171e-05,
+      "loss": 1.1675,
+      "step": 5191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999680063235674e-05,
+      "loss": 1.1066,
+      "step": 5192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999679937563501e-05,
+      "loss": 1.0981,
+      "step": 5193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999679811866652e-05,
+      "loss": 0.9234,
+      "step": 5194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996796861451275e-05,
+      "loss": 0.9038,
+      "step": 5195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999679560398927e-05,
+      "loss": 1.1016,
+      "step": 5196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996794346280505e-05,
+      "loss": 1.4688,
+      "step": 5197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999679308832499e-05,
+      "loss": 1.0326,
+      "step": 5198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999679183012271e-05,
+      "loss": 0.9382,
+      "step": 5199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999679057167368e-05,
+      "loss": 1.2755,
+      "step": 5200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999678931297789e-05,
+      "loss": 1.295,
+      "step": 5201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999678805403534e-05,
+      "loss": 1.2337,
+      "step": 5202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999678679484603e-05,
+      "loss": 1.0238,
+      "step": 5203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999678553540996e-05,
+      "loss": 1.218,
+      "step": 5204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999678427572714e-05,
+      "loss": 1.5987,
+      "step": 5205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999678301579755e-05,
+      "loss": 1.1188,
+      "step": 5206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996781755621216e-05,
+      "loss": 1.1451,
+      "step": 5207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999678049519812e-05,
+      "loss": 1.0454,
+      "step": 5208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999677923452827e-05,
+      "loss": 1.0136,
+      "step": 5209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999677797361165e-05,
+      "loss": 1.1909,
+      "step": 5210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996776712448285e-05,
+      "loss": 0.9326,
+      "step": 5211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999677545103816e-05,
+      "loss": 1.2454,
+      "step": 5212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999677418938127e-05,
+      "loss": 0.8697,
+      "step": 5213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999677292747763e-05,
+      "loss": 1.1012,
+      "step": 5214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999677166532723e-05,
+      "loss": 0.8678,
+      "step": 5215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999677040293007e-05,
+      "loss": 1.119,
+      "step": 5216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996769140286156e-05,
+      "loss": 1.3689,
+      "step": 5217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996767877395485e-05,
+      "loss": 1.0955,
+      "step": 5218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999676661425805e-05,
+      "loss": 1.2735,
+      "step": 5219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996765350873864e-05,
+      "loss": 1.2391,
+      "step": 5220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999676408724292e-05,
+      "loss": 0.8565,
+      "step": 5221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999676282336522e-05,
+      "loss": 0.9062,
+      "step": 5222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999676155924076e-05,
+      "loss": 1.0853,
+      "step": 5223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996760294869536e-05,
+      "loss": 1.4615,
+      "step": 5224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999675903025156e-05,
+      "loss": 0.9734,
+      "step": 5225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999675776538682e-05,
+      "loss": 1.1386,
+      "step": 5226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999675650027534e-05,
+      "loss": 1.1606,
+      "step": 5227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999675523491709e-05,
+      "loss": 1.2047,
+      "step": 5228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999675396931208e-05,
+      "loss": 1.385,
+      "step": 5229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999675270346032e-05,
+      "loss": 1.4627,
+      "step": 5230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99967514373618e-05,
+      "loss": 0.9382,
+      "step": 5231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999675017101653e-05,
+      "loss": 0.9209,
+      "step": 5232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996748904424484e-05,
+      "loss": 0.8517,
+      "step": 5233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99967476375857e-05,
+      "loss": 0.8616,
+      "step": 5234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999674637050015e-05,
+      "loss": 1.0328,
+      "step": 5235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999674510316784e-05,
+      "loss": 1.0035,
+      "step": 5236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999674383558877e-05,
+      "loss": 0.9074,
+      "step": 5237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996742567762956e-05,
+      "loss": 0.9313,
+      "step": 5238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996741299690374e-05,
+      "loss": 1.2877,
+      "step": 5239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996740031371035e-05,
+      "loss": 1.197,
+      "step": 5240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999673876280494e-05,
+      "loss": 1.0386,
+      "step": 5241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999673749399209e-05,
+      "loss": 1.2689,
+      "step": 5242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996736224932486e-05,
+      "loss": 1.027,
+      "step": 5243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999673495562612e-05,
+      "loss": 1.3627,
+      "step": 5244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996733686072996e-05,
+      "loss": 1.1203,
+      "step": 5245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999673241627312e-05,
+      "loss": 1.0976,
+      "step": 5246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996731146226476e-05,
+      "loss": 1.2692,
+      "step": 5247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999672987593308e-05,
+      "loss": 0.8317,
+      "step": 5248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999672860539293e-05,
+      "loss": 0.8718,
+      "step": 5249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996727334606016e-05,
+      "loss": 0.961,
+      "step": 5250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999672606357236e-05,
+      "loss": 1.0874,
+      "step": 5251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999672479229193e-05,
+      "loss": 1.4313,
+      "step": 5252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999672352076475e-05,
+      "loss": 1.0807,
+      "step": 5253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999672224899081e-05,
+      "loss": 1.7528,
+      "step": 5254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999672097697012e-05,
+      "loss": 1.6995,
+      "step": 5255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996719704702675e-05,
+      "loss": 1.2513,
+      "step": 5256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999671843218846e-05,
+      "loss": 1.2308,
+      "step": 5257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99967171594275e-05,
+      "loss": 1.3474,
+      "step": 5258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999671588641978e-05,
+      "loss": 1.0074,
+      "step": 5259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996714613165294e-05,
+      "loss": 1.315,
+      "step": 5260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999671333966406e-05,
+      "loss": 1.0775,
+      "step": 5261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996712065916065e-05,
+      "loss": 1.1714,
+      "step": 5262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999671079192131e-05,
+      "loss": 1.1032,
+      "step": 5263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996709517679805e-05,
+      "loss": 0.8256,
+      "step": 5264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999670824319154e-05,
+      "loss": 1.2036,
+      "step": 5265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999670696845652e-05,
+      "loss": 1.1551,
+      "step": 5266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999670569347474e-05,
+      "loss": 1.088,
+      "step": 5267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999670441824621e-05,
+      "loss": 1.1427,
+      "step": 5268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99967031427709e-05,
+      "loss": 1.334,
+      "step": 5269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999670186704886e-05,
+      "loss": 0.8837,
+      "step": 5270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999670059108005e-05,
+      "loss": 1.1194,
+      "step": 5271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996699314864494e-05,
+      "loss": 1.1684,
+      "step": 5272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996698038402166e-05,
+      "loss": 0.5408,
+      "step": 5273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996696761693095e-05,
+      "loss": 0.364,
+      "step": 5274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999669548473726e-05,
+      "loss": 1.16,
+      "step": 5275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996694207534665e-05,
+      "loss": 0.7768,
+      "step": 5276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999669293008532e-05,
+      "loss": 1.2486,
+      "step": 5277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999669165238922e-05,
+      "loss": 0.9353,
+      "step": 5278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999669037444635e-05,
+      "loss": 1.4989,
+      "step": 5279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999668909625674e-05,
+      "loss": 1.5388,
+      "step": 5280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996687817820366e-05,
+      "loss": 1.2112,
+      "step": 5281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999668653913723e-05,
+      "loss": 0.9339,
+      "step": 5282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999668526020734e-05,
+      "loss": 1.2608,
+      "step": 5283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99966839810307e-05,
+      "loss": 1.1584,
+      "step": 5284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99966827016073e-05,
+      "loss": 0.8799,
+      "step": 5285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999668142193714e-05,
+      "loss": 1.1249,
+      "step": 5286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996680142020226e-05,
+      "loss": 0.9992,
+      "step": 5287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999667886185655e-05,
+      "loss": 0.7114,
+      "step": 5288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996677581446125e-05,
+      "loss": 1.2043,
+      "step": 5289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999667630078894e-05,
+      "loss": 0.8246,
+      "step": 5290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996675019885e-05,
+      "loss": 0.8577,
+      "step": 5291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99966737387343e-05,
+      "loss": 0.9255,
+      "step": 5292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999667245733684e-05,
+      "loss": 1.7593,
+      "step": 5293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999667117569263e-05,
+      "loss": 1.4936,
+      "step": 5294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996669893801665e-05,
+      "loss": 1.2999,
+      "step": 5295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999666861166393e-05,
+      "loss": 1.7748,
+      "step": 5296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996667329279455e-05,
+      "loss": 4.5491,
+      "step": 5297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996666046648214e-05,
+      "loss": 1.2272,
+      "step": 5298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996664763770215e-05,
+      "loss": 1.1773,
+      "step": 5299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996663480645465e-05,
+      "loss": 1.0385,
+      "step": 5300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999666219727396e-05,
+      "loss": 0.9458,
+      "step": 5301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99966609136557e-05,
+      "loss": 1.2472,
+      "step": 5302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999665962979068e-05,
+      "loss": 1.0255,
+      "step": 5303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996658345678896e-05,
+      "loss": 0.7737,
+      "step": 5304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996657061320365e-05,
+      "loss": 1.0313,
+      "step": 5305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996655776715076e-05,
+      "loss": 1.3828,
+      "step": 5306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999665449186303e-05,
+      "loss": 0.9499,
+      "step": 5307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996653206764225e-05,
+      "loss": 1.0777,
+      "step": 5308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999665192141867e-05,
+      "loss": 1.1835,
+      "step": 5309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999665063582635e-05,
+      "loss": 0.9606,
+      "step": 5310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999664934998728e-05,
+      "loss": 1.0589,
+      "step": 5311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999664806390145e-05,
+      "loss": 1.1687,
+      "step": 5312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999664677756887e-05,
+      "loss": 0.9827,
+      "step": 5313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996645490989524e-05,
+      "loss": 0.867,
+      "step": 5314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999664420416342e-05,
+      "loss": 1.2698,
+      "step": 5315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999664291709057e-05,
+      "loss": 1.3419,
+      "step": 5316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996641629770955e-05,
+      "loss": 1.3358,
+      "step": 5317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996640342204595e-05,
+      "loss": 0.9363,
+      "step": 5318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999663905439147e-05,
+      "loss": 0.6667,
+      "step": 5319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999663776633159e-05,
+      "loss": 0.7523,
+      "step": 5320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996636478024955e-05,
+      "loss": 1.2398,
+      "step": 5321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996635189471564e-05,
+      "loss": 1.123,
+      "step": 5322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996633900671416e-05,
+      "loss": 1.1395,
+      "step": 5323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999663261162451e-05,
+      "loss": 1.1425,
+      "step": 5324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999663132233085e-05,
+      "loss": 1.1347,
+      "step": 5325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999663003279043e-05,
+      "loss": 0.7952,
+      "step": 5326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999662874300326e-05,
+      "loss": 1.2129,
+      "step": 5327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999662745296933e-05,
+      "loss": 1.2426,
+      "step": 5328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999662616268864e-05,
+      "loss": 1.0447,
+      "step": 5329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996624872161204e-05,
+      "loss": 1.2094,
+      "step": 5330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996623581387e-05,
+      "loss": 1.3758,
+      "step": 5331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996622290366054e-05,
+      "loss": 1.2301,
+      "step": 5332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999662099909834e-05,
+      "loss": 1.4833,
+      "step": 5333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996619707583874e-05,
+      "loss": 1.1072,
+      "step": 5334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999661841582265e-05,
+      "loss": 1.3784,
+      "step": 5335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999661712381467e-05,
+      "loss": 1.0261,
+      "step": 5336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999661583155994e-05,
+      "loss": 0.9837,
+      "step": 5337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999661453905845e-05,
+      "loss": 1.0831,
+      "step": 5338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996613246310205e-05,
+      "loss": 1.2405,
+      "step": 5339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99966119533152e-05,
+      "loss": 0.954,
+      "step": 5340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999661066007344e-05,
+      "loss": 1.0908,
+      "step": 5341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999660936658492e-05,
+      "loss": 1.2137,
+      "step": 5342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999660807284966e-05,
+      "loss": 1.3878,
+      "step": 5343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999660677886763e-05,
+      "loss": 0.8494,
+      "step": 5344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999660548463885e-05,
+      "loss": 1.1138,
+      "step": 5345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999660419016331e-05,
+      "loss": 1.2996,
+      "step": 5346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996602895441016e-05,
+      "loss": 2.2952,
+      "step": 5347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999660160047197e-05,
+      "loss": 3.8883,
+      "step": 5348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999660030525616e-05,
+      "loss": 1.2231,
+      "step": 5349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965990097936e-05,
+      "loss": 1.2186,
+      "step": 5350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999659771408428e-05,
+      "loss": 1.0336,
+      "step": 5351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965964181282e-05,
+      "loss": 1.1406,
+      "step": 5352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999659512192537e-05,
+      "loss": 1.2545,
+      "step": 5353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999659382547579e-05,
+      "loss": 0.7299,
+      "step": 5354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999659252877945e-05,
+      "loss": 0.9641,
+      "step": 5355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999659123183636e-05,
+      "loss": 0.8678,
+      "step": 5356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996589934646496e-05,
+      "loss": 1.1279,
+      "step": 5357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996588637209885e-05,
+      "loss": 0.6869,
+      "step": 5358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999658733952652e-05,
+      "loss": 0.8084,
+      "step": 5359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965860415964e-05,
+      "loss": 1.1045,
+      "step": 5360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996584743419525e-05,
+      "loss": 1.1457,
+      "step": 5361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999658344499589e-05,
+      "loss": 1.0041,
+      "step": 5362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999658214632551e-05,
+      "loss": 1.2116,
+      "step": 5363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996580847408366e-05,
+      "loss": 1.0101,
+      "step": 5364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996579548244465e-05,
+      "loss": 0.4227,
+      "step": 5365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996578248833806e-05,
+      "loss": 0.871,
+      "step": 5366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965769491764e-05,
+      "loss": 1.2822,
+      "step": 5367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996575649272235e-05,
+      "loss": 3.2187,
+      "step": 5368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999657434912131e-05,
+      "loss": 1.2285,
+      "step": 5369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996573048723633e-05,
+      "loss": 1.0293,
+      "step": 5370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965717480792e-05,
+      "loss": 1.2138,
+      "step": 5371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999657044718801e-05,
+      "loss": 1.1882,
+      "step": 5372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996569146050066e-05,
+      "loss": 1.2422,
+      "step": 5373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996567844665366e-05,
+      "loss": 1.2266,
+      "step": 5374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999656654303391e-05,
+      "loss": 1.0654,
+      "step": 5375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965652411557e-05,
+      "loss": 1.1164,
+      "step": 5376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996563939030734e-05,
+      "loss": 1.0866,
+      "step": 5377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999656263665901e-05,
+      "loss": 1.1725,
+      "step": 5378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999656133404053e-05,
+      "loss": 0.8584,
+      "step": 5379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99965600311753e-05,
+      "loss": 1.1981,
+      "step": 5380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996558728063314e-05,
+      "loss": 1.1273,
+      "step": 5381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999655742470457e-05,
+      "loss": 1.2806,
+      "step": 5382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999655612109907e-05,
+      "loss": 1.8106,
+      "step": 5383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999655481724681e-05,
+      "loss": 1.6897,
+      "step": 5384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996553513147806e-05,
+      "loss": 1.2792,
+      "step": 5385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996552208802035e-05,
+      "loss": 1.7448,
+      "step": 5386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999655090420951e-05,
+      "loss": 1.6907,
+      "step": 5387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999654959937023e-05,
+      "loss": 1.7045,
+      "step": 5388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999654829428421e-05,
+      "loss": 1.6589,
+      "step": 5389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996546988951414e-05,
+      "loss": 1.6445,
+      "step": 5390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996545683371875e-05,
+      "loss": 1.6535,
+      "step": 5391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999654437754557e-05,
+      "loss": 1.6749,
+      "step": 5392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996543071472524e-05,
+      "loss": 1.65,
+      "step": 5393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999654176515271e-05,
+      "loss": 1.6437,
+      "step": 5394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999654045858615e-05,
+      "loss": 1.6235,
+      "step": 5395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999653915177283e-05,
+      "loss": 1.6279,
+      "step": 5396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999653784471276e-05,
+      "loss": 1.645,
+      "step": 5397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999653653740592e-05,
+      "loss": 1.6092,
+      "step": 5398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996535229852335e-05,
+      "loss": 1.6206,
+      "step": 5399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999653392205199e-05,
+      "loss": 1.6062,
+      "step": 5400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996532614004896e-05,
+      "loss": 1.6103,
+      "step": 5401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999653130571105e-05,
+      "loss": 1.5852,
+      "step": 5402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652999717044e-05,
+      "loss": 1.5889,
+      "step": 5403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652868838308e-05,
+      "loss": 1.632,
+      "step": 5404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652737934896e-05,
+      "loss": 1.5914,
+      "step": 5405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652607006808e-05,
+      "loss": 1.5206,
+      "step": 5406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652476054046e-05,
+      "loss": 1.5644,
+      "step": 5407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652345076608e-05,
+      "loss": 1.3976,
+      "step": 5408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652214074494e-05,
+      "loss": 1.5316,
+      "step": 5409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999652083047705e-05,
+      "loss": 1.5824,
+      "step": 5410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996519519962396e-05,
+      "loss": 1.4277,
+      "step": 5411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996518209200994e-05,
+      "loss": 1.5752,
+      "step": 5412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996516898192835e-05,
+      "loss": 1.5158,
+      "step": 5413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999651558693792e-05,
+      "loss": 1.5819,
+      "step": 5414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996514275436256e-05,
+      "loss": 1.4186,
+      "step": 5415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999651296368783e-05,
+      "loss": 1.3164,
+      "step": 5416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999651165169266e-05,
+      "loss": 1.4611,
+      "step": 5417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999651033945072e-05,
+      "loss": 1.3652,
+      "step": 5418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999650902696203e-05,
+      "loss": 1.2459,
+      "step": 5419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996507714226584e-05,
+      "loss": 1.4631,
+      "step": 5420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999650640124439e-05,
+      "loss": 1.4849,
+      "step": 5421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999650508801543e-05,
+      "loss": 1.5499,
+      "step": 5422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996503774539724e-05,
+      "loss": 1.5977,
+      "step": 5423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996502460817264e-05,
+      "loss": 1.5931,
+      "step": 5424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999650114684804e-05,
+      "loss": 1.5806,
+      "step": 5425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999649983263207e-05,
+      "loss": 1.544,
+      "step": 5426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996498518169345e-05,
+      "loss": 1.6057,
+      "step": 5427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996497203459855e-05,
+      "loss": 1.6214,
+      "step": 5428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999649588850363e-05,
+      "loss": 1.1424,
+      "step": 5429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999649457330063e-05,
+      "loss": 1.5469,
+      "step": 5430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996493257850885e-05,
+      "loss": 1.6572,
+      "step": 5431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999649194215438e-05,
+      "loss": 1.6586,
+      "step": 5432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999649062621112e-05,
+      "loss": 1.6485,
+      "step": 5433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648931002111e-05,
+      "loss": 1.6322,
+      "step": 5434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648799358434e-05,
+      "loss": 1.6366,
+      "step": 5435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648667690082e-05,
+      "loss": 1.6318,
+      "step": 5436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996485359970544e-05,
+      "loss": 1.3802,
+      "step": 5437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648404279351e-05,
+      "loss": 1.635,
+      "step": 5438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648272536973e-05,
+      "loss": 1.6398,
+      "step": 5439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648140769918e-05,
+      "loss": 1.6355,
+      "step": 5440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648008978189e-05,
+      "loss": 1.455,
+      "step": 5441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999647877161784e-05,
+      "loss": 1.6273,
+      "step": 5442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999647745320704e-05,
+      "loss": 1.6324,
+      "step": 5443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999647613454947e-05,
+      "loss": 1.6379,
+      "step": 5444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996474815645156e-05,
+      "loss": 1.6175,
+      "step": 5445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999647349649409e-05,
+      "loss": 1.4542,
+      "step": 5446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996472177096264e-05,
+      "loss": 1.3211,
+      "step": 5447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999647085745169e-05,
+      "loss": 1.626,
+      "step": 5448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996469537560356e-05,
+      "loss": 0.9563,
+      "step": 5449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996468217422265e-05,
+      "loss": 1.6406,
+      "step": 5450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996466897037424e-05,
+      "loss": 1.5165,
+      "step": 5451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996465576405825e-05,
+      "loss": 1.6257,
+      "step": 5452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996464255527475e-05,
+      "loss": 1.6199,
+      "step": 5453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996462934402374e-05,
+      "loss": 1.6214,
+      "step": 5454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999646161303051e-05,
+      "loss": 1.6068,
+      "step": 5455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999646029141189e-05,
+      "loss": 1.6175,
+      "step": 5456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996458969546525e-05,
+      "loss": 1.6042,
+      "step": 5457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99964576474344e-05,
+      "loss": 1.3519,
+      "step": 5458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996456325075525e-05,
+      "loss": 1.6099,
+      "step": 5459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996455002469885e-05,
+      "loss": 1.6056,
+      "step": 5460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999645367961751e-05,
+      "loss": 1.6091,
+      "step": 5461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999645235651836e-05,
+      "loss": 1.6222,
+      "step": 5462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999645103317247e-05,
+      "loss": 1.601,
+      "step": 5463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999644970957981e-05,
+      "loss": 1.5557,
+      "step": 5464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999644838574041e-05,
+      "loss": 1.6104,
+      "step": 5465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999644706165425e-05,
+      "loss": 1.616,
+      "step": 5466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996445737321334e-05,
+      "loss": 1.611,
+      "step": 5467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999644441274166e-05,
+      "loss": 1.6151,
+      "step": 5468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999644308791524e-05,
+      "loss": 1.2763,
+      "step": 5469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999644176284206e-05,
+      "loss": 1.4887,
+      "step": 5470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999644043752213e-05,
+      "loss": 1.363,
+      "step": 5471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999643911195545e-05,
+      "loss": 1.6844,
+      "step": 5472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996437786142e-05,
+      "loss": 1.5473,
+      "step": 5473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996436460081815e-05,
+      "loss": 1.5769,
+      "step": 5474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996435133774865e-05,
+      "loss": 1.3247,
+      "step": 5475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999643380722116e-05,
+      "loss": 0.9342,
+      "step": 5476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996432480420705e-05,
+      "loss": 1.0491,
+      "step": 5477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999643115337349e-05,
+      "loss": 1.2968,
+      "step": 5478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999642982607953e-05,
+      "loss": 1.0755,
+      "step": 5479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99964284985388e-05,
+      "loss": 1.312,
+      "step": 5480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996427170751334e-05,
+      "loss": 1.1861,
+      "step": 5481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99964258427171e-05,
+      "loss": 1.1286,
+      "step": 5482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999642451443612e-05,
+      "loss": 1.3266,
+      "step": 5483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999642318590838e-05,
+      "loss": 1.486,
+      "step": 5484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999642185713389e-05,
+      "loss": 0.9346,
+      "step": 5485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996420528112645e-05,
+      "loss": 0.8062,
+      "step": 5486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996419198844643e-05,
+      "loss": 1.1346,
+      "step": 5487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999641786932989e-05,
+      "loss": 1.3305,
+      "step": 5488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999641653956839e-05,
+      "loss": 1.0559,
+      "step": 5489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999641520956013e-05,
+      "loss": 0.8833,
+      "step": 5490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999641387930511e-05,
+      "loss": 1.0513,
+      "step": 5491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999641254880334e-05,
+      "loss": 1.0562,
+      "step": 5492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999641121805482e-05,
+      "loss": 1.0458,
+      "step": 5493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996409887059536e-05,
+      "loss": 1.0736,
+      "step": 5494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999640855581751e-05,
+      "loss": 1.0123,
+      "step": 5495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999640722432872e-05,
+      "loss": 1.218,
+      "step": 5496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999640589259318e-05,
+      "loss": 1.1162,
+      "step": 5497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999640456061089e-05,
+      "loss": 0.9334,
+      "step": 5498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999640322838184e-05,
+      "loss": 1.1247,
+      "step": 5499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999640189590604e-05,
+      "loss": 1.2491,
+      "step": 5500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0697647333145142,
+      "eval_runtime": 83.3184,
+      "eval_samples_per_second": 16.623,
+      "eval_steps_per_second": 4.165,
+      "step": 5500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999640056318349e-05,
+      "loss": 1.0866,
+      "step": 5501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996399230214175e-05,
+      "loss": 1.1621,
+      "step": 5502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999639789699811e-05,
+      "loss": 1.9449,
+      "step": 5503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999639656353529e-05,
+      "loss": 1.3423,
+      "step": 5504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999639522982572e-05,
+      "loss": 1.0964,
+      "step": 5505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99963938958694e-05,
+      "loss": 1.1129,
+      "step": 5506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996392561666313e-05,
+      "loss": 1.3408,
+      "step": 5507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999639122721649e-05,
+      "loss": 0.9812,
+      "step": 5508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99963898925199e-05,
+      "loss": 0.3642,
+      "step": 5509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996388557576556e-05,
+      "loss": 0.8257,
+      "step": 5510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996387222386466e-05,
+      "loss": 1.2286,
+      "step": 5511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999638588694961e-05,
+      "loss": 1.333,
+      "step": 5512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996384551266014e-05,
+      "loss": 1.2729,
+      "step": 5513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999638321533566e-05,
+      "loss": 1.1928,
+      "step": 5514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999638187915855e-05,
+      "loss": 1.3171,
+      "step": 5515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999638054273468e-05,
+      "loss": 1.176,
+      "step": 5516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999637920606407e-05,
+      "loss": 0.9842,
+      "step": 5517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999637786914669e-05,
+      "loss": 1.0741,
+      "step": 5518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999637653198257e-05,
+      "loss": 1.2538,
+      "step": 5519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996375194571694e-05,
+      "loss": 0.947,
+      "step": 5520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999637385691406e-05,
+      "loss": 3.6488,
+      "step": 5521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999637251900968e-05,
+      "loss": 2.0481,
+      "step": 5522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999637118085854e-05,
+      "loss": 1.2494,
+      "step": 5523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996369842460646e-05,
+      "loss": 0.9502,
+      "step": 5524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996368503816e-05,
+      "loss": 1.3348,
+      "step": 5525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99963671649246e-05,
+      "loss": 1.2725,
+      "step": 5526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996365825786446e-05,
+      "loss": 0.4649,
+      "step": 5527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999636448640154e-05,
+      "loss": 0.2174,
+      "step": 5528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999636314676988e-05,
+      "loss": 0.1566,
+      "step": 5529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999636180689146e-05,
+      "loss": 0.0937,
+      "step": 5530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999636046676629e-05,
+      "loss": 0.0804,
+      "step": 5531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996359126394374e-05,
+      "loss": 0.0901,
+      "step": 5532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999635778577569e-05,
+      "loss": 0.1146,
+      "step": 5533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999635644491027e-05,
+      "loss": 0.0508,
+      "step": 5534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996355103798085e-05,
+      "loss": 0.0421,
+      "step": 5535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999635376243915e-05,
+      "loss": 0.0781,
+      "step": 5536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996352420833456e-05,
+      "loss": 0.061,
+      "step": 5537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999635107898102e-05,
+      "loss": 0.2096,
+      "step": 5538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999634973688182e-05,
+      "loss": 0.1037,
+      "step": 5539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999634839453588e-05,
+      "loss": 0.204,
+      "step": 5540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996347051943174e-05,
+      "loss": 0.1553,
+      "step": 5541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999634570910372e-05,
+      "loss": 0.2328,
+      "step": 5542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999634436601751e-05,
+      "loss": 0.1369,
+      "step": 5543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996343022684544e-05,
+      "loss": 0.1622,
+      "step": 5544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999634167910483e-05,
+      "loss": 0.0946,
+      "step": 5545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999634033527836e-05,
+      "loss": 0.812,
+      "step": 5546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996338991205136e-05,
+      "loss": 1.375,
+      "step": 5547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999633764688516e-05,
+      "loss": 1.1506,
+      "step": 5548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996336302318425e-05,
+      "loss": 1.2242,
+      "step": 5549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999633495750494e-05,
+      "loss": 1.2195,
+      "step": 5550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999633361244471e-05,
+      "loss": 1.1318,
+      "step": 5551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999633226713773e-05,
+      "loss": 1.2764,
+      "step": 5552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999633092158398e-05,
+      "loss": 1.1542,
+      "step": 5553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999632957578349e-05,
+      "loss": 1.4049,
+      "step": 5554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999632822973623e-05,
+      "loss": 1.8587,
+      "step": 5555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999632688344223e-05,
+      "loss": 1.7064,
+      "step": 5556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999632553690148e-05,
+      "loss": 1.2494,
+      "step": 5557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999632419011396e-05,
+      "loss": 1.3254,
+      "step": 5558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996322843079703e-05,
+      "loss": 0.8808,
+      "step": 5559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999632149579869e-05,
+      "loss": 1.0223,
+      "step": 5560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999632014827092e-05,
+      "loss": 0.8446,
+      "step": 5561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99963188004964e-05,
+      "loss": 1.0649,
+      "step": 5562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999631745247513e-05,
+      "loss": 0.8702,
+      "step": 5563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996316104207094e-05,
+      "loss": 1.0529,
+      "step": 5564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999631475569232e-05,
+      "loss": 1.352,
+      "step": 5565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996313406930786e-05,
+      "loss": 1.5323,
+      "step": 5566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999631205792249e-05,
+      "loss": 1.368,
+      "step": 5567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996310708667454e-05,
+      "loss": 1.0888,
+      "step": 5568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999630935916566e-05,
+      "loss": 0.9347,
+      "step": 5569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999630800941711e-05,
+      "loss": 0.9903,
+      "step": 5570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999630665942182e-05,
+      "loss": 1.0515,
+      "step": 5571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996305309179767e-05,
+      "loss": 1.1067,
+      "step": 5572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999630395869096e-05,
+      "loss": 1.2955,
+      "step": 5573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996302607955404e-05,
+      "loss": 1.2846,
+      "step": 5574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999630125697309e-05,
+      "loss": 1.0916,
+      "step": 5575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996299905744025e-05,
+      "loss": 1.1508,
+      "step": 5576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999629855426821e-05,
+      "loss": 1.2063,
+      "step": 5577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999629720254564e-05,
+      "loss": 1.0702,
+      "step": 5578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999629585057632e-05,
+      "loss": 1.0311,
+      "step": 5579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999629449836024e-05,
+      "loss": 1.3127,
+      "step": 5580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999629314589741e-05,
+      "loss": 1.3089,
+      "step": 5581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999629179318783e-05,
+      "loss": 1.0694,
+      "step": 5582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962904402315e-05,
+      "loss": 1.1529,
+      "step": 5583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999628908702841e-05,
+      "loss": 1.1264,
+      "step": 5584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996287733578575e-05,
+      "loss": 1.1999,
+      "step": 5585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999628637988198e-05,
+      "loss": 1.1031,
+      "step": 5586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996285025938636e-05,
+      "loss": 0.7861,
+      "step": 5587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999628367174853e-05,
+      "loss": 0.4965,
+      "step": 5588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999628231731168e-05,
+      "loss": 0.8337,
+      "step": 5589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996280962628075e-05,
+      "loss": 0.8052,
+      "step": 5590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999627960769772e-05,
+      "loss": 1.2772,
+      "step": 5591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996278252520614e-05,
+      "loss": 0.9121,
+      "step": 5592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999627689709675e-05,
+      "loss": 1.7697,
+      "step": 5593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996275541426136e-05,
+      "loss": 1.6754,
+      "step": 5594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999627418550877e-05,
+      "loss": 0.9327,
+      "step": 5595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999627282934465e-05,
+      "loss": 0.7492,
+      "step": 5596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996271472933773e-05,
+      "loss": 0.6132,
+      "step": 5597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999627011627615e-05,
+      "loss": 0.7812,
+      "step": 5598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999626875937177e-05,
+      "loss": 1.0239,
+      "step": 5599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999626740222064e-05,
+      "loss": 1.3136,
+      "step": 5600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999626604482276e-05,
+      "loss": 1.5179,
+      "step": 5601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999626468717812e-05,
+      "loss": 1.271,
+      "step": 5602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999626332928673e-05,
+      "loss": 1.0892,
+      "step": 5603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962619711486e-05,
+      "loss": 1.113,
+      "step": 5604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996260612763704e-05,
+      "loss": 1.1502,
+      "step": 5605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999625925413205e-05,
+      "loss": 1.1396,
+      "step": 5606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999625789525365e-05,
+      "loss": 1.56,
+      "step": 5607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962565361285e-05,
+      "loss": 1.2087,
+      "step": 5608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962551767566e-05,
+      "loss": 1.268,
+      "step": 5609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999625381713794e-05,
+      "loss": 1.2953,
+      "step": 5610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999625245727253e-05,
+      "loss": 1.222,
+      "step": 5611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999625109716037e-05,
+      "loss": 1.1168,
+      "step": 5612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999624973680146e-05,
+      "loss": 1.1631,
+      "step": 5613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996248376195784e-05,
+      "loss": 1.4415,
+      "step": 5614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999624701534337e-05,
+      "loss": 1.1527,
+      "step": 5615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962456542442e-05,
+      "loss": 1.1625,
+      "step": 5616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996244292898275e-05,
+      "loss": 1.3437,
+      "step": 5617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996242931305606e-05,
+      "loss": 0.7632,
+      "step": 5618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999624156946617e-05,
+      "loss": 1.2084,
+      "step": 5619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999624020737999e-05,
+      "loss": 0.9181,
+      "step": 5620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999623884504706e-05,
+      "loss": 1.1415,
+      "step": 5621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996237482467374e-05,
+      "loss": 1.2355,
+      "step": 5622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996236119640937e-05,
+      "loss": 1.2088,
+      "step": 5623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999623475656774e-05,
+      "loss": 0.5592,
+      "step": 5624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962333932478e-05,
+      "loss": 1.0354,
+      "step": 5625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996232029681114e-05,
+      "loss": 0.9641,
+      "step": 5626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999623066586766e-05,
+      "loss": 1.1484,
+      "step": 5627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999622930180746e-05,
+      "loss": 1.1756,
+      "step": 5628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999622793750051e-05,
+      "loss": 1.022,
+      "step": 5629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996226572946806e-05,
+      "loss": 1.2777,
+      "step": 5630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996225208146355e-05,
+      "loss": 1.4627,
+      "step": 5631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999622384309914e-05,
+      "loss": 1.3966,
+      "step": 5632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999622247780518e-05,
+      "loss": 1.1712,
+      "step": 5633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999622111226447e-05,
+      "loss": 1.1774,
+      "step": 5634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999621974647701e-05,
+      "loss": 1.6404,
+      "step": 5635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999621838044279e-05,
+      "loss": 1.4061,
+      "step": 5636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999621701416182e-05,
+      "loss": 1.254,
+      "step": 5637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962156476341e-05,
+      "loss": 1.154,
+      "step": 5638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996214280859624e-05,
+      "loss": 0.999,
+      "step": 5639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962129138384e-05,
+      "loss": 1.443,
+      "step": 5640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999621154657043e-05,
+      "loss": 1.3345,
+      "step": 5641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962101790557e-05,
+      "loss": 1.4813,
+      "step": 5642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999620881129422e-05,
+      "loss": 0.4938,
+      "step": 5643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999620744328598e-05,
+      "loss": 0.8844,
+      "step": 5644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996206075030996e-05,
+      "loss": 1.2019,
+      "step": 5645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999620470652926e-05,
+      "loss": 1.3681,
+      "step": 5646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999620333778077e-05,
+      "loss": 1.3369,
+      "step": 5647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999620196878553e-05,
+      "loss": 0.9395,
+      "step": 5648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999620059954353e-05,
+      "loss": 1.4062,
+      "step": 5649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996199230054784e-05,
+      "loss": 1.3625,
+      "step": 5650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996197860319295e-05,
+      "loss": 1.1258,
+      "step": 5651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999619649033704e-05,
+      "loss": 0.9185,
+      "step": 5652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999619512010804e-05,
+      "loss": 1.1147,
+      "step": 5653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999619374963229e-05,
+      "loss": 1.2958,
+      "step": 5654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996192378909786e-05,
+      "loss": 0.9991,
+      "step": 5655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999619100794053e-05,
+      "loss": 0.9079,
+      "step": 5656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999618963672452e-05,
+      "loss": 1.1405,
+      "step": 5657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999618826526176e-05,
+      "loss": 1.2317,
+      "step": 5658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999618689355224e-05,
+      "loss": 1.2066,
+      "step": 5659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996185521595984e-05,
+      "loss": 1.3058,
+      "step": 5660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996184149392967e-05,
+      "loss": 0.8117,
+      "step": 5661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99961827769432e-05,
+      "loss": 0.6044,
+      "step": 5662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999618140424668e-05,
+      "loss": 1.3677,
+      "step": 5663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999618003130341e-05,
+      "loss": 1.2567,
+      "step": 5664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999617865811338e-05,
+      "loss": 1.283,
+      "step": 5665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999617728467661e-05,
+      "loss": 1.1749,
+      "step": 5666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999617591099308e-05,
+      "loss": 0.912,
+      "step": 5667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99961745370628e-05,
+      "loss": 1.248,
+      "step": 5668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999617316288577e-05,
+      "loss": 1.3435,
+      "step": 5669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996171788461996e-05,
+      "loss": 1.1158,
+      "step": 5670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999617041379146e-05,
+      "loss": 0.7348,
+      "step": 5671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999616903887417e-05,
+      "loss": 1.1834,
+      "step": 5672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999616766371014e-05,
+      "loss": 1.0043,
+      "step": 5673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996166288299354e-05,
+      "loss": 1.0576,
+      "step": 5674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996164912641804e-05,
+      "loss": 1.4277,
+      "step": 5675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999616353673752e-05,
+      "loss": 0.969,
+      "step": 5676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999616216058647e-05,
+      "loss": 0.719,
+      "step": 5677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999616078418868e-05,
+      "loss": 1.0071,
+      "step": 5678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999615940754413e-05,
+      "loss": 1.0275,
+      "step": 5679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996158030652834e-05,
+      "loss": 1.1718,
+      "step": 5680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996156653514786e-05,
+      "loss": 1.0183,
+      "step": 5681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999615527612999e-05,
+      "loss": 1.116,
+      "step": 5682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999615389849843e-05,
+      "loss": 1.2664,
+      "step": 5683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999615252062013e-05,
+      "loss": 1.2419,
+      "step": 5684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999615114249507e-05,
+      "loss": 1.298,
+      "step": 5685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999614976412327e-05,
+      "loss": 1.3061,
+      "step": 5686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99961483855047e-05,
+      "loss": 1.0749,
+      "step": 5687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99961470066394e-05,
+      "loss": 0.6447,
+      "step": 5688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999614562752734e-05,
+      "loss": 0.4944,
+      "step": 5689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996144248168525e-05,
+      "loss": 0.7365,
+      "step": 5690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999614286856296e-05,
+      "loss": 1.5031,
+      "step": 5691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999614148871065e-05,
+      "loss": 1.3854,
+      "step": 5692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996140108611576e-05,
+      "loss": 1.0011,
+      "step": 5693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999613872826576e-05,
+      "loss": 1.2076,
+      "step": 5694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996137347673186e-05,
+      "loss": 0.9505,
+      "step": 5695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999613596683387e-05,
+      "loss": 1.2965,
+      "step": 5696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99961345857478e-05,
+      "loss": 1.14,
+      "step": 5697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996133204414974e-05,
+      "loss": 1.3744,
+      "step": 5698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996131822835403e-05,
+      "loss": 1.2792,
+      "step": 5699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999613044100907e-05,
+      "loss": 1.2024,
+      "step": 5700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996129058936e-05,
+      "loss": 1.1759,
+      "step": 5701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999612767661617e-05,
+      "loss": 1.1439,
+      "step": 5702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999612629404959e-05,
+      "loss": 1.1925,
+      "step": 5703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996124911236255e-05,
+      "loss": 0.9742,
+      "step": 5704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999612352817617e-05,
+      "loss": 1.1068,
+      "step": 5705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999612214486934e-05,
+      "loss": 0.5886,
+      "step": 5706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996120761315756e-05,
+      "loss": 0.991,
+      "step": 5707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999611937751542e-05,
+      "loss": 1.0096,
+      "step": 5708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996117993468336e-05,
+      "loss": 0.9139,
+      "step": 5709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999611660917449e-05,
+      "loss": 1.185,
+      "step": 5710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999611522463391e-05,
+      "loss": 1.2934,
+      "step": 5711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999611383984657e-05,
+      "loss": 0.9533,
+      "step": 5712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999611245481247e-05,
+      "loss": 1.0366,
+      "step": 5713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999611106953164e-05,
+      "loss": 0.8981,
+      "step": 5714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999610968400404e-05,
+      "loss": 0.8346,
+      "step": 5715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999610829822969e-05,
+      "loss": 1.0552,
+      "step": 5716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999610691220859e-05,
+      "loss": 1.1626,
+      "step": 5717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999610552594075e-05,
+      "loss": 1.4871,
+      "step": 5718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999610413942616e-05,
+      "loss": 1.0045,
+      "step": 5719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999610275266481e-05,
+      "loss": 1.1701,
+      "step": 5720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999610136565671e-05,
+      "loss": 1.1458,
+      "step": 5721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999609997840185e-05,
+      "loss": 1.342,
+      "step": 5722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996098590900255e-05,
+      "loss": 1.2274,
+      "step": 5723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999609720315191e-05,
+      "loss": 0.5639,
+      "step": 5724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996095815156794e-05,
+      "loss": 0.4767,
+      "step": 5725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996094426914943e-05,
+      "loss": 1.0088,
+      "step": 5726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999609303842634e-05,
+      "loss": 1.4282,
+      "step": 5727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999609164969098e-05,
+      "loss": 0.7006,
+      "step": 5728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996090260708874e-05,
+      "loss": 1.1023,
+      "step": 5729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999608887148002e-05,
+      "loss": 1.2316,
+      "step": 5730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999608748200441e-05,
+      "loss": 1.4644,
+      "step": 5731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996086092282046e-05,
+      "loss": 0.9136,
+      "step": 5732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999608470231294e-05,
+      "loss": 1.2339,
+      "step": 5733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996083312097076e-05,
+      "loss": 1.3003,
+      "step": 5734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999608192163446e-05,
+      "loss": 1.3405,
+      "step": 5735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99960805309251e-05,
+      "loss": 1.1211,
+      "step": 5736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996079139968986e-05,
+      "loss": 1.167,
+      "step": 5737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996077748766125e-05,
+      "loss": 1.0671,
+      "step": 5738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999607635731651e-05,
+      "loss": 1.2298,
+      "step": 5739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996074965620144e-05,
+      "loss": 1.0296,
+      "step": 5740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996073573677024e-05,
+      "loss": 1.244,
+      "step": 5741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996072181487154e-05,
+      "loss": 1.1463,
+      "step": 5742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999607078905054e-05,
+      "loss": 1.0963,
+      "step": 5743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999606939636717e-05,
+      "loss": 1.1444,
+      "step": 5744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999606800343705e-05,
+      "loss": 1.1883,
+      "step": 5745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999606661026018e-05,
+      "loss": 1.0707,
+      "step": 5746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996065216836563e-05,
+      "loss": 1.1605,
+      "step": 5747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999606382316618e-05,
+      "loss": 1.0791,
+      "step": 5748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999606242924907e-05,
+      "loss": 0.7398,
+      "step": 5749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996061035085196e-05,
+      "loss": 1.1299,
+      "step": 5750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999605964067457e-05,
+      "loss": 1.389,
+      "step": 5751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996058246017195e-05,
+      "loss": 1.3896,
+      "step": 5752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999605685111307e-05,
+      "loss": 0.9519,
+      "step": 5753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99960554559622e-05,
+      "loss": 1.1582,
+      "step": 5754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999605406056457e-05,
+      "loss": 1.1379,
+      "step": 5755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996052664920195e-05,
+      "loss": 1.1026,
+      "step": 5756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999605126902907e-05,
+      "loss": 0.9483,
+      "step": 5757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999604987289119e-05,
+      "loss": 1.1033,
+      "step": 5758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999604847650656e-05,
+      "loss": 1.5733,
+      "step": 5759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999604707987519e-05,
+      "loss": 1.185,
+      "step": 5760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999604568299706e-05,
+      "loss": 1.3739,
+      "step": 5761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999604428587218e-05,
+      "loss": 1.0264,
+      "step": 5762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999604288850055e-05,
+      "loss": 1.1799,
+      "step": 5763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996041490882175e-05,
+      "loss": 1.2137,
+      "step": 5764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999604009301704e-05,
+      "loss": 2.0663,
+      "step": 5765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996038694905163e-05,
+      "loss": 1.1117,
+      "step": 5766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996037296546535e-05,
+      "loss": 1.0114,
+      "step": 5767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999603589794115e-05,
+      "loss": 0.7678,
+      "step": 5768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999603449908902e-05,
+      "loss": 0.844,
+      "step": 5769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996033099990136e-05,
+      "loss": 0.9896,
+      "step": 5770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999603170064451e-05,
+      "loss": 0.3288,
+      "step": 5771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999603030105212e-05,
+      "loss": 0.2919,
+      "step": 5772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996028901212987e-05,
+      "loss": 0.9983,
+      "step": 5773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999602750112711e-05,
+      "loss": 1.1373,
+      "step": 5774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999602610079447e-05,
+      "loss": 1.32,
+      "step": 5775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999602470021509e-05,
+      "loss": 1.1605,
+      "step": 5776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999602329938896e-05,
+      "loss": 1.2259,
+      "step": 5777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996021898316074e-05,
+      "loss": 0.7605,
+      "step": 5778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999602049699644e-05,
+      "loss": 1.2776,
+      "step": 5779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999601909543006e-05,
+      "loss": 1.2831,
+      "step": 5780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996017693616924e-05,
+      "loss": 1.0186,
+      "step": 5781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999601629155703e-05,
+      "loss": 1.2855,
+      "step": 5782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99960148892504e-05,
+      "loss": 1.0829,
+      "step": 5783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996013486697016e-05,
+      "loss": 1.1834,
+      "step": 5784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999601208389688e-05,
+      "loss": 1.0825,
+      "step": 5785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996010680850004e-05,
+      "loss": 1.3013,
+      "step": 5786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996009277556364e-05,
+      "loss": 1.1277,
+      "step": 5787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999600787401598e-05,
+      "loss": 1.436,
+      "step": 5788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999600647022885e-05,
+      "loss": 1.2099,
+      "step": 5789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999600506619496e-05,
+      "loss": 1.4521,
+      "step": 5790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996003661914326e-05,
+      "loss": 1.0093,
+      "step": 5791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996002257386945e-05,
+      "loss": 1.2973,
+      "step": 5792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99960008526128e-05,
+      "loss": 1.0852,
+      "step": 5793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995999447591926e-05,
+      "loss": 1.2553,
+      "step": 5794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995998042324286e-05,
+      "loss": 0.214,
+      "step": 5795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99959966368099e-05,
+      "loss": 0.1374,
+      "step": 5796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999599523104876e-05,
+      "loss": 0.4754,
+      "step": 5797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995993825040876e-05,
+      "loss": 0.4334,
+      "step": 5798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999599241878624e-05,
+      "loss": 0.441,
+      "step": 5799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999599101228486e-05,
+      "loss": 0.3289,
+      "step": 5800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999598960553672e-05,
+      "loss": 0.3035,
+      "step": 5801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999598819854184e-05,
+      "loss": 0.2614,
+      "step": 5802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999598679130021e-05,
+      "loss": 0.2639,
+      "step": 5803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995985383811824e-05,
+      "loss": 0.3609,
+      "step": 5804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999598397607669e-05,
+      "loss": 0.5269,
+      "step": 5805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995982568094804e-05,
+      "loss": 0.2979,
+      "step": 5806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999598115986617e-05,
+      "loss": 0.3059,
+      "step": 5807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999597975139079e-05,
+      "loss": 0.6379,
+      "step": 5808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995978342668656e-05,
+      "loss": 0.9649,
+      "step": 5809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995976933699774e-05,
+      "loss": 0.9759,
+      "step": 5810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999597552448414e-05,
+      "loss": 0.6076,
+      "step": 5811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999597411502176e-05,
+      "loss": 1.2914,
+      "step": 5812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999597270531262e-05,
+      "loss": 1.1692,
+      "step": 5813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995971295356744e-05,
+      "loss": 1.1402,
+      "step": 5814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999596988515411e-05,
+      "loss": 1.1395,
+      "step": 5815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999596847470473e-05,
+      "loss": 1.0746,
+      "step": 5816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995967064008596e-05,
+      "loss": 1.2578,
+      "step": 5817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995965653065714e-05,
+      "loss": 1.0937,
+      "step": 5818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999596424187609e-05,
+      "loss": 1.114,
+      "step": 5819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995962830439704e-05,
+      "loss": 1.1062,
+      "step": 5820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995961418756576e-05,
+      "loss": 0.3034,
+      "step": 5821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99959600068267e-05,
+      "loss": 1.3744,
+      "step": 5822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995958594650074e-05,
+      "loss": 1.0629,
+      "step": 5823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995957182226693e-05,
+      "loss": 1.1352,
+      "step": 5824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999595576955656e-05,
+      "loss": 1.1774,
+      "step": 5825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995954356639686e-05,
+      "loss": 0.9296,
+      "step": 5826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999595294347605e-05,
+      "loss": 1.1025,
+      "step": 5827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999595153006568e-05,
+      "loss": 1.1803,
+      "step": 5828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995950116408555e-05,
+      "loss": 1.2585,
+      "step": 5829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999594870250468e-05,
+      "loss": 0.8967,
+      "step": 5830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999594728835405e-05,
+      "loss": 1.0024,
+      "step": 5831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995945873956686e-05,
+      "loss": 1.1998,
+      "step": 5832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999594445931256e-05,
+      "loss": 1.1819,
+      "step": 5833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995943044421685e-05,
+      "loss": 1.3396,
+      "step": 5834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999594162928406e-05,
+      "loss": 1.2523,
+      "step": 5835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999594021389969e-05,
+      "loss": 1.3324,
+      "step": 5836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995938798268573e-05,
+      "loss": 1.0069,
+      "step": 5837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99959373823907e-05,
+      "loss": 1.0306,
+      "step": 5838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999593596626608e-05,
+      "loss": 0.9872,
+      "step": 5839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995934549894704e-05,
+      "loss": 1.0294,
+      "step": 5840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999593313327659e-05,
+      "loss": 1.067,
+      "step": 5841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999593171641172e-05,
+      "loss": 1.1862,
+      "step": 5842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995930299300104e-05,
+      "loss": 1.0927,
+      "step": 5843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999592888194173e-05,
+      "loss": 1.2252,
+      "step": 5844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995927464336625e-05,
+      "loss": 1.2999,
+      "step": 5845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999592604648475e-05,
+      "loss": 0.9361,
+      "step": 5846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995924628386136e-05,
+      "loss": 1.0318,
+      "step": 5847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995923210040776e-05,
+      "loss": 1.0358,
+      "step": 5848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995921791448664e-05,
+      "loss": 0.9336,
+      "step": 5849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99959203726098e-05,
+      "loss": 0.7572,
+      "step": 5850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999591895352419e-05,
+      "loss": 0.7159,
+      "step": 5851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995917534191825e-05,
+      "loss": 1.3437,
+      "step": 5852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995916114612717e-05,
+      "loss": 1.0615,
+      "step": 5853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999591469478686e-05,
+      "loss": 1.2398,
+      "step": 5854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995913274714254e-05,
+      "loss": 4.6326,
+      "step": 5855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995911854394894e-05,
+      "loss": 3.5142,
+      "step": 5856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999591043382878e-05,
+      "loss": 3.019,
+      "step": 5857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999590901301593e-05,
+      "loss": 2.6754,
+      "step": 5858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999590759195633e-05,
+      "loss": 2.7023,
+      "step": 5859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995906170649976e-05,
+      "loss": 2.5355,
+      "step": 5860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995904749096875e-05,
+      "loss": 2.4428,
+      "step": 5861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999590332729702e-05,
+      "loss": 2.4379,
+      "step": 5862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999590190525042e-05,
+      "loss": 2.3608,
+      "step": 5863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999590048295707e-05,
+      "loss": 2.1782,
+      "step": 5864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995899060416974e-05,
+      "loss": 2.2068,
+      "step": 5865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999589763763013e-05,
+      "loss": 2.152,
+      "step": 5866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995896214596525e-05,
+      "loss": 2.0718,
+      "step": 5867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999589479131618e-05,
+      "loss": 2.2563,
+      "step": 5868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995893367789086e-05,
+      "loss": 2.1127,
+      "step": 5869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999589194401525e-05,
+      "loss": 2.1021,
+      "step": 5870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999589051999465e-05,
+      "loss": 2.1175,
+      "step": 5871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999588909572731e-05,
+      "loss": 1.365,
+      "step": 5872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999588767121322e-05,
+      "loss": 1.2585,
+      "step": 5873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999588624645238e-05,
+      "loss": 1.2671,
+      "step": 5874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995884821444797e-05,
+      "loss": 0.9925,
+      "step": 5875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999588339619046e-05,
+      "loss": 0.9696,
+      "step": 5876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999588197068937e-05,
+      "loss": 1.2444,
+      "step": 5877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999588054494154e-05,
+      "loss": 0.7877,
+      "step": 5878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995879118946956e-05,
+      "loss": 1.2257,
+      "step": 5879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999587769270562e-05,
+      "loss": 1.4817,
+      "step": 5880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995876266217544e-05,
+      "loss": 1.2465,
+      "step": 5881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995874839482715e-05,
+      "loss": 1.3474,
+      "step": 5882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995873412501135e-05,
+      "loss": 1.6764,
+      "step": 5883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995871985272804e-05,
+      "loss": 0.9174,
+      "step": 5884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999587055779773e-05,
+      "loss": 0.7554,
+      "step": 5885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999586913007591e-05,
+      "loss": 0.7866,
+      "step": 5886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995867702107334e-05,
+      "loss": 1.0494,
+      "step": 5887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995866273892014e-05,
+      "loss": 1.0432,
+      "step": 5888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999586484542994e-05,
+      "loss": 1.3386,
+      "step": 5889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999586341672112e-05,
+      "loss": 1.6875,
+      "step": 5890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999586198776556e-05,
+      "loss": 1.0927,
+      "step": 5891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999586055856324e-05,
+      "loss": 0.8526,
+      "step": 5892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995859129114175e-05,
+      "loss": 0.8085,
+      "step": 5893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995857699418357e-05,
+      "loss": 0.9571,
+      "step": 5894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99958562694758e-05,
+      "loss": 0.8421,
+      "step": 5895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999585483928649e-05,
+      "loss": 0.9443,
+      "step": 5896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999585340885043e-05,
+      "loss": 0.76,
+      "step": 5897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995851978167626e-05,
+      "loss": 0.7921,
+      "step": 5898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999585054723807e-05,
+      "loss": 0.5879,
+      "step": 5899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999584911606176e-05,
+      "loss": 0.6805,
+      "step": 5900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999584768463871e-05,
+      "loss": 0.7923,
+      "step": 5901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99958462529689e-05,
+      "loss": 0.6167,
+      "step": 5902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999584482105236e-05,
+      "loss": 0.9948,
+      "step": 5903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999584338888906e-05,
+      "loss": 0.3271,
+      "step": 5904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999584195647901e-05,
+      "loss": 1.1693,
+      "step": 5905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999584052382221e-05,
+      "loss": 0.9345,
+      "step": 5906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995839090918673e-05,
+      "loss": 1.5019,
+      "step": 5907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995837657768376e-05,
+      "loss": 1.1689,
+      "step": 5908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999583622437134e-05,
+      "loss": 1.1574,
+      "step": 5909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999583479072755e-05,
+      "loss": 1.1267,
+      "step": 5910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995833356837014e-05,
+      "loss": 1.2111,
+      "step": 5911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999583192269973e-05,
+      "loss": 1.0961,
+      "step": 5912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995830488315696e-05,
+      "loss": 1.2409,
+      "step": 5913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995829053684914e-05,
+      "loss": 1.0839,
+      "step": 5914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999582761880738e-05,
+      "loss": 1.189,
+      "step": 5915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995826183683104e-05,
+      "loss": 0.9599,
+      "step": 5916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995824748312076e-05,
+      "loss": 1.2058,
+      "step": 5917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995823312694304e-05,
+      "loss": 1.1992,
+      "step": 5918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995821876829775e-05,
+      "loss": 0.9585,
+      "step": 5919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999582044071851e-05,
+      "loss": 1.1868,
+      "step": 5920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995819004360484e-05,
+      "loss": 1.0446,
+      "step": 5921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995817567755715e-05,
+      "loss": 1.0714,
+      "step": 5922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99958161309042e-05,
+      "loss": 1.1892,
+      "step": 5923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999581469380593e-05,
+      "loss": 1.0839,
+      "step": 5924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995813256460925e-05,
+      "loss": 1.4469,
+      "step": 5925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995811818869166e-05,
+      "loss": 1.3577,
+      "step": 5926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999581038103065e-05,
+      "loss": 1.2339,
+      "step": 5927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99958089429454e-05,
+      "loss": 1.2053,
+      "step": 5928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999580750461339e-05,
+      "loss": 1.0734,
+      "step": 5929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999580606603463e-05,
+      "loss": 1.1296,
+      "step": 5930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999580462720914e-05,
+      "loss": 1.1232,
+      "step": 5931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999580318813689e-05,
+      "loss": 1.139,
+      "step": 5932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999580174881789e-05,
+      "loss": 1.1215,
+      "step": 5933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999580030925214e-05,
+      "loss": 0.9538,
+      "step": 5934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995798869439647e-05,
+      "loss": 0.9857,
+      "step": 5935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995797429380407e-05,
+      "loss": 1.0086,
+      "step": 5936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999579598907442e-05,
+      "loss": 1.2054,
+      "step": 5937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999579454852168e-05,
+      "loss": 1.0337,
+      "step": 5938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99957931077222e-05,
+      "loss": 1.5357,
+      "step": 5939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999579166667596e-05,
+      "loss": 1.2708,
+      "step": 5940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995790225382985e-05,
+      "loss": 1.1459,
+      "step": 5941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999578878384325e-05,
+      "loss": 2.3755,
+      "step": 5942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999578734205678e-05,
+      "loss": 1.2678,
+      "step": 5943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999578590002355e-05,
+      "loss": 1.4369,
+      "step": 5944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999578445774358e-05,
+      "loss": 1.0646,
+      "step": 5945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999578301521686e-05,
+      "loss": 1.3442,
+      "step": 5946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999578157244339e-05,
+      "loss": 1.3115,
+      "step": 5947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999578012942318e-05,
+      "loss": 1.0278,
+      "step": 5948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995778686156217e-05,
+      "loss": 0.9083,
+      "step": 5949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99957772426425e-05,
+      "loss": 1.231,
+      "step": 5950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995775798882036e-05,
+      "loss": 1.0653,
+      "step": 5951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999577435487484e-05,
+      "loss": 0.9043,
+      "step": 5952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999577291062087e-05,
+      "loss": 1.0906,
+      "step": 5953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999577146612018e-05,
+      "loss": 0.9863,
+      "step": 5954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995770021372726e-05,
+      "loss": 1.097,
+      "step": 5955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999576857637852e-05,
+      "loss": 1.0197,
+      "step": 5956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999576713113758e-05,
+      "loss": 1.1653,
+      "step": 5957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995765685649885e-05,
+      "loss": 1.1057,
+      "step": 5958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999576423991544e-05,
+      "loss": 1.0866,
+      "step": 5959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999576279393425e-05,
+      "loss": 1.7562,
+      "step": 5960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999576134770631e-05,
+      "loss": 1.2862,
+      "step": 5961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999575990123163e-05,
+      "loss": 1.2652,
+      "step": 5962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99957584545102e-05,
+      "loss": 1.1406,
+      "step": 5963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995757007542024e-05,
+      "loss": 1.3136,
+      "step": 5964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99957555603271e-05,
+      "loss": 1.0399,
+      "step": 5965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999575411286542e-05,
+      "loss": 1.0532,
+      "step": 5966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995752665157005e-05,
+      "loss": 1.0086,
+      "step": 5967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999575121720183e-05,
+      "loss": 1.0175,
+      "step": 5968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999574976899991e-05,
+      "loss": 1.0899,
+      "step": 5969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999574832055125e-05,
+      "loss": 1.1038,
+      "step": 5970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995746871855836e-05,
+      "loss": 1.323,
+      "step": 5971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995745422913674e-05,
+      "loss": 1.26,
+      "step": 5972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995743973724774e-05,
+      "loss": 1.1442,
+      "step": 5973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995742524289116e-05,
+      "loss": 1.0892,
+      "step": 5974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995741074606715e-05,
+      "loss": 1.3687,
+      "step": 5975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999573962467756e-05,
+      "loss": 1.1966,
+      "step": 5976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995738174501666e-05,
+      "loss": 1.1295,
+      "step": 5977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995736724079025e-05,
+      "loss": 1.2428,
+      "step": 5978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999573527340963e-05,
+      "loss": 0.745,
+      "step": 5979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99957338224935e-05,
+      "loss": 1.1817,
+      "step": 5980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995732371330605e-05,
+      "loss": 1.0781,
+      "step": 5981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999573091992097e-05,
+      "loss": 0.9299,
+      "step": 5982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999572946826459e-05,
+      "loss": 0.9942,
+      "step": 5983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999572801636147e-05,
+      "loss": 1.1955,
+      "step": 5984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995726564211585e-05,
+      "loss": 1.2949,
+      "step": 5985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995725111814964e-05,
+      "loss": 1.2096,
+      "step": 5986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999572365917159e-05,
+      "loss": 1.4178,
+      "step": 5987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999572220628148e-05,
+      "loss": 1.6477,
+      "step": 5988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999572075314462e-05,
+      "loss": 1.3698,
+      "step": 5989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995719299761e-05,
+      "loss": 0.9212,
+      "step": 5990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999571784613064e-05,
+      "loss": 1.149,
+      "step": 5991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995716392253535e-05,
+      "loss": 0.8656,
+      "step": 5992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999571493812968e-05,
+      "loss": 1.2262,
+      "step": 5993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999571348375908e-05,
+      "loss": 0.8487,
+      "step": 5994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995712029141734e-05,
+      "loss": 0.3838,
+      "step": 5995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999571057427764e-05,
+      "loss": 1.2446,
+      "step": 5996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999570911916679e-05,
+      "loss": 0.8315,
+      "step": 5997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999570766380921e-05,
+      "loss": 1.1521,
+      "step": 5998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995706208204875e-05,
+      "loss": 1.2743,
+      "step": 5999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999570475235379e-05,
+      "loss": 1.1449,
+      "step": 6000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0794674158096313,
+      "eval_runtime": 86.1885,
+      "eval_samples_per_second": 16.069,
+      "eval_steps_per_second": 4.026,
+      "step": 6000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995703296255955e-05,
+      "loss": 1.0349,
+      "step": 6001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999570183991138e-05,
+      "loss": 1.0534,
+      "step": 6002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999570038332006e-05,
+      "loss": 1.3352,
+      "step": 6003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999569892648198e-05,
+      "loss": 1.2638,
+      "step": 6004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999569746939716e-05,
+      "loss": 1.185,
+      "step": 6005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995696012065593e-05,
+      "loss": 1.1458,
+      "step": 6006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995694554487286e-05,
+      "loss": 1.2757,
+      "step": 6007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999569309666223e-05,
+      "loss": 1.2362,
+      "step": 6008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995691638590426e-05,
+      "loss": 1.2912,
+      "step": 6009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999569018027187e-05,
+      "loss": 1.4617,
+      "step": 6010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999568872170657e-05,
+      "loss": 1.0121,
+      "step": 6011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999568726289452e-05,
+      "loss": 1.678,
+      "step": 6012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999568580383572e-05,
+      "loss": 0.97,
+      "step": 6013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995684344530186e-05,
+      "loss": 0.9885,
+      "step": 6014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99956828849779e-05,
+      "loss": 0.9913,
+      "step": 6015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999568142517886e-05,
+      "loss": 1.0645,
+      "step": 6016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999567996513308e-05,
+      "loss": 1.6482,
+      "step": 6017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995678504840554e-05,
+      "loss": 0.8483,
+      "step": 6018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995677044301284e-05,
+      "loss": 0.9729,
+      "step": 6019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995675583515256e-05,
+      "loss": 0.7855,
+      "step": 6020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999567412248249e-05,
+      "loss": 1.9137,
+      "step": 6021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995672661202975e-05,
+      "loss": 1.5124,
+      "step": 6022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999567119967671e-05,
+      "loss": 1.4559,
+      "step": 6023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995669737903704e-05,
+      "loss": 1.3439,
+      "step": 6024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999566827588394e-05,
+      "loss": 1.2469,
+      "step": 6025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999566681361745e-05,
+      "loss": 1.208,
+      "step": 6026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995665351104193e-05,
+      "loss": 1.1692,
+      "step": 6027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99956638883442e-05,
+      "loss": 1.1804,
+      "step": 6028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999566242533746e-05,
+      "loss": 1.3017,
+      "step": 6029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995660962083966e-05,
+      "loss": 1.2244,
+      "step": 6030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999565949858373e-05,
+      "loss": 1.324,
+      "step": 6031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999565803483675e-05,
+      "loss": 1.6463,
+      "step": 6032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999565657084302e-05,
+      "loss": 1.201,
+      "step": 6033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999565510660255e-05,
+      "loss": 0.8303,
+      "step": 6034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995653642115326e-05,
+      "loss": 0.3958,
+      "step": 6035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999565217738136e-05,
+      "loss": 0.8038,
+      "step": 6036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999565071240064e-05,
+      "loss": 1.1105,
+      "step": 6037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999564924717318e-05,
+      "loss": 1.0792,
+      "step": 6038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999564778169897e-05,
+      "loss": 0.9225,
+      "step": 6039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995646315978016e-05,
+      "loss": 0.7761,
+      "step": 6040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995644850010315e-05,
+      "loss": 0.7188,
+      "step": 6041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995643383795864e-05,
+      "loss": 0.854,
+      "step": 6042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995641917334674e-05,
+      "loss": 1.1032,
+      "step": 6043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995640450626734e-05,
+      "loss": 0.8793,
+      "step": 6044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995638983672044e-05,
+      "loss": 0.5732,
+      "step": 6045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999563751647061e-05,
+      "loss": 1.0618,
+      "step": 6046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999563604902243e-05,
+      "loss": 1.3174,
+      "step": 6047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999563458132751e-05,
+      "loss": 1.0841,
+      "step": 6048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999563311338583e-05,
+      "loss": 0.999,
+      "step": 6049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995631645197414e-05,
+      "loss": 1.0769,
+      "step": 6050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999563017676225e-05,
+      "loss": 1.007,
+      "step": 6051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999562870808034e-05,
+      "loss": 1.052,
+      "step": 6052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995627239151675e-05,
+      "loss": 1.1273,
+      "step": 6053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995625769976274e-05,
+      "loss": 1.076,
+      "step": 6054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999562430055412e-05,
+      "loss": 0.9607,
+      "step": 6055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999562283088523e-05,
+      "loss": 1.3641,
+      "step": 6056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995621360969586e-05,
+      "loss": 1.1847,
+      "step": 6057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99956198908072e-05,
+      "loss": 1.026,
+      "step": 6058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999561842039806e-05,
+      "loss": 1.3139,
+      "step": 6059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999561694974218e-05,
+      "loss": 1.0536,
+      "step": 6060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999561547883955e-05,
+      "loss": 0.956,
+      "step": 6061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995614007690176e-05,
+      "loss": 1.3575,
+      "step": 6062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999561253629406e-05,
+      "loss": 1.0711,
+      "step": 6063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995611064651195e-05,
+      "loss": 1.0042,
+      "step": 6064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999560959276158e-05,
+      "loss": 0.8043,
+      "step": 6065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999560812062522e-05,
+      "loss": 1.4357,
+      "step": 6066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999560664824212e-05,
+      "loss": 1.1579,
+      "step": 6067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995605175612264e-05,
+      "loss": 1.1137,
+      "step": 6068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995603702735675e-05,
+      "loss": 1.3972,
+      "step": 6069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999560222961233e-05,
+      "loss": 0.9841,
+      "step": 6070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995600756242236e-05,
+      "loss": 1.4509,
+      "step": 6071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999559928262541e-05,
+      "loss": 1.2235,
+      "step": 6072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999559780876182e-05,
+      "loss": 1.0759,
+      "step": 6073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99955963346515e-05,
+      "loss": 1.0943,
+      "step": 6074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995594860294424e-05,
+      "loss": 0.8859,
+      "step": 6075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99955933856906e-05,
+      "loss": 1.0972,
+      "step": 6076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995591910840036e-05,
+      "loss": 1.0955,
+      "step": 6077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999559043574272e-05,
+      "loss": 1.2344,
+      "step": 6078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995588960398665e-05,
+      "loss": 1.1871,
+      "step": 6079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999558748480786e-05,
+      "loss": 1.0826,
+      "step": 6080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999558600897032e-05,
+      "loss": 1.0216,
+      "step": 6081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999558453288602e-05,
+      "loss": 0.9615,
+      "step": 6082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999558305655499e-05,
+      "loss": 1.3907,
+      "step": 6083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999558157997719e-05,
+      "loss": 1.0384,
+      "step": 6084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999558010315266e-05,
+      "loss": 1.2994,
+      "step": 6085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999557862608139e-05,
+      "loss": 1.2716,
+      "step": 6086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999557714876336e-05,
+      "loss": 1.1883,
+      "step": 6087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999557567119859e-05,
+      "loss": 1.1904,
+      "step": 6088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995574193387077e-05,
+      "loss": 1.3526,
+      "step": 6089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995572715328814e-05,
+      "loss": 1.2066,
+      "step": 6090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999557123702381e-05,
+      "loss": 0.7818,
+      "step": 6091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995569758472056e-05,
+      "loss": 0.7816,
+      "step": 6092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995568279673554e-05,
+      "loss": 1.0388,
+      "step": 6093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995566800628315e-05,
+      "loss": 1.3529,
+      "step": 6094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995565321336325e-05,
+      "loss": 2.0495,
+      "step": 6095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999556384179759e-05,
+      "loss": 1.6708,
+      "step": 6096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995562362012106e-05,
+      "loss": 1.2122,
+      "step": 6097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999556088197988e-05,
+      "loss": 1.3476,
+      "step": 6098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999555940170091e-05,
+      "loss": 1.5162,
+      "step": 6099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995557921175186e-05,
+      "loss": 1.7973,
+      "step": 6100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995556440402725e-05,
+      "loss": 1.6417,
+      "step": 6101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999555495938352e-05,
+      "loss": 1.5791,
+      "step": 6102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995553478117563e-05,
+      "loss": 0.623,
+      "step": 6103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999555199660486e-05,
+      "loss": 1.1451,
+      "step": 6104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999555051484541e-05,
+      "loss": 1.2129,
+      "step": 6105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999554903283922e-05,
+      "loss": 0.9466,
+      "step": 6106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995547550586284e-05,
+      "loss": 0.4574,
+      "step": 6107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99955460680866e-05,
+      "loss": 0.6577,
+      "step": 6108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999554458534017e-05,
+      "loss": 1.2173,
+      "step": 6109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995543102346995e-05,
+      "loss": 1.0804,
+      "step": 6110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999554161910708e-05,
+      "loss": 1.148,
+      "step": 6111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999554013562041e-05,
+      "loss": 1.3866,
+      "step": 6112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999553865188701e-05,
+      "loss": 1.078,
+      "step": 6113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999553716790685e-05,
+      "loss": 0.9918,
+      "step": 6114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999553568367995e-05,
+      "loss": 0.9809,
+      "step": 6115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99955341992063e-05,
+      "loss": 0.9585,
+      "step": 6116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999553271448591e-05,
+      "loss": 1.3153,
+      "step": 6117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999553122951878e-05,
+      "loss": 1.0406,
+      "step": 6118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999552974430489e-05,
+      "loss": 1.2737,
+      "step": 6119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999552825884426e-05,
+      "loss": 1.3108,
+      "step": 6120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999552677313689e-05,
+      "loss": 1.1334,
+      "step": 6121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995525287182775e-05,
+      "loss": 1.6797,
+      "step": 6122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999552380098191e-05,
+      "loss": 1.2596,
+      "step": 6123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99955223145343e-05,
+      "loss": 0.8459,
+      "step": 6124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999552082783995e-05,
+      "loss": 1.0164,
+      "step": 6125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999551934089884e-05,
+      "loss": 0.5692,
+      "step": 6126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995517853711e-05,
+      "loss": 0.6399,
+      "step": 6127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999551636627641e-05,
+      "loss": 1.1292,
+      "step": 6128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999551487859507e-05,
+      "loss": 0.9345,
+      "step": 6129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999551339066699e-05,
+      "loss": 1.1914,
+      "step": 6130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999551190249217e-05,
+      "loss": 1.0834,
+      "step": 6131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99955104140706e-05,
+      "loss": 1.3149,
+      "step": 6132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999550892540229e-05,
+      "loss": 1.2369,
+      "step": 6133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999550743648722e-05,
+      "loss": 1.0676,
+      "step": 6134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995505947325415e-05,
+      "loss": 1.2102,
+      "step": 6135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995504457916864e-05,
+      "loss": 1.1858,
+      "step": 6136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999550296826157e-05,
+      "loss": 0.9689,
+      "step": 6137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999550147835953e-05,
+      "loss": 1.1618,
+      "step": 6138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995499988210744e-05,
+      "loss": 1.1267,
+      "step": 6139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999549849781521e-05,
+      "loss": 0.9775,
+      "step": 6140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999549700717293e-05,
+      "loss": 1.2295,
+      "step": 6141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995495516283914e-05,
+      "loss": 1.0337,
+      "step": 6142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999549402514815e-05,
+      "loss": 0.7293,
+      "step": 6143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999549253376563e-05,
+      "loss": 0.9217,
+      "step": 6144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999549104213638e-05,
+      "loss": 0.9321,
+      "step": 6145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995489550260374e-05,
+      "loss": 0.7627,
+      "step": 6146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999548805813763e-05,
+      "loss": 0.8338,
+      "step": 6147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999548656576814e-05,
+      "loss": 0.8278,
+      "step": 6148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995485073151905e-05,
+      "loss": 1.0563,
+      "step": 6149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999548358028893e-05,
+      "loss": 1.2953,
+      "step": 6150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995482087179196e-05,
+      "loss": 0.9959,
+      "step": 6151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999548059382273e-05,
+      "loss": 1.1194,
+      "step": 6152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999547910021951e-05,
+      "loss": 1.236,
+      "step": 6153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999547760636955e-05,
+      "loss": 1.0783,
+      "step": 6154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999547611227284e-05,
+      "loss": 1.3052,
+      "step": 6155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995474617929395e-05,
+      "loss": 0.9864,
+      "step": 6156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99954731233392e-05,
+      "loss": 1.2228,
+      "step": 6157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999547162850226e-05,
+      "loss": 1.0359,
+      "step": 6158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995470133418576e-05,
+      "loss": 0.8855,
+      "step": 6159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999546863808815e-05,
+      "loss": 1.1619,
+      "step": 6160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999546714251098e-05,
+      "loss": 1.0703,
+      "step": 6161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999546564668705e-05,
+      "loss": 1.0541,
+      "step": 6162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999546415061639e-05,
+      "loss": 1.301,
+      "step": 6163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995462654298984e-05,
+      "loss": 1.2866,
+      "step": 6164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999546115773483e-05,
+      "loss": 1.3945,
+      "step": 6165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999545966092393e-05,
+      "loss": 0.7452,
+      "step": 6166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995458163866294e-05,
+      "loss": 0.3172,
+      "step": 6167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995456666561904e-05,
+      "loss": 0.2841,
+      "step": 6168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999545516901078e-05,
+      "loss": 0.2657,
+      "step": 6169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99954536712129e-05,
+      "loss": 1.2747,
+      "step": 6170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999545217316828e-05,
+      "loss": 0.9224,
+      "step": 6171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999545067487691e-05,
+      "loss": 1.1763,
+      "step": 6172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99954491763388e-05,
+      "loss": 1.1748,
+      "step": 6173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999544767755395e-05,
+      "loss": 1.0592,
+      "step": 6174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999544617852235e-05,
+      "loss": 0.9498,
+      "step": 6175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999544467924401e-05,
+      "loss": 1.3276,
+      "step": 6176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995443179718914e-05,
+      "loss": 1.1507,
+      "step": 6177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995441679947084e-05,
+      "loss": 0.4551,
+      "step": 6178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999544017992851e-05,
+      "loss": 0.8058,
+      "step": 6179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995438679663184e-05,
+      "loss": 0.847,
+      "step": 6180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999543717915112e-05,
+      "loss": 1.1359,
+      "step": 6181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999543567839231e-05,
+      "loss": 0.873,
+      "step": 6182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999543417738676e-05,
+      "loss": 0.9266,
+      "step": 6183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995432676134455e-05,
+      "loss": 0.8686,
+      "step": 6184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999543117463542e-05,
+      "loss": 1.9153,
+      "step": 6185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995429672889626e-05,
+      "loss": 1.4137,
+      "step": 6186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99954281708971e-05,
+      "loss": 1.2259,
+      "step": 6187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999542666865782e-05,
+      "loss": 1.1254,
+      "step": 6188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995425166171805e-05,
+      "loss": 1.0492,
+      "step": 6189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999542366343903e-05,
+      "loss": 1.0931,
+      "step": 6190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999542216045953e-05,
+      "loss": 1.1489,
+      "step": 6191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995420657233274e-05,
+      "loss": 1.3093,
+      "step": 6192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999541915376028e-05,
+      "loss": 1.1749,
+      "step": 6193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999541765004053e-05,
+      "loss": 1.2643,
+      "step": 6194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999541614607405e-05,
+      "loss": 1.4281,
+      "step": 6195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995414641860816e-05,
+      "loss": 1.284,
+      "step": 6196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999541313740085e-05,
+      "loss": 1.1549,
+      "step": 6197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999541163269412e-05,
+      "loss": 1.0731,
+      "step": 6198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999541012774066e-05,
+      "loss": 0.891,
+      "step": 6199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999540862254046e-05,
+      "loss": 0.7303,
+      "step": 6200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999540711709351e-05,
+      "loss": 0.8618,
+      "step": 6201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999540561139981e-05,
+      "loss": 0.6789,
+      "step": 6202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999540410545937e-05,
+      "loss": 0.9251,
+      "step": 6203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999540259927219e-05,
+      "loss": 0.9536,
+      "step": 6204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995401092838255e-05,
+      "loss": 0.9837,
+      "step": 6205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999539958615759e-05,
+      "loss": 0.7343,
+      "step": 6206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999539807923017e-05,
+      "loss": 0.9371,
+      "step": 6207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995396572056015e-05,
+      "loss": 0.9017,
+      "step": 6208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999539506463511e-05,
+      "loss": 0.9413,
+      "step": 6209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999539355696746e-05,
+      "loss": 1.1699,
+      "step": 6210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999539204905307e-05,
+      "loss": 1.277,
+      "step": 6211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995390540891935e-05,
+      "loss": 1.2824,
+      "step": 6212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995389032484055e-05,
+      "loss": 1.1497,
+      "step": 6213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999538752382943e-05,
+      "loss": 1.1358,
+      "step": 6214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999538601492806e-05,
+      "loss": 0.8836,
+      "step": 6215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999538450577995e-05,
+      "loss": 0.3426,
+      "step": 6216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99953829963851e-05,
+      "loss": 0.9849,
+      "step": 6217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99953814867435e-05,
+      "loss": 0.6789,
+      "step": 6218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999537997685515e-05,
+      "loss": 0.6195,
+      "step": 6219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995378466720066e-05,
+      "loss": 0.9728,
+      "step": 6220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999537695633823e-05,
+      "loss": 1.2863,
+      "step": 6221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995375445709656e-05,
+      "loss": 1.1435,
+      "step": 6222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999537393483434e-05,
+      "loss": 1.0346,
+      "step": 6223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999537242371228e-05,
+      "loss": 1.0292,
+      "step": 6224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995370912343474e-05,
+      "loss": 1.1251,
+      "step": 6225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999536940072793e-05,
+      "loss": 1.1762,
+      "step": 6226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999536788886563e-05,
+      "loss": 1.1432,
+      "step": 6227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999536637675659e-05,
+      "loss": 2.3195,
+      "step": 6228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999536486440081e-05,
+      "loss": 1.9143,
+      "step": 6229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999536335179829e-05,
+      "loss": 1.2253,
+      "step": 6230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999536183894902e-05,
+      "loss": 1.1785,
+      "step": 6231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999536032585301e-05,
+      "loss": 1.0539,
+      "step": 6232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995358812510254e-05,
+      "loss": 0.9327,
+      "step": 6233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995357298920754e-05,
+      "loss": 0.5802,
+      "step": 6234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999535578508451e-05,
+      "loss": 1.1302,
+      "step": 6235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995354271001524e-05,
+      "loss": 1.0786,
+      "step": 6236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999535275667179e-05,
+      "loss": 1.0324,
+      "step": 6237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999535124209532e-05,
+      "loss": 1.0391,
+      "step": 6238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995349727272104e-05,
+      "loss": 0.9431,
+      "step": 6239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999534821220214e-05,
+      "loss": 0.9996,
+      "step": 6240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999534669688544e-05,
+      "loss": 1.2086,
+      "step": 6241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999534518132199e-05,
+      "loss": 1.4165,
+      "step": 6242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99953436655118e-05,
+      "loss": 1.1405,
+      "step": 6243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999534214945486e-05,
+      "loss": 1.2725,
+      "step": 6244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999534063315118e-05,
+      "loss": 1.151,
+      "step": 6245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999533911660076e-05,
+      "loss": 1.1695,
+      "step": 6246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999533759980359e-05,
+      "loss": 1.0715,
+      "step": 6247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995336082759683e-05,
+      "loss": 1.0895,
+      "step": 6248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999533456546903e-05,
+      "loss": 1.3081,
+      "step": 6249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999533304793164e-05,
+      "loss": 1.5056,
+      "step": 6250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99953315301475e-05,
+      "loss": 1.3054,
+      "step": 6251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999533001211662e-05,
+      "loss": 1.1802,
+      "step": 6252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999532849383899e-05,
+      "loss": 1.2632,
+      "step": 6253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999532697531462e-05,
+      "loss": 1.1067,
+      "step": 6254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999532545654351e-05,
+      "loss": 1.1343,
+      "step": 6255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999532393752565e-05,
+      "loss": 1.2952,
+      "step": 6256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999532241826105e-05,
+      "loss": 1.4313,
+      "step": 6257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999532089874971e-05,
+      "loss": 1.7484,
+      "step": 6258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999531937899162e-05,
+      "loss": 1.668,
+      "step": 6259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99953178589868e-05,
+      "loss": 1.3347,
+      "step": 6260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999531633873522e-05,
+      "loss": 1.0661,
+      "step": 6261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999531481823691e-05,
+      "loss": 0.9182,
+      "step": 6262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999531329749185e-05,
+      "loss": 1.0667,
+      "step": 6263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999531177650005e-05,
+      "loss": 1.1956,
+      "step": 6264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99953102552615e-05,
+      "loss": 1.3031,
+      "step": 6265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999530873377621e-05,
+      "loss": 1.0669,
+      "step": 6266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995307212044186e-05,
+      "loss": 1.2452,
+      "step": 6267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999530569006541e-05,
+      "loss": 1.4003,
+      "step": 6268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995304167839893e-05,
+      "loss": 1.3337,
+      "step": 6269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999530264536763e-05,
+      "loss": 1.2999,
+      "step": 6270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995301122648624e-05,
+      "loss": 1.2661,
+      "step": 6271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999529959968288e-05,
+      "loss": 0.9675,
+      "step": 6272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995298076470385e-05,
+      "loss": 1.1439,
+      "step": 6273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995296553011157e-05,
+      "loss": 1.282,
+      "step": 6274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999529502930518e-05,
+      "loss": 1.2966,
+      "step": 6275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999529350535246e-05,
+      "loss": 1.1969,
+      "step": 6276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995291981153e-05,
+      "loss": 1.2212,
+      "step": 6277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999529045670679e-05,
+      "loss": 0.8573,
+      "step": 6278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995288932013845e-05,
+      "loss": 1.2187,
+      "step": 6279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999528740707415e-05,
+      "loss": 1.0123,
+      "step": 6280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999528588188772e-05,
+      "loss": 1.3662,
+      "step": 6281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999528435645454e-05,
+      "loss": 1.915,
+      "step": 6282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999528283077463e-05,
+      "loss": 0.9141,
+      "step": 6283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995281304847964e-05,
+      "loss": 1.4028,
+      "step": 6284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999527977867455e-05,
+      "loss": 1.143,
+      "step": 6285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995278252254406e-05,
+      "loss": 1.1971,
+      "step": 6286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999527672558751e-05,
+      "loss": 1.0219,
+      "step": 6287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999527519867388e-05,
+      "loss": 0.7101,
+      "step": 6288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995273671513496e-05,
+      "loss": 0.8893,
+      "step": 6289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999527214410639e-05,
+      "loss": 1.2082,
+      "step": 6290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999527061645252e-05,
+      "loss": 1.1676,
+      "step": 6291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999526908855191e-05,
+      "loss": 0.8196,
+      "step": 6292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999526756040457e-05,
+      "loss": 1.2572,
+      "step": 6293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999526603201048e-05,
+      "loss": 1.1991,
+      "step": 6294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995264503369645e-05,
+      "loss": 1.1967,
+      "step": 6295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995262974482066e-05,
+      "loss": 0.994,
+      "step": 6296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995261445347744e-05,
+      "loss": 1.1037,
+      "step": 6297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995259915966684e-05,
+      "loss": 1.2701,
+      "step": 6298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999525838633888e-05,
+      "loss": 0.863,
+      "step": 6299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999525685646433e-05,
+      "loss": 1.1585,
+      "step": 6300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999525532634304e-05,
+      "loss": 1.436,
+      "step": 6301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999525379597501e-05,
+      "loss": 0.9212,
+      "step": 6302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999525226536024e-05,
+      "loss": 1.8257,
+      "step": 6303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995250734498713e-05,
+      "loss": 0.5991,
+      "step": 6304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999524920339046e-05,
+      "loss": 1.1361,
+      "step": 6305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999524767203545e-05,
+      "loss": 1.1269,
+      "step": 6306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999524614043371e-05,
+      "loss": 1.1514,
+      "step": 6307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999524460858522e-05,
+      "loss": 1.3033,
+      "step": 6308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999524307648999e-05,
+      "loss": 0.9704,
+      "step": 6309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999524154414802e-05,
+      "loss": 1.1856,
+      "step": 6310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995240011559304e-05,
+      "loss": 1.4247,
+      "step": 6311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999523847872385e-05,
+      "loss": 1.0136,
+      "step": 6312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999523694564165e-05,
+      "loss": 1.3243,
+      "step": 6313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99952354123127e-05,
+      "loss": 0.9522,
+      "step": 6314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999523387873701e-05,
+      "loss": 1.2351,
+      "step": 6315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999523234491459e-05,
+      "loss": 1.0311,
+      "step": 6316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999523081084542e-05,
+      "loss": 1.3266,
+      "step": 6317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999522927652951e-05,
+      "loss": 1.2291,
+      "step": 6318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999522774196686e-05,
+      "loss": 1.4014,
+      "step": 6319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995226207157455e-05,
+      "loss": 1.4934,
+      "step": 6320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995224672101315e-05,
+      "loss": 1.2602,
+      "step": 6321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999522313679843e-05,
+      "loss": 1.2214,
+      "step": 6322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999522160124881e-05,
+      "loss": 1.1039,
+      "step": 6323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995220065452445e-05,
+      "loss": 1.1839,
+      "step": 6324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995218529409336e-05,
+      "loss": 1.3264,
+      "step": 6325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999521699311949e-05,
+      "loss": 1.4254,
+      "step": 6326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999521545658289e-05,
+      "loss": 0.9513,
+      "step": 6327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995213919799556e-05,
+      "loss": 1.7214,
+      "step": 6328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999521238276947e-05,
+      "loss": 2.0432,
+      "step": 6329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995210845492654e-05,
+      "loss": 1.57,
+      "step": 6330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995209307969094e-05,
+      "loss": 1.2588,
+      "step": 6331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995207770198796e-05,
+      "loss": 1.3624,
+      "step": 6332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999520623218175e-05,
+      "loss": 1.1314,
+      "step": 6333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995204693917955e-05,
+      "loss": 0.9187,
+      "step": 6334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999520315540742e-05,
+      "loss": 1.0166,
+      "step": 6335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999520161665016e-05,
+      "loss": 0.987,
+      "step": 6336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999520007764614e-05,
+      "loss": 1.1031,
+      "step": 6337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995198538395385e-05,
+      "loss": 1.2838,
+      "step": 6338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999519699889788e-05,
+      "loss": 0.6294,
+      "step": 6339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999519545915364e-05,
+      "loss": 0.9347,
+      "step": 6340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999519391916266e-05,
+      "loss": 0.9834,
+      "step": 6341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999519237892493e-05,
+      "loss": 1.083,
+      "step": 6342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999519083844047e-05,
+      "loss": 1.3272,
+      "step": 6343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995189297709254e-05,
+      "loss": 1.0224,
+      "step": 6344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99951877567313e-05,
+      "loss": 1.2144,
+      "step": 6345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999518621550661e-05,
+      "loss": 1.216,
+      "step": 6346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999518467403518e-05,
+      "loss": 1.0545,
+      "step": 6347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995183132317e-05,
+      "loss": 1.6558,
+      "step": 6348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999518159035208e-05,
+      "loss": 1.3072,
+      "step": 6349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995180048140424e-05,
+      "loss": 1.5042,
+      "step": 6350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995178505682015e-05,
+      "loss": 1.491,
+      "step": 6351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999517696297688e-05,
+      "loss": 1.4511,
+      "step": 6352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999517542002499e-05,
+      "loss": 1.2005,
+      "step": 6353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999517387682636e-05,
+      "loss": 0.913,
+      "step": 6354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999517233338099e-05,
+      "loss": 0.9846,
+      "step": 6355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999517078968888e-05,
+      "loss": 0.995,
+      "step": 6356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999516924575002e-05,
+      "loss": 1.0608,
+      "step": 6357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995167701564425e-05,
+      "loss": 0.9245,
+      "step": 6358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999516615713209e-05,
+      "loss": 0.947,
+      "step": 6359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995164612453014e-05,
+      "loss": 0.8862,
+      "step": 6360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99951630675272e-05,
+      "loss": 0.9554,
+      "step": 6361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995161522354626e-05,
+      "loss": 1.039,
+      "step": 6362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999515997693532e-05,
+      "loss": 1.1183,
+      "step": 6363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999515843126928e-05,
+      "loss": 1.1794,
+      "step": 6364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999515688535649e-05,
+      "loss": 1.4328,
+      "step": 6365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999515533919696e-05,
+      "loss": 0.9539,
+      "step": 6366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995153792790696e-05,
+      "loss": 0.9478,
+      "step": 6367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999515224613768e-05,
+      "loss": 0.9248,
+      "step": 6368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995150699237924e-05,
+      "loss": 0.7609,
+      "step": 6369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999514915209143e-05,
+      "loss": 1.0044,
+      "step": 6370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999514760469819e-05,
+      "loss": 1.1496,
+      "step": 6371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999514605705822e-05,
+      "loss": 0.828,
+      "step": 6372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999514450917149e-05,
+      "loss": 1.1735,
+      "step": 6373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999514296103803e-05,
+      "loss": 1.0782,
+      "step": 6374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995141412657834e-05,
+      "loss": 1.0884,
+      "step": 6375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995139864030884e-05,
+      "loss": 0.8416,
+      "step": 6376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99951383151572e-05,
+      "loss": 1.1527,
+      "step": 6377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995136766036774e-05,
+      "loss": 1.0447,
+      "step": 6378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99951352166696e-05,
+      "loss": 1.3278,
+      "step": 6379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995133667055694e-05,
+      "loss": 0.9146,
+      "step": 6380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999513211719504e-05,
+      "loss": 1.1876,
+      "step": 6381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995130567087644e-05,
+      "loss": 1.1195,
+      "step": 6382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999512901673351e-05,
+      "loss": 0.8774,
+      "step": 6383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999512746613263e-05,
+      "loss": 1.2939,
+      "step": 6384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999512591528501e-05,
+      "loss": 1.1973,
+      "step": 6385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999512436419066e-05,
+      "loss": 1.1773,
+      "step": 6386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999512281284956e-05,
+      "loss": 1.3966,
+      "step": 6387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999512126126171e-05,
+      "loss": 0.8522,
+      "step": 6388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999511970942713e-05,
+      "loss": 1.1953,
+      "step": 6389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999511815734581e-05,
+      "loss": 1.2966,
+      "step": 6390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999511660501775e-05,
+      "loss": 1.1662,
+      "step": 6391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999511505244294e-05,
+      "loss": 1.4128,
+      "step": 6392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999511349962139e-05,
+      "loss": 1.075,
+      "step": 6393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99951119465531e-05,
+      "loss": 1.1241,
+      "step": 6394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999511039323807e-05,
+      "loss": 1.1761,
+      "step": 6395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99951088396763e-05,
+      "loss": 1.3527,
+      "step": 6396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999510728586778e-05,
+      "loss": 1.2662,
+      "step": 6397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999510573181253e-05,
+      "loss": 0.9103,
+      "step": 6398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999510417751053e-05,
+      "loss": 0.9209,
+      "step": 6399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995102622961796e-05,
+      "loss": 0.6531,
+      "step": 6400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999510106816632e-05,
+      "loss": 0.145,
+      "step": 6401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99950995131241e-05,
+      "loss": 0.0826,
+      "step": 6402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999509795783515e-05,
+      "loss": 0.0694,
+      "step": 6403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999509640229944e-05,
+      "loss": 0.2337,
+      "step": 6404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995094846517e-05,
+      "loss": 0.2951,
+      "step": 6405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995093290487814e-05,
+      "loss": 1.0009,
+      "step": 6406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999509173421189e-05,
+      "loss": 0.9331,
+      "step": 6407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999509017768923e-05,
+      "loss": 1.3308,
+      "step": 6408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995088620919825e-05,
+      "loss": 1.3025,
+      "step": 6409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999508706390368e-05,
+      "loss": 0.9508,
+      "step": 6410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999508550664079e-05,
+      "loss": 0.854,
+      "step": 6411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999508394913116e-05,
+      "loss": 0.9445,
+      "step": 6412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995082391374795e-05,
+      "loss": 1.2569,
+      "step": 6413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999508083337168e-05,
+      "loss": 1.3479,
+      "step": 6414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995079275121825e-05,
+      "loss": 1.3147,
+      "step": 6415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999507771662524e-05,
+      "loss": 1.3282,
+      "step": 6416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995076157881906e-05,
+      "loss": 1.3538,
+      "step": 6417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999507459889183e-05,
+      "loss": 1.3551,
+      "step": 6418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999507303965502e-05,
+      "loss": 1.2779,
+      "step": 6419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999507148017146e-05,
+      "loss": 1.3151,
+      "step": 6420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999506992044116e-05,
+      "loss": 1.1586,
+      "step": 6421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995068360464125e-05,
+      "loss": 0.8012,
+      "step": 6422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999506680024034e-05,
+      "loss": 0.5621,
+      "step": 6423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999506523976982e-05,
+      "loss": 0.6656,
+      "step": 6424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999506367905257e-05,
+      "loss": 0.6819,
+      "step": 6425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995062118088564e-05,
+      "loss": 0.9057,
+      "step": 6426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999506055687783e-05,
+      "loss": 1.3069,
+      "step": 6427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999505899542034e-05,
+      "loss": 0.8545,
+      "step": 6428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999505743371612e-05,
+      "loss": 1.0369,
+      "step": 6429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999505587176516e-05,
+      "loss": 1.0941,
+      "step": 6430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995054309567455e-05,
+      "loss": 1.3182,
+      "step": 6431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995052747123006e-05,
+      "loss": 1.4654,
+      "step": 6432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999505118443183e-05,
+      "loss": 1.0973,
+      "step": 6433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99950496214939e-05,
+      "loss": 0.9623,
+      "step": 6434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995048058309236e-05,
+      "loss": 1.0791,
+      "step": 6435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999504649487783e-05,
+      "loss": 1.4668,
+      "step": 6436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995044931199676e-05,
+      "loss": 1.3924,
+      "step": 6437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999504336727478e-05,
+      "loss": 0.6485,
+      "step": 6438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999504180310316e-05,
+      "loss": 1.164,
+      "step": 6439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995040238684785e-05,
+      "loss": 1.101,
+      "step": 6440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999503867401968e-05,
+      "loss": 1.0594,
+      "step": 6441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999503710910783e-05,
+      "loss": 1.0415,
+      "step": 6442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995035543949245e-05,
+      "loss": 1.0195,
+      "step": 6443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999503397854391e-05,
+      "loss": 0.9997,
+      "step": 6444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999503241289184e-05,
+      "loss": 1.1615,
+      "step": 6445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999503084699303e-05,
+      "loss": 0.9525,
+      "step": 6446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999502928084747e-05,
+      "loss": 1.1436,
+      "step": 6447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999502771445518e-05,
+      "loss": 0.9101,
+      "step": 6448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999502614781615e-05,
+      "loss": 1.2262,
+      "step": 6449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995024580930375e-05,
+      "loss": 0.9769,
+      "step": 6450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995023013797856e-05,
+      "loss": 0.797,
+      "step": 6451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995021446418614e-05,
+      "loss": 1.1696,
+      "step": 6452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995019878792614e-05,
+      "loss": 1.226,
+      "step": 6453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995018310919877e-05,
+      "loss": 1.2363,
+      "step": 6454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99950167428004e-05,
+      "loss": 1.0814,
+      "step": 6455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999501517443419e-05,
+      "loss": 1.03,
+      "step": 6456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999501360582123e-05,
+      "loss": 1.0106,
+      "step": 6457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995012036961534e-05,
+      "loss": 0.7034,
+      "step": 6458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995010467855097e-05,
+      "loss": 0.5473,
+      "step": 6459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999500889850193e-05,
+      "loss": 0.5888,
+      "step": 6460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999500732890201e-05,
+      "loss": 0.2767,
+      "step": 6461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999500575905535e-05,
+      "loss": 0.2416,
+      "step": 6462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995004188961954e-05,
+      "loss": 0.1992,
+      "step": 6463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995002618621816e-05,
+      "loss": 0.4523,
+      "step": 6464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999500104803494e-05,
+      "loss": 0.262,
+      "step": 6465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999499947720132e-05,
+      "loss": 0.177,
+      "step": 6466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994997906120966e-05,
+      "loss": 0.1315,
+      "step": 6467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994996334793866e-05,
+      "loss": 0.1851,
+      "step": 6468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999499476322003e-05,
+      "loss": 0.3109,
+      "step": 6469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994993191399453e-05,
+      "loss": 0.8243,
+      "step": 6470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994991619332134e-05,
+      "loss": 1.1854,
+      "step": 6471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999499004701808e-05,
+      "loss": 1.1112,
+      "step": 6472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994988474457285e-05,
+      "loss": 1.0755,
+      "step": 6473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999498690164974e-05,
+      "loss": 1.141,
+      "step": 6474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994985328595465e-05,
+      "loss": 1.4014,
+      "step": 6475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994983755294446e-05,
+      "loss": 1.1114,
+      "step": 6476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999498218174668e-05,
+      "loss": 1.0627,
+      "step": 6477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999498060795219e-05,
+      "loss": 1.2869,
+      "step": 6478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999497903391096e-05,
+      "loss": 1.2929,
+      "step": 6479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994977459622976e-05,
+      "loss": 1.0122,
+      "step": 6480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999497588508826e-05,
+      "loss": 0.7117,
+      "step": 6481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99949743103068e-05,
+      "loss": 0.7217,
+      "step": 6482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999497273527861e-05,
+      "loss": 1.2543,
+      "step": 6483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999497116000367e-05,
+      "loss": 1.2273,
+      "step": 6484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994969584481994e-05,
+      "loss": 1.0454,
+      "step": 6485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994968008713574e-05,
+      "loss": 1.1889,
+      "step": 6486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999496643269842e-05,
+      "loss": 1.0451,
+      "step": 6487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994964856436524e-05,
+      "loss": 1.039,
+      "step": 6488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994963279927886e-05,
+      "loss": 0.8729,
+      "step": 6489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999496170317252e-05,
+      "loss": 0.7305,
+      "step": 6490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99949601261704e-05,
+      "loss": 1.1235,
+      "step": 6491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999495854892154e-05,
+      "loss": 1.2587,
+      "step": 6492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999495697142595e-05,
+      "loss": 1.1407,
+      "step": 6493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994955393683616e-05,
+      "loss": 1.2415,
+      "step": 6494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999495381569455e-05,
+      "loss": 1.3505,
+      "step": 6495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999495223745873e-05,
+      "loss": 1.1493,
+      "step": 6496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999495065897618e-05,
+      "loss": 1.2276,
+      "step": 6497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999494908024688e-05,
+      "loss": 0.9747,
+      "step": 6498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999494750127085e-05,
+      "loss": 1.2284,
+      "step": 6499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999494592204808e-05,
+      "loss": 1.2709,
+      "step": 6500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0758188962936401,
+      "eval_runtime": 94.9288,
+      "eval_samples_per_second": 14.59,
+      "eval_steps_per_second": 3.655,
+      "step": 6500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994944342578564e-05,
+      "loss": 1.2316,
+      "step": 6501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999494276286232e-05,
+      "loss": 1.0427,
+      "step": 6502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994941182899326e-05,
+      "loss": 1.5905,
+      "step": 6503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994939602689595e-05,
+      "loss": 0.9685,
+      "step": 6504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994938022233126e-05,
+      "loss": 1.0835,
+      "step": 6505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999493644152991e-05,
+      "loss": 1.1555,
+      "step": 6506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999493486057997e-05,
+      "loss": 1.1553,
+      "step": 6507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994933279383275e-05,
+      "loss": 1.2677,
+      "step": 6508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999493169793985e-05,
+      "loss": 1.3516,
+      "step": 6509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999493011624968e-05,
+      "loss": 1.1805,
+      "step": 6510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999492853431278e-05,
+      "loss": 1.0536,
+      "step": 6511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999492695212913e-05,
+      "loss": 0.7601,
+      "step": 6512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994925369698744e-05,
+      "loss": 1.234,
+      "step": 6513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999492378702162e-05,
+      "loss": 1.1395,
+      "step": 6514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994922204097764e-05,
+      "loss": 1.0047,
+      "step": 6515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999492062092716e-05,
+      "loss": 1.2291,
+      "step": 6516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994919037509814e-05,
+      "loss": 1.1659,
+      "step": 6517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999491745384573e-05,
+      "loss": 1.0848,
+      "step": 6518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999491586993491e-05,
+      "loss": 1.1353,
+      "step": 6519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999491428577735e-05,
+      "loss": 0.9852,
+      "step": 6520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994912701373046e-05,
+      "loss": 1.0104,
+      "step": 6521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999491111672201e-05,
+      "loss": 1.2367,
+      "step": 6522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999490953182423e-05,
+      "loss": 1.0964,
+      "step": 6523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994907946679716e-05,
+      "loss": 0.6671,
+      "step": 6524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994906361288454e-05,
+      "loss": 1.2634,
+      "step": 6525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999490477565046e-05,
+      "loss": 1.1709,
+      "step": 6526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999490318976573e-05,
+      "loss": 0.6647,
+      "step": 6527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999490160363425e-05,
+      "loss": 0.5002,
+      "step": 6528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999490001725604e-05,
+      "loss": 1.1257,
+      "step": 6529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999489843063109e-05,
+      "loss": 1.2702,
+      "step": 6530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994896843759396e-05,
+      "loss": 1.1731,
+      "step": 6531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999489525664096e-05,
+      "loss": 1.0127,
+      "step": 6532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999489366927579e-05,
+      "loss": 1.1434,
+      "step": 6533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994892081663886e-05,
+      "loss": 1.0425,
+      "step": 6534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999489049380524e-05,
+      "loss": 1.4964,
+      "step": 6535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999488890569986e-05,
+      "loss": 1.5346,
+      "step": 6536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999488731734773e-05,
+      "loss": 1.4415,
+      "step": 6537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994885728748866e-05,
+      "loss": 1.3784,
+      "step": 6538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994884139903254e-05,
+      "loss": 1.4162,
+      "step": 6539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999488255081092e-05,
+      "loss": 1.6596,
+      "step": 6540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999488096147183e-05,
+      "loss": 1.2537,
+      "step": 6541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999487937188601e-05,
+      "loss": 0.9764,
+      "step": 6542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994877782053453e-05,
+      "loss": 1.1744,
+      "step": 6543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994876191974155e-05,
+      "loss": 1.3614,
+      "step": 6544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999487460164812e-05,
+      "loss": 1.3881,
+      "step": 6545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999487301107534e-05,
+      "loss": 1.0576,
+      "step": 6546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999487142025583e-05,
+      "loss": 1.1086,
+      "step": 6547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994869829189575e-05,
+      "loss": 1.1733,
+      "step": 6548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999486823787658e-05,
+      "loss": 1.1962,
+      "step": 6549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999486664631685e-05,
+      "loss": 1.044,
+      "step": 6550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994865054510374e-05,
+      "loss": 0.9311,
+      "step": 6551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994863462457164e-05,
+      "loss": 0.9222,
+      "step": 6552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994861870157216e-05,
+      "loss": 1.154,
+      "step": 6553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999486027761053e-05,
+      "loss": 1.0585,
+      "step": 6554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999485868481711e-05,
+      "loss": 1.2625,
+      "step": 6555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999485709177695e-05,
+      "loss": 1.243,
+      "step": 6556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999485549849004e-05,
+      "loss": 1.2738,
+      "step": 6557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99948539049564e-05,
+      "loss": 1.4934,
+      "step": 6558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999485231117602e-05,
+      "loss": 1.1014,
+      "step": 6559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99948507171489e-05,
+      "loss": 1.0812,
+      "step": 6560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999484912287504e-05,
+      "loss": 1.0902,
+      "step": 6561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994847528354446e-05,
+      "loss": 1.2505,
+      "step": 6562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999484593358711e-05,
+      "loss": 1.286,
+      "step": 6563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994844338573047e-05,
+      "loss": 1.432,
+      "step": 6564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999484274331223e-05,
+      "loss": 1.2748,
+      "step": 6565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999484114780468e-05,
+      "loss": 1.4985,
+      "step": 6566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999483955205039e-05,
+      "loss": 1.5063,
+      "step": 6567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999483795604936e-05,
+      "loss": 1.1407,
+      "step": 6568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994836359801594e-05,
+      "loss": 1.2549,
+      "step": 6569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999483476330709e-05,
+      "loss": 1.1062,
+      "step": 6570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994833166565846e-05,
+      "loss": 1.0548,
+      "step": 6571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999483156957787e-05,
+      "loss": 1.0383,
+      "step": 6572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999482997234315e-05,
+      "loss": 1.208,
+      "step": 6573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999482837486169e-05,
+      "loss": 0.8818,
+      "step": 6574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994826777133494e-05,
+      "loss": 0.8663,
+      "step": 6575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994825179158554e-05,
+      "loss": 1.0257,
+      "step": 6576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994823580936884e-05,
+      "loss": 0.9159,
+      "step": 6577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994821982468476e-05,
+      "loss": 1.2628,
+      "step": 6578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994820383753324e-05,
+      "loss": 1.2665,
+      "step": 6579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999481878479143e-05,
+      "loss": 0.9341,
+      "step": 6580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999481718558281e-05,
+      "loss": 1.3301,
+      "step": 6581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999481558612744e-05,
+      "loss": 0.5321,
+      "step": 6582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999481398642534e-05,
+      "loss": 0.5291,
+      "step": 6583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99948123864765e-05,
+      "loss": 0.745,
+      "step": 6584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994810786280924e-05,
+      "loss": 1.0752,
+      "step": 6585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99948091858386e-05,
+      "loss": 1.2102,
+      "step": 6586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999480758514954e-05,
+      "loss": 0.5819,
+      "step": 6587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994805984213746e-05,
+      "loss": 0.9426,
+      "step": 6588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994804383031215e-05,
+      "loss": 1.0205,
+      "step": 6589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994802781601946e-05,
+      "loss": 1.0928,
+      "step": 6590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999480117992593e-05,
+      "loss": 1.1653,
+      "step": 6591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999479957800319e-05,
+      "loss": 1.3666,
+      "step": 6592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999479797583371e-05,
+      "loss": 1.3309,
+      "step": 6593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999479637341748e-05,
+      "loss": 0.9612,
+      "step": 6594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994794770754515e-05,
+      "loss": 1.2552,
+      "step": 6595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999479316784482e-05,
+      "loss": 1.3941,
+      "step": 6596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999479156468838e-05,
+      "loss": 1.3753,
+      "step": 6597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994789961285204e-05,
+      "loss": 1.4621,
+      "step": 6598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999478835763529e-05,
+      "loss": 0.9614,
+      "step": 6599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994786753738644e-05,
+      "loss": 1.1641,
+      "step": 6600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994785149595244e-05,
+      "loss": 1.1776,
+      "step": 6601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999478354520513e-05,
+      "loss": 1.2815,
+      "step": 6602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999478194056826e-05,
+      "loss": 0.6628,
+      "step": 6603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994780335684656e-05,
+      "loss": 0.8115,
+      "step": 6604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994778730554314e-05,
+      "loss": 1.2254,
+      "step": 6605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999477712517723e-05,
+      "loss": 1.3675,
+      "step": 6606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999477551955342e-05,
+      "loss": 1.2994,
+      "step": 6607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999477391368286e-05,
+      "loss": 0.9946,
+      "step": 6608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994772307565566e-05,
+      "loss": 1.0522,
+      "step": 6609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999477070120153e-05,
+      "loss": 1.0744,
+      "step": 6610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994769094590764e-05,
+      "loss": 1.4907,
+      "step": 6611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999476748773326e-05,
+      "loss": 1.18,
+      "step": 6612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994765880629014e-05,
+      "loss": 0.9152,
+      "step": 6613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994764273278036e-05,
+      "loss": 1.127,
+      "step": 6614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994762665680314e-05,
+      "loss": 1.1186,
+      "step": 6615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994761057835855e-05,
+      "loss": 1.0582,
+      "step": 6616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999475944974466e-05,
+      "loss": 1.4079,
+      "step": 6617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994757841406724e-05,
+      "loss": 1.3987,
+      "step": 6618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999475623282205e-05,
+      "loss": 1.1375,
+      "step": 6619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994754623990645e-05,
+      "loss": 1.0294,
+      "step": 6620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99947530149125e-05,
+      "loss": 1.163,
+      "step": 6621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994751405587616e-05,
+      "loss": 0.1906,
+      "step": 6622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994749796015995e-05,
+      "loss": 0.6676,
+      "step": 6623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999474818619763e-05,
+      "loss": 1.3446,
+      "step": 6624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994746576132536e-05,
+      "loss": 1.1012,
+      "step": 6625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994744965820703e-05,
+      "loss": 1.0001,
+      "step": 6626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999474335526213e-05,
+      "loss": 1.0914,
+      "step": 6627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999474174445682e-05,
+      "loss": 0.9416,
+      "step": 6628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999474013340477e-05,
+      "loss": 1.1493,
+      "step": 6629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999473852210599e-05,
+      "loss": 1.2614,
+      "step": 6630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999473691056046e-05,
+      "loss": 1.3037,
+      "step": 6631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994735298768206e-05,
+      "loss": 0.8561,
+      "step": 6632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994733686729205e-05,
+      "loss": 1.146,
+      "step": 6633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994732074443474e-05,
+      "loss": 1.0051,
+      "step": 6634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994730461911e-05,
+      "loss": 1.1089,
+      "step": 6635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999472884913179e-05,
+      "loss": 1.3316,
+      "step": 6636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994727236105845e-05,
+      "loss": 1.1779,
+      "step": 6637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999472562283316e-05,
+      "loss": 1.2062,
+      "step": 6638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994724009313734e-05,
+      "loss": 1.2757,
+      "step": 6639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999472239554758e-05,
+      "loss": 1.1214,
+      "step": 6640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994720781534673e-05,
+      "loss": 1.0394,
+      "step": 6641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994719167275044e-05,
+      "loss": 1.045,
+      "step": 6642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999471755276867e-05,
+      "loss": 0.9549,
+      "step": 6643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999471593801556e-05,
+      "loss": 0.9414,
+      "step": 6644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999471432301572e-05,
+      "loss": 1.0907,
+      "step": 6645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994712707769134e-05,
+      "loss": 1.0651,
+      "step": 6646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999471109227582e-05,
+      "loss": 0.9648,
+      "step": 6647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999470947653576e-05,
+      "loss": 1.051,
+      "step": 6648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999470786054896e-05,
+      "loss": 0.9602,
+      "step": 6649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994706244315426e-05,
+      "loss": 1.0035,
+      "step": 6650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999470462783516e-05,
+      "loss": 0.6232,
+      "step": 6651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999470301110815e-05,
+      "loss": 0.6789,
+      "step": 6652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994701394134405e-05,
+      "loss": 0.8334,
+      "step": 6653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999469977691392e-05,
+      "loss": 1.5565,
+      "step": 6654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999469815944671e-05,
+      "loss": 1.1051,
+      "step": 6655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994696541732756e-05,
+      "loss": 1.239,
+      "step": 6656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999469492377205e-05,
+      "loss": 1.2687,
+      "step": 6657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994693305564634e-05,
+      "loss": 1.0522,
+      "step": 6658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999469168711046e-05,
+      "loss": 1.0463,
+      "step": 6659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994690068409556e-05,
+      "loss": 1.0889,
+      "step": 6660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999468844946192e-05,
+      "loss": 0.9888,
+      "step": 6661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999468683026754e-05,
+      "loss": 1.3083,
+      "step": 6662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999468521082642e-05,
+      "loss": 0.9751,
+      "step": 6663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999468359113857e-05,
+      "loss": 0.9221,
+      "step": 6664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994681971203985e-05,
+      "loss": 0.7942,
+      "step": 6665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999468035102266e-05,
+      "loss": 0.6858,
+      "step": 6666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99946787305946e-05,
+      "loss": 1.0173,
+      "step": 6667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99946771099198e-05,
+      "loss": 0.4216,
+      "step": 6668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999467548899826e-05,
+      "loss": 0.2308,
+      "step": 6669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999467386782999e-05,
+      "loss": 0.1515,
+      "step": 6670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999467224641498e-05,
+      "loss": 0.0879,
+      "step": 6671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999467062475323e-05,
+      "loss": 0.0501,
+      "step": 6672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999466900284474e-05,
+      "loss": 0.0293,
+      "step": 6673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999466738068952e-05,
+      "loss": 0.0257,
+      "step": 6674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999466575828756e-05,
+      "loss": 0.0204,
+      "step": 6675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999466413563887e-05,
+      "loss": 0.0162,
+      "step": 6676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999466251274344e-05,
+      "loss": 0.0418,
+      "step": 6677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994660889601274e-05,
+      "loss": 0.0339,
+      "step": 6678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994659266212364e-05,
+      "loss": 0.0208,
+      "step": 6679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994657642576724e-05,
+      "loss": 0.0155,
+      "step": 6680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994656018694346e-05,
+      "loss": 0.0171,
+      "step": 6681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999465439456523e-05,
+      "loss": 0.012,
+      "step": 6682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999465277018938e-05,
+      "loss": 0.2595,
+      "step": 6683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999465114556679e-05,
+      "loss": 0.4083,
+      "step": 6684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999464952069747e-05,
+      "loss": 0.3142,
+      "step": 6685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994647895581406e-05,
+      "loss": 0.191,
+      "step": 6686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994646270218604e-05,
+      "loss": 0.1369,
+      "step": 6687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994644644609065e-05,
+      "loss": 0.0933,
+      "step": 6688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994643018752796e-05,
+      "loss": 0.0805,
+      "step": 6689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999464139264979e-05,
+      "loss": 0.0519,
+      "step": 6690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994639766300046e-05,
+      "loss": 0.0392,
+      "step": 6691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994638139703564e-05,
+      "loss": 0.0669,
+      "step": 6692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994636512860346e-05,
+      "loss": 0.0817,
+      "step": 6693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999463488577039e-05,
+      "loss": 0.0558,
+      "step": 6694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99946332584337e-05,
+      "loss": 0.0373,
+      "step": 6695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999463163085027e-05,
+      "loss": 0.0279,
+      "step": 6696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994630003020106e-05,
+      "loss": 0.0529,
+      "step": 6697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999462837494321e-05,
+      "loss": 0.7594,
+      "step": 6698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999462674661957e-05,
+      "loss": 1.082,
+      "step": 6699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99946251180492e-05,
+      "loss": 1.4619,
+      "step": 6700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999462348923208e-05,
+      "loss": 1.1149,
+      "step": 6701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999462186016825e-05,
+      "loss": 1.2684,
+      "step": 6702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994620230857656e-05,
+      "loss": 0.9541,
+      "step": 6703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999461860130034e-05,
+      "loss": 1.1064,
+      "step": 6704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999461697149629e-05,
+      "loss": 1.3397,
+      "step": 6705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999461534144549e-05,
+      "loss": 1.1285,
+      "step": 6706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994613711147964e-05,
+      "loss": 1.1078,
+      "step": 6707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994612080603706e-05,
+      "loss": 1.2667,
+      "step": 6708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994610449812704e-05,
+      "loss": 1.1437,
+      "step": 6709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994608818774965e-05,
+      "loss": 1.0543,
+      "step": 6710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994607187490495e-05,
+      "loss": 1.1706,
+      "step": 6711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999460555595928e-05,
+      "loss": 1.1993,
+      "step": 6712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999460392418134e-05,
+      "loss": 1.1689,
+      "step": 6713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994602292156655e-05,
+      "loss": 0.9531,
+      "step": 6714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994600659885237e-05,
+      "loss": 0.4351,
+      "step": 6715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999459902736709e-05,
+      "loss": 0.3077,
+      "step": 6716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999459739460219e-05,
+      "loss": 1.1547,
+      "step": 6717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999459576159057e-05,
+      "loss": 1.0603,
+      "step": 6718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999459412833221e-05,
+      "loss": 1.2145,
+      "step": 6719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999459249482711e-05,
+      "loss": 1.0326,
+      "step": 6720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999459086107527e-05,
+      "loss": 4.1597,
+      "step": 6721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994589227076706e-05,
+      "loss": 4.1984,
+      "step": 6722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994587592831396e-05,
+      "loss": 3.4404,
+      "step": 6723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994585958339354e-05,
+      "loss": 4.4833,
+      "step": 6724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999458432360058e-05,
+      "loss": 4.0146,
+      "step": 6725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999458268861506e-05,
+      "loss": 1.5474,
+      "step": 6726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994581053382814e-05,
+      "loss": 0.9322,
+      "step": 6727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999457941790382e-05,
+      "loss": 0.4904,
+      "step": 6728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994577782178095e-05,
+      "loss": 0.3933,
+      "step": 6729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994576146205644e-05,
+      "loss": 0.3312,
+      "step": 6730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999457450998645e-05,
+      "loss": 0.2082,
+      "step": 6731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994572873520515e-05,
+      "loss": 1.03,
+      "step": 6732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999457123680785e-05,
+      "loss": 1.4107,
+      "step": 6733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999456959984845e-05,
+      "loss": 1.2318,
+      "step": 6734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999456796264231e-05,
+      "loss": 0.5894,
+      "step": 6735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999456632518944e-05,
+      "loss": 0.5863,
+      "step": 6736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999456468748983e-05,
+      "loss": 1.0186,
+      "step": 6737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999456304954349e-05,
+      "loss": 1.0029,
+      "step": 6738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999456141135041e-05,
+      "loss": 0.848,
+      "step": 6739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999455977291059e-05,
+      "loss": 1.549,
+      "step": 6740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994558134224035e-05,
+      "loss": 1.1585,
+      "step": 6741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999455649529074e-05,
+      "loss": 1.1006,
+      "step": 6742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999455485611072e-05,
+      "loss": 1.3462,
+      "step": 6743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999455321668396e-05,
+      "loss": 1.1244,
+      "step": 6744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999455157701046e-05,
+      "loss": 1.3523,
+      "step": 6745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999454993709023e-05,
+      "loss": 1.0386,
+      "step": 6746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999454829692327e-05,
+      "loss": 1.4276,
+      "step": 6747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994546656509565e-05,
+      "loss": 1.3696,
+      "step": 6748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994545015849125e-05,
+      "loss": 1.2155,
+      "step": 6749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999454337494195e-05,
+      "loss": 1.1756,
+      "step": 6750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999454173378804e-05,
+      "loss": 1.3096,
+      "step": 6751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99945400923874e-05,
+      "loss": 1.0379,
+      "step": 6752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999453845074001e-05,
+      "loss": 1.0734,
+      "step": 6753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994536808845906e-05,
+      "loss": 0.8223,
+      "step": 6754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999453516670505e-05,
+      "loss": 0.8731,
+      "step": 6755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999453352431746e-05,
+      "loss": 0.7837,
+      "step": 6756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994531881683136e-05,
+      "loss": 0.9755,
+      "step": 6757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999453023880209e-05,
+      "loss": 1.4549,
+      "step": 6758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999452859567428e-05,
+      "loss": 1.209,
+      "step": 6759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999452695229976e-05,
+      "loss": 0.7333,
+      "step": 6760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99945253086785e-05,
+      "loss": 1.0987,
+      "step": 6761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999452366481049e-05,
+      "loss": 1.315,
+      "step": 6762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994522020695757e-05,
+      "loss": 1.3147,
+      "step": 6763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994520376334284e-05,
+      "loss": 1.0884,
+      "step": 6764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999451873172608e-05,
+      "loss": 1.1147,
+      "step": 6765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999451708687114e-05,
+      "loss": 1.1792,
+      "step": 6766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999451544176946e-05,
+      "loss": 1.3728,
+      "step": 6767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999451379642105e-05,
+      "loss": 1.1996,
+      "step": 6768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99945121508259e-05,
+      "loss": 0.7847,
+      "step": 6769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999451050498402e-05,
+      "loss": 0.4295,
+      "step": 6770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994508858895406e-05,
+      "loss": 0.6031,
+      "step": 6771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999450721256005e-05,
+      "loss": 0.8572,
+      "step": 6772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999450556597796e-05,
+      "loss": 1.1249,
+      "step": 6773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999450391914914e-05,
+      "loss": 0.8992,
+      "step": 6774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999450227207358e-05,
+      "loss": 1.182,
+      "step": 6775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994500624751285e-05,
+      "loss": 1.1329,
+      "step": 6776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994498977182255e-05,
+      "loss": 1.2008,
+      "step": 6777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994497329366494e-05,
+      "loss": 1.1274,
+      "step": 6778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999449568130399e-05,
+      "loss": 4.991,
+      "step": 6779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999449403299475e-05,
+      "loss": 2.3363,
+      "step": 6780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999449238443879e-05,
+      "loss": 1.1702,
+      "step": 6781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994490735636076e-05,
+      "loss": 1.3478,
+      "step": 6782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994489086586636e-05,
+      "loss": 1.3045,
+      "step": 6783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994487437290464e-05,
+      "loss": 0.9367,
+      "step": 6784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994485787747556e-05,
+      "loss": 1.5058,
+      "step": 6785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999448413795791e-05,
+      "loss": 1.0185,
+      "step": 6786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994482487921526e-05,
+      "loss": 1.3135,
+      "step": 6787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999448083763841e-05,
+      "loss": 1.1197,
+      "step": 6788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999447918710857e-05,
+      "loss": 0.989,
+      "step": 6789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999447753633198e-05,
+      "loss": 1.0585,
+      "step": 6790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994475885308654e-05,
+      "loss": 1.1427,
+      "step": 6791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994474234038604e-05,
+      "loss": 1.2104,
+      "step": 6792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999447258252181e-05,
+      "loss": 1.1049,
+      "step": 6793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999447093075829e-05,
+      "loss": 1.2225,
+      "step": 6794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999446927874803e-05,
+      "loss": 1.271,
+      "step": 6795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999446762649104e-05,
+      "loss": 1.402,
+      "step": 6796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994465973987304e-05,
+      "loss": 1.1503,
+      "step": 6797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999446432123684e-05,
+      "loss": 1.2106,
+      "step": 6798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999446266823964e-05,
+      "loss": 1.0387,
+      "step": 6799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994461014995706e-05,
+      "loss": 1.1426,
+      "step": 6800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999445936150504e-05,
+      "loss": 1.159,
+      "step": 6801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999445770776764e-05,
+      "loss": 1.0738,
+      "step": 6802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99944560537835e-05,
+      "loss": 1.0195,
+      "step": 6803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999445439955263e-05,
+      "loss": 1.1082,
+      "step": 6804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994452745075015e-05,
+      "loss": 1.1216,
+      "step": 6805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999445109035068e-05,
+      "loss": 0.9553,
+      "step": 6806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994449435379595e-05,
+      "loss": 0.9602,
+      "step": 6807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999444778016178e-05,
+      "loss": 1.156,
+      "step": 6808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999444612469724e-05,
+      "loss": 0.9277,
+      "step": 6809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999444446898596e-05,
+      "loss": 0.9691,
+      "step": 6810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999444281302794e-05,
+      "loss": 1.2191,
+      "step": 6811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999444115682319e-05,
+      "loss": 1.2958,
+      "step": 6812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999443950037171e-05,
+      "loss": 1.5685,
+      "step": 6813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994437843673484e-05,
+      "loss": 1.3699,
+      "step": 6814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999443618672853e-05,
+      "loss": 1.0349,
+      "step": 6815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994434529536847e-05,
+      "loss": 1.0596,
+      "step": 6816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999443287209842e-05,
+      "loss": 0.853,
+      "step": 6817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999443121441326e-05,
+      "loss": 0.9858,
+      "step": 6818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999442955648137e-05,
+      "loss": 1.1507,
+      "step": 6819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994427898302744e-05,
+      "loss": 1.123,
+      "step": 6820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999442623987739e-05,
+      "loss": 0.9609,
+      "step": 6821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999442458120529e-05,
+      "loss": 1.2603,
+      "step": 6822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994422922286455e-05,
+      "loss": 0.9679,
+      "step": 6823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999442126312089e-05,
+      "loss": 1.1821,
+      "step": 6824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99944196037086e-05,
+      "loss": 1.0709,
+      "step": 6825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994417944049563e-05,
+      "loss": 1.222,
+      "step": 6826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994416284143796e-05,
+      "loss": 1.2295,
+      "step": 6827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999441462399129e-05,
+      "loss": 0.9857,
+      "step": 6828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994412963592057e-05,
+      "loss": 0.7959,
+      "step": 6829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994411302946084e-05,
+      "loss": 0.9756,
+      "step": 6830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999440964205338e-05,
+      "loss": 1.1994,
+      "step": 6831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999440798091394e-05,
+      "loss": 1.4401,
+      "step": 6832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999440631952777e-05,
+      "loss": 0.9325,
+      "step": 6833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999440465789486e-05,
+      "loss": 0.7858,
+      "step": 6834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994402996015223e-05,
+      "loss": 0.8887,
+      "step": 6835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999440133388884e-05,
+      "loss": 1.0236,
+      "step": 6836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994399671515734e-05,
+      "loss": 1.1955,
+      "step": 6837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999439800889589e-05,
+      "loss": 0.9023,
+      "step": 6838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999439634602931e-05,
+      "loss": 1.3026,
+      "step": 6839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994394682916005e-05,
+      "loss": 1.0394,
+      "step": 6840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999439301955595e-05,
+      "loss": 0.9914,
+      "step": 6841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999439135594917e-05,
+      "loss": 1.3485,
+      "step": 6842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999438969209565e-05,
+      "loss": 1.207,
+      "step": 6843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999438802799541e-05,
+      "loss": 0.8097,
+      "step": 6844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999438636364843e-05,
+      "loss": 1.2124,
+      "step": 6845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994384699054706e-05,
+      "loss": 1.2324,
+      "step": 6846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999438303421425e-05,
+      "loss": 1.0811,
+      "step": 6847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999438136912707e-05,
+      "loss": 0.9873,
+      "step": 6848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999437970379315e-05,
+      "loss": 1.0589,
+      "step": 6849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99943780382125e-05,
+      "loss": 1.2425,
+      "step": 6850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999437637238511e-05,
+      "loss": 0.6193,
+      "step": 6851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994374706310985e-05,
+      "loss": 0.4063,
+      "step": 6852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994373039990136e-05,
+      "loss": 0.3648,
+      "step": 6853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999437137342254e-05,
+      "loss": 0.3846,
+      "step": 6854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994369706608226e-05,
+      "loss": 0.3362,
+      "step": 6855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994368039547165e-05,
+      "loss": 0.3245,
+      "step": 6856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999436637223937e-05,
+      "loss": 0.3606,
+      "step": 6857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999436470468485e-05,
+      "loss": 0.3314,
+      "step": 6858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999436303688359e-05,
+      "loss": 0.4419,
+      "step": 6859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994361368835595e-05,
+      "loss": 0.9462,
+      "step": 6860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994359700540874e-05,
+      "loss": 1.5972,
+      "step": 6861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999435803199941e-05,
+      "loss": 1.2746,
+      "step": 6862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999435636321122e-05,
+      "loss": 1.2198,
+      "step": 6863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994354694176295e-05,
+      "loss": 0.869,
+      "step": 6864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999435302489463e-05,
+      "loss": 0.4223,
+      "step": 6865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999435135536624e-05,
+      "loss": 0.3452,
+      "step": 6866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999434968559111e-05,
+      "loss": 0.6081,
+      "step": 6867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994348015569246e-05,
+      "loss": 1.0303,
+      "step": 6868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999434634530065e-05,
+      "loss": 1.0843,
+      "step": 6869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999434467478532e-05,
+      "loss": 0.9482,
+      "step": 6870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999434300402326e-05,
+      "loss": 0.6892,
+      "step": 6871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999434133301446e-05,
+      "loss": 0.5223,
+      "step": 6872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999433966175893e-05,
+      "loss": 1.151,
+      "step": 6873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999433799025667e-05,
+      "loss": 1.3094,
+      "step": 6874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994336318507667e-05,
+      "loss": 1.0836,
+      "step": 6875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994334646511934e-05,
+      "loss": 1.6371,
+      "step": 6876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999433297426948e-05,
+      "loss": 1.6683,
+      "step": 6877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999433130178027e-05,
+      "loss": 1.7682,
+      "step": 6878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994329629044345e-05,
+      "loss": 1.4243,
+      "step": 6879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994327956061684e-05,
+      "loss": 1.7879,
+      "step": 6880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994326282832285e-05,
+      "loss": 1.7397,
+      "step": 6881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999432460935615e-05,
+      "loss": 1.1371,
+      "step": 6882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999432293563328e-05,
+      "loss": 1.1943,
+      "step": 6883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999432126166368e-05,
+      "loss": 1.2871,
+      "step": 6884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999431958744735e-05,
+      "loss": 1.2688,
+      "step": 6885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994317912984284e-05,
+      "loss": 1.5491,
+      "step": 6886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999431623827449e-05,
+      "loss": 1.1822,
+      "step": 6887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994314563317955e-05,
+      "loss": 1.2313,
+      "step": 6888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994312888114685e-05,
+      "loss": 1.0805,
+      "step": 6889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999431121266469e-05,
+      "loss": 1.0169,
+      "step": 6890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999430953696795e-05,
+      "loss": 1.4716,
+      "step": 6891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999430786102449e-05,
+      "loss": 1.3546,
+      "step": 6892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999430618483429e-05,
+      "loss": 1.2015,
+      "step": 6893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999430450839736e-05,
+      "loss": 0.8801,
+      "step": 6894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999430283171369e-05,
+      "loss": 1.2549,
+      "step": 6895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99943011547833e-05,
+      "loss": 1.0718,
+      "step": 6896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999429947760617e-05,
+      "loss": 1.2091,
+      "step": 6897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994297800182296e-05,
+      "loss": 1.1126,
+      "step": 6898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994296122511694e-05,
+      "loss": 0.8508,
+      "step": 6899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999429444459437e-05,
+      "loss": 1.2032,
+      "step": 6900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99942927664303e-05,
+      "loss": 1.1239,
+      "step": 6901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999429108801951e-05,
+      "loss": 1.1106,
+      "step": 6902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999428940936197e-05,
+      "loss": 1.2155,
+      "step": 6903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994287730457715e-05,
+      "loss": 1.0073,
+      "step": 6904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994286051306715e-05,
+      "loss": 0.7326,
+      "step": 6905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994284371908985e-05,
+      "loss": 1.1721,
+      "step": 6906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994282692264525e-05,
+      "loss": 1.2096,
+      "step": 6907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999428101237332e-05,
+      "loss": 0.9284,
+      "step": 6908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999427933223539e-05,
+      "loss": 1.0399,
+      "step": 6909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999427765185073e-05,
+      "loss": 1.6415,
+      "step": 6910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999427597121934e-05,
+      "loss": 1.1169,
+      "step": 6911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999427429034121e-05,
+      "loss": 1.2315,
+      "step": 6912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999427260921635e-05,
+      "loss": 1.3721,
+      "step": 6913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999427092784476e-05,
+      "loss": 1.1605,
+      "step": 6914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994269246226425e-05,
+      "loss": 1.0359,
+      "step": 6915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999426756436137e-05,
+      "loss": 1.2378,
+      "step": 6916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994265882249576e-05,
+      "loss": 1.1071,
+      "step": 6917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999426419989105e-05,
+      "loss": 0.9853,
+      "step": 6918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999426251728579e-05,
+      "loss": 0.9752,
+      "step": 6919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99942608344338e-05,
+      "loss": 1.0607,
+      "step": 6920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999425915133508e-05,
+      "loss": 1.1958,
+      "step": 6921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999425746798962e-05,
+      "loss": 1.3917,
+      "step": 6922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999425578439743e-05,
+      "loss": 0.859,
+      "step": 6923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999425410055851e-05,
+      "loss": 0.5413,
+      "step": 6924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999425241647285e-05,
+      "loss": 0.7141,
+      "step": 6925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999425073214047e-05,
+      "loss": 0.6767,
+      "step": 6926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999424904756135e-05,
+      "loss": 1.3069,
+      "step": 6927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999424736273549e-05,
+      "loss": 1.1934,
+      "step": 6928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99942456776629e-05,
+      "loss": 1.2299,
+      "step": 6929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994243992343585e-05,
+      "loss": 0.6811,
+      "step": 6930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999424230677754e-05,
+      "loss": 0.117,
+      "step": 6931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999424062096475e-05,
+      "loss": 0.1434,
+      "step": 6932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999423893490524e-05,
+      "loss": 0.0817,
+      "step": 6933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994237248598984e-05,
+      "loss": 0.0638,
+      "step": 6934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999423556204601e-05,
+      "loss": 0.234,
+      "step": 6935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999423387524629e-05,
+      "loss": 0.0815,
+      "step": 6936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999423218819984e-05,
+      "loss": 0.3098,
+      "step": 6937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999423050090667e-05,
+      "loss": 0.9545,
+      "step": 6938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999422881336675e-05,
+      "loss": 1.0009,
+      "step": 6939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994227125580106e-05,
+      "loss": 0.3902,
+      "step": 6940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999422543754673e-05,
+      "loss": 0.1013,
+      "step": 6941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999422374926662e-05,
+      "loss": 3.111,
+      "step": 6942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999422206073978e-05,
+      "loss": 5.4106,
+      "step": 6943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99942203719662e-05,
+      "loss": 4.7675,
+      "step": 6944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999421868294591e-05,
+      "loss": 4.1092,
+      "step": 6945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999421699367887e-05,
+      "loss": 0.9569,
+      "step": 6946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999421530416509e-05,
+      "loss": 1.146,
+      "step": 6947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999421361440458e-05,
+      "loss": 0.9052,
+      "step": 6948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999421192439735e-05,
+      "loss": 0.9786,
+      "step": 6949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999421023414338e-05,
+      "loss": 1.1034,
+      "step": 6950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994208543642685e-05,
+      "loss": 1.2519,
+      "step": 6951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994206852895255e-05,
+      "loss": 1.1061,
+      "step": 6952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999420516190109e-05,
+      "loss": 0.933,
+      "step": 6953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994203470660184e-05,
+      "loss": 0.9922,
+      "step": 6954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994201779172556e-05,
+      "loss": 0.9366,
+      "step": 6955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999420008743819e-05,
+      "loss": 1.1203,
+      "step": 6956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99941983954571e-05,
+      "loss": 1.1666,
+      "step": 6957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994196703229276e-05,
+      "loss": 1.0677,
+      "step": 6958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999419501075472e-05,
+      "loss": 1.1357,
+      "step": 6959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994193318033425e-05,
+      "loss": 1.1193,
+      "step": 6960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99941916250654e-05,
+      "loss": 1.3423,
+      "step": 6961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999418993185065e-05,
+      "loss": 1.2765,
+      "step": 6962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999418823838916e-05,
+      "loss": 0.8398,
+      "step": 6963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994186544680943e-05,
+      "loss": 0.3674,
+      "step": 6964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999418485072599e-05,
+      "loss": 0.4106,
+      "step": 6965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999418315652431e-05,
+      "loss": 0.5539,
+      "step": 6966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99941814620759e-05,
+      "loss": 1.1158,
+      "step": 6967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994179767380746e-05,
+      "loss": 1.0879,
+      "step": 6968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999417807243887e-05,
+      "loss": 0.9524,
+      "step": 6969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999417637725026e-05,
+      "loss": 0.9135,
+      "step": 6970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999417468181491e-05,
+      "loss": 0.9985,
+      "step": 6971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999417298613284e-05,
+      "loss": 1.372,
+      "step": 6972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999417129020403e-05,
+      "loss": 1.4207,
+      "step": 6973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999416959402849e-05,
+      "loss": 1.0777,
+      "step": 6974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999416789760622e-05,
+      "loss": 1.1585,
+      "step": 6975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999416620093722e-05,
+      "loss": 1.0714,
+      "step": 6976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999416450402148e-05,
+      "loss": 1.1122,
+      "step": 6977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994162806859014e-05,
+      "loss": 1.3217,
+      "step": 6978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999416110944982e-05,
+      "loss": 1.0181,
+      "step": 6979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999415941179388e-05,
+      "loss": 1.3369,
+      "step": 6980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999415771389122e-05,
+      "loss": 1.1256,
+      "step": 6981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999415601574183e-05,
+      "loss": 1.1828,
+      "step": 6982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99941543173457e-05,
+      "loss": 1.3077,
+      "step": 6983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999415261870285e-05,
+      "loss": 0.9496,
+      "step": 6984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999415091981326e-05,
+      "loss": 1.2384,
+      "step": 6985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994149220676935e-05,
+      "loss": 0.9296,
+      "step": 6986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999414752129389e-05,
+      "loss": 1.3176,
+      "step": 6987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99941458216641e-05,
+      "loss": 0.3747,
+      "step": 6988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994144121787586e-05,
+      "loss": 0.1655,
+      "step": 6989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994142421664336e-05,
+      "loss": 0.2205,
+      "step": 6990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999414072129436e-05,
+      "loss": 0.2001,
+      "step": 6991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999413902067764e-05,
+      "loss": 0.2171,
+      "step": 6992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999413731981421e-05,
+      "loss": 0.8767,
+      "step": 6993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999413561870403e-05,
+      "loss": 1.1458,
+      "step": 6994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994133917347124e-05,
+      "loss": 1.0598,
+      "step": 6995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999413221574348e-05,
+      "loss": 0.576,
+      "step": 6996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999413051389312e-05,
+      "loss": 0.6216,
+      "step": 6997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999412881179602e-05,
+      "loss": 0.5405,
+      "step": 6998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994127109452185e-05,
+      "loss": 0.9278,
+      "step": 6999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999412540686162e-05,
+      "loss": 0.918,
+      "step": 7000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0878294706344604,
+      "eval_runtime": 84.2959,
+      "eval_samples_per_second": 16.43,
+      "eval_steps_per_second": 4.116,
+      "step": 7000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999412370402433e-05,
+      "loss": 1.0767,
+      "step": 7001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994122000940305e-05,
+      "loss": 1.2599,
+      "step": 7002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994120297609543e-05,
+      "loss": 1.2067,
+      "step": 7003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999411859403206e-05,
+      "loss": 1.2545,
+      "step": 7004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999411689020783e-05,
+      "loss": 1.119,
+      "step": 7005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999411518613688e-05,
+      "loss": 1.1328,
+      "step": 7006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99941134818192e-05,
+      "loss": 1.1584,
+      "step": 7007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999411177725478e-05,
+      "loss": 1.1062,
+      "step": 7008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994110072443635e-05,
+      "loss": 1.0393,
+      "step": 7009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999410836738576e-05,
+      "loss": 0.9956,
+      "step": 7010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994106662081155e-05,
+      "loss": 1.1625,
+      "step": 7011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994104956529805e-05,
+      "loss": 0.7091,
+      "step": 7012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999410325073174e-05,
+      "loss": 1.0893,
+      "step": 7013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999410154468694e-05,
+      "loss": 1.008,
+      "step": 7014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99940998383954e-05,
+      "loss": 1.069,
+      "step": 7015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994098131857135e-05,
+      "loss": 1.0903,
+      "step": 7016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994096425072147e-05,
+      "loss": 1.1329,
+      "step": 7017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999409471804042e-05,
+      "loss": 1.099,
+      "step": 7018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999409301076196e-05,
+      "loss": 0.8939,
+      "step": 7019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999409130323677e-05,
+      "loss": 1.1571,
+      "step": 7020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994089595464846e-05,
+      "loss": 1.0265,
+      "step": 7021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99940878874462e-05,
+      "loss": 1.7861,
+      "step": 7022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999408617918081e-05,
+      "loss": 1.5894,
+      "step": 7023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99940844706687e-05,
+      "loss": 1.3719,
+      "step": 7024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999408276190986e-05,
+      "loss": 1.1614,
+      "step": 7025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999408105290427e-05,
+      "loss": 1.1558,
+      "step": 7026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994079343651965e-05,
+      "loss": 1.1941,
+      "step": 7027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994077634152934e-05,
+      "loss": 1.3192,
+      "step": 7028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994075924407165e-05,
+      "loss": 1.0487,
+      "step": 7029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994074214414666e-05,
+      "loss": 1.1189,
+      "step": 7030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999407250417543e-05,
+      "loss": 1.2138,
+      "step": 7031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999407079368947e-05,
+      "loss": 0.9906,
+      "step": 7032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999406908295677e-05,
+      "loss": 1.2017,
+      "step": 7033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999406737197735e-05,
+      "loss": 1.2657,
+      "step": 7034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99940656607512e-05,
+      "loss": 1.2592,
+      "step": 7035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994063949278316e-05,
+      "loss": 1.2166,
+      "step": 7036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999406223755869e-05,
+      "loss": 1.1549,
+      "step": 7037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994060525592347e-05,
+      "loss": 0.8705,
+      "step": 7038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994058813379266e-05,
+      "loss": 1.2059,
+      "step": 7039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999405710091946e-05,
+      "loss": 1.0886,
+      "step": 7040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994055388212926e-05,
+      "loss": 0.1385,
+      "step": 7041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999405367525965e-05,
+      "loss": 0.1784,
+      "step": 7042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999405196205965e-05,
+      "loss": 1.1128,
+      "step": 7043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999405024861292e-05,
+      "loss": 1.0444,
+      "step": 7044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994048534919454e-05,
+      "loss": 0.9553,
+      "step": 7045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999404682097926e-05,
+      "loss": 1.2984,
+      "step": 7046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994045106792334e-05,
+      "loss": 1.1051,
+      "step": 7047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999404339235868e-05,
+      "loss": 1.0497,
+      "step": 7048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999404167767829e-05,
+      "loss": 0.6193,
+      "step": 7049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994039962751176e-05,
+      "loss": 1.0276,
+      "step": 7050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999403824757733e-05,
+      "loss": 1.0596,
+      "step": 7051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999403653215675e-05,
+      "loss": 0.7923,
+      "step": 7052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999403481648944e-05,
+      "loss": 1.1786,
+      "step": 7053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994033100575404e-05,
+      "loss": 1.2927,
+      "step": 7054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994031384414635e-05,
+      "loss": 1.0812,
+      "step": 7055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994029668007135e-05,
+      "loss": 1.1084,
+      "step": 7056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994027951352905e-05,
+      "loss": 1.2078,
+      "step": 7057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994026234451944e-05,
+      "loss": 0.9318,
+      "step": 7058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999402451730425e-05,
+      "loss": 1.1528,
+      "step": 7059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999402279990984e-05,
+      "loss": 1.398,
+      "step": 7060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994021082268685e-05,
+      "loss": 1.0382,
+      "step": 7061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994019364380795e-05,
+      "loss": 0.694,
+      "step": 7062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999401764624618e-05,
+      "loss": 0.7811,
+      "step": 7063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999401592786484e-05,
+      "loss": 1.1132,
+      "step": 7064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999401420923676e-05,
+      "loss": 0.9212,
+      "step": 7065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994012490361965e-05,
+      "loss": 0.4159,
+      "step": 7066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999401077124043e-05,
+      "loss": 0.5159,
+      "step": 7067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994009051872156e-05,
+      "loss": 1.156,
+      "step": 7068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994007332257166e-05,
+      "loss": 0.8949,
+      "step": 7069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999400561239544e-05,
+      "loss": 0.7439,
+      "step": 7070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999400389228698e-05,
+      "loss": 1.122,
+      "step": 7071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99940021719318e-05,
+      "loss": 1.1978,
+      "step": 7072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999400045132988e-05,
+      "loss": 1.0629,
+      "step": 7073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999399873048123e-05,
+      "loss": 0.4868,
+      "step": 7074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999399700938586e-05,
+      "loss": 1.0516,
+      "step": 7075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999399528804375e-05,
+      "loss": 1.1544,
+      "step": 7076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993993566454913e-05,
+      "loss": 1.2188,
+      "step": 7077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999399184461935e-05,
+      "loss": 0.8363,
+      "step": 7078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999399012253705e-05,
+      "loss": 0.7173,
+      "step": 7079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999398840020802e-05,
+      "loss": 0.535,
+      "step": 7080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999398667763226e-05,
+      "loss": 0.6233,
+      "step": 7081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999398495480977e-05,
+      "loss": 0.4848,
+      "step": 7082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999398323174055e-05,
+      "loss": 1.0284,
+      "step": 7083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993981508424605e-05,
+      "loss": 1.0248,
+      "step": 7084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999397978486193e-05,
+      "loss": 1.3033,
+      "step": 7085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999397806105252e-05,
+      "loss": 1.3483,
+      "step": 7086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993976336996385e-05,
+      "loss": 1.154,
+      "step": 7087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999397461269351e-05,
+      "loss": 1.2915,
+      "step": 7088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993972888143914e-05,
+      "loss": 1.0423,
+      "step": 7089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999397116334758e-05,
+      "loss": 1.1034,
+      "step": 7090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999396943830453e-05,
+      "loss": 0.9868,
+      "step": 7091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993967713014736e-05,
+      "loss": 0.2809,
+      "step": 7092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999396598747822e-05,
+      "loss": 0.3869,
+      "step": 7093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999396426169497e-05,
+      "loss": 1.2138,
+      "step": 7094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993962535665e-05,
+      "loss": 1.1352,
+      "step": 7095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999396080938828e-05,
+      "loss": 0.7317,
+      "step": 7096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999395908286485e-05,
+      "loss": 1.0277,
+      "step": 7097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993957356094676e-05,
+      "loss": 1.166,
+      "step": 7098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999395562907778e-05,
+      "loss": 1.1118,
+      "step": 7099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999395390181416e-05,
+      "loss": 1.0271,
+      "step": 7100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999395217430381e-05,
+      "loss": 1.1238,
+      "step": 7101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999395044654671e-05,
+      "loss": 0.994,
+      "step": 7102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99939487185429e-05,
+      "loss": 1.1137,
+      "step": 7103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999394699029235e-05,
+      "loss": 1.2896,
+      "step": 7104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993945261795074e-05,
+      "loss": 1.1667,
+      "step": 7105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999394353305107e-05,
+      "loss": 1.2486,
+      "step": 7106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999394180406034e-05,
+      "loss": 1.0502,
+      "step": 7107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999394007482287e-05,
+      "loss": 1.0564,
+      "step": 7108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999393834533867e-05,
+      "loss": 2.3266,
+      "step": 7109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999393661560775e-05,
+      "loss": 1.1655,
+      "step": 7110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99939348856301e-05,
+      "loss": 1.1056,
+      "step": 7111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999393315540572e-05,
+      "loss": 0.9551,
+      "step": 7112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993931424934604e-05,
+      "loss": 1.2253,
+      "step": 7113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993929694216754e-05,
+      "loss": 1.2538,
+      "step": 7114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993927963252186e-05,
+      "loss": 1.1117,
+      "step": 7115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999392623204089e-05,
+      "loss": 1.1508,
+      "step": 7116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999392450058285e-05,
+      "loss": 1.3282,
+      "step": 7117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993922768878094e-05,
+      "loss": 1.2517,
+      "step": 7118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993921036926605e-05,
+      "loss": 0.919,
+      "step": 7119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993919304728385e-05,
+      "loss": 1.1984,
+      "step": 7120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993917572283434e-05,
+      "loss": 1.0078,
+      "step": 7121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999391583959175e-05,
+      "loss": 1.0748,
+      "step": 7122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999391410665335e-05,
+      "loss": 1.374,
+      "step": 7123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999391237346821e-05,
+      "loss": 1.2001,
+      "step": 7124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999391064003635e-05,
+      "loss": 1.3737,
+      "step": 7125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999390890635775e-05,
+      "loss": 1.2602,
+      "step": 7126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999390717243242e-05,
+      "loss": 1.1102,
+      "step": 7127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993905438260366e-05,
+      "loss": 1.1347,
+      "step": 7128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993903703841584e-05,
+      "loss": 0.888,
+      "step": 7129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999390196917607e-05,
+      "loss": 0.718,
+      "step": 7130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999390023426383e-05,
+      "loss": 0.4841,
+      "step": 7131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999389849910486e-05,
+      "loss": 0.3925,
+      "step": 7132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993896763699154e-05,
+      "loss": 0.3584,
+      "step": 7133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999389502804673e-05,
+      "loss": 0.2742,
+      "step": 7134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999389329214756e-05,
+      "loss": 0.5736,
+      "step": 7135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999389155600168e-05,
+      "loss": 1.1826,
+      "step": 7136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999388981960906e-05,
+      "loss": 1.0934,
+      "step": 7137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999388808296971e-05,
+      "loss": 1.0974,
+      "step": 7138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999388634608364e-05,
+      "loss": 1.0185,
+      "step": 7139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993884608950836e-05,
+      "loss": 1.0648,
+      "step": 7140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99938828715713e-05,
+      "loss": 1.1015,
+      "step": 7141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999388113394504e-05,
+      "loss": 1.2112,
+      "step": 7142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999387939607205e-05,
+      "loss": 1.2462,
+      "step": 7143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999387765795232e-05,
+      "loss": 1.2127,
+      "step": 7144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993875919585874e-05,
+      "loss": 1.1119,
+      "step": 7145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993874180972694e-05,
+      "loss": 1.1006,
+      "step": 7146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999387244211279e-05,
+      "loss": 1.3689,
+      "step": 7147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999387070300615e-05,
+      "loss": 1.1952,
+      "step": 7148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999386896365278e-05,
+      "loss": 1.1658,
+      "step": 7149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999386722405269e-05,
+      "loss": 1.2052,
+      "step": 7150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999386548420586e-05,
+      "loss": 1.1605,
+      "step": 7151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999386374411231e-05,
+      "loss": 0.5754,
+      "step": 7152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999386200377203e-05,
+      "loss": 1.0458,
+      "step": 7153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993860263185015e-05,
+      "loss": 1.1622,
+      "step": 7154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999385852235128e-05,
+      "loss": 1.0383,
+      "step": 7155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999385678127081e-05,
+      "loss": 0.9998,
+      "step": 7156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993855039943613e-05,
+      "loss": 0.9812,
+      "step": 7157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999385329836969e-05,
+      "loss": 1.1803,
+      "step": 7158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999385155654903e-05,
+      "loss": 1.0188,
+      "step": 7159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993849814481644e-05,
+      "loss": 0.9353,
+      "step": 7160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999384807216754e-05,
+      "loss": 0.4585,
+      "step": 7161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999384632960669e-05,
+      "loss": 0.2882,
+      "step": 7162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999384458679912e-05,
+      "loss": 0.2708,
+      "step": 7163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993842843744823e-05,
+      "loss": 0.2178,
+      "step": 7164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99938411004438e-05,
+      "loss": 0.3011,
+      "step": 7165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999383935689605e-05,
+      "loss": 0.2672,
+      "step": 7166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999383761310156e-05,
+      "loss": 0.1637,
+      "step": 7167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999383586906035e-05,
+      "loss": 0.156,
+      "step": 7168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99938341247724e-05,
+      "loss": 0.1112,
+      "step": 7169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993832380237735e-05,
+      "loss": 0.0966,
+      "step": 7170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999383063545633e-05,
+      "loss": 0.071,
+      "step": 7171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993828890428206e-05,
+      "loss": 0.6251,
+      "step": 7172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999382714515335e-05,
+      "loss": 0.5264,
+      "step": 7173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999382539963177e-05,
+      "loss": 0.9327,
+      "step": 7174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999382365386345e-05,
+      "loss": 1.2255,
+      "step": 7175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999382190784841e-05,
+      "loss": 1.1865,
+      "step": 7176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993820161586636e-05,
+      "loss": 1.1373,
+      "step": 7177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999381841507814e-05,
+      "loss": 0.9976,
+      "step": 7178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999381666832291e-05,
+      "loss": 1.1255,
+      "step": 7179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993814921320956e-05,
+      "loss": 1.2367,
+      "step": 7180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993813174072275e-05,
+      "loss": 1.0031,
+      "step": 7181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993811426576864e-05,
+      "loss": 1.0087,
+      "step": 7182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999380967883472e-05,
+      "loss": 1.1194,
+      "step": 7183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999380793084585e-05,
+      "loss": 1.1359,
+      "step": 7184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999380618261025e-05,
+      "loss": 1.5818,
+      "step": 7185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993804434127926e-05,
+      "loss": 1.4419,
+      "step": 7186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999380268539887e-05,
+      "loss": 1.1241,
+      "step": 7187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999380093642309e-05,
+      "loss": 1.1122,
+      "step": 7188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999379918720058e-05,
+      "loss": 1.3919,
+      "step": 7189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999379743773134e-05,
+      "loss": 1.2955,
+      "step": 7190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993795688015375e-05,
+      "loss": 1.2744,
+      "step": 7191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999379393805268e-05,
+      "loss": 1.4393,
+      "step": 7192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999379218784326e-05,
+      "loss": 1.3003,
+      "step": 7193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999379043738711e-05,
+      "loss": 1.6819,
+      "step": 7194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993788686684226e-05,
+      "loss": 1.8338,
+      "step": 7195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999378693573462e-05,
+      "loss": 1.4827,
+      "step": 7196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999378518453828e-05,
+      "loss": 0.9903,
+      "step": 7197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999378343309522e-05,
+      "loss": 1.3554,
+      "step": 7198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999378168140543e-05,
+      "loss": 1.2596,
+      "step": 7199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99937799294689e-05,
+      "loss": 1.0637,
+      "step": 7200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999377817728566e-05,
+      "loss": 1.4157,
+      "step": 7201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999377642485568e-05,
+      "loss": 1.2872,
+      "step": 7202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999377467217897e-05,
+      "loss": 5.7376,
+      "step": 7203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999377291925554e-05,
+      "loss": 5.4455,
+      "step": 7204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993771166085376e-05,
+      "loss": 1.0564,
+      "step": 7205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999376941266849e-05,
+      "loss": 0.7091,
+      "step": 7206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999376765900487e-05,
+      "loss": 1.0153,
+      "step": 7207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993765905094535e-05,
+      "loss": 1.5328,
+      "step": 7208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999376415093746e-05,
+      "loss": 0.9528,
+      "step": 7209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999376239653366e-05,
+      "loss": 1.1227,
+      "step": 7210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999376064188313e-05,
+      "loss": 0.9175,
+      "step": 7211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999375888698587e-05,
+      "loss": 1.1229,
+      "step": 7212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999375713184189e-05,
+      "loss": 1.1384,
+      "step": 7213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999375537645118e-05,
+      "loss": 1.18,
+      "step": 7214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993753620813744e-05,
+      "loss": 1.2178,
+      "step": 7215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999375186492957e-05,
+      "loss": 1.4092,
+      "step": 7216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999375010879868e-05,
+      "loss": 0.9861,
+      "step": 7217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993748352421055e-05,
+      "loss": 1.0413,
+      "step": 7218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99937465957967e-05,
+      "loss": 1.1335,
+      "step": 7219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993744838925625e-05,
+      "loss": 0.6983,
+      "step": 7220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999374308180782e-05,
+      "loss": 0.8774,
+      "step": 7221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993741324443286e-05,
+      "loss": 1.1254,
+      "step": 7222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999373956683202e-05,
+      "loss": 0.427,
+      "step": 7223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999373780897404e-05,
+      "loss": 0.367,
+      "step": 7224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999373605086932e-05,
+      "loss": 0.356,
+      "step": 7225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999373429251788e-05,
+      "loss": 0.3391,
+      "step": 7226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99937325339197e-05,
+      "loss": 0.377,
+      "step": 7227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99937307750748e-05,
+      "loss": 0.323,
+      "step": 7228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993729015983174e-05,
+      "loss": 0.293,
+      "step": 7229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999372725664482e-05,
+      "loss": 0.2573,
+      "step": 7230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999372549705974e-05,
+      "loss": 0.9169,
+      "step": 7231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993723737227924e-05,
+      "loss": 0.7361,
+      "step": 7232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999372197714939e-05,
+      "loss": 0.9766,
+      "step": 7233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999372021682413e-05,
+      "loss": 1.1306,
+      "step": 7234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999371845625213e-05,
+      "loss": 1.3027,
+      "step": 7235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993716695433415e-05,
+      "loss": 0.7542,
+      "step": 7236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999371493436796e-05,
+      "loss": 1.1492,
+      "step": 7237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999371317305579e-05,
+      "loss": 1.1588,
+      "step": 7238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999371141149689e-05,
+      "loss": 0.913,
+      "step": 7239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999370964969126e-05,
+      "loss": 1.1554,
+      "step": 7240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999370788763891e-05,
+      "loss": 1.176,
+      "step": 7241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999370612533982e-05,
+      "loss": 1.0936,
+      "step": 7242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999370436279401e-05,
+      "loss": 1.0287,
+      "step": 7243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999370260000147e-05,
+      "loss": 1.0089,
+      "step": 7244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99937008369622e-05,
+      "loss": 1.0935,
+      "step": 7245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993699073676206e-05,
+      "loss": 1.1778,
+      "step": 7246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993697310143486e-05,
+      "loss": 0.9498,
+      "step": 7247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999369554636404e-05,
+      "loss": 1.0388,
+      "step": 7248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999369378233787e-05,
+      "loss": 1.2277,
+      "step": 7249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999369201806496e-05,
+      "loss": 1.0388,
+      "step": 7250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993690253545335e-05,
+      "loss": 1.2701,
+      "step": 7251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993688488778976e-05,
+      "loss": 1.0766,
+      "step": 7252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99936867237659e-05,
+      "loss": 0.5179,
+      "step": 7253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999368495850608e-05,
+      "loss": 0.7307,
+      "step": 7254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999368319299954e-05,
+      "loss": 1.1805,
+      "step": 7255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999368142724628e-05,
+      "loss": 0.9093,
+      "step": 7256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993679661246284e-05,
+      "loss": 1.1146,
+      "step": 7257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999367789499957e-05,
+      "loss": 1.3203,
+      "step": 7258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993676128506116e-05,
+      "loss": 1.4558,
+      "step": 7259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993674361765947e-05,
+      "loss": 0.7744,
+      "step": 7260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999367259477905e-05,
+      "loss": 0.8,
+      "step": 7261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993670827545416e-05,
+      "loss": 0.8254,
+      "step": 7262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999366906006506e-05,
+      "loss": 0.9704,
+      "step": 7263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993667292337984e-05,
+      "loss": 1.1857,
+      "step": 7264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993665524364175e-05,
+      "loss": 1.0887,
+      "step": 7265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993663756143636e-05,
+      "loss": 1.0521,
+      "step": 7266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993661987676373e-05,
+      "loss": 1.1798,
+      "step": 7267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993660218962394e-05,
+      "loss": 0.878,
+      "step": 7268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999365845000167e-05,
+      "loss": 1.4214,
+      "step": 7269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999365668079423e-05,
+      "loss": 1.3913,
+      "step": 7270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999365491134006e-05,
+      "loss": 1.4542,
+      "step": 7271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999365314163916e-05,
+      "loss": 1.3582,
+      "step": 7272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993651371691543e-05,
+      "loss": 1.3169,
+      "step": 7273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999364960149719e-05,
+      "loss": 0.9608,
+      "step": 7274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993647831056114e-05,
+      "loss": 1.1639,
+      "step": 7275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993646060368303e-05,
+      "loss": 1.1301,
+      "step": 7276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993644289433776e-05,
+      "loss": 1.0542,
+      "step": 7277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999364251825252e-05,
+      "loss": 1.14,
+      "step": 7278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999364074682453e-05,
+      "loss": 1.0161,
+      "step": 7279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993638975149824e-05,
+      "loss": 0.9362,
+      "step": 7280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999363720322839e-05,
+      "loss": 0.9912,
+      "step": 7281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999363543106023e-05,
+      "loss": 0.5056,
+      "step": 7282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999363365864533e-05,
+      "loss": 0.7172,
+      "step": 7283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999363188598372e-05,
+      "loss": 1.1353,
+      "step": 7284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999363011307537e-05,
+      "loss": 0.7113,
+      "step": 7285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993628339920304e-05,
+      "loss": 0.8874,
+      "step": 7286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993626566518505e-05,
+      "loss": 0.7167,
+      "step": 7287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999362479286999e-05,
+      "loss": 1.0762,
+      "step": 7288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993623018974736e-05,
+      "loss": 1.2297,
+      "step": 7289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999362124483276e-05,
+      "loss": 1.1268,
+      "step": 7290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999361947044405e-05,
+      "loss": 1.0713,
+      "step": 7291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993617695808626e-05,
+      "loss": 0.9842,
+      "step": 7292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993615920926465e-05,
+      "loss": 1.1844,
+      "step": 7293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993614145797586e-05,
+      "loss": 1.1295,
+      "step": 7294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993612370421977e-05,
+      "loss": 1.3198,
+      "step": 7295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993610594799643e-05,
+      "loss": 1.1933,
+      "step": 7296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999360881893058e-05,
+      "loss": 1.1275,
+      "step": 7297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999360704281479e-05,
+      "loss": 0.9787,
+      "step": 7298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993605266452275e-05,
+      "loss": 0.9672,
+      "step": 7299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999360348984303e-05,
+      "loss": 1.1196,
+      "step": 7300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999360171298707e-05,
+      "loss": 1.2345,
+      "step": 7301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999359993588437e-05,
+      "loss": 1.1671,
+      "step": 7302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999359815853495e-05,
+      "loss": 0.6403,
+      "step": 7303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993596380938815e-05,
+      "loss": 1.2603,
+      "step": 7304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999359460309594e-05,
+      "loss": 1.2867,
+      "step": 7305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999359282500634e-05,
+      "loss": 1.336,
+      "step": 7306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999359104667002e-05,
+      "loss": 1.2513,
+      "step": 7307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999358926808696e-05,
+      "loss": 1.0588,
+      "step": 7308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999358748925719e-05,
+      "loss": 1.2615,
+      "step": 7309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999358571018069e-05,
+      "loss": 1.0442,
+      "step": 7310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999358393085746e-05,
+      "loss": 1.1173,
+      "step": 7311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99935821512875e-05,
+      "loss": 0.9781,
+      "step": 7312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999358037147082e-05,
+      "loss": 1.3245,
+      "step": 7313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999357859140741e-05,
+      "loss": 1.2609,
+      "step": 7314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999357681109727e-05,
+      "loss": 0.613,
+      "step": 7315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999357503054042e-05,
+      "loss": 0.7257,
+      "step": 7316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999357324973683e-05,
+      "loss": 1.2196,
+      "step": 7317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999357146868652e-05,
+      "loss": 1.1914,
+      "step": 7318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999356968738948e-05,
+      "loss": 1.2337,
+      "step": 7319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993567905845716e-05,
+      "loss": 1.654,
+      "step": 7320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999356612405522e-05,
+      "loss": 1.7418,
+      "step": 7321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999356434201801e-05,
+      "loss": 1.6275,
+      "step": 7322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999356255973407e-05,
+      "loss": 1.6289,
+      "step": 7323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99935607772034e-05,
+      "loss": 1.5845,
+      "step": 7324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993558994426005e-05,
+      "loss": 1.559,
+      "step": 7325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993557211401884e-05,
+      "loss": 1.5541,
+      "step": 7326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999355542813105e-05,
+      "loss": 1.5677,
+      "step": 7327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999355364461347e-05,
+      "loss": 1.5506,
+      "step": 7328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993551860849174e-05,
+      "loss": 1.5629,
+      "step": 7329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999355007683815e-05,
+      "loss": 1.5482,
+      "step": 7330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993548292580406e-05,
+      "loss": 1.5299,
+      "step": 7331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999354650807593e-05,
+      "loss": 1.5587,
+      "step": 7332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999354472332473e-05,
+      "loss": 1.4898,
+      "step": 7333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993542938326805e-05,
+      "loss": 1.5037,
+      "step": 7334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999354115308215e-05,
+      "loss": 1.1999,
+      "step": 7335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999353936759077e-05,
+      "loss": 1.4152,
+      "step": 7336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999353758185267e-05,
+      "loss": 1.5625,
+      "step": 7337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999353579586784e-05,
+      "loss": 1.3958,
+      "step": 7338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999353400963629e-05,
+      "loss": 1.3154,
+      "step": 7339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993532223158015e-05,
+      "loss": 1.3841,
+      "step": 7340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993530436433e-05,
+      "loss": 1.541,
+      "step": 7341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999352864946128e-05,
+      "loss": 1.4549,
+      "step": 7342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999352686224281e-05,
+      "loss": 1.405,
+      "step": 7343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999352507477764e-05,
+      "loss": 1.5148,
+      "step": 7344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993523287065725e-05,
+      "loss": 1.5382,
+      "step": 7345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993521499107096e-05,
+      "loss": 1.3132,
+      "step": 7346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993519710901736e-05,
+      "loss": 1.3158,
+      "step": 7347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999351792244965e-05,
+      "loss": 1.178,
+      "step": 7348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999351613375084e-05,
+      "loss": 1.2829,
+      "step": 7349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99935143448053e-05,
+      "loss": 1.3511,
+      "step": 7350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993512555613045e-05,
+      "loss": 1.2952,
+      "step": 7351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999351076617406e-05,
+      "loss": 1.3797,
+      "step": 7352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993508976488344e-05,
+      "loss": 1.5269,
+      "step": 7353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999350718655591e-05,
+      "loss": 1.5188,
+      "step": 7354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999350539637675e-05,
+      "loss": 1.4032,
+      "step": 7355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999350360595086e-05,
+      "loss": 1.6251,
+      "step": 7356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999350181527824e-05,
+      "loss": 1.6092,
+      "step": 7357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999350002435891e-05,
+      "loss": 1.2604,
+      "step": 7358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999349823319284e-05,
+      "loss": 1.0308,
+      "step": 7359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999349644178005e-05,
+      "loss": 1.3368,
+      "step": 7360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999349465012054e-05,
+      "loss": 1.6052,
+      "step": 7361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993492858214304e-05,
+      "loss": 1.3298,
+      "step": 7362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999349106606134e-05,
+      "loss": 1.5927,
+      "step": 7363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993489273661646e-05,
+      "loss": 1.3533,
+      "step": 7364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999348748101523e-05,
+      "loss": 1.6058,
+      "step": 7365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993485688122086e-05,
+      "loss": 1.5934,
+      "step": 7366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999348389498223e-05,
+      "loss": 1.3531,
+      "step": 7367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999348210159563e-05,
+      "loss": 1.366,
+      "step": 7368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999348030796232e-05,
+      "loss": 1.3443,
+      "step": 7369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999347851408228e-05,
+      "loss": 1.5869,
+      "step": 7370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999347671995551e-05,
+      "loss": 1.5461,
+      "step": 7371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999347492558202e-05,
+      "loss": 1.593,
+      "step": 7372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999347313096181e-05,
+      "loss": 1.4488,
+      "step": 7373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999347133609486e-05,
+      "loss": 1.613,
+      "step": 7374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99934695409812e-05,
+      "loss": 1.4047,
+      "step": 7375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999346774562081e-05,
+      "loss": 1.6145,
+      "step": 7376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993465950013694e-05,
+      "loss": 1.3279,
+      "step": 7377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999346415415985e-05,
+      "loss": 1.509,
+      "step": 7378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999346235805928e-05,
+      "loss": 1.4429,
+      "step": 7379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999346056171199e-05,
+      "loss": 1.5884,
+      "step": 7380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999345876511798e-05,
+      "loss": 1.5937,
+      "step": 7381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993456968277245e-05,
+      "loss": 1.607,
+      "step": 7382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993455171189775e-05,
+      "loss": 1.5965,
+      "step": 7383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999345337385558e-05,
+      "loss": 1.5968,
+      "step": 7384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993451576274665e-05,
+      "loss": 1.5513,
+      "step": 7385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993449778447025e-05,
+      "loss": 1.222,
+      "step": 7386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999344798037267e-05,
+      "loss": 1.3366,
+      "step": 7387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999344618205157e-05,
+      "loss": 1.3441,
+      "step": 7388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993444383483754e-05,
+      "loss": 1.6091,
+      "step": 7389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999344258466921e-05,
+      "loss": 1.3621,
+      "step": 7390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999344078560795e-05,
+      "loss": 1.5896,
+      "step": 7391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999343898629997e-05,
+      "loss": 1.5968,
+      "step": 7392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999343718674525e-05,
+      "loss": 1.3465,
+      "step": 7393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999343538694382e-05,
+      "loss": 1.5791,
+      "step": 7394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993433586895656e-05,
+      "loss": 1.3193,
+      "step": 7395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993431786600765e-05,
+      "loss": 1.5984,
+      "step": 7396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993429986059156e-05,
+      "loss": 1.0591,
+      "step": 7397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999342818527082e-05,
+      "loss": 1.6761,
+      "step": 7398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993426384235754e-05,
+      "loss": 1.3947,
+      "step": 7399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999342458295397e-05,
+      "loss": 1.146,
+      "step": 7400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993422781425456e-05,
+      "loss": 1.0649,
+      "step": 7401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999342097965022e-05,
+      "loss": 1.2759,
+      "step": 7402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993419177628264e-05,
+      "loss": 1.3245,
+      "step": 7403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999341737535958e-05,
+      "loss": 1.3419,
+      "step": 7404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999341557284417e-05,
+      "loss": 1.2456,
+      "step": 7405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999341377008204e-05,
+      "loss": 1.2134,
+      "step": 7406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999341196707318e-05,
+      "loss": 1.0812,
+      "step": 7407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99934101638176e-05,
+      "loss": 0.9918,
+      "step": 7408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99934083603153e-05,
+      "loss": 0.9686,
+      "step": 7409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999340655656627e-05,
+      "loss": 1.0861,
+      "step": 7410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999340475257051e-05,
+      "loss": 1.0647,
+      "step": 7411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993402948328025e-05,
+      "loss": 1.311,
+      "step": 7412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999340114383882e-05,
+      "loss": 1.3783,
+      "step": 7413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99933993391029e-05,
+      "loss": 1.2996,
+      "step": 7414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999339753412025e-05,
+      "loss": 1.0855,
+      "step": 7415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993395728890866e-05,
+      "loss": 1.189,
+      "step": 7416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999339392341477e-05,
+      "loss": 1.3219,
+      "step": 7417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999339211769195e-05,
+      "loss": 0.8666,
+      "step": 7418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993390311722396e-05,
+      "loss": 1.1259,
+      "step": 7419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999338850550612e-05,
+      "loss": 1.1693,
+      "step": 7420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999338669904312e-05,
+      "loss": 1.8688,
+      "step": 7421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993384892333405e-05,
+      "loss": 1.2217,
+      "step": 7422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999338308537695e-05,
+      "loss": 0.9835,
+      "step": 7423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999338127817379e-05,
+      "loss": 1.1256,
+      "step": 7424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999337947072389e-05,
+      "loss": 1.0414,
+      "step": 7425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993377663027276e-05,
+      "loss": 0.865,
+      "step": 7426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993375855083934e-05,
+      "loss": 1.0807,
+      "step": 7427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999337404689386e-05,
+      "loss": 1.2146,
+      "step": 7428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999337223845707e-05,
+      "loss": 1.1836,
+      "step": 7429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999337042977355e-05,
+      "loss": 1.1424,
+      "step": 7430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999336862084332e-05,
+      "loss": 1.043,
+      "step": 7431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999336681166636e-05,
+      "loss": 1.2966,
+      "step": 7432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999336500224268e-05,
+      "loss": 1.5118,
+      "step": 7433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999336319257226e-05,
+      "loss": 1.3014,
+      "step": 7434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993361382655125e-05,
+      "loss": 1.1033,
+      "step": 7435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999335957249127e-05,
+      "loss": 1.3528,
+      "step": 7436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999335776208068e-05,
+      "loss": 1.185,
+      "step": 7437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999335595142338e-05,
+      "loss": 1.1749,
+      "step": 7438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993354140519346e-05,
+      "loss": 1.037,
+      "step": 7439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993352329368596e-05,
+      "loss": 0.9035,
+      "step": 7440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993350517971115e-05,
+      "loss": 1.3873,
+      "step": 7441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993348706326916e-05,
+      "loss": 1.1147,
+      "step": 7442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999334689443599e-05,
+      "loss": 1.264,
+      "step": 7443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993345082298335e-05,
+      "loss": 2.0064,
+      "step": 7444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993343269913966e-05,
+      "loss": 1.4161,
+      "step": 7445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999334145728287e-05,
+      "loss": 0.7776,
+      "step": 7446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999333964440505e-05,
+      "loss": 0.6111,
+      "step": 7447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999333783128051e-05,
+      "loss": 0.6493,
+      "step": 7448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993336017909244e-05,
+      "loss": 0.6455,
+      "step": 7449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999333420429125e-05,
+      "loss": 0.602,
+      "step": 7450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999333239042653e-05,
+      "loss": 0.603,
+      "step": 7451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999333057631509e-05,
+      "loss": 0.5794,
+      "step": 7452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999332876195693e-05,
+      "loss": 0.5581,
+      "step": 7453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993326947352046e-05,
+      "loss": 0.5843,
+      "step": 7454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999332513250043e-05,
+      "loss": 0.5669,
+      "step": 7455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993323317402094e-05,
+      "loss": 0.5727,
+      "step": 7456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999332150205705e-05,
+      "loss": 0.5761,
+      "step": 7457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999331968646526e-05,
+      "loss": 0.5629,
+      "step": 7458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999331787062676e-05,
+      "loss": 0.5799,
+      "step": 7459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999331605454153e-05,
+      "loss": 0.5618,
+      "step": 7460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993314238209585e-05,
+      "loss": 0.8993,
+      "step": 7461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993312421630904e-05,
+      "loss": 1.4123,
+      "step": 7462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999331060480551e-05,
+      "loss": 1.0034,
+      "step": 7463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999330878773339e-05,
+      "loss": 1.3514,
+      "step": 7464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999330697041454e-05,
+      "loss": 1.3569,
+      "step": 7465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993305152848974e-05,
+      "loss": 1.0425,
+      "step": 7466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999330333503668e-05,
+      "loss": 1.0502,
+      "step": 7467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993301516977666e-05,
+      "loss": 1.1806,
+      "step": 7468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993299698671933e-05,
+      "loss": 1.0627,
+      "step": 7469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999329788011947e-05,
+      "loss": 0.986,
+      "step": 7470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999329606132028e-05,
+      "loss": 0.9962,
+      "step": 7471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999329424227437e-05,
+      "loss": 1.2365,
+      "step": 7472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993292422981744e-05,
+      "loss": 1.1068,
+      "step": 7473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993290603442386e-05,
+      "loss": 1.3851,
+      "step": 7474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999328878365631e-05,
+      "loss": 1.3683,
+      "step": 7475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999328696362351e-05,
+      "loss": 1.3053,
+      "step": 7476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999328514334398e-05,
+      "loss": 0.757,
+      "step": 7477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999328332281774e-05,
+      "loss": 0.6441,
+      "step": 7478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993281502044765e-05,
+      "loss": 0.9481,
+      "step": 7479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999327968102507e-05,
+      "loss": 1.1469,
+      "step": 7480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999327785975866e-05,
+      "loss": 1.0571,
+      "step": 7481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999327603824552e-05,
+      "loss": 0.7546,
+      "step": 7482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999327421648565e-05,
+      "loss": 0.6546,
+      "step": 7483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999327239447907e-05,
+      "loss": 1.1642,
+      "step": 7484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999327057222576e-05,
+      "loss": 1.0698,
+      "step": 7485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999326874972573e-05,
+      "loss": 1.1704,
+      "step": 7486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999326692697897e-05,
+      "loss": 1.2712,
+      "step": 7487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999326510398549e-05,
+      "loss": 0.9271,
+      "step": 7488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993263280745296e-05,
+      "loss": 1.1715,
+      "step": 7489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993261457258364e-05,
+      "loss": 1.149,
+      "step": 7490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999325963352472e-05,
+      "loss": 1.0354,
+      "step": 7491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999325780954435e-05,
+      "loss": 1.1546,
+      "step": 7492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993255985317254e-05,
+      "loss": 1.2658,
+      "step": 7493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999325416084344e-05,
+      "loss": 1.1783,
+      "step": 7494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993252336122904e-05,
+      "loss": 1.0157,
+      "step": 7495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999325051115564e-05,
+      "loss": 1.0256,
+      "step": 7496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999324868594166e-05,
+      "loss": 1.5544,
+      "step": 7497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993246860480944e-05,
+      "loss": 1.1515,
+      "step": 7498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999324503477352e-05,
+      "loss": 1.2725,
+      "step": 7499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999324320881936e-05,
+      "loss": 1.0838,
+      "step": 7500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.067463994026184,
+      "eval_runtime": 84.948,
+      "eval_samples_per_second": 16.304,
+      "eval_steps_per_second": 4.085,
+      "step": 7500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99932413826185e-05,
+      "loss": 1.2704,
+      "step": 7501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999323955617089e-05,
+      "loss": 1.2755,
+      "step": 7502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993237729476574e-05,
+      "loss": 1.0452,
+      "step": 7503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999323590253553e-05,
+      "loss": 1.1438,
+      "step": 7504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999323407534776e-05,
+      "loss": 1.1556,
+      "step": 7505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999323224791327e-05,
+      "loss": 1.3714,
+      "step": 7506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999323042023206e-05,
+      "loss": 1.3532,
+      "step": 7507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993228592304124e-05,
+      "loss": 0.9893,
+      "step": 7508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999322676412947e-05,
+      "loss": 0.9889,
+      "step": 7509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999322493570809e-05,
+      "loss": 1.1921,
+      "step": 7510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999322310703999e-05,
+      "loss": 1.367,
+      "step": 7511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993221278125164e-05,
+      "loss": 0.8645,
+      "step": 7512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999321944896361e-05,
+      "loss": 0.8964,
+      "step": 7513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993217619555345e-05,
+      "loss": 1.1824,
+      "step": 7514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999321578990035e-05,
+      "loss": 1.1325,
+      "step": 7515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999321395999864e-05,
+      "loss": 0.7096,
+      "step": 7516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99932121298502e-05,
+      "loss": 1.5663,
+      "step": 7517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999321029945504e-05,
+      "loss": 1.1504,
+      "step": 7518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993208468813156e-05,
+      "loss": 1.3688,
+      "step": 7519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999320663792455e-05,
+      "loss": 1.1016,
+      "step": 7520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999320480678923e-05,
+      "loss": 1.009,
+      "step": 7521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999320297540718e-05,
+      "loss": 1.3089,
+      "step": 7522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99932011437784e-05,
+      "loss": 1.3837,
+      "step": 7523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999319931190291e-05,
+      "loss": 0.9332,
+      "step": 7524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99931974797807e-05,
+      "loss": 1.2406,
+      "step": 7525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999319564741176e-05,
+      "loss": 1.1936,
+      "step": 7526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999319381479609e-05,
+      "loss": 1.6506,
+      "step": 7527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999319198193371e-05,
+      "loss": 0.859,
+      "step": 7528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993190148824604e-05,
+      "loss": 1.2011,
+      "step": 7529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993188315468776e-05,
+      "loss": 0.672,
+      "step": 7530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993186481866225e-05,
+      "loss": 0.8646,
+      "step": 7531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993184648016956e-05,
+      "loss": 0.9417,
+      "step": 7532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999318281392096e-05,
+      "loss": 1.31,
+      "step": 7533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999318097957824e-05,
+      "loss": 1.1293,
+      "step": 7534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99931791449888e-05,
+      "loss": 1.1137,
+      "step": 7535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999317731015264e-05,
+      "loss": 1.2222,
+      "step": 7536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999317547506976e-05,
+      "loss": 0.9789,
+      "step": 7537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993173639740153e-05,
+      "loss": 1.0231,
+      "step": 7538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999317180416383e-05,
+      "loss": 0.7891,
+      "step": 7539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999316996834078e-05,
+      "loss": 1.3479,
+      "step": 7540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993168132271005e-05,
+      "loss": 1.1883,
+      "step": 7541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999316629595451e-05,
+      "loss": 1.1731,
+      "step": 7542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993164459391296e-05,
+      "loss": 0.9551,
+      "step": 7543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993162622581356e-05,
+      "loss": 1.1011,
+      "step": 7544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99931607855247e-05,
+      "loss": 1.026,
+      "step": 7545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999315894822132e-05,
+      "loss": 1.1553,
+      "step": 7546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999315711067122e-05,
+      "loss": 1.0465,
+      "step": 7547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999315527287439e-05,
+      "loss": 0.8922,
+      "step": 7548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999315343483084e-05,
+      "loss": 1.1713,
+      "step": 7549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999315159654058e-05,
+      "loss": 1.1499,
+      "step": 7550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999314975800358e-05,
+      "loss": 1.4271,
+      "step": 7551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993147919219876e-05,
+      "loss": 1.0735,
+      "step": 7552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993146080189436e-05,
+      "loss": 1.0994,
+      "step": 7553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999314424091228e-05,
+      "loss": 1.016,
+      "step": 7554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993142401388404e-05,
+      "loss": 0.9924,
+      "step": 7555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99931405616178e-05,
+      "loss": 1.2288,
+      "step": 7556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999313872160048e-05,
+      "loss": 1.1756,
+      "step": 7557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999313688133643e-05,
+      "loss": 1.1061,
+      "step": 7558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999313504082567e-05,
+      "loss": 1.1704,
+      "step": 7559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999313320006819e-05,
+      "loss": 0.9434,
+      "step": 7560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999313135906398e-05,
+      "loss": 0.926,
+      "step": 7561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999312951781305e-05,
+      "loss": 1.0371,
+      "step": 7562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993127676315396e-05,
+      "loss": 0.769,
+      "step": 7563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999312583457102e-05,
+      "loss": 1.1982,
+      "step": 7564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993123992579923e-05,
+      "loss": 1.1619,
+      "step": 7565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999312215034211e-05,
+      "loss": 1.3264,
+      "step": 7566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999312030785757e-05,
+      "loss": 1.7573,
+      "step": 7567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993118465126307e-05,
+      "loss": 0.8679,
+      "step": 7568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999311662214833e-05,
+      "loss": 1.2439,
+      "step": 7569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999311477892363e-05,
+      "loss": 1.1812,
+      "step": 7570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99931129354522e-05,
+      "loss": 1.0988,
+      "step": 7571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999311109173406e-05,
+      "loss": 1.2373,
+      "step": 7572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999310924776919e-05,
+      "loss": 1.9465,
+      "step": 7573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993107403557606e-05,
+      "loss": 1.4244,
+      "step": 7574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99931055590993e-05,
+      "loss": 1.3393,
+      "step": 7575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993103714394265e-05,
+      "loss": 1.2637,
+      "step": 7576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99931018694425e-05,
+      "loss": 1.245,
+      "step": 7577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993100024244035e-05,
+      "loss": 0.4278,
+      "step": 7578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999309817879884e-05,
+      "loss": 1.1911,
+      "step": 7579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999309633310692e-05,
+      "loss": 0.746,
+      "step": 7580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999309448716828e-05,
+      "loss": 1.1593,
+      "step": 7581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993092640982925e-05,
+      "loss": 1.1381,
+      "step": 7582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999309079455085e-05,
+      "loss": 1.0607,
+      "step": 7583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999308894787204e-05,
+      "loss": 1.0424,
+      "step": 7584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999308710094652e-05,
+      "loss": 1.1306,
+      "step": 7585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993085253774276e-05,
+      "loss": 1.1298,
+      "step": 7586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999308340635531e-05,
+      "loss": 0.9243,
+      "step": 7587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999308155868962e-05,
+      "loss": 0.8916,
+      "step": 7588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999307971077721e-05,
+      "loss": 0.6633,
+      "step": 7589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999307786261808e-05,
+      "loss": 1.0586,
+      "step": 7590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999307601421224e-05,
+      "loss": 1.1927,
+      "step": 7591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999307416555966e-05,
+      "loss": 0.7294,
+      "step": 7592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999307231666037e-05,
+      "loss": 0.5216,
+      "step": 7593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999307046751436e-05,
+      "loss": 1.1658,
+      "step": 7594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999306861812162e-05,
+      "loss": 1.1383,
+      "step": 7595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999306676848217e-05,
+      "loss": 0.7021,
+      "step": 7596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999306491859599e-05,
+      "loss": 1.2803,
+      "step": 7597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993063068463084e-05,
+      "loss": 1.1629,
+      "step": 7598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999306121808347e-05,
+      "loss": 0.7347,
+      "step": 7599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999305936745713e-05,
+      "loss": 1.154,
+      "step": 7600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999305751658407e-05,
+      "loss": 1.0441,
+      "step": 7601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999305566546429e-05,
+      "loss": 1.0921,
+      "step": 7602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999305381409779e-05,
+      "loss": 1.0165,
+      "step": 7603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999305196248456e-05,
+      "loss": 0.9674,
+      "step": 7604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993050110624614e-05,
+      "loss": 1.2161,
+      "step": 7605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999304825851796e-05,
+      "loss": 0.8684,
+      "step": 7606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999304640616457e-05,
+      "loss": 1.3836,
+      "step": 7607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999304455356446e-05,
+      "loss": 1.0503,
+      "step": 7608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999304270071763e-05,
+      "loss": 1.0424,
+      "step": 7609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993040847624085e-05,
+      "loss": 0.7922,
+      "step": 7610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993038994283816e-05,
+      "loss": 1.1937,
+      "step": 7611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999303714069682e-05,
+      "loss": 1.0771,
+      "step": 7612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999303528686311e-05,
+      "loss": 0.7628,
+      "step": 7613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999303343278268e-05,
+      "loss": 1.0711,
+      "step": 7614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999303157845553e-05,
+      "loss": 0.9608,
+      "step": 7615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993029723881655e-05,
+      "loss": 1.0593,
+      "step": 7616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993027869061063e-05,
+      "loss": 1.1047,
+      "step": 7617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999302601399375e-05,
+      "loss": 0.8421,
+      "step": 7618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993024158679716e-05,
+      "loss": 0.4275,
+      "step": 7619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999302230311896e-05,
+      "loss": 0.4701,
+      "step": 7620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999302044731149e-05,
+      "loss": 1.1274,
+      "step": 7621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999301859125729e-05,
+      "loss": 1.1504,
+      "step": 7622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999301673495637e-05,
+      "loss": 1.0298,
+      "step": 7623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993014878408734e-05,
+      "loss": 1.2913,
+      "step": 7624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999301302161438e-05,
+      "loss": 1.1684,
+      "step": 7625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99930111645733e-05,
+      "loss": 1.1173,
+      "step": 7626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99930093072855e-05,
+      "loss": 1.2299,
+      "step": 7627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999300744975098e-05,
+      "loss": 1.2688,
+      "step": 7628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999300559196974e-05,
+      "loss": 0.7244,
+      "step": 7629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999300373394178e-05,
+      "loss": 1.1436,
+      "step": 7630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99930018756671e-05,
+      "loss": 1.3276,
+      "step": 7631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993000017145696e-05,
+      "loss": 1.2979,
+      "step": 7632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999299815837758e-05,
+      "loss": 1.2512,
+      "step": 7633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999299629936274e-05,
+      "loss": 1.1245,
+      "step": 7634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999299444010117e-05,
+      "loss": 1.122,
+      "step": 7635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999299258059289e-05,
+      "loss": 1.0844,
+      "step": 7636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992990720837886e-05,
+      "loss": 0.9044,
+      "step": 7637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992988860836163e-05,
+      "loss": 1.0386,
+      "step": 7638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992987000587724e-05,
+      "loss": 1.1474,
+      "step": 7639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999298514009256e-05,
+      "loss": 1.2707,
+      "step": 7640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992983279350674e-05,
+      "loss": 0.9895,
+      "step": 7641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999298141836207e-05,
+      "loss": 1.152,
+      "step": 7642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999297955712674e-05,
+      "loss": 1.1866,
+      "step": 7643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992977695644705e-05,
+      "loss": 1.3486,
+      "step": 7644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992975833915936e-05,
+      "loss": 0.8838,
+      "step": 7645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999297397194045e-05,
+      "loss": 1.1444,
+      "step": 7646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999297210971825e-05,
+      "loss": 1.7278,
+      "step": 7647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999297024724933e-05,
+      "loss": 1.3322,
+      "step": 7648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999296838453368e-05,
+      "loss": 1.0926,
+      "step": 7649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999296652157132e-05,
+      "loss": 1.049,
+      "step": 7650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999296465836223e-05,
+      "loss": 1.2226,
+      "step": 7651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999296279490642e-05,
+      "loss": 1.1249,
+      "step": 7652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99929609312039e-05,
+      "loss": 1.1267,
+      "step": 7653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992959067254655e-05,
+      "loss": 1.1067,
+      "step": 7654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999295720305869e-05,
+      "loss": 1.2643,
+      "step": 7655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999295533861601e-05,
+      "loss": 1.0183,
+      "step": 7656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992953473926595e-05,
+      "loss": 1.0721,
+      "step": 7657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999295160899047e-05,
+      "loss": 1.1275,
+      "step": 7658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992949743807625e-05,
+      "loss": 1.1531,
+      "step": 7659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999294787837807e-05,
+      "loss": 1.0552,
+      "step": 7660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999294601270178e-05,
+      "loss": 1.0566,
+      "step": 7661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992944146778776e-05,
+      "loss": 1.3829,
+      "step": 7662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992942280609054e-05,
+      "loss": 1.2207,
+      "step": 7663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992940414192616e-05,
+      "loss": 1.0252,
+      "step": 7664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999293854752945e-05,
+      "loss": 0.7548,
+      "step": 7665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999293668061956e-05,
+      "loss": 1.0979,
+      "step": 7666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999293481346296e-05,
+      "loss": 1.8026,
+      "step": 7667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999293294605964e-05,
+      "loss": 1.3354,
+      "step": 7668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99929310784096e-05,
+      "loss": 1.3458,
+      "step": 7669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992929210512834e-05,
+      "loss": 1.0722,
+      "step": 7670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992927342369356e-05,
+      "loss": 0.8646,
+      "step": 7671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992925473979155e-05,
+      "loss": 1.4073,
+      "step": 7672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999292360534223e-05,
+      "loss": 1.0497,
+      "step": 7673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999292173645859e-05,
+      "loss": 0.5976,
+      "step": 7674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999291986732823e-05,
+      "loss": 1.0724,
+      "step": 7675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999291799795115e-05,
+      "loss": 1.0314,
+      "step": 7676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999291612832735e-05,
+      "loss": 1.1537,
+      "step": 7677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992914258456836e-05,
+      "loss": 1.3026,
+      "step": 7678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992912388339595e-05,
+      "loss": 0.8799,
+      "step": 7679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999291051797563e-05,
+      "loss": 1.2481,
+      "step": 7680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992908647364956e-05,
+      "loss": 1.0373,
+      "step": 7681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992906776507565e-05,
+      "loss": 1.3132,
+      "step": 7682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999290490540345e-05,
+      "loss": 1.036,
+      "step": 7683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999290303405261e-05,
+      "loss": 0.9312,
+      "step": 7684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992901162455054e-05,
+      "loss": 1.787,
+      "step": 7685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999289929061078e-05,
+      "loss": 1.1312,
+      "step": 7686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992897418519784e-05,
+      "loss": 1.0114,
+      "step": 7687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999289554618207e-05,
+      "loss": 1.1453,
+      "step": 7688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999289367359764e-05,
+      "loss": 1.143,
+      "step": 7689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999289180076649e-05,
+      "loss": 1.1481,
+      "step": 7690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992889927688614e-05,
+      "loss": 0.7681,
+      "step": 7691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992888054364026e-05,
+      "loss": 0.9479,
+      "step": 7692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999288618079272e-05,
+      "loss": 1.0966,
+      "step": 7693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992884306974684e-05,
+      "loss": 1.1311,
+      "step": 7694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999288243290994e-05,
+      "loss": 1.9402,
+      "step": 7695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992880558598475e-05,
+      "loss": 0.9394,
+      "step": 7696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999287868404029e-05,
+      "loss": 1.3466,
+      "step": 7697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999287680923538e-05,
+      "loss": 0.9499,
+      "step": 7698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992874934183756e-05,
+      "loss": 1.1177,
+      "step": 7699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999287305888541e-05,
+      "loss": 1.1028,
+      "step": 7700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999287118334035e-05,
+      "loss": 1.7157,
+      "step": 7701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992869307548565e-05,
+      "loss": 1.1457,
+      "step": 7702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999286743151006e-05,
+      "loss": 0.9201,
+      "step": 7703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999286555522484e-05,
+      "loss": 0.9718,
+      "step": 7704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99928636786929e-05,
+      "loss": 1.4427,
+      "step": 7705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999286180191425e-05,
+      "loss": 1.1255,
+      "step": 7706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999285992488887e-05,
+      "loss": 1.0025,
+      "step": 7707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992858047616776e-05,
+      "loss": 0.9853,
+      "step": 7708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999285617009796e-05,
+      "loss": 0.8761,
+      "step": 7709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999285429233242e-05,
+      "loss": 0.9686,
+      "step": 7710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992852414320165e-05,
+      "loss": 1.4701,
+      "step": 7711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992850536061196e-05,
+      "loss": 1.2018,
+      "step": 7712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999284865755551e-05,
+      "loss": 1.1506,
+      "step": 7713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99928467788031e-05,
+      "loss": 1.236,
+      "step": 7714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999284489980397e-05,
+      "loss": 1.0325,
+      "step": 7715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992843020558124e-05,
+      "loss": 1.0386,
+      "step": 7716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999284114106556e-05,
+      "loss": 1.0008,
+      "step": 7717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999283926132627e-05,
+      "loss": 0.8698,
+      "step": 7718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992837381340265e-05,
+      "loss": 0.6912,
+      "step": 7719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999283550110754e-05,
+      "loss": 0.861,
+      "step": 7720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99928336206281e-05,
+      "loss": 0.6516,
+      "step": 7721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992831739901946e-05,
+      "loss": 0.9092,
+      "step": 7722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992829858929064e-05,
+      "loss": 0.9458,
+      "step": 7723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992827977709464e-05,
+      "loss": 1.2814,
+      "step": 7724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999282609624315e-05,
+      "loss": 1.0289,
+      "step": 7725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992824214530114e-05,
+      "loss": 1.1974,
+      "step": 7726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992822332570363e-05,
+      "loss": 0.9542,
+      "step": 7727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999282045036389e-05,
+      "loss": 1.0783,
+      "step": 7728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99928185679107e-05,
+      "loss": 1.1048,
+      "step": 7729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999281668521078e-05,
+      "loss": 1.5022,
+      "step": 7730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992814802264164e-05,
+      "loss": 0.9485,
+      "step": 7731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999281291907082e-05,
+      "loss": 1.0487,
+      "step": 7732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999281103563075e-05,
+      "loss": 0.9813,
+      "step": 7733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992809151943967e-05,
+      "loss": 1.4835,
+      "step": 7734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999280726801047e-05,
+      "loss": 0.905,
+      "step": 7735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999280538383024e-05,
+      "loss": 1.3004,
+      "step": 7736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999280349940331e-05,
+      "loss": 1.1508,
+      "step": 7737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999280161472965e-05,
+      "loss": 0.7234,
+      "step": 7738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999279972980927e-05,
+      "loss": 0.7698,
+      "step": 7739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999279784464218e-05,
+      "loss": 1.1764,
+      "step": 7740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999279595922837e-05,
+      "loss": 1.2074,
+      "step": 7741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999279407356784e-05,
+      "loss": 1.1786,
+      "step": 7742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999279218766059e-05,
+      "loss": 1.2808,
+      "step": 7743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999279030150662e-05,
+      "loss": 1.3466,
+      "step": 7744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999278841510593e-05,
+      "loss": 0.9158,
+      "step": 7745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999278652845854e-05,
+      "loss": 1.2887,
+      "step": 7746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999278464156441e-05,
+      "loss": 1.0082,
+      "step": 7747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999278275442357e-05,
+      "loss": 0.1872,
+      "step": 7748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999278086703601e-05,
+      "loss": 1.0771,
+      "step": 7749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992778979401734e-05,
+      "loss": 1.3316,
+      "step": 7750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992777091520735e-05,
+      "loss": 1.2208,
+      "step": 7751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992775203393026e-05,
+      "loss": 1.2078,
+      "step": 7752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999277331501859e-05,
+      "loss": 1.1975,
+      "step": 7753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992771426397444e-05,
+      "loss": 0.9454,
+      "step": 7754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999276953752958e-05,
+      "loss": 1.2826,
+      "step": 7755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999276764841499e-05,
+      "loss": 1.4365,
+      "step": 7756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992765759053686e-05,
+      "loss": 1.0539,
+      "step": 7757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999276386944567e-05,
+      "loss": 1.0879,
+      "step": 7758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999276197959093e-05,
+      "loss": 0.7525,
+      "step": 7759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999276008948947e-05,
+      "loss": 1.0603,
+      "step": 7760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992758199141293e-05,
+      "loss": 1.0382,
+      "step": 7761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992756308546394e-05,
+      "loss": 1.4256,
+      "step": 7762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999275441770479e-05,
+      "loss": 1.701,
+      "step": 7763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999275252661646e-05,
+      "loss": 1.3756,
+      "step": 7764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999275063528141e-05,
+      "loss": 1.5012,
+      "step": 7765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999274874369965e-05,
+      "loss": 1.3719,
+      "step": 7766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992746851871165e-05,
+      "loss": 0.9829,
+      "step": 7767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999274495979596e-05,
+      "loss": 1.0734,
+      "step": 7768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999274306747404e-05,
+      "loss": 1.3196,
+      "step": 7769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992741174905404e-05,
+      "loss": 1.0388,
+      "step": 7770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999273928209005e-05,
+      "loss": 1.3656,
+      "step": 7771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992737389027977e-05,
+      "loss": 1.1544,
+      "step": 7772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999273549571919e-05,
+      "loss": 0.8635,
+      "step": 7773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992733602163674e-05,
+      "loss": 1.4682,
+      "step": 7774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999273170836145e-05,
+      "loss": 1.3557,
+      "step": 7775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999272981431251e-05,
+      "loss": 1.0475,
+      "step": 7776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992727920016844e-05,
+      "loss": 1.0465,
+      "step": 7777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992726025474466e-05,
+      "loss": 1.0858,
+      "step": 7778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999272413068538e-05,
+      "loss": 1.0212,
+      "step": 7779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999272223564956e-05,
+      "loss": 1.0863,
+      "step": 7780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992720340367025e-05,
+      "loss": 0.8423,
+      "step": 7781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999271844483778e-05,
+      "loss": 0.7978,
+      "step": 7782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999271654906181e-05,
+      "loss": 1.2256,
+      "step": 7783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999271465303913e-05,
+      "loss": 0.8826,
+      "step": 7784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999271275676972e-05,
+      "loss": 0.8992,
+      "step": 7785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999271086025361e-05,
+      "loss": 1.2419,
+      "step": 7786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992708963490764e-05,
+      "loss": 1.3075,
+      "step": 7787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992707066481216e-05,
+      "loss": 0.6001,
+      "step": 7788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999270516922494e-05,
+      "loss": 0.762,
+      "step": 7789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999270327172195e-05,
+      "loss": 0.955,
+      "step": 7790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992701373972245e-05,
+      "loss": 1.1614,
+      "step": 7791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992699475975817e-05,
+      "loss": 1.1102,
+      "step": 7792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999269757773268e-05,
+      "loss": 1.5231,
+      "step": 7793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992695679242815e-05,
+      "loss": 0.9765,
+      "step": 7794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999269378050624e-05,
+      "loss": 0.8491,
+      "step": 7795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992691881522945e-05,
+      "loss": 1.0887,
+      "step": 7796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999268998229294e-05,
+      "loss": 1.1864,
+      "step": 7797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999268808281621e-05,
+      "loss": 1.0629,
+      "step": 7798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999268618309276e-05,
+      "loss": 1.305,
+      "step": 7799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99926842831226e-05,
+      "loss": 1.5089,
+      "step": 7800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999268238290572e-05,
+      "loss": 1.2749,
+      "step": 7801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999268048244212e-05,
+      "loss": 1.1039,
+      "step": 7802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999267858173181e-05,
+      "loss": 1.1027,
+      "step": 7803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992676680774773e-05,
+      "loss": 1.5077,
+      "step": 7804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992674779571024e-05,
+      "loss": 1.3028,
+      "step": 7805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992672878120564e-05,
+      "loss": 1.2525,
+      "step": 7806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992670976423374e-05,
+      "loss": 1.3055,
+      "step": 7807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999266907447947e-05,
+      "loss": 1.0859,
+      "step": 7808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999266717228886e-05,
+      "loss": 0.8956,
+      "step": 7809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999266526985152e-05,
+      "loss": 0.9879,
+      "step": 7810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999266336716747e-05,
+      "loss": 1.046,
+      "step": 7811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99926614642367e-05,
+      "loss": 1.4059,
+      "step": 7812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992659561059216e-05,
+      "loss": 1.278,
+      "step": 7813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992657657635014e-05,
+      "loss": 1.0187,
+      "step": 7814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992655753964094e-05,
+      "loss": 1.1006,
+      "step": 7815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999265385004646e-05,
+      "loss": 1.1579,
+      "step": 7816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992651945882105e-05,
+      "loss": 1.092,
+      "step": 7817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992650041471034e-05,
+      "loss": 0.9898,
+      "step": 7818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999264813681325e-05,
+      "loss": 1.1935,
+      "step": 7819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999264623190874e-05,
+      "loss": 0.9533,
+      "step": 7820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999264432675752e-05,
+      "loss": 1.0438,
+      "step": 7821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992642421359583e-05,
+      "loss": 1.2504,
+      "step": 7822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999264051571493e-05,
+      "loss": 0.7561,
+      "step": 7823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992638609823556e-05,
+      "loss": 1.0104,
+      "step": 7824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999263670368547e-05,
+      "loss": 1.2529,
+      "step": 7825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999263479730066e-05,
+      "loss": 1.1248,
+      "step": 7826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999263289066914e-05,
+      "loss": 1.0451,
+      "step": 7827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99926309837909e-05,
+      "loss": 1.1424,
+      "step": 7828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999262907666594e-05,
+      "loss": 0.9508,
+      "step": 7829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999262716929427e-05,
+      "loss": 1.1694,
+      "step": 7830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999262526167588e-05,
+      "loss": 1.0555,
+      "step": 7831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992623353810774e-05,
+      "loss": 1.0308,
+      "step": 7832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992621445698956e-05,
+      "loss": 0.9165,
+      "step": 7833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992619537340414e-05,
+      "loss": 1.1295,
+      "step": 7834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999261762873516e-05,
+      "loss": 0.9402,
+      "step": 7835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999261571988319e-05,
+      "loss": 1.0604,
+      "step": 7836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99926138107845e-05,
+      "loss": 1.0774,
+      "step": 7837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992611901439096e-05,
+      "loss": 0.8003,
+      "step": 7838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999260999184697e-05,
+      "loss": 0.8127,
+      "step": 7839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999260808200813e-05,
+      "loss": 0.5362,
+      "step": 7840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999260617192258e-05,
+      "loss": 0.4679,
+      "step": 7841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999260426159031e-05,
+      "loss": 1.2592,
+      "step": 7842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999260235101132e-05,
+      "loss": 1.0108,
+      "step": 7843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992600440185614e-05,
+      "loss": 0.7332,
+      "step": 7844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99925985291132e-05,
+      "loss": 1.0624,
+      "step": 7845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999259661779406e-05,
+      "loss": 0.7453,
+      "step": 7846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992594706228204e-05,
+      "loss": 0.8955,
+      "step": 7847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999259279441564e-05,
+      "loss": 0.8746,
+      "step": 7848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999259088235635e-05,
+      "loss": 1.0011,
+      "step": 7849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999258897005035e-05,
+      "loss": 0.8298,
+      "step": 7850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999258705749763e-05,
+      "loss": 0.9943,
+      "step": 7851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992585144698197e-05,
+      "loss": 1.1311,
+      "step": 7852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992583231652046e-05,
+      "loss": 0.849,
+      "step": 7853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999258131835918e-05,
+      "loss": 0.9639,
+      "step": 7854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999257940481959e-05,
+      "loss": 1.1424,
+      "step": 7855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999257749103329e-05,
+      "loss": 1.0167,
+      "step": 7856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999257557700028e-05,
+      "loss": 1.1265,
+      "step": 7857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999257366272054e-05,
+      "loss": 0.9859,
+      "step": 7858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992571748194096e-05,
+      "loss": 1.2524,
+      "step": 7859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999256983342093e-05,
+      "loss": 1.0596,
+      "step": 7860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992567918401046e-05,
+      "loss": 0.7429,
+      "step": 7861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992566003134456e-05,
+      "loss": 1.1427,
+      "step": 7862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999256408762114e-05,
+      "loss": 1.18,
+      "step": 7863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999256217186111e-05,
+      "loss": 1.3083,
+      "step": 7864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999256025585436e-05,
+      "loss": 1.0804,
+      "step": 7865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992558339600904e-05,
+      "loss": 1.179,
+      "step": 7866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999255642310073e-05,
+      "loss": 1.2044,
+      "step": 7867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999255450635384e-05,
+      "loss": 0.9728,
+      "step": 7868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999255258936023e-05,
+      "loss": 1.3624,
+      "step": 7869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992550672119894e-05,
+      "loss": 1.4786,
+      "step": 7870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992548754632865e-05,
+      "loss": 1.0779,
+      "step": 7871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99925468368991e-05,
+      "loss": 1.2894,
+      "step": 7872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999254491891863e-05,
+      "loss": 1.2372,
+      "step": 7873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999254300069144e-05,
+      "loss": 1.0072,
+      "step": 7874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992541082217536e-05,
+      "loss": 1.0588,
+      "step": 7875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999253916349692e-05,
+      "loss": 1.3491,
+      "step": 7876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999253724452958e-05,
+      "loss": 1.5759,
+      "step": 7877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999253532531553e-05,
+      "loss": 0.7801,
+      "step": 7878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992533405854756e-05,
+      "loss": 1.0617,
+      "step": 7879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999253148614727e-05,
+      "loss": 1.0505,
+      "step": 7880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999252956619308e-05,
+      "loss": 1.1608,
+      "step": 7881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992527645992164e-05,
+      "loss": 1.2429,
+      "step": 7882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999252572554454e-05,
+      "loss": 1.1126,
+      "step": 7883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999252380485019e-05,
+      "loss": 0.9354,
+      "step": 7884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992521883909127e-05,
+      "loss": 1.076,
+      "step": 7885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999251996272135e-05,
+      "loss": 1.2101,
+      "step": 7886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999251804128685e-05,
+      "loss": 1.2844,
+      "step": 7887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999251611960565e-05,
+      "loss": 1.1769,
+      "step": 7888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999251419767772e-05,
+      "loss": 1.2664,
+      "step": 7889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992512275503076e-05,
+      "loss": 1.0522,
+      "step": 7890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999251035308172e-05,
+      "loss": 1.0897,
+      "step": 7891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992508430413654e-05,
+      "loss": 1.0961,
+      "step": 7892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999250650749887e-05,
+      "loss": 1.3068,
+      "step": 7893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992504584337365e-05,
+      "loss": 0.4432,
+      "step": 7894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992502660929155e-05,
+      "loss": 0.7508,
+      "step": 7895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992500737274214e-05,
+      "loss": 0.9558,
+      "step": 7896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999249881337257e-05,
+      "loss": 0.8569,
+      "step": 7897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99924968892242e-05,
+      "loss": 0.9947,
+      "step": 7898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999249496482913e-05,
+      "loss": 1.2156,
+      "step": 7899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999249304018733e-05,
+      "loss": 1.2722,
+      "step": 7900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992491115298824e-05,
+      "loss": 0.9461,
+      "step": 7901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992489190163595e-05,
+      "loss": 0.937,
+      "step": 7902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999248726478165e-05,
+      "loss": 1.1194,
+      "step": 7903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992485339152993e-05,
+      "loss": 1.1772,
+      "step": 7904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999248341327763e-05,
+      "loss": 1.2923,
+      "step": 7905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999248148715554e-05,
+      "loss": 1.09,
+      "step": 7906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992479560786744e-05,
+      "loss": 2.6286,
+      "step": 7907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999247763417122e-05,
+      "loss": 3.4125,
+      "step": 7908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999247570730899e-05,
+      "loss": 1.1838,
+      "step": 7909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999247378020004e-05,
+      "loss": 1.379,
+      "step": 7910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999247185284438e-05,
+      "loss": 1.099,
+      "step": 7911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992469925242e-05,
+      "loss": 1.0116,
+      "step": 7912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999246799739291e-05,
+      "loss": 1.0267,
+      "step": 7913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999246606929711e-05,
+      "loss": 1.4954,
+      "step": 7914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999246414095458e-05,
+      "loss": 1.6992,
+      "step": 7915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992462212365345e-05,
+      "loss": 1.3373,
+      "step": 7916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999246028352939e-05,
+      "loss": 0.8221,
+      "step": 7917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999245835444672e-05,
+      "loss": 1.0465,
+      "step": 7918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999245642511734e-05,
+      "loss": 1.1051,
+      "step": 7919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992454495541244e-05,
+      "loss": 1.0927,
+      "step": 7920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992452565718426e-05,
+      "loss": 1.1845,
+      "step": 7921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99924506356489e-05,
+      "loss": 1.0775,
+      "step": 7922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992448705332654e-05,
+      "loss": 1.3101,
+      "step": 7923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99924467747697e-05,
+      "loss": 1.2129,
+      "step": 7924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992444843960027e-05,
+      "loss": 1.3299,
+      "step": 7925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999244291290364e-05,
+      "loss": 1.1885,
+      "step": 7926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999244098160053e-05,
+      "loss": 0.8202,
+      "step": 7927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992439050050715e-05,
+      "loss": 1.169,
+      "step": 7928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999243711825419e-05,
+      "loss": 0.9432,
+      "step": 7929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999243518621094e-05,
+      "loss": 1.1669,
+      "step": 7930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999243325392098e-05,
+      "loss": 1.0737,
+      "step": 7931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992431321384294e-05,
+      "loss": 1.1336,
+      "step": 7932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99924293886009e-05,
+      "loss": 1.4216,
+      "step": 7933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99924274555708e-05,
+      "loss": 1.34,
+      "step": 7934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999242552229398e-05,
+      "loss": 0.8725,
+      "step": 7935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999242358877044e-05,
+      "loss": 1.1615,
+      "step": 7936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992421655000194e-05,
+      "loss": 1.2269,
+      "step": 7937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999241972098323e-05,
+      "loss": 1.1951,
+      "step": 7938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992417786719545e-05,
+      "loss": 1.058,
+      "step": 7939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999241585220915e-05,
+      "loss": 1.3327,
+      "step": 7940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999241391745205e-05,
+      "loss": 1.2085,
+      "step": 7941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999241198244822e-05,
+      "loss": 0.9829,
+      "step": 7942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992410047197685e-05,
+      "loss": 1.2403,
+      "step": 7943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999240811170043e-05,
+      "loss": 1.1999,
+      "step": 7944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999240617595646e-05,
+      "loss": 0.9268,
+      "step": 7945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992404239965786e-05,
+      "loss": 1.1453,
+      "step": 7946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992402303728395e-05,
+      "loss": 0.9478,
+      "step": 7947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999240036724427e-05,
+      "loss": 0.852,
+      "step": 7948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992398430513454e-05,
+      "loss": 1.402,
+      "step": 7949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999239649353591e-05,
+      "loss": 1.0803,
+      "step": 7950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999239455631166e-05,
+      "loss": 1.0025,
+      "step": 7951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999239261884069e-05,
+      "loss": 1.3306,
+      "step": 7952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999239068112302e-05,
+      "loss": 1.1896,
+      "step": 7953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999238874315861e-05,
+      "loss": 0.9684,
+      "step": 7954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99923868049475e-05,
+      "loss": 1.3475,
+      "step": 7955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992384866489676e-05,
+      "loss": 1.0657,
+      "step": 7956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992382927785135e-05,
+      "loss": 1.1366,
+      "step": 7957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992380988833884e-05,
+      "loss": 1.3813,
+      "step": 7958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992379049635916e-05,
+      "loss": 1.1277,
+      "step": 7959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999237711019124e-05,
+      "loss": 1.2634,
+      "step": 7960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999237517049984e-05,
+      "loss": 1.0193,
+      "step": 7961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999237323056173e-05,
+      "loss": 1.0104,
+      "step": 7962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99923712903769e-05,
+      "loss": 1.8146,
+      "step": 7963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999236934994536e-05,
+      "loss": 0.9461,
+      "step": 7964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992367409267105e-05,
+      "loss": 1.1662,
+      "step": 7965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999236546834214e-05,
+      "loss": 0.9277,
+      "step": 7966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999236352717046e-05,
+      "loss": 1.0035,
+      "step": 7967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999236158575206e-05,
+      "loss": 0.9615,
+      "step": 7968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999235964408695e-05,
+      "loss": 0.8536,
+      "step": 7969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999235770217513e-05,
+      "loss": 1.0511,
+      "step": 7970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999235576001659e-05,
+      "loss": 0.8384,
+      "step": 7971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992353817611336e-05,
+      "loss": 0.7573,
+      "step": 7972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999235187495938e-05,
+      "loss": 1.2414,
+      "step": 7973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992349932060696e-05,
+      "loss": 1.2377,
+      "step": 7974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99923479889153e-05,
+      "loss": 1.0569,
+      "step": 7975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992346045523194e-05,
+      "loss": 0.9342,
+      "step": 7976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999234410188437e-05,
+      "loss": 1.0519,
+      "step": 7977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999234215799884e-05,
+      "loss": 1.0635,
+      "step": 7978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992340213866585e-05,
+      "loss": 0.9977,
+      "step": 7979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999233826948763e-05,
+      "loss": 1.084,
+      "step": 7980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999233632486194e-05,
+      "loss": 1.1179,
+      "step": 7981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999233437998956e-05,
+      "loss": 1.1019,
+      "step": 7982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999233243487045e-05,
+      "loss": 0.6534,
+      "step": 7983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999233048950464e-05,
+      "loss": 1.2761,
+      "step": 7984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992328543892096e-05,
+      "loss": 0.9804,
+      "step": 7985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999232659803286e-05,
+      "loss": 0.9749,
+      "step": 7986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992324651926895e-05,
+      "loss": 1.5774,
+      "step": 7987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999232270557422e-05,
+      "loss": 1.1378,
+      "step": 7988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999232075897483e-05,
+      "loss": 0.9541,
+      "step": 7989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992318812128734e-05,
+      "loss": 1.1965,
+      "step": 7990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992316865035924e-05,
+      "loss": 0.8706,
+      "step": 7991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999231491769639e-05,
+      "loss": 0.7041,
+      "step": 7992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992312970110153e-05,
+      "loss": 0.6082,
+      "step": 7993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999231102227719e-05,
+      "loss": 1.1324,
+      "step": 7994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999230907419753e-05,
+      "loss": 1.0204,
+      "step": 7995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999230712587115e-05,
+      "loss": 0.9703,
+      "step": 7996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999230517729805e-05,
+      "loss": 1.0797,
+      "step": 7997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999230322847824e-05,
+      "loss": 1.1712,
+      "step": 7998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992301279411716e-05,
+      "loss": 1.0651,
+      "step": 7999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999229933009848e-05,
+      "loss": 0.9631,
+      "step": 8000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.054444432258606,
+      "eval_runtime": 84.4105,
+      "eval_samples_per_second": 16.408,
+      "eval_steps_per_second": 4.111,
+      "step": 8000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992297380538534e-05,
+      "loss": 1.1107,
+      "step": 8001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999229543073187e-05,
+      "loss": 1.0455,
+      "step": 8002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999229348067849e-05,
+      "loss": 1.0669,
+      "step": 8003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99922915303784e-05,
+      "loss": 1.3531,
+      "step": 8004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992289579831595e-05,
+      "loss": 1.266,
+      "step": 8005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999228762903808e-05,
+      "loss": 1.009,
+      "step": 8006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999228567799785e-05,
+      "loss": 0.7729,
+      "step": 8007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999228372671091e-05,
+      "loss": 1.1114,
+      "step": 8008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999228177517725e-05,
+      "loss": 0.8466,
+      "step": 8009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999227982339688e-05,
+      "loss": 1.5055,
+      "step": 8010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992277871369806e-05,
+      "loss": 1.2978,
+      "step": 8011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999227591909601e-05,
+      "loss": 1.2087,
+      "step": 8012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99922739665755e-05,
+      "loss": 1.776,
+      "step": 8013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999227201380827e-05,
+      "loss": 1.6039,
+      "step": 8014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999227006079434e-05,
+      "loss": 1.1489,
+      "step": 8015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999226810753369e-05,
+      "loss": 1.2858,
+      "step": 8016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999226615402632e-05,
+      "loss": 1.231,
+      "step": 8017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992264200272254e-05,
+      "loss": 1.1254,
+      "step": 8018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999226224627146e-05,
+      "loss": 1.0541,
+      "step": 8019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999226029202396e-05,
+      "loss": 1.0562,
+      "step": 8020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992258337529745e-05,
+      "loss": 1.3727,
+      "step": 8021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999225638278882e-05,
+      "loss": 1.0765,
+      "step": 8022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999225442780118e-05,
+      "loss": 1.0304,
+      "step": 8023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992252472566825e-05,
+      "loss": 0.962,
+      "step": 8024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999225051708576e-05,
+      "loss": 0.9971,
+      "step": 8025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999224856135798e-05,
+      "loss": 1.0339,
+      "step": 8026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992246605383485e-05,
+      "loss": 1.0693,
+      "step": 8027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992244649162276e-05,
+      "loss": 1.1078,
+      "step": 8028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999224269269436e-05,
+      "loss": 1.3662,
+      "step": 8029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999224073597973e-05,
+      "loss": 1.4999,
+      "step": 8030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992238779018386e-05,
+      "loss": 1.6307,
+      "step": 8031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992236821810336e-05,
+      "loss": 1.8442,
+      "step": 8032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992234864355556e-05,
+      "loss": 1.3682,
+      "step": 8033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999223290665408e-05,
+      "loss": 1.2394,
+      "step": 8034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992230948705884e-05,
+      "loss": 1.0143,
+      "step": 8035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999222899051097e-05,
+      "loss": 0.952,
+      "step": 8036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999222703206936e-05,
+      "loss": 0.9989,
+      "step": 8037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999222507338102e-05,
+      "loss": 1.2007,
+      "step": 8038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999222311444597e-05,
+      "loss": 1.1594,
+      "step": 8039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999222115526422e-05,
+      "loss": 1.6533,
+      "step": 8040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999221919583574e-05,
+      "loss": 1.2455,
+      "step": 8041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999221723616056e-05,
+      "loss": 1.1915,
+      "step": 8042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999221527623866e-05,
+      "loss": 0.9804,
+      "step": 8043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992213316070054e-05,
+      "loss": 1.0802,
+      "step": 8044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999221135565473e-05,
+      "loss": 1.2212,
+      "step": 8045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99922093949927e-05,
+      "loss": 1.0551,
+      "step": 8046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992207434083946e-05,
+      "loss": 1.2246,
+      "step": 8047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992205472928485e-05,
+      "loss": 1.357,
+      "step": 8048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992203511526314e-05,
+      "loss": 1.0874,
+      "step": 8049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999220154987743e-05,
+      "loss": 1.0338,
+      "step": 8050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992199587981835e-05,
+      "loss": 1.3102,
+      "step": 8051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999219762583952e-05,
+      "loss": 0.9823,
+      "step": 8052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99921956634505e-05,
+      "loss": 1.0771,
+      "step": 8053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999219370081477e-05,
+      "loss": 0.9605,
+      "step": 8054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999219173793232e-05,
+      "loss": 0.8811,
+      "step": 8055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992189774803163e-05,
+      "loss": 3.6941,
+      "step": 8056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999218781142729e-05,
+      "loss": 6.1163,
+      "step": 8057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999218584780471e-05,
+      "loss": 4.5305,
+      "step": 8058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999218388393541e-05,
+      "loss": 5.9376,
+      "step": 8059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992181919819403e-05,
+      "loss": 5.7547,
+      "step": 8060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992179955456676e-05,
+      "loss": 5.3154,
+      "step": 8061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992177990847245e-05,
+      "loss": 3.2164,
+      "step": 8062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99921760259911e-05,
+      "loss": 1.0417,
+      "step": 8063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999217406088824e-05,
+      "loss": 1.563,
+      "step": 8064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999217209553867e-05,
+      "loss": 1.1756,
+      "step": 8065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999217012994239e-05,
+      "loss": 1.4225,
+      "step": 8066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99921681640994e-05,
+      "loss": 1.0189,
+      "step": 8067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992166198009685e-05,
+      "loss": 0.7917,
+      "step": 8068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999216423167327e-05,
+      "loss": 1.1507,
+      "step": 8069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992162265090137e-05,
+      "loss": 1.0214,
+      "step": 8070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992160298260294e-05,
+      "loss": 0.8676,
+      "step": 8071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999215833118374e-05,
+      "loss": 1.2787,
+      "step": 8072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999215636386048e-05,
+      "loss": 1.5972,
+      "step": 8073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999215439629049e-05,
+      "loss": 1.3106,
+      "step": 8074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992152428473806e-05,
+      "loss": 1.2539,
+      "step": 8075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992150460410405e-05,
+      "loss": 1.1408,
+      "step": 8076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999214849210029e-05,
+      "loss": 1.209,
+      "step": 8077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999214652354346e-05,
+      "loss": 1.2816,
+      "step": 8078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992144554739915e-05,
+      "loss": 1.1278,
+      "step": 8079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999214258568967e-05,
+      "loss": 1.2074,
+      "step": 8080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99921406163927e-05,
+      "loss": 1.1355,
+      "step": 8081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999213864684903e-05,
+      "loss": 0.933,
+      "step": 8082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992136677058646e-05,
+      "loss": 1.1354,
+      "step": 8083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992134707021544e-05,
+      "loss": 1.0388,
+      "step": 8084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999213273673774e-05,
+      "loss": 3.3175,
+      "step": 8085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992130766207214e-05,
+      "loss": 1.2026,
+      "step": 8086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999212879542998e-05,
+      "loss": 1.2803,
+      "step": 8087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999212682440604e-05,
+      "loss": 1.1053,
+      "step": 8088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992124853135384e-05,
+      "loss": 1.3526,
+      "step": 8089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992122881618006e-05,
+      "loss": 1.3599,
+      "step": 8090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992120909853925e-05,
+      "loss": 0.9951,
+      "step": 8091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992118937843134e-05,
+      "loss": 1.264,
+      "step": 8092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999211696558563e-05,
+      "loss": 0.7493,
+      "step": 8093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999211499308142e-05,
+      "loss": 1.1717,
+      "step": 8094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992113020330486e-05,
+      "loss": 1.0709,
+      "step": 8095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999211104733285e-05,
+      "loss": 1.1257,
+      "step": 8096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99921090740885e-05,
+      "loss": 0.7699,
+      "step": 8097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999210710059743e-05,
+      "loss": 2.4386,
+      "step": 8098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999210512685966e-05,
+      "loss": 1.8469,
+      "step": 8099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999210315287518e-05,
+      "loss": 1.0731,
+      "step": 8100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999210117864398e-05,
+      "loss": 1.175,
+      "step": 8101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999209920416607e-05,
+      "loss": 1.1062,
+      "step": 8102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992097229441456e-05,
+      "loss": 0.981,
+      "step": 8103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992095254470115e-05,
+      "loss": 0.514,
+      "step": 8104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999209327925208e-05,
+      "loss": 0.6874,
+      "step": 8105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992091303787324e-05,
+      "loss": 1.045,
+      "step": 8106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999208932807586e-05,
+      "loss": 1.1178,
+      "step": 8107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992087352117684e-05,
+      "loss": 1.1132,
+      "step": 8108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999208537591279e-05,
+      "loss": 0.8647,
+      "step": 8109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99920833994612e-05,
+      "loss": 1.5936,
+      "step": 8110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992081422762885e-05,
+      "loss": 0.9714,
+      "step": 8111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999207944581786e-05,
+      "loss": 0.8233,
+      "step": 8112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999207746862613e-05,
+      "loss": 1.031,
+      "step": 8113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999207549118768e-05,
+      "loss": 0.4453,
+      "step": 8114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999207351350252e-05,
+      "loss": 0.3886,
+      "step": 8115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992071535570665e-05,
+      "loss": 1.0795,
+      "step": 8116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992069557392085e-05,
+      "loss": 0.8687,
+      "step": 8117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992067578966794e-05,
+      "loss": 1.0988,
+      "step": 8118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992065600294794e-05,
+      "loss": 1.2491,
+      "step": 8119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999206362137608e-05,
+      "loss": 1.3787,
+      "step": 8120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992061642210655e-05,
+      "loss": 1.224,
+      "step": 8121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992059662798524e-05,
+      "loss": 1.234,
+      "step": 8122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999205768313968e-05,
+      "loss": 1.1286,
+      "step": 8123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999205570323412e-05,
+      "loss": 0.8692,
+      "step": 8124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992053723081856e-05,
+      "loss": 1.1534,
+      "step": 8125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999205174268288e-05,
+      "loss": 1.1841,
+      "step": 8126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999204976203719e-05,
+      "loss": 1.2964,
+      "step": 8127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999204778114479e-05,
+      "loss": 1.3143,
+      "step": 8128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999204580000568e-05,
+      "loss": 1.3841,
+      "step": 8129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999204381861985e-05,
+      "loss": 1.3286,
+      "step": 8130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992041836987323e-05,
+      "loss": 1.1715,
+      "step": 8131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999203985510808e-05,
+      "loss": 1.2359,
+      "step": 8132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999203787298212e-05,
+      "loss": 1.3515,
+      "step": 8133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999203589060946e-05,
+      "loss": 1.2288,
+      "step": 8134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999203390799008e-05,
+      "loss": 1.0974,
+      "step": 8135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992031925123994e-05,
+      "loss": 1.0515,
+      "step": 8136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992029942011197e-05,
+      "loss": 1.1022,
+      "step": 8137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999202795865169e-05,
+      "loss": 1.0654,
+      "step": 8138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992025975045464e-05,
+      "loss": 1.0587,
+      "step": 8139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992023991192535e-05,
+      "loss": 0.9211,
+      "step": 8140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99920220070929e-05,
+      "loss": 1.0319,
+      "step": 8141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999202002274654e-05,
+      "loss": 1.1965,
+      "step": 8142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999201803815348e-05,
+      "loss": 1.1252,
+      "step": 8143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992016053313706e-05,
+      "loss": 1.1956,
+      "step": 8144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999201406822722e-05,
+      "loss": 1.0656,
+      "step": 8145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999201208289403e-05,
+      "loss": 1.1604,
+      "step": 8146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992010097314124e-05,
+      "loss": 0.9163,
+      "step": 8147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999200811148751e-05,
+      "loss": 1.1694,
+      "step": 8148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999200612541418e-05,
+      "loss": 1.1699,
+      "step": 8149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999200413909414e-05,
+      "loss": 1.3381,
+      "step": 8150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992002152527395e-05,
+      "loss": 1.2251,
+      "step": 8151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999200016571394e-05,
+      "loss": 1.0544,
+      "step": 8152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999199817865377e-05,
+      "loss": 1.4957,
+      "step": 8153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999199619134689e-05,
+      "loss": 1.6528,
+      "step": 8154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991994203793303e-05,
+      "loss": 1.618,
+      "step": 8155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991992215993e-05,
+      "loss": 1.1554,
+      "step": 8156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991990227945996e-05,
+      "loss": 1.1553,
+      "step": 8157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999198823965228e-05,
+      "loss": 1.0059,
+      "step": 8158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999198625111184e-05,
+      "loss": 1.2516,
+      "step": 8159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991984262324696e-05,
+      "loss": 1.2199,
+      "step": 8160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991982273290846e-05,
+      "loss": 1.0004,
+      "step": 8161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991980284010287e-05,
+      "loss": 1.0749,
+      "step": 8162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999197829448302e-05,
+      "loss": 0.9781,
+      "step": 8163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999197630470903e-05,
+      "loss": 0.3949,
+      "step": 8164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999197431468834e-05,
+      "loss": 1.1743,
+      "step": 8165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999197232442093e-05,
+      "loss": 0.8457,
+      "step": 8166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999197033390682e-05,
+      "loss": 0.8697,
+      "step": 8167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991968343146e-05,
+      "loss": 1.1964,
+      "step": 8168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999196635213846e-05,
+      "loss": 0.86,
+      "step": 8169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999196436088422e-05,
+      "loss": 1.0926,
+      "step": 8170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999196236938327e-05,
+      "loss": 1.0888,
+      "step": 8171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999196037763561e-05,
+      "loss": 0.7222,
+      "step": 8172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999195838564123e-05,
+      "loss": 1.1379,
+      "step": 8173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999195639340015e-05,
+      "loss": 1.0324,
+      "step": 8174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991954400912355e-05,
+      "loss": 1.1589,
+      "step": 8175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991952408177846e-05,
+      "loss": 1.0928,
+      "step": 8176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999195041519663e-05,
+      "loss": 0.9772,
+      "step": 8177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999194842196871e-05,
+      "loss": 1.2171,
+      "step": 8178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999194642849408e-05,
+      "loss": 1.1504,
+      "step": 8179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991944434772734e-05,
+      "loss": 1.19,
+      "step": 8180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999194244080467e-05,
+      "loss": 1.1003,
+      "step": 8181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991940446589916e-05,
+      "loss": 1.0247,
+      "step": 8182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999193845212844e-05,
+      "loss": 1.1476,
+      "step": 8183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999193645742025e-05,
+      "loss": 1.2773,
+      "step": 8184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991934462465364e-05,
+      "loss": 1.5125,
+      "step": 8185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999193246726376e-05,
+      "loss": 1.5328,
+      "step": 8186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999193047181544e-05,
+      "loss": 0.9536,
+      "step": 8187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999192847612042e-05,
+      "loss": 0.8888,
+      "step": 8188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991926480178684e-05,
+      "loss": 1.2458,
+      "step": 8189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991924483990246e-05,
+      "loss": 1.0091,
+      "step": 8190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999192248755509e-05,
+      "loss": 0.1777,
+      "step": 8191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999192049087323e-05,
+      "loss": 0.1513,
+      "step": 8192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991918493944655e-05,
+      "loss": 0.2262,
+      "step": 8193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991916496769376e-05,
+      "loss": 1.1995,
+      "step": 8194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999191449934738e-05,
+      "loss": 1.1373,
+      "step": 8195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991912501678686e-05,
+      "loss": 1.0447,
+      "step": 8196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991910503763276e-05,
+      "loss": 1.281,
+      "step": 8197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999190850560115e-05,
+      "loss": 0.925,
+      "step": 8198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999190650719232e-05,
+      "loss": 1.2642,
+      "step": 8199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999190450853678e-05,
+      "loss": 1.0075,
+      "step": 8200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991902509634534e-05,
+      "loss": 1.1623,
+      "step": 8201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999190051048557e-05,
+      "loss": 0.9961,
+      "step": 8202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999189851108991e-05,
+      "loss": 1.1757,
+      "step": 8203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999189651144753e-05,
+      "loss": 1.3607,
+      "step": 8204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999189451155844e-05,
+      "loss": 1.2876,
+      "step": 8205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999189251142265e-05,
+      "loss": 1.4278,
+      "step": 8206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999189051104014e-05,
+      "loss": 1.3018,
+      "step": 8207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999188851041092e-05,
+      "loss": 1.1068,
+      "step": 8208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991886509535e-05,
+      "loss": 1.1355,
+      "step": 8209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991884508412364e-05,
+      "loss": 0.9839,
+      "step": 8210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999188250704302e-05,
+      "loss": 1.1312,
+      "step": 8211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999188050542697e-05,
+      "loss": 0.9444,
+      "step": 8212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99918785035642e-05,
+      "loss": 0.7048,
+      "step": 8213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999187650145474e-05,
+      "loss": 0.9837,
+      "step": 8214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999187449909856e-05,
+      "loss": 1.1194,
+      "step": 8215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999187249649566e-05,
+      "loss": 1.1048,
+      "step": 8216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999187049364607e-05,
+      "loss": 0.9613,
+      "step": 8217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999186849054975e-05,
+      "loss": 1.1043,
+      "step": 8218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999186648720674e-05,
+      "loss": 1.1788,
+      "step": 8219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999186448361701e-05,
+      "loss": 1.0607,
+      "step": 8220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999186247978058e-05,
+      "loss": 1.3091,
+      "step": 8221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999186047569743e-05,
+      "loss": 1.0853,
+      "step": 8222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991858471367584e-05,
+      "loss": 0.5792,
+      "step": 8223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999185646679101e-05,
+      "loss": 0.6422,
+      "step": 8224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999185446196774e-05,
+      "loss": 2.3584,
+      "step": 8225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999185245689776e-05,
+      "loss": 2.8526,
+      "step": 8226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999185045158107e-05,
+      "loss": 2.4602,
+      "step": 8227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999184844601767e-05,
+      "loss": 2.0139,
+      "step": 8228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999184644020756e-05,
+      "loss": 1.1379,
+      "step": 8229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999184443415075e-05,
+      "loss": 1.3758,
+      "step": 8230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999184242784722e-05,
+      "loss": 1.0097,
+      "step": 8231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991840421296984e-05,
+      "loss": 1.2272,
+      "step": 8232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991838414500034e-05,
+      "loss": 1.3314,
+      "step": 8233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999183640745638e-05,
+      "loss": 0.9933,
+      "step": 8234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999183440016603e-05,
+      "loss": 0.9973,
+      "step": 8235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991832392628955e-05,
+      "loss": 1.0917,
+      "step": 8236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999183038484517e-05,
+      "loss": 0.9456,
+      "step": 8237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999182837681469e-05,
+      "loss": 1.409,
+      "step": 8238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991826368537485e-05,
+      "loss": 1.0647,
+      "step": 8239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991824360013584e-05,
+      "loss": 0.9981,
+      "step": 8240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991822351242965e-05,
+      "loss": 1.1877,
+      "step": 8241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991820342225644e-05,
+      "loss": 1.3668,
+      "step": 8242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999181833296162e-05,
+      "loss": 1.2226,
+      "step": 8243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999181632345087e-05,
+      "loss": 1.1048,
+      "step": 8244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999181431369342e-05,
+      "loss": 0.9423,
+      "step": 8245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999181230368926e-05,
+      "loss": 1.2384,
+      "step": 8246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991810293438395e-05,
+      "loss": 1.7832,
+      "step": 8247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999180828294082e-05,
+      "loss": 0.9334,
+      "step": 8248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999180627219653e-05,
+      "loss": 1.1745,
+      "step": 8249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999180426120554e-05,
+      "loss": 0.8047,
+      "step": 8250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999180224996784e-05,
+      "loss": 1.095,
+      "step": 8251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999180023848343e-05,
+      "loss": 0.8205,
+      "step": 8252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999179822675231e-05,
+      "loss": 1.1301,
+      "step": 8253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999179621477449e-05,
+      "loss": 1.3518,
+      "step": 8254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991794202549954e-05,
+      "loss": 1.0477,
+      "step": 8255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999179219007871e-05,
+      "loss": 0.4998,
+      "step": 8256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991790177360755e-05,
+      "loss": 1.0548,
+      "step": 8257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999178816439609e-05,
+      "loss": 1.2165,
+      "step": 8258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999178615118473e-05,
+      "loss": 0.8835,
+      "step": 8259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999178413772665e-05,
+      "loss": 1.2443,
+      "step": 8260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999178212402186e-05,
+      "loss": 0.9742,
+      "step": 8261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999178011007037e-05,
+      "loss": 1.2147,
+      "step": 8262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999177809587217e-05,
+      "loss": 1.1196,
+      "step": 8263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991776081427255e-05,
+      "loss": 1.172,
+      "step": 8264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999177406673564e-05,
+      "loss": 1.1525,
+      "step": 8265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991772051797306e-05,
+      "loss": 0.9987,
+      "step": 8266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999177003661227e-05,
+      "loss": 1.1509,
+      "step": 8267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999176802118053e-05,
+      "loss": 1.2156,
+      "step": 8268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999176600550208e-05,
+      "loss": 1.0428,
+      "step": 8269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999176398957692e-05,
+      "loss": 1.0568,
+      "step": 8270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999176197340505e-05,
+      "loss": 1.1073,
+      "step": 8271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999175995698647e-05,
+      "loss": 1.2529,
+      "step": 8272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991757940321186e-05,
+      "loss": 1.3799,
+      "step": 8273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999175592340919e-05,
+      "loss": 1.3727,
+      "step": 8274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991753906250494e-05,
+      "loss": 1.2017,
+      "step": 8275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991751888845086e-05,
+      "loss": 0.9024,
+      "step": 8276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999174987119297e-05,
+      "loss": 0.7042,
+      "step": 8277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999174785329415e-05,
+      "loss": 0.962,
+      "step": 8278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999174583514861e-05,
+      "loss": 1.1191,
+      "step": 8279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991743816756367e-05,
+      "loss": 1.3123,
+      "step": 8280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999174179811742e-05,
+      "loss": 0.7662,
+      "step": 8281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999173977923176e-05,
+      "loss": 0.8643,
+      "step": 8282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99917377600994e-05,
+      "loss": 1.2898,
+      "step": 8283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991735740720324e-05,
+      "loss": 1.2237,
+      "step": 8284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991733721094545e-05,
+      "loss": 1.3147,
+      "step": 8285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991731701222056e-05,
+      "loss": 1.1869,
+      "step": 8286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999172968110286e-05,
+      "loss": 1.265,
+      "step": 8287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999172766073695e-05,
+      "loss": 1.3549,
+      "step": 8288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999172564012434e-05,
+      "loss": 1.2528,
+      "step": 8289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999172361926502e-05,
+      "loss": 1.1335,
+      "step": 8290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991721598158996e-05,
+      "loss": 1.05,
+      "step": 8291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999171957680626e-05,
+      "loss": 1.0144,
+      "step": 8292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999171755520682e-05,
+      "loss": 0.9581,
+      "step": 8293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991715533360674e-05,
+      "loss": 1.1402,
+      "step": 8294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999171351126781e-05,
+      "loss": 1.1666,
+      "step": 8295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999171148892824e-05,
+      "loss": 0.9455,
+      "step": 8296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991709466341974e-05,
+      "loss": 1.3822,
+      "step": 8297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999170744350899e-05,
+      "loss": 1.3498,
+      "step": 8298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99917054204293e-05,
+      "loss": 1.2778,
+      "step": 8299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999170339710291e-05,
+      "loss": 1.682,
+      "step": 8300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99917013735298e-05,
+      "loss": 1.2964,
+      "step": 8301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999169934970999e-05,
+      "loss": 1.1784,
+      "step": 8302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999169732564347e-05,
+      "loss": 1.1179,
+      "step": 8303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999169530133024e-05,
+      "loss": 1.1623,
+      "step": 8304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999169327677031e-05,
+      "loss": 1.2041,
+      "step": 8305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991691251963665e-05,
+      "loss": 1.1615,
+      "step": 8306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999168922691032e-05,
+      "loss": 1.0726,
+      "step": 8307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999168720161026e-05,
+      "loss": 0.8384,
+      "step": 8308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99916851760635e-05,
+      "loss": 0.9102,
+      "step": 8309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999168315027003e-05,
+      "loss": 0.8444,
+      "step": 8310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999168112422985e-05,
+      "loss": 0.7279,
+      "step": 8311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999167909794296e-05,
+      "loss": 0.5946,
+      "step": 8312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999167707140937e-05,
+      "loss": 0.8794,
+      "step": 8313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999167504462907e-05,
+      "loss": 1.1536,
+      "step": 8314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999167301760206e-05,
+      "loss": 0.9535,
+      "step": 8315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991670990328346e-05,
+      "loss": 1.1651,
+      "step": 8316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999166896280793e-05,
+      "loss": 0.8064,
+      "step": 8317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991666935040793e-05,
+      "loss": 0.8431,
+      "step": 8318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991664907026955e-05,
+      "loss": 1.3013,
+      "step": 8319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999166287876641e-05,
+      "loss": 1.0501,
+      "step": 8320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999166085025916e-05,
+      "loss": 0.9354,
+      "step": 8321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999165882150521e-05,
+      "loss": 0.9781,
+      "step": 8322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991656792504535e-05,
+      "loss": 0.8342,
+      "step": 8323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991654763257166e-05,
+      "loss": 1.1559,
+      "step": 8324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999165273376309e-05,
+      "loss": 1.1642,
+      "step": 8325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99916507040223e-05,
+      "loss": 1.2031,
+      "step": 8326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991648674034805e-05,
+      "loss": 1.2286,
+      "step": 8327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99916466438006e-05,
+      "loss": 1.1127,
+      "step": 8328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99916446133197e-05,
+      "loss": 1.3282,
+      "step": 8329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999164258259208e-05,
+      "loss": 1.2678,
+      "step": 8330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991640551617766e-05,
+      "loss": 1.3532,
+      "step": 8331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991638520396736e-05,
+      "loss": 1.1879,
+      "step": 8332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991636488928996e-05,
+      "loss": 1.2062,
+      "step": 8333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999163445721455e-05,
+      "loss": 1.2149,
+      "step": 8334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999163242525341e-05,
+      "loss": 1.1491,
+      "step": 8335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991630393045554e-05,
+      "loss": 1.0911,
+      "step": 8336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991628360590986e-05,
+      "loss": 1.0811,
+      "step": 8337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991626327889715e-05,
+      "loss": 1.2723,
+      "step": 8338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999162429494174e-05,
+      "loss": 0.9685,
+      "step": 8339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991622261747056e-05,
+      "loss": 0.91,
+      "step": 8340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999162022830567e-05,
+      "loss": 1.0759,
+      "step": 8341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991618194617577e-05,
+      "loss": 1.2971,
+      "step": 8342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999161616068276e-05,
+      "loss": 1.0533,
+      "step": 8343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991614126501256e-05,
+      "loss": 0.9694,
+      "step": 8344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991612092073034e-05,
+      "loss": 1.0244,
+      "step": 8345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999161005739812e-05,
+      "loss": 1.284,
+      "step": 8346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991608022476486e-05,
+      "loss": 1.184,
+      "step": 8347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999160598730815e-05,
+      "loss": 1.2793,
+      "step": 8348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991603951893104e-05,
+      "loss": 0.9557,
+      "step": 8349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999160191623136e-05,
+      "loss": 1.1593,
+      "step": 8350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991599880322895e-05,
+      "loss": 1.137,
+      "step": 8351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991597844167736e-05,
+      "loss": 0.6558,
+      "step": 8352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991595807765866e-05,
+      "loss": 1.1623,
+      "step": 8353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991593771117286e-05,
+      "loss": 1.269,
+      "step": 8354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999159173422201e-05,
+      "loss": 1.1349,
+      "step": 8355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991589697080015e-05,
+      "loss": 1.1626,
+      "step": 8356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999158765969132e-05,
+      "loss": 1.279,
+      "step": 8357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999158562205592e-05,
+      "loss": 1.2476,
+      "step": 8358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999158358417381e-05,
+      "loss": 1.1814,
+      "step": 8359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991581546045e-05,
+      "loss": 1.097,
+      "step": 8360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991579507669475e-05,
+      "loss": 1.1707,
+      "step": 8361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999157746904725e-05,
+      "loss": 1.0193,
+      "step": 8362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991575430178316e-05,
+      "loss": 0.79,
+      "step": 8363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991573391062674e-05,
+      "loss": 1.1222,
+      "step": 8364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999157135170033e-05,
+      "loss": 1.1173,
+      "step": 8365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991569312091274e-05,
+      "loss": 1.5073,
+      "step": 8366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991567272235516e-05,
+      "loss": 0.9067,
+      "step": 8367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991565232133054e-05,
+      "loss": 0.6524,
+      "step": 8368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991563191783875e-05,
+      "loss": 0.7113,
+      "step": 8369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991561151188e-05,
+      "loss": 0.7702,
+      "step": 8370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991559110345413e-05,
+      "loss": 1.1213,
+      "step": 8371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991557069256124e-05,
+      "loss": 1.0213,
+      "step": 8372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999155502792013e-05,
+      "loss": 1.1007,
+      "step": 8373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999155298633743e-05,
+      "loss": 1.0572,
+      "step": 8374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999155094450802e-05,
+      "loss": 0.9129,
+      "step": 8375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991548902431906e-05,
+      "loss": 1.0989,
+      "step": 8376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991546860109086e-05,
+      "loss": 0.918,
+      "step": 8377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999154481753956e-05,
+      "loss": 1.1115,
+      "step": 8378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999154277472333e-05,
+      "loss": 1.1857,
+      "step": 8379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999154073166039e-05,
+      "loss": 1.066,
+      "step": 8380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991538688350745e-05,
+      "loss": 0.7562,
+      "step": 8381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991536644794394e-05,
+      "loss": 1.1435,
+      "step": 8382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999153460099134e-05,
+      "loss": 0.8951,
+      "step": 8383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991532556941576e-05,
+      "loss": 1.1395,
+      "step": 8384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999153051264511e-05,
+      "loss": 0.9641,
+      "step": 8385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999152846810193e-05,
+      "loss": 1.1511,
+      "step": 8386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999152642331205e-05,
+      "loss": 0.9422,
+      "step": 8387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991524378275464e-05,
+      "loss": 1.1244,
+      "step": 8388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999152233299218e-05,
+      "loss": 1.1136,
+      "step": 8389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999152028746218e-05,
+      "loss": 1.1821,
+      "step": 8390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991518241685476e-05,
+      "loss": 1.1484,
+      "step": 8391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999151619566207e-05,
+      "loss": 0.9898,
+      "step": 8392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991514149391955e-05,
+      "loss": 0.9929,
+      "step": 8393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991512102875137e-05,
+      "loss": 1.2335,
+      "step": 8394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991510056111614e-05,
+      "loss": 0.8588,
+      "step": 8395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999150800910138e-05,
+      "loss": 0.5444,
+      "step": 8396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991505961844446e-05,
+      "loss": 0.6808,
+      "step": 8397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99915039143408e-05,
+      "loss": 0.3225,
+      "step": 8398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999150186659045e-05,
+      "loss": 0.5494,
+      "step": 8399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991499818593404e-05,
+      "loss": 0.6306,
+      "step": 8400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999149777034964e-05,
+      "loss": 1.2615,
+      "step": 8401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991495721859174e-05,
+      "loss": 1.2836,
+      "step": 8402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991493673122004e-05,
+      "loss": 1.2379,
+      "step": 8403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999149162413813e-05,
+      "loss": 1.2976,
+      "step": 8404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999148957490755e-05,
+      "loss": 0.9837,
+      "step": 8405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999148752543026e-05,
+      "loss": 1.1341,
+      "step": 8406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999148547570627e-05,
+      "loss": 0.6938,
+      "step": 8407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991483425735575e-05,
+      "loss": 1.1328,
+      "step": 8408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991481375518165e-05,
+      "loss": 1.086,
+      "step": 8409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999147932505406e-05,
+      "loss": 1.2981,
+      "step": 8410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991477274343246e-05,
+      "loss": 1.4599,
+      "step": 8411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999147522338573e-05,
+      "loss": 1.4413,
+      "step": 8412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99914731721815e-05,
+      "loss": 0.7609,
+      "step": 8413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999147112073057e-05,
+      "loss": 0.608,
+      "step": 8414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991469069032934e-05,
+      "loss": 0.3055,
+      "step": 8415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99914670170886e-05,
+      "loss": 0.4677,
+      "step": 8416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999146496489755e-05,
+      "loss": 0.4433,
+      "step": 8417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991462912459805e-05,
+      "loss": 0.3726,
+      "step": 8418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991460859775346e-05,
+      "loss": 0.2769,
+      "step": 8419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991458806844184e-05,
+      "loss": 0.3368,
+      "step": 8420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999145675366632e-05,
+      "loss": 0.3348,
+      "step": 8421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999145470024175e-05,
+      "loss": 0.3991,
+      "step": 8422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999145264657047e-05,
+      "loss": 0.9928,
+      "step": 8423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999145059265249e-05,
+      "loss": 1.1942,
+      "step": 8424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999144853848781e-05,
+      "loss": 1.1893,
+      "step": 8425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999144648407641e-05,
+      "loss": 0.9798,
+      "step": 8426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999144442941832e-05,
+      "loss": 0.7083,
+      "step": 8427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991442374513516e-05,
+      "loss": 0.942,
+      "step": 8428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999144031936201e-05,
+      "loss": 1.0437,
+      "step": 8429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99914382639638e-05,
+      "loss": 0.7463,
+      "step": 8430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999143620831888e-05,
+      "loss": 0.7416,
+      "step": 8431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999143415242726e-05,
+      "loss": 1.4202,
+      "step": 8432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991432096288935e-05,
+      "loss": 2.1545,
+      "step": 8433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991430039903904e-05,
+      "loss": 2.3004,
+      "step": 8434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991427983272164e-05,
+      "loss": 1.0673,
+      "step": 8435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991425926393726e-05,
+      "loss": 1.0957,
+      "step": 8436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991423869268585e-05,
+      "loss": 0.9421,
+      "step": 8437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991421811896734e-05,
+      "loss": 0.3061,
+      "step": 8438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999141975427817e-05,
+      "loss": 1.0725,
+      "step": 8439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991417696412915e-05,
+      "loss": 1.3373,
+      "step": 8440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999141563830095e-05,
+      "loss": 1.6996,
+      "step": 8441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999141357994228e-05,
+      "loss": 1.1349,
+      "step": 8442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999141152133691e-05,
+      "loss": 1.2943,
+      "step": 8443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999140946248483e-05,
+      "loss": 1.1657,
+      "step": 8444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999140740338604e-05,
+      "loss": 1.0395,
+      "step": 8445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991405344040555e-05,
+      "loss": 1.4422,
+      "step": 8446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999140328444836e-05,
+      "loss": 1.5469,
+      "step": 8447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999140122460946e-05,
+      "loss": 1.4338,
+      "step": 8448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999139916452386e-05,
+      "loss": 1.5494,
+      "step": 8449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999139710419155e-05,
+      "loss": 1.0026,
+      "step": 8450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991395043612544e-05,
+      "loss": 0.7611,
+      "step": 8451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991392982786825e-05,
+      "loss": 0.9159,
+      "step": 8452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999139092171441e-05,
+      "loss": 1.0654,
+      "step": 8453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999138886039528e-05,
+      "loss": 1.3249,
+      "step": 8454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999138679882945e-05,
+      "loss": 1.4767,
+      "step": 8455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999138473701692e-05,
+      "loss": 1.1754,
+      "step": 8456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999138267495769e-05,
+      "loss": 1.2345,
+      "step": 8457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999138061265174e-05,
+      "loss": 1.1498,
+      "step": 8458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991378550099093e-05,
+      "loss": 1.229,
+      "step": 8459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999137648729975e-05,
+      "loss": 1.3139,
+      "step": 8460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999137442425369e-05,
+      "loss": 0.9882,
+      "step": 8461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999137236096093e-05,
+      "loss": 1.2208,
+      "step": 8462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991370297421466e-05,
+      "loss": 1.1347,
+      "step": 8463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999136823363529e-05,
+      "loss": 1.2023,
+      "step": 8464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999136616960243e-05,
+      "loss": 1.6499,
+      "step": 8465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999136410532285e-05,
+      "loss": 1.1429,
+      "step": 8466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999136204079656e-05,
+      "loss": 1.0542,
+      "step": 8467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991359976023575e-05,
+      "loss": 1.0439,
+      "step": 8468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999135791100389e-05,
+      "loss": 0.9817,
+      "step": 8469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991355845737496e-05,
+      "loss": 0.9858,
+      "step": 8470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991353780224405e-05,
+      "loss": 1.2853,
+      "step": 8471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991351714464596e-05,
+      "loss": 1.2696,
+      "step": 8472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999134964845809e-05,
+      "loss": 1.5929,
+      "step": 8473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999134758220488e-05,
+      "loss": 1.0557,
+      "step": 8474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999134551570497e-05,
+      "loss": 1.1169,
+      "step": 8475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999134344895835e-05,
+      "loss": 0.8522,
+      "step": 8476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999134138196503e-05,
+      "loss": 1.2347,
+      "step": 8477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991339314725e-05,
+      "loss": 1.1851,
+      "step": 8478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999133724723827e-05,
+      "loss": 1.2193,
+      "step": 8479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999133517950484e-05,
+      "loss": 0.9337,
+      "step": 8480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991333111524695e-05,
+      "loss": 1.1871,
+      "step": 8481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999133104329786e-05,
+      "loss": 1.1762,
+      "step": 8482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991328974824314e-05,
+      "loss": 1.2306,
+      "step": 8483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991326906104064e-05,
+      "loss": 0.8215,
+      "step": 8484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999132483713711e-05,
+      "loss": 1.1186,
+      "step": 8485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991322767923455e-05,
+      "loss": 1.0112,
+      "step": 8486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999132069846309e-05,
+      "loss": 1.1783,
+      "step": 8487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991318628756026e-05,
+      "loss": 1.3128,
+      "step": 8488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999131655880226e-05,
+      "loss": 1.1166,
+      "step": 8489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999131448860178e-05,
+      "loss": 0.995,
+      "step": 8490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999131241815461e-05,
+      "loss": 1.2413,
+      "step": 8491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999131034746073e-05,
+      "loss": 1.0379,
+      "step": 8492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991308276520145e-05,
+      "loss": 1.1876,
+      "step": 8493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991306205332855e-05,
+      "loss": 1.2118,
+      "step": 8494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999130413389887e-05,
+      "loss": 1.316,
+      "step": 8495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999130206221817e-05,
+      "loss": 0.973,
+      "step": 8496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999129999029078e-05,
+      "loss": 1.3465,
+      "step": 8497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999129791811667e-05,
+      "loss": 0.9186,
+      "step": 8498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991295845695865e-05,
+      "loss": 1.2283,
+      "step": 8499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999129377302836e-05,
+      "loss": 1.1846,
+      "step": 8500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0542309284210205,
+      "eval_runtime": 99.0013,
+      "eval_samples_per_second": 13.99,
+      "eval_steps_per_second": 3.505,
+      "step": 8500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999129170011415e-05,
+      "loss": 0.6902,
+      "step": 8501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999128962695323e-05,
+      "loss": 0.8112,
+      "step": 8502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991287553545614e-05,
+      "loss": 1.0512,
+      "step": 8503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999128547989129e-05,
+      "loss": 1.0738,
+      "step": 8504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999128340599026e-05,
+      "loss": 1.0696,
+      "step": 8505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999128133184253e-05,
+      "loss": 0.9503,
+      "step": 8506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99912792574481e-05,
+      "loss": 1.1886,
+      "step": 8507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999127718280697e-05,
+      "loss": 1.2016,
+      "step": 8508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999127510791913e-05,
+      "loss": 1.1265,
+      "step": 8509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999127303278458e-05,
+      "loss": 1.1183,
+      "step": 8510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991270957403334e-05,
+      "loss": 1.6124,
+      "step": 8511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999126888177539e-05,
+      "loss": 1.1347,
+      "step": 8512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999126680590074e-05,
+      "loss": 1.2228,
+      "step": 8513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991264729779374e-05,
+      "loss": 1.1211,
+      "step": 8514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999126265341132e-05,
+      "loss": 1.1596,
+      "step": 8515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999126057679655e-05,
+      "loss": 0.8949,
+      "step": 8516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999125849993509e-05,
+      "loss": 0.7996,
+      "step": 8517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999125642282692e-05,
+      "loss": 0.7547,
+      "step": 8518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999125434547205e-05,
+      "loss": 0.7877,
+      "step": 8519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991252267870475e-05,
+      "loss": 0.7012,
+      "step": 8520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99912501900222e-05,
+      "loss": 0.7266,
+      "step": 8521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999124811192721e-05,
+      "loss": 0.7987,
+      "step": 8522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999124603358553e-05,
+      "loss": 0.6141,
+      "step": 8523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999124395499715e-05,
+      "loss": 0.6994,
+      "step": 8524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991241876162053e-05,
+      "loss": 0.6509,
+      "step": 8525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999123979708026e-05,
+      "loss": 0.8349,
+      "step": 8526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991237717751757e-05,
+      "loss": 0.3005,
+      "step": 8527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991235638176556e-05,
+      "loss": 1.373,
+      "step": 8528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999123355835466e-05,
+      "loss": 1.057,
+      "step": 8529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999123147828605e-05,
+      "loss": 1.438,
+      "step": 8530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999122939797075e-05,
+      "loss": 1.1025,
+      "step": 8531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991227317408735e-05,
+      "loss": 1.017,
+      "step": 8532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991225236600024e-05,
+      "loss": 1.1234,
+      "step": 8533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999122315554461e-05,
+      "loss": 1.251,
+      "step": 8534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991221074242486e-05,
+      "loss": 1.1081,
+      "step": 8535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999121899269366e-05,
+      "loss": 1.1494,
+      "step": 8536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991216910898134e-05,
+      "loss": 1.2125,
+      "step": 8537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991214828855907e-05,
+      "loss": 1.5262,
+      "step": 8538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999121274656698e-05,
+      "loss": 0.9857,
+      "step": 8539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999121066403134e-05,
+      "loss": 1.1487,
+      "step": 8540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991208581249e-05,
+      "loss": 1.0883,
+      "step": 8541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999120649821997e-05,
+      "loss": 1.5234,
+      "step": 8542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999120441494422e-05,
+      "loss": 2.2477,
+      "step": 8543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999120233142178e-05,
+      "loss": 2.9172,
+      "step": 8544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999120024765264e-05,
+      "loss": 2.8604,
+      "step": 8545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999119816363679e-05,
+      "loss": 2.2149,
+      "step": 8546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999119607937424e-05,
+      "loss": 0.9414,
+      "step": 8547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999119399486498e-05,
+      "loss": 1.1388,
+      "step": 8548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999119191010902e-05,
+      "loss": 0.8004,
+      "step": 8549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999118982510637e-05,
+      "loss": 1.2408,
+      "step": 8550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991187739857e-05,
+      "loss": 1.2502,
+      "step": 8551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999118565436094e-05,
+      "loss": 1.3305,
+      "step": 8552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991183568618175e-05,
+      "loss": 0.9993,
+      "step": 8553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99911814826287e-05,
+      "loss": 0.1552,
+      "step": 8554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991179396392526e-05,
+      "loss": 0.1205,
+      "step": 8555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999117730990965e-05,
+      "loss": 0.1116,
+      "step": 8556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999117522318008e-05,
+      "loss": 1.049,
+      "step": 8557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991173136203795e-05,
+      "loss": 1.0186,
+      "step": 8558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991171048980816e-05,
+      "loss": 1.154,
+      "step": 8559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999116896151114e-05,
+      "loss": 1.5328,
+      "step": 8560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999116687379475e-05,
+      "loss": 1.1713,
+      "step": 8561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999116478583166e-05,
+      "loss": 1.2756,
+      "step": 8562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999116269762187e-05,
+      "loss": 1.2244,
+      "step": 8563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991160609165374e-05,
+      "loss": 1.2125,
+      "step": 8564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999115852046218e-05,
+      "loss": 1.0166,
+      "step": 8565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991156431512285e-05,
+      "loss": 0.795,
+      "step": 8566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991154342315685e-05,
+      "loss": 1.2471,
+      "step": 8567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999115225287239e-05,
+      "loss": 0.8797,
+      "step": 8568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991150163182374e-05,
+      "loss": 0.3646,
+      "step": 8569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999114807324568e-05,
+      "loss": 1.0403,
+      "step": 8570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999114598306227e-05,
+      "loss": 1.2882,
+      "step": 8571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999114389263216e-05,
+      "loss": 1.0653,
+      "step": 8572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991141801955347e-05,
+      "loss": 1.1562,
+      "step": 8573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999113971103183e-05,
+      "loss": 1.317,
+      "step": 8574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991137619861615e-05,
+      "loss": 1.2079,
+      "step": 8575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99911355284447e-05,
+      "loss": 1.2094,
+      "step": 8576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999113343678108e-05,
+      "loss": 1.0105,
+      "step": 8577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999113134487075e-05,
+      "loss": 0.6547,
+      "step": 8578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999112925271373e-05,
+      "loss": 0.8973,
+      "step": 8579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999112716031001e-05,
+      "loss": 1.044,
+      "step": 8580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999112506765958e-05,
+      "loss": 1.189,
+      "step": 8581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999112297476245e-05,
+      "loss": 1.3591,
+      "step": 8582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999112088161862e-05,
+      "loss": 1.0588,
+      "step": 8583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991118788228084e-05,
+      "loss": 0.9702,
+      "step": 8584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991116694590856e-05,
+      "loss": 0.9531,
+      "step": 8585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999111460070691e-05,
+      "loss": 0.8241,
+      "step": 8586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991112506576276e-05,
+      "loss": 1.1639,
+      "step": 8587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999111041219894e-05,
+      "loss": 1.12,
+      "step": 8588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99911083175749e-05,
+      "loss": 1.3793,
+      "step": 8589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999110622270415e-05,
+      "loss": 0.9946,
+      "step": 8590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999110412758671e-05,
+      "loss": 0.8347,
+      "step": 8591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999110203222256e-05,
+      "loss": 0.4633,
+      "step": 8592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999109993661171e-05,
+      "loss": 0.2479,
+      "step": 8593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991097840754165e-05,
+      "loss": 0.373,
+      "step": 8594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999109574464991e-05,
+      "loss": 0.8449,
+      "step": 8595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999109364829896e-05,
+      "loss": 1.3162,
+      "step": 8596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99910915517013e-05,
+      "loss": 1.2068,
+      "step": 8597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999108945485695e-05,
+      "loss": 1.4203,
+      "step": 8598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999108735776589e-05,
+      "loss": 1.4168,
+      "step": 8599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999108526042813e-05,
+      "loss": 1.3545,
+      "step": 8600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991083162843666e-05,
+      "loss": 0.9488,
+      "step": 8601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999108106501251e-05,
+      "loss": 1.1733,
+      "step": 8602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991078966934645e-05,
+      "loss": 1.254,
+      "step": 8603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999107686861008e-05,
+      "loss": 1.0556,
+      "step": 8604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991074770038817e-05,
+      "loss": 1.1763,
+      "step": 8605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991072671220844e-05,
+      "loss": 1.1645,
+      "step": 8606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991070572156174e-05,
+      "loss": 1.0256,
+      "step": 8607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99910684728448e-05,
+      "loss": 1.0187,
+      "step": 8608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999106637328673e-05,
+      "loss": 1.0299,
+      "step": 8609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991064273481966e-05,
+      "loss": 0.8712,
+      "step": 8610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999106217343049e-05,
+      "loss": 1.0586,
+      "step": 8611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999106007313231e-05,
+      "loss": 0.9963,
+      "step": 8612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999105797258744e-05,
+      "loss": 1.1931,
+      "step": 8613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991055871795854e-05,
+      "loss": 1.1627,
+      "step": 8614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999105377075758e-05,
+      "loss": 1.0302,
+      "step": 8615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99910516694726e-05,
+      "loss": 1.0384,
+      "step": 8616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991049567940914e-05,
+      "loss": 1.2491,
+      "step": 8617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991047466162534e-05,
+      "loss": 0.9819,
+      "step": 8618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999104536413746e-05,
+      "loss": 0.6744,
+      "step": 8619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991043261865664e-05,
+      "loss": 1.2716,
+      "step": 8620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999104115934719e-05,
+      "loss": 1.1201,
+      "step": 8621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991039056581994e-05,
+      "loss": 0.8674,
+      "step": 8622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999103695357011e-05,
+      "loss": 1.0039,
+      "step": 8623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999103485031152e-05,
+      "loss": 1.1552,
+      "step": 8624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999103274680623e-05,
+      "loss": 1.206,
+      "step": 8625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999103064305424e-05,
+      "loss": 1.3881,
+      "step": 8626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999102853905555e-05,
+      "loss": 0.9618,
+      "step": 8627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991026434810154e-05,
+      "loss": 1.0148,
+      "step": 8628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991024330318064e-05,
+      "loss": 0.8362,
+      "step": 8629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999102222557927e-05,
+      "loss": 0.9687,
+      "step": 8630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999102012059378e-05,
+      "loss": 1.278,
+      "step": 8631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999101801536158e-05,
+      "loss": 0.9909,
+      "step": 8632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999101590988268e-05,
+      "loss": 0.8792,
+      "step": 8633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999101380415708e-05,
+      "loss": 1.0751,
+      "step": 8634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999101169818479e-05,
+      "loss": 1.0353,
+      "step": 8635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999100959196579e-05,
+      "loss": 1.1233,
+      "step": 8636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999100748550009e-05,
+      "loss": 1.1861,
+      "step": 8637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999100537878769e-05,
+      "loss": 1.0026,
+      "step": 8638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991003271828585e-05,
+      "loss": 1.0155,
+      "step": 8639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991001164622784e-05,
+      "loss": 1.0189,
+      "step": 8640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990999057170286e-05,
+      "loss": 1.4193,
+      "step": 8641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999099694947108e-05,
+      "loss": 0.6558,
+      "step": 8642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999099484152518e-05,
+      "loss": 0.3156,
+      "step": 8643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999099273333258e-05,
+      "loss": 0.2008,
+      "step": 8644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990990624893274e-05,
+      "loss": 0.4412,
+      "step": 8645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990988516207265e-05,
+      "loss": 1.0611,
+      "step": 8646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999098640727456e-05,
+      "loss": 1.0295,
+      "step": 8647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999098429809516e-05,
+      "loss": 1.2576,
+      "step": 8648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999098218866905e-05,
+      "loss": 1.1036,
+      "step": 8649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999098007899624e-05,
+      "loss": 1.1413,
+      "step": 8650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990977969076734e-05,
+      "loss": 1.3274,
+      "step": 8651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990975858910526e-05,
+      "loss": 0.8323,
+      "step": 8652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990973748497614e-05,
+      "loss": 1.0233,
+      "step": 8653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999097163783801e-05,
+      "loss": 0.6758,
+      "step": 8654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99909695269317e-05,
+      "loss": 1.018,
+      "step": 8655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999096741577869e-05,
+      "loss": 1.0897,
+      "step": 8656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990965304378986e-05,
+      "loss": 0.9251,
+      "step": 8657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999096319273258e-05,
+      "loss": 0.9358,
+      "step": 8658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999096108083946e-05,
+      "loss": 1.3181,
+      "step": 8659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999095896869965e-05,
+      "loss": 1.065,
+      "step": 8660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990956856313144e-05,
+      "loss": 1.2166,
+      "step": 8661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999095474367993e-05,
+      "loss": 1.2498,
+      "step": 8662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990952630800023e-05,
+      "loss": 1.0486,
+      "step": 8663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999095051767341e-05,
+      "loss": 1.0942,
+      "step": 8664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990948404300095e-05,
+      "loss": 1.3692,
+      "step": 8665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990946290680086e-05,
+      "loss": 1.2716,
+      "step": 8666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990944176813374e-05,
+      "loss": 1.1882,
+      "step": 8667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990942062699965e-05,
+      "loss": 1.1674,
+      "step": 8668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999093994833985e-05,
+      "loss": 0.8575,
+      "step": 8669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999093783373304e-05,
+      "loss": 0.8663,
+      "step": 8670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999093571887953e-05,
+      "loss": 1.1732,
+      "step": 8671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999093360377932e-05,
+      "loss": 1.4423,
+      "step": 8672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999093148843241e-05,
+      "loss": 1.1841,
+      "step": 8673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99909293728388e-05,
+      "loss": 1.1292,
+      "step": 8674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999092725699849e-05,
+      "loss": 0.7134,
+      "step": 8675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999092514091148e-05,
+      "loss": 1.5599,
+      "step": 8676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990923024577764e-05,
+      "loss": 0.8031,
+      "step": 8677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990920907997355e-05,
+      "loss": 1.0661,
+      "step": 8678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999091879117024e-05,
+      "loss": 1.6182,
+      "step": 8679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999091667409643e-05,
+      "loss": 1.2834,
+      "step": 8680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990914556775926e-05,
+      "loss": 0.929,
+      "step": 8681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999091243920872e-05,
+      "loss": 0.9997,
+      "step": 8682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990910321394804e-05,
+      "loss": 0.9971,
+      "step": 8683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990908203334194e-05,
+      "loss": 1.1674,
+      "step": 8684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999090608502688e-05,
+      "loss": 1.2116,
+      "step": 8685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999090396647287e-05,
+      "loss": 1.0061,
+      "step": 8686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990901847672165e-05,
+      "loss": 1.3171,
+      "step": 8687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990899728624755e-05,
+      "loss": 1.1013,
+      "step": 8688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999089760933065e-05,
+      "loss": 1.5107,
+      "step": 8689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999089548978984e-05,
+      "loss": 0.9969,
+      "step": 8690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999089337000233e-05,
+      "loss": 1.0103,
+      "step": 8691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999089124996813e-05,
+      "loss": 1.3219,
+      "step": 8692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999088912968722e-05,
+      "loss": 1.202,
+      "step": 8693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999088700915961e-05,
+      "loss": 0.8583,
+      "step": 8694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999088488838531e-05,
+      "loss": 0.652,
+      "step": 8695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999088276736431e-05,
+      "loss": 0.5651,
+      "step": 8696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999088064609661e-05,
+      "loss": 0.5008,
+      "step": 8697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99908785245822e-05,
+      "loss": 1.1029,
+      "step": 8698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99908764028211e-05,
+      "loss": 1.2576,
+      "step": 8699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990874280813294e-05,
+      "loss": 1.2909,
+      "step": 8700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999087215855879e-05,
+      "loss": 1.0101,
+      "step": 8701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999087003605759e-05,
+      "loss": 1.3033,
+      "step": 8702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999086791330969e-05,
+      "loss": 1.1506,
+      "step": 8703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990865790315086e-05,
+      "loss": 1.0781,
+      "step": 8704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999086366707379e-05,
+      "loss": 1.2098,
+      "step": 8705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999086154358579e-05,
+      "loss": 1.1135,
+      "step": 8706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990859419851095e-05,
+      "loss": 0.985,
+      "step": 8707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99908572958697e-05,
+      "loss": 1.0368,
+      "step": 8708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99908551716416e-05,
+      "loss": 1.3571,
+      "step": 8709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999085304716681e-05,
+      "loss": 1.4024,
+      "step": 8710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990850922445306e-05,
+      "loss": 1.2478,
+      "step": 8711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990848797477116e-05,
+      "loss": 1.0357,
+      "step": 8712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999084667226223e-05,
+      "loss": 1.0545,
+      "step": 8713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999084454680063e-05,
+      "loss": 0.9262,
+      "step": 8714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990842421092344e-05,
+      "loss": 0.9385,
+      "step": 8715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999084029513735e-05,
+      "loss": 1.1249,
+      "step": 8716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999083816893566e-05,
+      "loss": 1.2108,
+      "step": 8717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990836042487275e-05,
+      "loss": 0.9183,
+      "step": 8718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999083391579219e-05,
+      "loss": 1.0582,
+      "step": 8719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990831788850404e-05,
+      "loss": 1.0237,
+      "step": 8720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990829661661916e-05,
+      "loss": 1.0782,
+      "step": 8721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999082753422673e-05,
+      "loss": 1.018,
+      "step": 8722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999082540654485e-05,
+      "loss": 1.084,
+      "step": 8723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999082327861627e-05,
+      "loss": 1.151,
+      "step": 8724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990821150440986e-05,
+      "loss": 0.7537,
+      "step": 8725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999081902201901e-05,
+      "loss": 1.154,
+      "step": 8726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999081689335032e-05,
+      "loss": 1.5061,
+      "step": 8727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990814764434944e-05,
+      "loss": 1.2121,
+      "step": 8728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990812635272876e-05,
+      "loss": 0.9162,
+      "step": 8729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99908105058641e-05,
+      "loss": 1.1134,
+      "step": 8730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990808376208624e-05,
+      "loss": 1.0876,
+      "step": 8731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999080624630645e-05,
+      "loss": 0.8893,
+      "step": 8732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990804116157586e-05,
+      "loss": 1.0382,
+      "step": 8733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990801985762014e-05,
+      "loss": 1.1064,
+      "step": 8734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999079985511974e-05,
+      "loss": 1.2253,
+      "step": 8735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990797724230776e-05,
+      "loss": 1.4358,
+      "step": 8736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999079559309511e-05,
+      "loss": 1.6728,
+      "step": 8737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999079346171275e-05,
+      "loss": 1.3647,
+      "step": 8738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999079133008369e-05,
+      "loss": 1.1633,
+      "step": 8739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990789198207925e-05,
+      "loss": 1.621,
+      "step": 8740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990787066085463e-05,
+      "loss": 1.548,
+      "step": 8741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990784933716306e-05,
+      "loss": 1.3156,
+      "step": 8742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990782801100444e-05,
+      "loss": 1.0951,
+      "step": 8743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999078066823789e-05,
+      "loss": 1.5253,
+      "step": 8744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999077853512863e-05,
+      "loss": 0.2305,
+      "step": 8745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999077640177269e-05,
+      "loss": 0.3264,
+      "step": 8746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999077426817004e-05,
+      "loss": 0.3777,
+      "step": 8747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999077213432068e-05,
+      "loss": 0.0945,
+      "step": 8748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999077000022464e-05,
+      "loss": 0.0586,
+      "step": 8749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990767865881896e-05,
+      "loss": 0.0599,
+      "step": 8750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999076573129245e-05,
+      "loss": 1.2068,
+      "step": 8751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99907635964563e-05,
+      "loss": 1.1719,
+      "step": 8752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999076146137347e-05,
+      "loss": 1.2801,
+      "step": 8753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990759326043924e-05,
+      "loss": 1.1722,
+      "step": 8754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999075719046769e-05,
+      "loss": 1.0336,
+      "step": 8755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999075505464476e-05,
+      "loss": 1.0087,
+      "step": 8756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999075291857512e-05,
+      "loss": 1.2859,
+      "step": 8757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990750782258785e-05,
+      "loss": 1.125,
+      "step": 8758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990748645695764e-05,
+      "loss": 1.0779,
+      "step": 8759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999074650888603e-05,
+      "loss": 1.3599,
+      "step": 8760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99907443718296e-05,
+      "loss": 1.9964,
+      "step": 8761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999074223452648e-05,
+      "loss": 3.0764,
+      "step": 8762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999074009697666e-05,
+      "loss": 0.4825,
+      "step": 8763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999073795918013e-05,
+      "loss": 0.9809,
+      "step": 8764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999073582113692e-05,
+      "loss": 1.3737,
+      "step": 8765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990733682847e-05,
+      "loss": 1.3429,
+      "step": 8766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990731544310386e-05,
+      "loss": 1.1932,
+      "step": 8767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990729405527074e-05,
+      "loss": 1.2749,
+      "step": 8768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999072726649706e-05,
+      "loss": 1.0659,
+      "step": 8769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990725127220355e-05,
+      "loss": 1.0079,
+      "step": 8770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990722987696947e-05,
+      "loss": 1.2343,
+      "step": 8771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999072084792684e-05,
+      "loss": 0.8325,
+      "step": 8772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999071870791005e-05,
+      "loss": 1.0094,
+      "step": 8773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999071656764654e-05,
+      "loss": 1.2184,
+      "step": 8774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999071442713634e-05,
+      "loss": 0.8084,
+      "step": 8775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999071228637945e-05,
+      "loss": 0.5433,
+      "step": 8776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990710145375854e-05,
+      "loss": 0.7831,
+      "step": 8777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999070800412556e-05,
+      "loss": 1.2789,
+      "step": 8778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990705862628575e-05,
+      "loss": 1.3226,
+      "step": 8779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999070372088489e-05,
+      "loss": 0.8119,
+      "step": 8780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999070157889451e-05,
+      "loss": 1.311,
+      "step": 8781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990699436657424e-05,
+      "loss": 1.1202,
+      "step": 8782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999069729417364e-05,
+      "loss": 1.1229,
+      "step": 8783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999069515144317e-05,
+      "loss": 1.1352,
+      "step": 8784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990693008465996e-05,
+      "loss": 0.9946,
+      "step": 8785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999069086524212e-05,
+      "loss": 1.2596,
+      "step": 8786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999068872177155e-05,
+      "loss": 1.1081,
+      "step": 8787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999068657805428e-05,
+      "loss": 0.9151,
+      "step": 8788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990684434090316e-05,
+      "loss": 0.9901,
+      "step": 8789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999068228987965e-05,
+      "loss": 1.2045,
+      "step": 8790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99906801454223e-05,
+      "loss": 0.8862,
+      "step": 8791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999067800071824e-05,
+      "loss": 1.1898,
+      "step": 8792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990675855767484e-05,
+      "loss": 0.9776,
+      "step": 8793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999067371057003e-05,
+      "loss": 1.017,
+      "step": 8794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999067156512588e-05,
+      "loss": 1.2088,
+      "step": 8795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999066941943503e-05,
+      "loss": 1.3442,
+      "step": 8796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999066727349749e-05,
+      "loss": 1.1206,
+      "step": 8797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990665127313246e-05,
+      "loss": 1.4715,
+      "step": 8798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990662980882305e-05,
+      "loss": 1.6074,
+      "step": 8799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990660834204673e-05,
+      "loss": 1.2598,
+      "step": 8800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999065868728033e-05,
+      "loss": 1.1772,
+      "step": 8801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999065654010931e-05,
+      "loss": 1.1079,
+      "step": 8802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990654392691586e-05,
+      "loss": 0.9349,
+      "step": 8803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990652245027154e-05,
+      "loss": 0.7483,
+      "step": 8804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999065009711603e-05,
+      "loss": 0.6774,
+      "step": 8805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990647948958215e-05,
+      "loss": 0.6102,
+      "step": 8806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999064580055369e-05,
+      "loss": 0.4901,
+      "step": 8807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999064365190248e-05,
+      "loss": 0.6469,
+      "step": 8808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999064150300457e-05,
+      "loss": 0.7031,
+      "step": 8809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990639353859956e-05,
+      "loss": 1.2631,
+      "step": 8810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990637204468654e-05,
+      "loss": 0.9765,
+      "step": 8811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990635054830656e-05,
+      "loss": 1.6765,
+      "step": 8812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990632904945955e-05,
+      "loss": 1.2318,
+      "step": 8813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990630754814564e-05,
+      "loss": 1.4459,
+      "step": 8814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999062860443646e-05,
+      "loss": 1.3727,
+      "step": 8815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999062645381167e-05,
+      "loss": 1.4865,
+      "step": 8816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999062430294019e-05,
+      "loss": 1.0549,
+      "step": 8817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990622151822005e-05,
+      "loss": 1.2053,
+      "step": 8818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990620000457123e-05,
+      "loss": 1.2681,
+      "step": 8819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990617848845545e-05,
+      "loss": 1.2183,
+      "step": 8820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999061569698727e-05,
+      "loss": 1.1343,
+      "step": 8821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99906135448823e-05,
+      "loss": 1.234,
+      "step": 8822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999061139253063e-05,
+      "loss": 1.0511,
+      "step": 8823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999060923993226e-05,
+      "loss": 1.2056,
+      "step": 8824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990607087087206e-05,
+      "loss": 1.6887,
+      "step": 8825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999060493399544e-05,
+      "loss": 1.3319,
+      "step": 8826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990602780656993e-05,
+      "loss": 1.0182,
+      "step": 8827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990600627071835e-05,
+      "loss": 1.0378,
+      "step": 8828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990598473239994e-05,
+      "loss": 1.4682,
+      "step": 8829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999059631916144e-05,
+      "loss": 1.3639,
+      "step": 8830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99905941648362e-05,
+      "loss": 1.1836,
+      "step": 8831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999059201026426e-05,
+      "loss": 1.3155,
+      "step": 8832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999058985544562e-05,
+      "loss": 1.128,
+      "step": 8833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99905877003803e-05,
+      "loss": 1.2607,
+      "step": 8834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999058554506827e-05,
+      "loss": 1.1759,
+      "step": 8835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999058338950954e-05,
+      "loss": 0.9747,
+      "step": 8836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999058123370412e-05,
+      "loss": 1.1861,
+      "step": 8837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990579077652e-05,
+      "loss": 0.6354,
+      "step": 8838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990576921353187e-05,
+      "loss": 0.6298,
+      "step": 8839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990574764807675e-05,
+      "loss": 0.5622,
+      "step": 8840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999057260801547e-05,
+      "loss": 0.3542,
+      "step": 8841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990570450976564e-05,
+      "loss": 0.4076,
+      "step": 8842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999056829369096e-05,
+      "loss": 0.5068,
+      "step": 8843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990566136158665e-05,
+      "loss": 0.6871,
+      "step": 8844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999056397837968e-05,
+      "loss": 1.1034,
+      "step": 8845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990561820353986e-05,
+      "loss": 1.3705,
+      "step": 8846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990559662081606e-05,
+      "loss": 1.0481,
+      "step": 8847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999055750356252e-05,
+      "loss": 1.5051,
+      "step": 8848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999055534479674e-05,
+      "loss": 1.2912,
+      "step": 8849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999055318578427e-05,
+      "loss": 1.1822,
+      "step": 8850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99905510265251e-05,
+      "loss": 1.1203,
+      "step": 8851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999054886701923e-05,
+      "loss": 0.8883,
+      "step": 8852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999054670726667e-05,
+      "loss": 1.2353,
+      "step": 8853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999054454726741e-05,
+      "loss": 0.9927,
+      "step": 8854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999054238702145e-05,
+      "loss": 0.6175,
+      "step": 8855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99905402265288e-05,
+      "loss": 1.0129,
+      "step": 8856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999053806578946e-05,
+      "loss": 1.0805,
+      "step": 8857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999053590480342e-05,
+      "loss": 1.1087,
+      "step": 8858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999053374357068e-05,
+      "loss": 1.2239,
+      "step": 8859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999053158209124e-05,
+      "loss": 1.129,
+      "step": 8860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990529420365106e-05,
+      "loss": 0.9646,
+      "step": 8861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999052725839228e-05,
+      "loss": 1.0584,
+      "step": 8862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990525096172755e-05,
+      "loss": 1.085,
+      "step": 8863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999052293370654e-05,
+      "loss": 1.0483,
+      "step": 8864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990520770993624e-05,
+      "loss": 1.0508,
+      "step": 8865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999051860803401e-05,
+      "loss": 1.2326,
+      "step": 8866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990516444827706e-05,
+      "loss": 1.1117,
+      "step": 8867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99905142813747e-05,
+      "loss": 0.9983,
+      "step": 8868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990512117675e-05,
+      "loss": 0.9904,
+      "step": 8869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999050995372861e-05,
+      "loss": 1.2398,
+      "step": 8870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990507789535525e-05,
+      "loss": 1.4326,
+      "step": 8871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990505625095735e-05,
+      "loss": 1.0302,
+      "step": 8872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990503460409255e-05,
+      "loss": 1.2675,
+      "step": 8873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999050129547608e-05,
+      "loss": 0.8434,
+      "step": 8874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990499130296204e-05,
+      "loss": 0.3367,
+      "step": 8875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999049696486964e-05,
+      "loss": 1.0727,
+      "step": 8876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999049479919637e-05,
+      "loss": 1.4354,
+      "step": 8877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999049263327641e-05,
+      "loss": 1.0382,
+      "step": 8878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999049046710975e-05,
+      "loss": 0.8141,
+      "step": 8879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99904883006964e-05,
+      "loss": 0.7363,
+      "step": 8880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990486134036354e-05,
+      "loss": 1.6423,
+      "step": 8881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999048396712962e-05,
+      "loss": 2.6529,
+      "step": 8882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999048179997618e-05,
+      "loss": 1.0003,
+      "step": 8883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999047963257604e-05,
+      "loss": 0.9851,
+      "step": 8884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990477464929214e-05,
+      "loss": 0.9985,
+      "step": 8885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999047529703569e-05,
+      "loss": 1.1115,
+      "step": 8886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999047312889547e-05,
+      "loss": 1.3173,
+      "step": 8887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990470960508555e-05,
+      "loss": 0.7853,
+      "step": 8888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999046879187494e-05,
+      "loss": 0.6068,
+      "step": 8889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999046662299464e-05,
+      "loss": 0.5536,
+      "step": 8890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999046445386763e-05,
+      "loss": 0.5109,
+      "step": 8891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999046228449394e-05,
+      "loss": 0.4583,
+      "step": 8892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999046011487355e-05,
+      "loss": 0.4123,
+      "step": 8893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999045794500646e-05,
+      "loss": 0.4004,
+      "step": 8894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999045577489268e-05,
+      "loss": 0.3739,
+      "step": 8895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990453604532205e-05,
+      "loss": 0.3816,
+      "step": 8896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999045143392503e-05,
+      "loss": 0.4391,
+      "step": 8897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990449263071156e-05,
+      "loss": 1.1637,
+      "step": 8898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99904470919706e-05,
+      "loss": 1.235,
+      "step": 8899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990444920623334e-05,
+      "loss": 0.955,
+      "step": 8900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999044274902939e-05,
+      "loss": 0.7153,
+      "step": 8901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999044057718873e-05,
+      "loss": 0.9503,
+      "step": 8902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990438405101394e-05,
+      "loss": 1.3083,
+      "step": 8903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999043623276736e-05,
+      "loss": 0.7842,
+      "step": 8904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999043406018662e-05,
+      "loss": 0.6211,
+      "step": 8905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999043188735919e-05,
+      "loss": 0.996,
+      "step": 8906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999042971428507e-05,
+      "loss": 0.991,
+      "step": 8907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999042754096425e-05,
+      "loss": 1.1821,
+      "step": 8908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999042536739673e-05,
+      "loss": 1.0599,
+      "step": 8909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999042319358253e-05,
+      "loss": 1.1233,
+      "step": 8910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999042101952162e-05,
+      "loss": 1.0006,
+      "step": 8911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990418845214026e-05,
+      "loss": 1.2382,
+      "step": 8912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999041667065973e-05,
+      "loss": 1.105,
+      "step": 8913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999041449585874e-05,
+      "loss": 1.1125,
+      "step": 8914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999041232081106e-05,
+      "loss": 1.1714,
+      "step": 8915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999041014551668e-05,
+      "loss": 1.079,
+      "step": 8916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990407969975606e-05,
+      "loss": 0.4372,
+      "step": 8917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999040579418784e-05,
+      "loss": 0.5173,
+      "step": 8918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999040361815338e-05,
+      "loss": 1.5945,
+      "step": 8919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999040144187222e-05,
+      "loss": 1.1579,
+      "step": 8920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990399265344366e-05,
+      "loss": 0.8848,
+      "step": 8921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999039708856982e-05,
+      "loss": 1.0272,
+      "step": 8922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999039491154858e-05,
+      "loss": 1.1456,
+      "step": 8923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990392734280644e-05,
+      "loss": 0.9839,
+      "step": 8924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999039055676601e-05,
+      "loss": 0.9463,
+      "step": 8925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999038837900469e-05,
+      "loss": 1.1473,
+      "step": 8926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990386200996666e-05,
+      "loss": 1.6757,
+      "step": 8927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999038402274195e-05,
+      "loss": 2.2957,
+      "step": 8928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999038184424054e-05,
+      "loss": 0.9495,
+      "step": 8929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990379665492446e-05,
+      "loss": 1.2136,
+      "step": 8930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999037748649764e-05,
+      "loss": 1.3899,
+      "step": 8931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999037530725615e-05,
+      "loss": 1.2546,
+      "step": 8932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999037312776796e-05,
+      "loss": 1.3167,
+      "step": 8933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999037094803308e-05,
+      "loss": 1.0479,
+      "step": 8934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999036876805151e-05,
+      "loss": 0.8051,
+      "step": 8935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990366587823236e-05,
+      "loss": 0.6917,
+      "step": 8936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999036440734827e-05,
+      "loss": 0.6909,
+      "step": 8937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999036222662661e-05,
+      "loss": 0.7093,
+      "step": 8938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990360045658266e-05,
+      "loss": 1.1164,
+      "step": 8939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999035786444322e-05,
+      "loss": 1.2123,
+      "step": 8940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999035568298147e-05,
+      "loss": 1.0801,
+      "step": 8941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990353501273034e-05,
+      "loss": 1.0261,
+      "step": 8942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999035131931791e-05,
+      "loss": 0.8396,
+      "step": 8943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999034913711608e-05,
+      "loss": 1.0039,
+      "step": 8944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990346954667566e-05,
+      "loss": 1.102,
+      "step": 8945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999034477197235e-05,
+      "loss": 0.9566,
+      "step": 8946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999034258903045e-05,
+      "loss": 1.3018,
+      "step": 8947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999034040584184e-05,
+      "loss": 1.2828,
+      "step": 8948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999033822240655e-05,
+      "loss": 1.1951,
+      "step": 8949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990336038724564e-05,
+      "loss": 1.3283,
+      "step": 8950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999033385479588e-05,
+      "loss": 1.3253,
+      "step": 8951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99903316706205e-05,
+      "loss": 1.4602,
+      "step": 8952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999032948619843e-05,
+      "loss": 0.837,
+      "step": 8953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999032730152966e-05,
+      "loss": 1.0146,
+      "step": 8954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99903251166142e-05,
+      "loss": 1.0879,
+      "step": 8955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999032293145205e-05,
+      "loss": 1.4743,
+      "step": 8956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990320746043204e-05,
+      "loss": 1.0642,
+      "step": 8957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990318560387664e-05,
+      "loss": 1.0941,
+      "step": 8958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999031637448542e-05,
+      "loss": 0.8432,
+      "step": 8959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990314188336495e-05,
+      "loss": 1.264,
+      "step": 8960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999031200194088e-05,
+      "loss": 0.8801,
+      "step": 8961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999030981529856e-05,
+      "loss": 0.9694,
+      "step": 8962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999030762840955e-05,
+      "loss": 1.2519,
+      "step": 8963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990305441273844e-05,
+      "loss": 1.1655,
+      "step": 8964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999030325389145e-05,
+      "loss": 1.7452,
+      "step": 8965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990301066262355e-05,
+      "loss": 1.1885,
+      "step": 8966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990298878386566e-05,
+      "loss": 1.1091,
+      "step": 8967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990296690264086e-05,
+      "loss": 1.0145,
+      "step": 8968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999029450189492e-05,
+      "loss": 1.1637,
+      "step": 8969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999029231327905e-05,
+      "loss": 0.8908,
+      "step": 8970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999029012441649e-05,
+      "loss": 1.0207,
+      "step": 8971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999028793530724e-05,
+      "loss": 1.0373,
+      "step": 8972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999028574595129e-05,
+      "loss": 1.2048,
+      "step": 8973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999028355634865e-05,
+      "loss": 1.6187,
+      "step": 8974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990281366499314e-05,
+      "loss": 1.1053,
+      "step": 8975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999027917640329e-05,
+      "loss": 1.4927,
+      "step": 8976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999027698606056e-05,
+      "loss": 1.1898,
+      "step": 8977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999027479547115e-05,
+      "loss": 1.4062,
+      "step": 8978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999027260463503e-05,
+      "loss": 0.9272,
+      "step": 8979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990270413552235e-05,
+      "loss": 0.9686,
+      "step": 8980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999026822222274e-05,
+      "loss": 1.0166,
+      "step": 8981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990266030646555e-05,
+      "loss": 1.2849,
+      "step": 8982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999026383882367e-05,
+      "loss": 0.7406,
+      "step": 8983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99902616467541e-05,
+      "loss": 0.7619,
+      "step": 8984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999025945443783e-05,
+      "loss": 1.2153,
+      "step": 8985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999025726187486e-05,
+      "loss": 0.8999,
+      "step": 8986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990255069065215e-05,
+      "loss": 0.9631,
+      "step": 8987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999025287600886e-05,
+      "loss": 0.8158,
+      "step": 8988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999025068270582e-05,
+      "loss": 1.0232,
+      "step": 8989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999024848915609e-05,
+      "loss": 1.1381,
+      "step": 8990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999024629535966e-05,
+      "loss": 0.9671,
+      "step": 8991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999024410131653e-05,
+      "loss": 1.0486,
+      "step": 8992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999024190702672e-05,
+      "loss": 1.0093,
+      "step": 8993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990239712490215e-05,
+      "loss": 0.9589,
+      "step": 8994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990237517707014e-05,
+      "loss": 0.9503,
+      "step": 8995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999023532267712e-05,
+      "loss": 1.0971,
+      "step": 8996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999023312740053e-05,
+      "loss": 1.1159,
+      "step": 8997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990230931877246e-05,
+      "loss": 0.8094,
+      "step": 8998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999022873610728e-05,
+      "loss": 1.1743,
+      "step": 8999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990226540090615e-05,
+      "loss": 1.1199,
+      "step": 9000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0580137968063354,
+      "eval_runtime": 83.5773,
+      "eval_samples_per_second": 16.571,
+      "eval_steps_per_second": 4.152,
+      "step": 9000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990224343827255e-05,
+      "loss": 0.9495,
+      "step": 9001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99902221473172e-05,
+      "loss": 1.0997,
+      "step": 9002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999021995056046e-05,
+      "loss": 1.0835,
+      "step": 9003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990217753557015e-05,
+      "loss": 1.1702,
+      "step": 9004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999021555630689e-05,
+      "loss": 1.4746,
+      "step": 9005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990213358810065e-05,
+      "loss": 4.1693,
+      "step": 9006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990211161066545e-05,
+      "loss": 1.4417,
+      "step": 9007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990208963076335e-05,
+      "loss": 1.0626,
+      "step": 9008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999020676483943e-05,
+      "loss": 1.1296,
+      "step": 9009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999020456635583e-05,
+      "loss": 1.0832,
+      "step": 9010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999020236762555e-05,
+      "loss": 1.0376,
+      "step": 9011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999020016864857e-05,
+      "loss": 1.1048,
+      "step": 9012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99901979694249e-05,
+      "loss": 1.0899,
+      "step": 9013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999019576995453e-05,
+      "loss": 1.1138,
+      "step": 9014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990193570237476e-05,
+      "loss": 1.126,
+      "step": 9015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999019137027372e-05,
+      "loss": 1.1415,
+      "step": 9016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999018917006328e-05,
+      "loss": 1.0015,
+      "step": 9017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990186969606146e-05,
+      "loss": 0.7915,
+      "step": 9018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990184768902314e-05,
+      "loss": 0.869,
+      "step": 9019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999018256795179e-05,
+      "loss": 0.9655,
+      "step": 9020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999018036675458e-05,
+      "loss": 1.0987,
+      "step": 9021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999017816531067e-05,
+      "loss": 1.1674,
+      "step": 9022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990175963620066e-05,
+      "loss": 1.2031,
+      "step": 9023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999017376168278e-05,
+      "loss": 0.9145,
+      "step": 9024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999017155949879e-05,
+      "loss": 0.8055,
+      "step": 9025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999016935706812e-05,
+      "loss": 0.8942,
+      "step": 9026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990167154390746e-05,
+      "loss": 1.1422,
+      "step": 9027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999016495146669e-05,
+      "loss": 1.146,
+      "step": 9028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999016274829593e-05,
+      "loss": 1.1712,
+      "step": 9029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990160544878485e-05,
+      "loss": 1.0368,
+      "step": 9030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999015834121435e-05,
+      "loss": 0.8829,
+      "step": 9031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999015613730352e-05,
+      "loss": 1.1787,
+      "step": 9032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990153933145995e-05,
+      "loss": 1.5084,
+      "step": 9033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999015172874178e-05,
+      "loss": 0.6153,
+      "step": 9034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999014952409087e-05,
+      "loss": 1.1142,
+      "step": 9035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999014731919327e-05,
+      "loss": 1.2513,
+      "step": 9036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999014511404898e-05,
+      "loss": 1.0711,
+      "step": 9037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999014290865799e-05,
+      "loss": 0.9331,
+      "step": 9038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990140703020315e-05,
+      "loss": 1.4594,
+      "step": 9039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990138497135954e-05,
+      "loss": 1.106,
+      "step": 9040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990136291004884e-05,
+      "loss": 1.03,
+      "step": 9041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999013408462714e-05,
+      "loss": 1.2,
+      "step": 9042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999013187800269e-05,
+      "loss": 0.9152,
+      "step": 9043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999012967113156e-05,
+      "loss": 0.9454,
+      "step": 9044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990127464013723e-05,
+      "loss": 1.101,
+      "step": 9045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990125256649204e-05,
+      "loss": 1.1816,
+      "step": 9046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999012304903799e-05,
+      "loss": 1.1778,
+      "step": 9047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999012084118009e-05,
+      "loss": 1.3422,
+      "step": 9048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999011863307549e-05,
+      "loss": 1.1377,
+      "step": 9049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99901164247242e-05,
+      "loss": 1.1026,
+      "step": 9050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990114216126216e-05,
+      "loss": 1.2882,
+      "step": 9051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999011200728155e-05,
+      "loss": 1.2365,
+      "step": 9052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999010979819018e-05,
+      "loss": 1.1337,
+      "step": 9053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999010758885213e-05,
+      "loss": 0.9899,
+      "step": 9054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990105379267386e-05,
+      "loss": 1.2565,
+      "step": 9055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999010316943594e-05,
+      "loss": 1.5078,
+      "step": 9056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999010095935781e-05,
+      "loss": 1.4685,
+      "step": 9057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990098749032986e-05,
+      "loss": 1.4006,
+      "step": 9058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999009653846147e-05,
+      "loss": 1.1974,
+      "step": 9059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999009432764326e-05,
+      "loss": 1.3823,
+      "step": 9060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999009211657836e-05,
+      "loss": 0.6458,
+      "step": 9061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999008990526677e-05,
+      "loss": 1.0128,
+      "step": 9062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999008769370849e-05,
+      "loss": 1.1074,
+      "step": 9063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999008548190351e-05,
+      "loss": 1.2405,
+      "step": 9064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999008326985185e-05,
+      "loss": 1.6292,
+      "step": 9065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999008105755349e-05,
+      "loss": 1.1772,
+      "step": 9066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990078845008444e-05,
+      "loss": 1.118,
+      "step": 9067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999007663221671e-05,
+      "loss": 1.3651,
+      "step": 9068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999007441917828e-05,
+      "loss": 2.1085,
+      "step": 9069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990072205893155e-05,
+      "loss": 1.1137,
+      "step": 9070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999006999236134e-05,
+      "loss": 1.2094,
+      "step": 9071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990067778582837e-05,
+      "loss": 0.8978,
+      "step": 9072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999006556455764e-05,
+      "loss": 0.7341,
+      "step": 9073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990063350285745e-05,
+      "loss": 1.0128,
+      "step": 9074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999006113576717e-05,
+      "loss": 1.4113,
+      "step": 9075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990058921001895e-05,
+      "loss": 1.2249,
+      "step": 9076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990056705989934e-05,
+      "loss": 1.3765,
+      "step": 9077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990054490731284e-05,
+      "loss": 1.348,
+      "step": 9078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999005227522594e-05,
+      "loss": 1.5085,
+      "step": 9079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99900500594739e-05,
+      "loss": 1.3146,
+      "step": 9080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990047843475174e-05,
+      "loss": 1.1631,
+      "step": 9081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999004562722975e-05,
+      "loss": 1.3994,
+      "step": 9082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999004341073764e-05,
+      "loss": 1.1258,
+      "step": 9083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999004119399884e-05,
+      "loss": 1.2075,
+      "step": 9084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990038977013356e-05,
+      "loss": 1.3821,
+      "step": 9085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999003675978117e-05,
+      "loss": 1.0025,
+      "step": 9086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990034542302294e-05,
+      "loss": 1.1581,
+      "step": 9087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999003232457673e-05,
+      "loss": 1.0726,
+      "step": 9088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990030106604466e-05,
+      "loss": 1.0915,
+      "step": 9089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990027888385524e-05,
+      "loss": 1.2369,
+      "step": 9090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990025669919885e-05,
+      "loss": 1.0006,
+      "step": 9091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990023451207557e-05,
+      "loss": 1.1857,
+      "step": 9092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999002123224854e-05,
+      "loss": 1.0311,
+      "step": 9093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999001901304282e-05,
+      "loss": 1.1941,
+      "step": 9094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990016793590425e-05,
+      "loss": 1.2601,
+      "step": 9095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999001457389133e-05,
+      "loss": 1.0129,
+      "step": 9096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990012353945545e-05,
+      "loss": 1.2379,
+      "step": 9097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999001013375307e-05,
+      "loss": 1.4153,
+      "step": 9098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99900079133139e-05,
+      "loss": 1.368,
+      "step": 9099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990005692628044e-05,
+      "loss": 1.4062,
+      "step": 9100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999000347169549e-05,
+      "loss": 0.7764,
+      "step": 9101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999000125051626e-05,
+      "loss": 0.9343,
+      "step": 9102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998999902909033e-05,
+      "loss": 1.0394,
+      "step": 9103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998999680741771e-05,
+      "loss": 0.9598,
+      "step": 9104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989994585498404e-05,
+      "loss": 1.0693,
+      "step": 9105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989992363332396e-05,
+      "loss": 0.4744,
+      "step": 9106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989990140919706e-05,
+      "loss": 0.3535,
+      "step": 9107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998998791826032e-05,
+      "loss": 0.3838,
+      "step": 9108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998998569535425e-05,
+      "loss": 0.2679,
+      "step": 9109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998998347220148e-05,
+      "loss": 0.2629,
+      "step": 9110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998998124880203e-05,
+      "loss": 0.2073,
+      "step": 9111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989979025155886e-05,
+      "loss": 0.7131,
+      "step": 9112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998997680126305e-05,
+      "loss": 0.7333,
+      "step": 9113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998997457712352e-05,
+      "loss": 1.3434,
+      "step": 9114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99899723527373e-05,
+      "loss": 1.5403,
+      "step": 9115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989970128104394e-05,
+      "loss": 1.2184,
+      "step": 9116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99899679032248e-05,
+      "loss": 1.1485,
+      "step": 9117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998996567809851e-05,
+      "loss": 1.2869,
+      "step": 9118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998996345272553e-05,
+      "loss": 1.7247,
+      "step": 9119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989961227105865e-05,
+      "loss": 1.1431,
+      "step": 9120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99899590012395e-05,
+      "loss": 1.1697,
+      "step": 9121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998995677512646e-05,
+      "loss": 1.182,
+      "step": 9122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989954548766715e-05,
+      "loss": 1.1294,
+      "step": 9123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989952322160284e-05,
+      "loss": 1.0839,
+      "step": 9124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998995009530716e-05,
+      "loss": 0.7609,
+      "step": 9125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998994786820736e-05,
+      "loss": 1.1856,
+      "step": 9126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989945640860856e-05,
+      "loss": 1.1002,
+      "step": 9127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998994341326766e-05,
+      "loss": 0.9823,
+      "step": 9128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989941185427777e-05,
+      "loss": 1.3698,
+      "step": 9129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998993895734121e-05,
+      "loss": 0.9,
+      "step": 9130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989936729007945e-05,
+      "loss": 1.2307,
+      "step": 9131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989934500427994e-05,
+      "loss": 1.1759,
+      "step": 9132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998993227160135e-05,
+      "loss": 0.8295,
+      "step": 9133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998993004252802e-05,
+      "loss": 1.1786,
+      "step": 9134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989927813207995e-05,
+      "loss": 1.3618,
+      "step": 9135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989925583641285e-05,
+      "loss": 1.3421,
+      "step": 9136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989923353827885e-05,
+      "loss": 1.1763,
+      "step": 9137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998992112376779e-05,
+      "loss": 1.0261,
+      "step": 9138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998991889346101e-05,
+      "loss": 1.1373,
+      "step": 9139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998991666290754e-05,
+      "loss": 1.1484,
+      "step": 9140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998991443210738e-05,
+      "loss": 1.1353,
+      "step": 9141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998991220106052e-05,
+      "loss": 1.166,
+      "step": 9142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989909969766976e-05,
+      "loss": 1.2454,
+      "step": 9143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998990773822675e-05,
+      "loss": 1.2827,
+      "step": 9144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998990550643983e-05,
+      "loss": 1.0495,
+      "step": 9145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998990327440621e-05,
+      "loss": 1.1319,
+      "step": 9146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998990104212591e-05,
+      "loss": 1.2523,
+      "step": 9147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989898809598924e-05,
+      "loss": 1.0989,
+      "step": 9148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989896576825246e-05,
+      "loss": 1.2735,
+      "step": 9149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998989434380487e-05,
+      "loss": 1.3394,
+      "step": 9150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998989211053782e-05,
+      "loss": 1.1166,
+      "step": 9151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989889877024064e-05,
+      "loss": 1.2377,
+      "step": 9152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989887643263625e-05,
+      "loss": 1.1499,
+      "step": 9153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99898854092565e-05,
+      "loss": 1.1686,
+      "step": 9154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998988317500267e-05,
+      "loss": 1.0717,
+      "step": 9155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989880940502165e-05,
+      "loss": 1.1107,
+      "step": 9156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998987870575497e-05,
+      "loss": 1.4478,
+      "step": 9157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998987647076108e-05,
+      "loss": 1.3044,
+      "step": 9158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998987423552051e-05,
+      "loss": 1.1052,
+      "step": 9159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998987200003324e-05,
+      "loss": 0.9653,
+      "step": 9160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998986976429928e-05,
+      "loss": 1.0159,
+      "step": 9161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998986752831864e-05,
+      "loss": 1.042,
+      "step": 9162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99898652920913e-05,
+      "loss": 1.109,
+      "step": 9163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998986305561728e-05,
+      "loss": 1.3906,
+      "step": 9164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989860818896564e-05,
+      "loss": 1.334,
+      "step": 9165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989858581929164e-05,
+      "loss": 1.0409,
+      "step": 9166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998985634471507e-05,
+      "loss": 0.8628,
+      "step": 9167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998985410725429e-05,
+      "loss": 0.5957,
+      "step": 9168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998985186954682e-05,
+      "loss": 1.0405,
+      "step": 9169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998984963159265e-05,
+      "loss": 1.1866,
+      "step": 9170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998984739339181e-05,
+      "loss": 1.0571,
+      "step": 9171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998984515494427e-05,
+      "loss": 1.0295,
+      "step": 9172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998984291625004e-05,
+      "loss": 1.2393,
+      "step": 9173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998984067730912e-05,
+      "loss": 0.9602,
+      "step": 9174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989838438121515e-05,
+      "loss": 1.2002,
+      "step": 9175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998983619868722e-05,
+      "loss": 1.4526,
+      "step": 9176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998983395900624e-05,
+      "loss": 1.1412,
+      "step": 9177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998983171907856e-05,
+      "loss": 1.3989,
+      "step": 9178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99898294789042e-05,
+      "loss": 1.0866,
+      "step": 9179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989827238483144e-05,
+      "loss": 1.22,
+      "step": 9180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99898249978154e-05,
+      "loss": 1.0868,
+      "step": 9181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998982275690097e-05,
+      "loss": 0.6667,
+      "step": 9182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998982051573986e-05,
+      "loss": 0.258,
+      "step": 9183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989818274332047e-05,
+      "loss": 0.5565,
+      "step": 9184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998981603267755e-05,
+      "loss": 1.094,
+      "step": 9185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998981379077636e-05,
+      "loss": 1.0107,
+      "step": 9186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998981154862849e-05,
+      "loss": 0.566,
+      "step": 9187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989809306233925e-05,
+      "loss": 1.1785,
+      "step": 9188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998980706359268e-05,
+      "loss": 0.5826,
+      "step": 9189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989804820704735e-05,
+      "loss": 0.9039,
+      "step": 9190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99898025775701e-05,
+      "loss": 1.2496,
+      "step": 9191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989800334188786e-05,
+      "loss": 1.0865,
+      "step": 9192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998979809056078e-05,
+      "loss": 0.7945,
+      "step": 9193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989795846686084e-05,
+      "loss": 0.9995,
+      "step": 9194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998979360256469e-05,
+      "loss": 1.2338,
+      "step": 9195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998979135819662e-05,
+      "loss": 1.2236,
+      "step": 9196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998978911358186e-05,
+      "loss": 1.3668,
+      "step": 9197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998978686872041e-05,
+      "loss": 1.1045,
+      "step": 9198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998978462361227e-05,
+      "loss": 1.1917,
+      "step": 9199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998978237825744e-05,
+      "loss": 1.1984,
+      "step": 9200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989780132655916e-05,
+      "loss": 1.2166,
+      "step": 9201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989777886807714e-05,
+      "loss": 1.0184,
+      "step": 9202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998977564071282e-05,
+      "loss": 1.1177,
+      "step": 9203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989773394371235e-05,
+      "loss": 1.1258,
+      "step": 9204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989771147782964e-05,
+      "loss": 1.5137,
+      "step": 9205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998976890094801e-05,
+      "loss": 1.0125,
+      "step": 9206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998976665386635e-05,
+      "loss": 1.0696,
+      "step": 9207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998976440653802e-05,
+      "loss": 1.1876,
+      "step": 9208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989762158962995e-05,
+      "loss": 1.1406,
+      "step": 9209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998975991114128e-05,
+      "loss": 0.9748,
+      "step": 9210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998975766307288e-05,
+      "loss": 0.9761,
+      "step": 9211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998975541475779e-05,
+      "loss": 1.0726,
+      "step": 9212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998975316619601e-05,
+      "loss": 0.8541,
+      "step": 9213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998975091738755e-05,
+      "loss": 1.1607,
+      "step": 9214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989748668332385e-05,
+      "loss": 1.1494,
+      "step": 9215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989746419030546e-05,
+      "loss": 1.4115,
+      "step": 9216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998974416948201e-05,
+      "loss": 1.1737,
+      "step": 9217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99897419196868e-05,
+      "loss": 1.1807,
+      "step": 9218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998973966964489e-05,
+      "loss": 1.0111,
+      "step": 9219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998973741935628e-05,
+      "loss": 0.8739,
+      "step": 9220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998973516882101e-05,
+      "loss": 0.6346,
+      "step": 9221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998973291803903e-05,
+      "loss": 1.0811,
+      "step": 9222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998973066701037e-05,
+      "loss": 0.8113,
+      "step": 9223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998972841573502e-05,
+      "loss": 1.1547,
+      "step": 9224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998972616421298e-05,
+      "loss": 1.2342,
+      "step": 9225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989723912444257e-05,
+      "loss": 1.4094,
+      "step": 9226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998972166042884e-05,
+      "loss": 1.1467,
+      "step": 9227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989719408166745e-05,
+      "loss": 0.7494,
+      "step": 9228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998971715565796e-05,
+      "loss": 1.2098,
+      "step": 9229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998971490290247e-05,
+      "loss": 1.2357,
+      "step": 9230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998971264990031e-05,
+      "loss": 1.1795,
+      "step": 9231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989710396651456e-05,
+      "loss": 0.8123,
+      "step": 9232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998970814315592e-05,
+      "loss": 0.8651,
+      "step": 9233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998970588941369e-05,
+      "loss": 1.2199,
+      "step": 9234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998970363542477e-05,
+      "loss": 0.5233,
+      "step": 9235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998970138118917e-05,
+      "loss": 1.163,
+      "step": 9236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998969912670688e-05,
+      "loss": 1.7315,
+      "step": 9237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99896968719779e-05,
+      "loss": 1.5246,
+      "step": 9238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998969461700223e-05,
+      "loss": 1.2901,
+      "step": 9239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998969236177987e-05,
+      "loss": 1.0149,
+      "step": 9240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998969010631083e-05,
+      "loss": 0.9027,
+      "step": 9241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989687850595105e-05,
+      "loss": 1.275,
+      "step": 9242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989685594632686e-05,
+      "loss": 1.1306,
+      "step": 9243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998968333842358e-05,
+      "loss": 1.2844,
+      "step": 9244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998968108196778e-05,
+      "loss": 1.2132,
+      "step": 9245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99896788252653e-05,
+      "loss": 1.1061,
+      "step": 9246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989676568316134e-05,
+      "loss": 0.9193,
+      "step": 9247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998967431112027e-05,
+      "loss": 1.0906,
+      "step": 9248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998967205367773e-05,
+      "loss": 0.9511,
+      "step": 9249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99896697959885e-05,
+      "loss": 1.1308,
+      "step": 9250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998966753805258e-05,
+      "loss": 0.7206,
+      "step": 9251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989665279869977e-05,
+      "loss": 1.072,
+      "step": 9252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998966302144068e-05,
+      "loss": 1.1586,
+      "step": 9253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99896607627647e-05,
+      "loss": 1.266,
+      "step": 9254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998965850384203e-05,
+      "loss": 1.3535,
+      "step": 9255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998965624467267e-05,
+      "loss": 1.1896,
+      "step": 9256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998965398525663e-05,
+      "loss": 0.433,
+      "step": 9257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99896517255939e-05,
+      "loss": 0.373,
+      "step": 9258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989649465684485e-05,
+      "loss": 0.321,
+      "step": 9259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998964720552838e-05,
+      "loss": 0.4569,
+      "step": 9260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998964494512558e-05,
+      "loss": 0.3219,
+      "step": 9261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99896426844761e-05,
+      "loss": 0.9249,
+      "step": 9262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989640423579936e-05,
+      "loss": 0.8442,
+      "step": 9263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998963816243708e-05,
+      "loss": 1.4251,
+      "step": 9264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989635901047536e-05,
+      "loss": 1.3881,
+      "step": 9265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998963363941131e-05,
+      "loss": 1.1606,
+      "step": 9266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99896313775284e-05,
+      "loss": 1.3775,
+      "step": 9267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998962911539879e-05,
+      "loss": 1.3913,
+      "step": 9268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989626853022505e-05,
+      "loss": 1.0168,
+      "step": 9269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998962459039953e-05,
+      "loss": 1.1377,
+      "step": 9270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998962232752986e-05,
+      "loss": 1.2716,
+      "step": 9271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998962006441351e-05,
+      "loss": 1.3371,
+      "step": 9272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998961780105047e-05,
+      "loss": 1.0409,
+      "step": 9273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989615537440746e-05,
+      "loss": 1.2988,
+      "step": 9274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998961327358433e-05,
+      "loss": 0.8929,
+      "step": 9275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989611009481236e-05,
+      "loss": 1.1388,
+      "step": 9276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998960874513145e-05,
+      "loss": 0.7059,
+      "step": 9277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998960648053498e-05,
+      "loss": 0.6566,
+      "step": 9278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998960421569182e-05,
+      "loss": 0.6692,
+      "step": 9279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998960195060197e-05,
+      "loss": 1.1011,
+      "step": 9280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998959968526544e-05,
+      "loss": 0.6837,
+      "step": 9281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998959741968222e-05,
+      "loss": 1.2752,
+      "step": 9282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989595153852306e-05,
+      "loss": 1.2245,
+      "step": 9283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998959288777572e-05,
+      "loss": 1.2041,
+      "step": 9284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998959062145244e-05,
+      "loss": 1.539,
+      "step": 9285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989588354882465e-05,
+      "loss": 0.8351,
+      "step": 9286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989586088065814e-05,
+      "loss": 1.0497,
+      "step": 9287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998958382100247e-05,
+      "loss": 1.0906,
+      "step": 9288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998958155369244e-05,
+      "loss": 0.7888,
+      "step": 9289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989579286135735e-05,
+      "loss": 0.9287,
+      "step": 9290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998957701833233e-05,
+      "loss": 1.2553,
+      "step": 9291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998957475028225e-05,
+      "loss": 1.2213,
+      "step": 9292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998957248198547e-05,
+      "loss": 1.154,
+      "step": 9293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989570213442014e-05,
+      "loss": 0.9525,
+      "step": 9294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989567944651864e-05,
+      "loss": 1.2354,
+      "step": 9295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998956567561504e-05,
+      "loss": 1.3407,
+      "step": 9296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989563406331516e-05,
+      "loss": 1.2901,
+      "step": 9297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998956113680131e-05,
+      "loss": 1.5461,
+      "step": 9298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989558867024414e-05,
+      "loss": 0.7708,
+      "step": 9299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989556597000835e-05,
+      "loss": 1.1066,
+      "step": 9300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998955432673057e-05,
+      "loss": 1.3445,
+      "step": 9301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989552056213615e-05,
+      "loss": 1.076,
+      "step": 9302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989549785449986e-05,
+      "loss": 1.2831,
+      "step": 9303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998954751443966e-05,
+      "loss": 1.116,
+      "step": 9304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998954524318265e-05,
+      "loss": 0.6092,
+      "step": 9305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998954297167895e-05,
+      "loss": 1.1854,
+      "step": 9306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989540699928575e-05,
+      "loss": 1.0404,
+      "step": 9307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989538427931504e-05,
+      "loss": 1.1735,
+      "step": 9308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989536155687744e-05,
+      "loss": 1.2108,
+      "step": 9309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998953388319731e-05,
+      "loss": 1.1294,
+      "step": 9310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998953161046017e-05,
+      "loss": 1.0185,
+      "step": 9311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998952933747636e-05,
+      "loss": 0.8502,
+      "step": 9312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989527064245864e-05,
+      "loss": 0.98,
+      "step": 9313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989524790768674e-05,
+      "loss": 1.2549,
+      "step": 9314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998952251704481e-05,
+      "loss": 1.2172,
+      "step": 9315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998952024307425e-05,
+      "loss": 0.9761,
+      "step": 9316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989517968857e-05,
+      "loss": 1.0534,
+      "step": 9317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998951569439308e-05,
+      "loss": 1.0157,
+      "step": 9318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998951341968245e-05,
+      "loss": 4.9269,
+      "step": 9319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998951114472516e-05,
+      "loss": 4.9151,
+      "step": 9320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989508869521165e-05,
+      "loss": 1.274,
+      "step": 9321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998950659407049e-05,
+      "loss": 0.8477,
+      "step": 9322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998950431837314e-05,
+      "loss": 0.8863,
+      "step": 9323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989502042429085e-05,
+      "loss": 1.1951,
+      "step": 9324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989499766238354e-05,
+      "loss": 1.0261,
+      "step": 9325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998949748980094e-05,
+      "loss": 1.0344,
+      "step": 9326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989495213116836e-05,
+      "loss": 1.8301,
+      "step": 9327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998949293618605e-05,
+      "loss": 1.6782,
+      "step": 9328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998949065900857e-05,
+      "loss": 1.6676,
+      "step": 9329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998948838158441e-05,
+      "loss": 1.5011,
+      "step": 9330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998948610391357e-05,
+      "loss": 1.6872,
+      "step": 9331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998948382599604e-05,
+      "loss": 1.1545,
+      "step": 9332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998948154783182e-05,
+      "loss": 1.0702,
+      "step": 9333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989479269420915e-05,
+      "loss": 1.3011,
+      "step": 9334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989476990763326e-05,
+      "loss": 0.8697,
+      "step": 9335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989474711859054e-05,
+      "loss": 0.5981,
+      "step": 9336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998947243270809e-05,
+      "loss": 0.7164,
+      "step": 9337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998947015331045e-05,
+      "loss": 1.2006,
+      "step": 9338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998946787366612e-05,
+      "loss": 1.3245,
+      "step": 9339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99894655937751e-05,
+      "loss": 1.077,
+      "step": 9340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99894633136374e-05,
+      "loss": 1.1189,
+      "step": 9341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989461033253016e-05,
+      "loss": 1.1796,
+      "step": 9342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998945875262194e-05,
+      "loss": 1.222,
+      "step": 9343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989456471744184e-05,
+      "loss": 1.2311,
+      "step": 9344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989454190619744e-05,
+      "loss": 1.1154,
+      "step": 9345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989451909248614e-05,
+      "loss": 0.9449,
+      "step": 9346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99894496276308e-05,
+      "loss": 1.1974,
+      "step": 9347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99894473457663e-05,
+      "loss": 1.0124,
+      "step": 9348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998944506365512e-05,
+      "loss": 1.0151,
+      "step": 9349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998944278129725e-05,
+      "loss": 0.7458,
+      "step": 9350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998944049869269e-05,
+      "loss": 0.9251,
+      "step": 9351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998943821584145e-05,
+      "loss": 1.1292,
+      "step": 9352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998943593274352e-05,
+      "loss": 1.2313,
+      "step": 9353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989433649398913e-05,
+      "loss": 0.7314,
+      "step": 9354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989431365807615e-05,
+      "loss": 1.3168,
+      "step": 9355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989429081969634e-05,
+      "loss": 0.7289,
+      "step": 9356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998942679788497e-05,
+      "loss": 1.2072,
+      "step": 9357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989424513553615e-05,
+      "loss": 1.1549,
+      "step": 9358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998942222897558e-05,
+      "loss": 1.0921,
+      "step": 9359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998941994415086e-05,
+      "loss": 0.9914,
+      "step": 9360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998941765907945e-05,
+      "loss": 1.1754,
+      "step": 9361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998941537376136e-05,
+      "loss": 1.2133,
+      "step": 9362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998941308819659e-05,
+      "loss": 1.4551,
+      "step": 9363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998941080238513e-05,
+      "loss": 1.2509,
+      "step": 9364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998940851632698e-05,
+      "loss": 1.1247,
+      "step": 9365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989406230022154e-05,
+      "loss": 1.0529,
+      "step": 9366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998940394347064e-05,
+      "loss": 1.0516,
+      "step": 9367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998940165667243e-05,
+      "loss": 0.8707,
+      "step": 9368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998939936962754e-05,
+      "loss": 1.4619,
+      "step": 9369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998939708233598e-05,
+      "loss": 1.1269,
+      "step": 9370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989394794797716e-05,
+      "loss": 1.0765,
+      "step": 9371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998939250701278e-05,
+      "loss": 1.1936,
+      "step": 9372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989390218981157e-05,
+      "loss": 1.3124,
+      "step": 9373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989387930702845e-05,
+      "loss": 1.4301,
+      "step": 9374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998938564217786e-05,
+      "loss": 0.4058,
+      "step": 9375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989383353406174e-05,
+      "loss": 0.9463,
+      "step": 9376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998938106438781e-05,
+      "loss": 1.3724,
+      "step": 9377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998937877512276e-05,
+      "loss": 0.7495,
+      "step": 9378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998937648561103e-05,
+      "loss": 1.0482,
+      "step": 9379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998937419585261e-05,
+      "loss": 1.1149,
+      "step": 9380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998937190584751e-05,
+      "loss": 1.0348,
+      "step": 9381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989369615595724e-05,
+      "loss": 0.7258,
+      "step": 9382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998936732509725e-05,
+      "loss": 0.2154,
+      "step": 9383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989365034352096e-05,
+      "loss": 0.3616,
+      "step": 9384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998936274336026e-05,
+      "loss": 1.0633,
+      "step": 9385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998936045212173e-05,
+      "loss": 1.0139,
+      "step": 9386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989358160636523e-05,
+      "loss": 1.0686,
+      "step": 9387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998935586890463e-05,
+      "loss": 1.367,
+      "step": 9388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998935357692605e-05,
+      "loss": 0.723,
+      "step": 9389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998935128470079e-05,
+      "loss": 0.5576,
+      "step": 9390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989348992228845e-05,
+      "loss": 0.7907,
+      "step": 9391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998934669951021e-05,
+      "loss": 1.1041,
+      "step": 9392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989344406544894e-05,
+      "loss": 1.063,
+      "step": 9393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989342113332894e-05,
+      "loss": 1.1615,
+      "step": 9394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998933981987422e-05,
+      "loss": 1.0552,
+      "step": 9395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989337526168844e-05,
+      "loss": 1.0791,
+      "step": 9396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989335232216795e-05,
+      "loss": 1.0922,
+      "step": 9397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989332938018055e-05,
+      "loss": 1.0177,
+      "step": 9398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998933064357263e-05,
+      "loss": 1.3498,
+      "step": 9399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989328348880534e-05,
+      "loss": 1.2368,
+      "step": 9400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989326053941745e-05,
+      "loss": 0.96,
+      "step": 9401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998932375875627e-05,
+      "loss": 0.6169,
+      "step": 9402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998932146332411e-05,
+      "loss": 0.5365,
+      "step": 9403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989319167645274e-05,
+      "loss": 0.4449,
+      "step": 9404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989316871719746e-05,
+      "loss": 0.9199,
+      "step": 9405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998931457554754e-05,
+      "loss": 0.6908,
+      "step": 9406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998931227912864e-05,
+      "loss": 0.4082,
+      "step": 9407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998930998246307e-05,
+      "loss": 0.6992,
+      "step": 9408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998930768555081e-05,
+      "loss": 0.5228,
+      "step": 9409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998930538839186e-05,
+      "loss": 1.0521,
+      "step": 9410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989303090986234e-05,
+      "loss": 0.9689,
+      "step": 9411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989300793333924e-05,
+      "loss": 0.9596,
+      "step": 9412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989298495434925e-05,
+      "loss": 0.6542,
+      "step": 9413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998929619728925e-05,
+      "loss": 1.0697,
+      "step": 9414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998929389889688e-05,
+      "loss": 1.5153,
+      "step": 9415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989291600257834e-05,
+      "loss": 1.2395,
+      "step": 9416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99892893013721e-05,
+      "loss": 1.1297,
+      "step": 9417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998928700223968e-05,
+      "loss": 1.3725,
+      "step": 9418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998928470286058e-05,
+      "loss": 1.166,
+      "step": 9419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99892824032348e-05,
+      "loss": 1.2635,
+      "step": 9420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998928010336234e-05,
+      "loss": 1.3999,
+      "step": 9421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998927780324318e-05,
+      "loss": 1.1487,
+      "step": 9422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998927550287735e-05,
+      "loss": 1.0197,
+      "step": 9423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998927320226483e-05,
+      "loss": 1.1196,
+      "step": 9424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989270901405636e-05,
+      "loss": 1.0885,
+      "step": 9425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989268600299756e-05,
+      "loss": 1.2092,
+      "step": 9426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989266298947187e-05,
+      "loss": 1.1147,
+      "step": 9427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989263997347934e-05,
+      "loss": 0.9987,
+      "step": 9428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989261695502e-05,
+      "loss": 0.9774,
+      "step": 9429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998925939340938e-05,
+      "loss": 1.0896,
+      "step": 9430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998925709107008e-05,
+      "loss": 1.327,
+      "step": 9431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99892547884841e-05,
+      "loss": 1.1995,
+      "step": 9432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998925248565143e-05,
+      "loss": 1.1151,
+      "step": 9433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998925018257207e-05,
+      "loss": 0.2965,
+      "step": 9434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998924787924604e-05,
+      "loss": 1.1415,
+      "step": 9435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989245575673314e-05,
+      "loss": 1.1174,
+      "step": 9436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998924327185391e-05,
+      "loss": 0.9524,
+      "step": 9437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998924096778783e-05,
+      "loss": 1.2086,
+      "step": 9438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998923866347506e-05,
+      "loss": 1.1131,
+      "step": 9439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998923635891561e-05,
+      "loss": 1.2125,
+      "step": 9440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998923405410948e-05,
+      "loss": 1.3067,
+      "step": 9441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998923174905666e-05,
+      "loss": 1.2938,
+      "step": 9442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998922944375716e-05,
+      "loss": 1.057,
+      "step": 9443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998922713821097e-05,
+      "loss": 1.1474,
+      "step": 9444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99892248324181e-05,
+      "loss": 1.0418,
+      "step": 9445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989222526378555e-05,
+      "loss": 0.9933,
+      "step": 9446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998922022009232e-05,
+      "loss": 1.1512,
+      "step": 9447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998921791355941e-05,
+      "loss": 0.9912,
+      "step": 9448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989215606779804e-05,
+      "loss": 1.1426,
+      "step": 9449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998921329975352e-05,
+      "loss": 1.3306,
+      "step": 9450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998921099248056e-05,
+      "loss": 1.353,
+      "step": 9451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998920868496091e-05,
+      "loss": 1.0916,
+      "step": 9452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989206377194575e-05,
+      "loss": 1.0523,
+      "step": 9453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998920406918156e-05,
+      "loss": 1.2159,
+      "step": 9454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989201760921865e-05,
+      "loss": 1.2316,
+      "step": 9455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998919945241548e-05,
+      "loss": 1.054,
+      "step": 9456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989197143662416e-05,
+      "loss": 1.178,
+      "step": 9457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998919483466267e-05,
+      "loss": 0.9162,
+      "step": 9458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998919252541624e-05,
+      "loss": 1.32,
+      "step": 9459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998919021592313e-05,
+      "loss": 0.9844,
+      "step": 9460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989187906183335e-05,
+      "loss": 1.0082,
+      "step": 9461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989185596196864e-05,
+      "loss": 1.0474,
+      "step": 9462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99891832859637e-05,
+      "loss": 1.077,
+      "step": 9463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989180975483865e-05,
+      "loss": 1.0237,
+      "step": 9464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998917866475733e-05,
+      "loss": 1.1396,
+      "step": 9465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998917635378413e-05,
+      "loss": 1.1195,
+      "step": 9466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989174042564234e-05,
+      "loss": 1.1072,
+      "step": 9467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989171731097664e-05,
+      "loss": 1.8758,
+      "step": 9468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998916941938441e-05,
+      "loss": 1.7659,
+      "step": 9469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989167107424475e-05,
+      "loss": 1.2814,
+      "step": 9470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998916479521785e-05,
+      "loss": 1.4703,
+      "step": 9471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989162482764546e-05,
+      "loss": 1.1845,
+      "step": 9472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998916017006456e-05,
+      "loss": 1.1917,
+      "step": 9473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998915785711789e-05,
+      "loss": 1.2706,
+      "step": 9474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989155543924534e-05,
+      "loss": 1.229,
+      "step": 9475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989153230484506e-05,
+      "loss": 0.9546,
+      "step": 9476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998915091679779e-05,
+      "loss": 0.955,
+      "step": 9477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989148602864394e-05,
+      "loss": 1.3085,
+      "step": 9478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998914628868432e-05,
+      "loss": 1.2113,
+      "step": 9479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998914397425755e-05,
+      "loss": 0.991,
+      "step": 9480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989141659584106e-05,
+      "loss": 1.1268,
+      "step": 9481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998913934466398e-05,
+      "loss": 1.2348,
+      "step": 9482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998913702949717e-05,
+      "loss": 1.2291,
+      "step": 9483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998913471408368e-05,
+      "loss": 1.0154,
+      "step": 9484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998913239842351e-05,
+      "loss": 1.1843,
+      "step": 9485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998913008251665e-05,
+      "loss": 1.2066,
+      "step": 9486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998912776636311e-05,
+      "loss": 1.1451,
+      "step": 9487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998912544996289e-05,
+      "loss": 0.9843,
+      "step": 9488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998912313331599e-05,
+      "loss": 0.9165,
+      "step": 9489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998912081642241e-05,
+      "loss": 1.1365,
+      "step": 9490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998911849928214e-05,
+      "loss": 1.0598,
+      "step": 9491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998911618189519e-05,
+      "loss": 1.1168,
+      "step": 9492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998911386426156e-05,
+      "loss": 1.021,
+      "step": 9493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998911154638125e-05,
+      "loss": 1.054,
+      "step": 9494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998910922825425e-05,
+      "loss": 0.9778,
+      "step": 9495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998910690988058e-05,
+      "loss": 1.0052,
+      "step": 9496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998910459126022e-05,
+      "loss": 1.1659,
+      "step": 9497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989102272393176e-05,
+      "loss": 1.3724,
+      "step": 9498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998909995327946e-05,
+      "loss": 2.5844,
+      "step": 9499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989097633919055e-05,
+      "loss": 1.0749,
+      "step": 9500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0519160032272339,
+      "eval_runtime": 86.0341,
+      "eval_samples_per_second": 16.098,
+      "eval_steps_per_second": 4.033,
+      "step": 9500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998909531431197e-05,
+      "loss": 1.1633,
+      "step": 9501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989092994458195e-05,
+      "loss": 0.8895,
+      "step": 9502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998909067435775e-05,
+      "loss": 1.1799,
+      "step": 9503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989088354010616e-05,
+      "loss": 0.8809,
+      "step": 9504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989086033416806e-05,
+      "loss": 0.9955,
+      "step": 9505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998908371257631e-05,
+      "loss": 1.0974,
+      "step": 9506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989081391489135e-05,
+      "loss": 1.0556,
+      "step": 9507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989079070155275e-05,
+      "loss": 0.9914,
+      "step": 9508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998907674857474e-05,
+      "loss": 1.2065,
+      "step": 9509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998907442674752e-05,
+      "loss": 1.0535,
+      "step": 9510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998907210467361e-05,
+      "loss": 1.067,
+      "step": 9511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989069782353024e-05,
+      "loss": 1.2242,
+      "step": 9512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998906745978576e-05,
+      "loss": 1.039,
+      "step": 9513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998906513697181e-05,
+      "loss": 0.7001,
+      "step": 9514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998906281391118e-05,
+      "loss": 0.5576,
+      "step": 9515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998906049060387e-05,
+      "loss": 0.5156,
+      "step": 9516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989058167049884e-05,
+      "loss": 0.303,
+      "step": 9517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998905584324921e-05,
+      "loss": 0.194,
+      "step": 9518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998905351920186e-05,
+      "loss": 1.6149,
+      "step": 9519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989051194907815e-05,
+      "loss": 5.9522,
+      "step": 9520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989048870367096e-05,
+      "loss": 5.3922,
+      "step": 9521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99890465455797e-05,
+      "loss": 5.8952,
+      "step": 9522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998904422054562e-05,
+      "loss": 5.8887,
+      "step": 9523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998904189526486e-05,
+      "loss": 5.8202,
+      "step": 9524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998903956973741e-05,
+      "loss": 5.7676,
+      "step": 9525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998903724396329e-05,
+      "loss": 5.8913,
+      "step": 9526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998903491794248e-05,
+      "loss": 5.7219,
+      "step": 9527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989032591675e-05,
+      "loss": 5.3908,
+      "step": 9528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998903026516083e-05,
+      "loss": 5.8612,
+      "step": 9529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998902793839998e-05,
+      "loss": 5.416,
+      "step": 9530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998902561139245e-05,
+      "loss": 5.6954,
+      "step": 9531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989023284138234e-05,
+      "loss": 5.797,
+      "step": 9532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989020956637344e-05,
+      "loss": 5.6869,
+      "step": 9533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998901862888977e-05,
+      "loss": 5.8472,
+      "step": 9534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989016300895514e-05,
+      "loss": 5.7293,
+      "step": 9535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989013972654575e-05,
+      "loss": 5.6829,
+      "step": 9536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998901164416696e-05,
+      "loss": 5.8825,
+      "step": 9537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998900931543267e-05,
+      "loss": 5.7073,
+      "step": 9538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989006986451685e-05,
+      "loss": 5.5586,
+      "step": 9539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998900465722403e-05,
+      "loss": 5.7517,
+      "step": 9540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989002327749686e-05,
+      "loss": 4.9932,
+      "step": 9541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998899999802866e-05,
+      "loss": 5.8724,
+      "step": 9542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998899766806096e-05,
+      "loss": 5.8379,
+      "step": 9543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998899533784658e-05,
+      "loss": 5.7381,
+      "step": 9544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998899300738551e-05,
+      "loss": 5.697,
+      "step": 9545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988990676677774e-05,
+      "loss": 5.7577,
+      "step": 9546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998898834572334e-05,
+      "loss": 5.6929,
+      "step": 9547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998898601452223e-05,
+      "loss": 5.6557,
+      "step": 9548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988983683074445e-05,
+      "loss": 5.7832,
+      "step": 9549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998898135137998e-05,
+      "loss": 5.6743,
+      "step": 9550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998897901943883e-05,
+      "loss": 5.754,
+      "step": 9551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988976687250996e-05,
+      "loss": 5.1066,
+      "step": 9552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998897435481649e-05,
+      "loss": 1.4271,
+      "step": 9553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988972022135304e-05,
+      "loss": 1.3299,
+      "step": 9554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998896968920743e-05,
+      "loss": 0.9807,
+      "step": 9555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998896735603288e-05,
+      "loss": 1.043,
+      "step": 9556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998896502261164e-05,
+      "loss": 0.8069,
+      "step": 9557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998896268894373e-05,
+      "loss": 1.1061,
+      "step": 9558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998896035502914e-05,
+      "loss": 1.2951,
+      "step": 9559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998895802086787e-05,
+      "loss": 0.7709,
+      "step": 9560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998895568645992e-05,
+      "loss": 0.327,
+      "step": 9561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998895335180528e-05,
+      "loss": 0.6259,
+      "step": 9562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998895101690396e-05,
+      "loss": 1.0665,
+      "step": 9563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988948681755965e-05,
+      "loss": 1.0852,
+      "step": 9564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998894634636129e-05,
+      "loss": 0.8451,
+      "step": 9565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998894401071993e-05,
+      "loss": 1.0385,
+      "step": 9566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99889416748319e-05,
+      "loss": 1.1364,
+      "step": 9567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998893933869718e-05,
+      "loss": 0.9963,
+      "step": 9568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998893700231578e-05,
+      "loss": 1.2045,
+      "step": 9569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988934665687704e-05,
+      "loss": 1.1008,
+      "step": 9570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988932328812945e-05,
+      "loss": 1.0951,
+      "step": 9571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998892999169151e-05,
+      "loss": 1.003,
+      "step": 9572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998892765432339e-05,
+      "loss": 1.1747,
+      "step": 9573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998892531670859e-05,
+      "loss": 1.0342,
+      "step": 9574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988922978847115e-05,
+      "loss": 0.7845,
+      "step": 9575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988920640738954e-05,
+      "loss": 1.0553,
+      "step": 9576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988918302384116e-05,
+      "loss": 0.9495,
+      "step": 9577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99889159637826e-05,
+      "loss": 1.1584,
+      "step": 9578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99889136249344e-05,
+      "loss": 0.8282,
+      "step": 9579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988911285839523e-05,
+      "loss": 1.1773,
+      "step": 9580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998890894649796e-05,
+      "loss": 1.0359,
+      "step": 9581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998890660690973e-05,
+      "loss": 1.1669,
+      "step": 9582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988904267074804e-05,
+      "loss": 0.765,
+      "step": 9583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988901926993204e-05,
+      "loss": 1.1385,
+      "step": 9584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998889958666493e-05,
+      "loss": 1.0772,
+      "step": 9585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998889724608997e-05,
+      "loss": 1.0362,
+      "step": 9586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998889490526833e-05,
+      "loss": 1.168,
+      "step": 9587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998889256420001e-05,
+      "loss": 1.2161,
+      "step": 9588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998889022288501e-05,
+      "loss": 1.0458,
+      "step": 9589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998888788132333e-05,
+      "loss": 1.1914,
+      "step": 9590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998888553951498e-05,
+      "loss": 1.0189,
+      "step": 9591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998888319745994e-05,
+      "loss": 0.2751,
+      "step": 9592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988880855158226e-05,
+      "loss": 0.9459,
+      "step": 9593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998887851260983e-05,
+      "loss": 1.2398,
+      "step": 9594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998887616981475e-05,
+      "loss": 1.1093,
+      "step": 9595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988873826772994e-05,
+      "loss": 1.1416,
+      "step": 9596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988871483484555e-05,
+      "loss": 0.9976,
+      "step": 9597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998886913994945e-05,
+      "loss": 0.247,
+      "step": 9598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998886679616765e-05,
+      "loss": 0.4209,
+      "step": 9599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988864452139174e-05,
+      "loss": 1.2544,
+      "step": 9600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998886210786402e-05,
+      "loss": 0.9261,
+      "step": 9601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998885976334219e-05,
+      "loss": 1.3198,
+      "step": 9602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998885741857368e-05,
+      "loss": 1.3817,
+      "step": 9603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988855073558484e-05,
+      "loss": 0.9208,
+      "step": 9604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998885272829661e-05,
+      "loss": 1.2108,
+      "step": 9605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988850382788055e-05,
+      "loss": 1.4153,
+      "step": 9606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988848037032825e-05,
+      "loss": 0.9594,
+      "step": 9607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998884569103092e-05,
+      "loss": 1.1309,
+      "step": 9608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988843344782324e-05,
+      "loss": 1.1749,
+      "step": 9609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998884099828706e-05,
+      "loss": 0.9763,
+      "step": 9610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998883865154511e-05,
+      "loss": 1.2261,
+      "step": 9611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988836304556487e-05,
+      "loss": 1.0687,
+      "step": 9612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998883395732118e-05,
+      "loss": 1.0324,
+      "step": 9613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988831609839195e-05,
+      "loss": 1.3793,
+      "step": 9614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998882926211053e-05,
+      "loss": 0.9648,
+      "step": 9615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998882691413518e-05,
+      "loss": 1.2078,
+      "step": 9616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998882456591316e-05,
+      "loss": 1.2542,
+      "step": 9617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988822217444455e-05,
+      "loss": 1.3269,
+      "step": 9618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988819868729076e-05,
+      "loss": 1.1625,
+      "step": 9619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988817519767014e-05,
+      "loss": 1.1604,
+      "step": 9620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988815170558276e-05,
+      "loss": 1.0048,
+      "step": 9621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988812821102854e-05,
+      "loss": 0.9177,
+      "step": 9622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988810471400756e-05,
+      "loss": 0.7633,
+      "step": 9623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998880812145198e-05,
+      "loss": 0.6968,
+      "step": 9624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998880577125652e-05,
+      "loss": 1.1713,
+      "step": 9625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998880342081439e-05,
+      "loss": 1.2567,
+      "step": 9626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988801070125574e-05,
+      "loss": 0.8987,
+      "step": 9627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988798719190074e-05,
+      "loss": 1.107,
+      "step": 9628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988796368007905e-05,
+      "loss": 1.1169,
+      "step": 9629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998879401657905e-05,
+      "loss": 1.0902,
+      "step": 9630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998879166490353e-05,
+      "loss": 1.0164,
+      "step": 9631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998878931298132e-05,
+      "loss": 0.9741,
+      "step": 9632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998878696081244e-05,
+      "loss": 0.9065,
+      "step": 9633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988784608396865e-05,
+      "loss": 1.2073,
+      "step": 9634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988782255734624e-05,
+      "loss": 1.1442,
+      "step": 9635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99887799028257e-05,
+      "loss": 1.2982,
+      "step": 9636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998877754967009e-05,
+      "loss": 1.0833,
+      "step": 9637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998877519626782e-05,
+      "loss": 1.2827,
+      "step": 9638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998877284261886e-05,
+      "loss": 1.3037,
+      "step": 9639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998877048872322e-05,
+      "loss": 1.1347,
+      "step": 9640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99887681345809e-05,
+      "loss": 1.0885,
+      "step": 9641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988765780191906e-05,
+      "loss": 1.1442,
+      "step": 9642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988763425556234e-05,
+      "loss": 0.748,
+      "step": 9643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988761070673886e-05,
+      "loss": 1.2061,
+      "step": 9644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998875871554485e-05,
+      "loss": 1.2798,
+      "step": 9645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998875636016914e-05,
+      "loss": 1.0119,
+      "step": 9646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998875400454676e-05,
+      "loss": 1.0513,
+      "step": 9647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998875164867769e-05,
+      "loss": 1.0932,
+      "step": 9648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998874929256194e-05,
+      "loss": 1.0462,
+      "step": 9649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998874693619952e-05,
+      "loss": 0.8771,
+      "step": 9650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998874457959042e-05,
+      "loss": 0.9754,
+      "step": 9651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998874222273464e-05,
+      "loss": 1.2265,
+      "step": 9652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998873986563218e-05,
+      "loss": 1.1009,
+      "step": 9653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988737508283044e-05,
+      "loss": 1.1044,
+      "step": 9654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998873515068723e-05,
+      "loss": 1.0872,
+      "step": 9655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998873279284474e-05,
+      "loss": 1.3058,
+      "step": 9656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988730434755564e-05,
+      "loss": 0.9225,
+      "step": 9657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988728076419714e-05,
+      "loss": 0.9627,
+      "step": 9658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998872571783719e-05,
+      "loss": 1.1607,
+      "step": 9659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988723359007983e-05,
+      "loss": 1.1587,
+      "step": 9660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99887209999321e-05,
+      "loss": 1.2425,
+      "step": 9661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988718640609535e-05,
+      "loss": 1.0909,
+      "step": 9662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988716281040296e-05,
+      "loss": 1.0877,
+      "step": 9663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998871392122438e-05,
+      "loss": 1.1519,
+      "step": 9664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998871156116178e-05,
+      "loss": 1.5058,
+      "step": 9665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998870920085251e-05,
+      "loss": 1.6072,
+      "step": 9666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998870684029655e-05,
+      "loss": 1.698,
+      "step": 9667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998870447949392e-05,
+      "loss": 1.301,
+      "step": 9668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998870211844462e-05,
+      "loss": 1.3828,
+      "step": 9669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988699757148625e-05,
+      "loss": 1.5221,
+      "step": 9670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998869739560597e-05,
+      "loss": 1.0533,
+      "step": 9671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998869503381662e-05,
+      "loss": 0.8912,
+      "step": 9672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99886926717806e-05,
+      "loss": 0.7509,
+      "step": 9673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99886903094979e-05,
+      "loss": 0.8431,
+      "step": 9674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988687946968526e-05,
+      "loss": 0.188,
+      "step": 9675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988685584192475e-05,
+      "loss": 0.6945,
+      "step": 9676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998868322116975e-05,
+      "loss": 1.049,
+      "step": 9677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988680857900336e-05,
+      "loss": 1.0678,
+      "step": 9678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998867849438425e-05,
+      "loss": 1.2841,
+      "step": 9679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988676130621485e-05,
+      "loss": 1.0755,
+      "step": 9680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988673766612045e-05,
+      "loss": 0.8511,
+      "step": 9681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998867140235592e-05,
+      "loss": 1.0256,
+      "step": 9682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998866903785312e-05,
+      "loss": 1.1308,
+      "step": 9683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988666673103654e-05,
+      "loss": 0.9747,
+      "step": 9684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99886643081075e-05,
+      "loss": 1.1631,
+      "step": 9685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998866194286467e-05,
+      "loss": 1.09,
+      "step": 9686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998865957737516e-05,
+      "loss": 1.8164,
+      "step": 9687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998865721163898e-05,
+      "loss": 1.7847,
+      "step": 9688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998865484565611e-05,
+      "loss": 1.7774,
+      "step": 9689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998865247942657e-05,
+      "loss": 1.3281,
+      "step": 9690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988650112950354e-05,
+      "loss": 1.3932,
+      "step": 9691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988647746227454e-05,
+      "loss": 1.4631,
+      "step": 9692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998864537925789e-05,
+      "loss": 1.2774,
+      "step": 9693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998864301204164e-05,
+      "loss": 1.2605,
+      "step": 9694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998864064457871e-05,
+      "loss": 0.9692,
+      "step": 9695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988638276869106e-05,
+      "loss": 1.1142,
+      "step": 9696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998863590891283e-05,
+      "loss": 1.2291,
+      "step": 9697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988633540709867e-05,
+      "loss": 1.0415,
+      "step": 9698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998863117226023e-05,
+      "loss": 1.2165,
+      "step": 9699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998862880356392e-05,
+      "loss": 1.092,
+      "step": 9700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998862643462092e-05,
+      "loss": 1.5197,
+      "step": 9701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998862406543125e-05,
+      "loss": 1.032,
+      "step": 9702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998862169599491e-05,
+      "loss": 1.2112,
+      "step": 9703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998861932631189e-05,
+      "loss": 1.0574,
+      "step": 9704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998861695638219e-05,
+      "loss": 1.1226,
+      "step": 9705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998861458620582e-05,
+      "loss": 1.1803,
+      "step": 9706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998861221578276e-05,
+      "loss": 1.3568,
+      "step": 9707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998860984511303e-05,
+      "loss": 1.6537,
+      "step": 9708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998860747419663e-05,
+      "loss": 1.0766,
+      "step": 9709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998860510303354e-05,
+      "loss": 1.119,
+      "step": 9710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988602731623776e-05,
+      "loss": 1.1105,
+      "step": 9711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998860035996734e-05,
+      "loss": 0.682,
+      "step": 9712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998859798806422e-05,
+      "loss": 1.0323,
+      "step": 9713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998859561591443e-05,
+      "loss": 0.7753,
+      "step": 9714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998859324351796e-05,
+      "loss": 1.0854,
+      "step": 9715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998859087087482e-05,
+      "loss": 0.9504,
+      "step": 9716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988588497984986e-05,
+      "loss": 1.2355,
+      "step": 9717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998858612484849e-05,
+      "loss": 1.2058,
+      "step": 9718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988583751465314e-05,
+      "loss": 1.6141,
+      "step": 9719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988581377835464e-05,
+      "loss": 1.2754,
+      "step": 9720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998857900395894e-05,
+      "loss": 0.976,
+      "step": 9721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998857662983573e-05,
+      "loss": 1.2198,
+      "step": 9722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998857425546584e-05,
+      "loss": 0.9265,
+      "step": 9723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998857188084928e-05,
+      "loss": 1.2309,
+      "step": 9724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998856950598604e-05,
+      "loss": 0.8348,
+      "step": 9725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998856713087613e-05,
+      "loss": 1.1515,
+      "step": 9726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998856475551954e-05,
+      "loss": 1.0123,
+      "step": 9727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998856237991627e-05,
+      "loss": 1.0243,
+      "step": 9728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998856000406633e-05,
+      "loss": 1.0206,
+      "step": 9729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998855762796971e-05,
+      "loss": 1.1601,
+      "step": 9730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998855525162641e-05,
+      "loss": 0.9926,
+      "step": 9731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988552875036446e-05,
+      "loss": 0.7021,
+      "step": 9732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988550498199796e-05,
+      "loss": 0.964,
+      "step": 9733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998854812111647e-05,
+      "loss": 0.972,
+      "step": 9734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998854574378646e-05,
+      "loss": 1.144,
+      "step": 9735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988543366209784e-05,
+      "loss": 1.2402,
+      "step": 9736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988540988386436e-05,
+      "loss": 1.0957,
+      "step": 9737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988538610316405e-05,
+      "loss": 1.081,
+      "step": 9738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998853623199969e-05,
+      "loss": 1.106,
+      "step": 9739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998853385343631e-05,
+      "loss": 0.6211,
+      "step": 9740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998853147462625e-05,
+      "loss": 1.0139,
+      "step": 9741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998852909556951e-05,
+      "loss": 1.1727,
+      "step": 9742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988526716266106e-05,
+      "loss": 0.9099,
+      "step": 9743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998852433671602e-05,
+      "loss": 0.9932,
+      "step": 9744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988521956919245e-05,
+      "loss": 0.7721,
+      "step": 9745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99885195768758e-05,
+      "loss": 1.2619,
+      "step": 9746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988517196585685e-05,
+      "loss": 1.2188,
+      "step": 9747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998851481604889e-05,
+      "loss": 1.2085,
+      "step": 9748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998851243526542e-05,
+      "loss": 0.9917,
+      "step": 9749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998851005423528e-05,
+      "loss": 1.3981,
+      "step": 9750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988507672958464e-05,
+      "loss": 1.2426,
+      "step": 9751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988505291434965e-05,
+      "loss": 1.0428,
+      "step": 9752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998850290966479e-05,
+      "loss": 1.1453,
+      "step": 9753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988500527647944e-05,
+      "loss": 1.0549,
+      "step": 9754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998849814538441e-05,
+      "loss": 1.0695,
+      "step": 9755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998849576287421e-05,
+      "loss": 1.1746,
+      "step": 9756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988493380117336e-05,
+      "loss": 1.2758,
+      "step": 9757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998849099711378e-05,
+      "loss": 1.1487,
+      "step": 9758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998848861386355e-05,
+      "loss": 1.1116,
+      "step": 9759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998848623036665e-05,
+      "loss": 1.1157,
+      "step": 9760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998848384662307e-05,
+      "loss": 1.0451,
+      "step": 9761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988481462632806e-05,
+      "loss": 0.8861,
+      "step": 9762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998847907839588e-05,
+      "loss": 1.2327,
+      "step": 9763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988476693912266e-05,
+      "loss": 1.5421,
+      "step": 9764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998847430918199e-05,
+      "loss": 0.8174,
+      "step": 9765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988471924205026e-05,
+      "loss": 0.9378,
+      "step": 9766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998846953898139e-05,
+      "loss": 1.1805,
+      "step": 9767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998846715351108e-05,
+      "loss": 1.1691,
+      "step": 9768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99884647677941e-05,
+      "loss": 1.0429,
+      "step": 9769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998846238183044e-05,
+      "loss": 1.0347,
+      "step": 9770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988459995620096e-05,
+      "loss": 1.0842,
+      "step": 9771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988457609163083e-05,
+      "loss": 1.4174,
+      "step": 9772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988455222459395e-05,
+      "loss": 1.3438,
+      "step": 9773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998845283550902e-05,
+      "loss": 1.3013,
+      "step": 9774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998845044831199e-05,
+      "loss": 0.4824,
+      "step": 9775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998844806086828e-05,
+      "loss": 0.2471,
+      "step": 9776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988445673177884e-05,
+      "loss": 0.2471,
+      "step": 9777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998844328524081e-05,
+      "loss": 0.1561,
+      "step": 9778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998844089705707e-05,
+      "loss": 0.1945,
+      "step": 9779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988438508626664e-05,
+      "loss": 0.1443,
+      "step": 9780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988436119949565e-05,
+      "loss": 0.1177,
+      "step": 9781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99884337310258e-05,
+      "loss": 0.1291,
+      "step": 9782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998843134185536e-05,
+      "loss": 0.2359,
+      "step": 9783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988428952438236e-05,
+      "loss": 0.3172,
+      "step": 9784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998842656277445e-05,
+      "loss": 0.2112,
+      "step": 9785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998842417286398e-05,
+      "loss": 0.8715,
+      "step": 9786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998842178270684e-05,
+      "loss": 1.3531,
+      "step": 9787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998841939230302e-05,
+      "loss": 1.3317,
+      "step": 9788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998841700165252e-05,
+      "loss": 1.3058,
+      "step": 9789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998841461075535e-05,
+      "loss": 1.2345,
+      "step": 9790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988412219611516e-05,
+      "loss": 0.7695,
+      "step": 9791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998840982822099e-05,
+      "loss": 0.8632,
+      "step": 9792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988407436583796e-05,
+      "loss": 1.1626,
+      "step": 9793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988405044699925e-05,
+      "loss": 1.0613,
+      "step": 9794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998840265256939e-05,
+      "loss": 1.042,
+      "step": 9795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998840026019217e-05,
+      "loss": 0.3696,
+      "step": 9796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998839786756827e-05,
+      "loss": 0.77,
+      "step": 9797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988395474697705e-05,
+      "loss": 1.2609,
+      "step": 9798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998839308158046e-05,
+      "loss": 1.0351,
+      "step": 9799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998839068821654e-05,
+      "loss": 2.8369,
+      "step": 9800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998838829460595e-05,
+      "loss": 1.3971,
+      "step": 9801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998838590074868e-05,
+      "loss": 1.3847,
+      "step": 9802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998838350664474e-05,
+      "loss": 1.1103,
+      "step": 9803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998838111229411e-05,
+      "loss": 1.404,
+      "step": 9804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998837871769683e-05,
+      "loss": 1.1215,
+      "step": 9805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998837632285286e-05,
+      "loss": 0.905,
+      "step": 9806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998837392776221e-05,
+      "loss": 1.1019,
+      "step": 9807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99883715324249e-05,
+      "loss": 0.7928,
+      "step": 9808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998836913684091e-05,
+      "loss": 0.9298,
+      "step": 9809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998836674101024e-05,
+      "loss": 1.22,
+      "step": 9810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99883643449329e-05,
+      "loss": 1.2782,
+      "step": 9811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988361948608885e-05,
+      "loss": 1.0752,
+      "step": 9812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99883595520382e-05,
+      "loss": 1.2075,
+      "step": 9813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998835715522083e-05,
+      "loss": 1.3142,
+      "step": 9814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998835475815679e-05,
+      "loss": 1.3853,
+      "step": 9815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998835236084608e-05,
+      "loss": 1.2796,
+      "step": 9816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998834996328869e-05,
+      "loss": 1.1469,
+      "step": 9817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998834756548463e-05,
+      "loss": 0.8047,
+      "step": 9818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988345167433884e-05,
+      "loss": 1.2134,
+      "step": 9819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988342769136476e-05,
+      "loss": 1.0523,
+      "step": 9820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998834037059239e-05,
+      "loss": 1.109,
+      "step": 9821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998833797180163e-05,
+      "loss": 1.2216,
+      "step": 9822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988335572764193e-05,
+      "loss": 1.0843,
+      "step": 9823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998833317348009e-05,
+      "loss": 1.0799,
+      "step": 9824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988330773949304e-05,
+      "loss": 1.2196,
+      "step": 9825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988328374171845e-05,
+      "loss": 1.1059,
+      "step": 9826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988325974147716e-05,
+      "loss": 1.1167,
+      "step": 9827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988323573876904e-05,
+      "loss": 1.1676,
+      "step": 9828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998832117335943e-05,
+      "loss": 0.9388,
+      "step": 9829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998831877259528e-05,
+      "loss": 0.8102,
+      "step": 9830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988316371584444e-05,
+      "loss": 0.9359,
+      "step": 9831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998831397032694e-05,
+      "loss": 1.1201,
+      "step": 9832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998831156882277e-05,
+      "loss": 1.09,
+      "step": 9833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998830916707191e-05,
+      "loss": 0.981,
+      "step": 9834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998830676507439e-05,
+      "loss": 0.9748,
+      "step": 9835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998830436283019e-05,
+      "loss": 1.1448,
+      "step": 9836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998830196033932e-05,
+      "loss": 0.9496,
+      "step": 9837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998829955760177e-05,
+      "loss": 1.2107,
+      "step": 9838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988297154617546e-05,
+      "loss": 0.8966,
+      "step": 9839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998829475138666e-05,
+      "loss": 0.731,
+      "step": 9840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998829234790908e-05,
+      "loss": 1.031,
+      "step": 9841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998828994418484e-05,
+      "loss": 0.3472,
+      "step": 9842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998828754021393e-05,
+      "loss": 0.3417,
+      "step": 9843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998828513599634e-05,
+      "loss": 0.3128,
+      "step": 9844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998828273153208e-05,
+      "loss": 0.3333,
+      "step": 9845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998828032682113e-05,
+      "loss": 0.5605,
+      "step": 9846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988277921863525e-05,
+      "loss": 0.2648,
+      "step": 9847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998827551665924e-05,
+      "loss": 0.2204,
+      "step": 9848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998827311120828e-05,
+      "loss": 0.2649,
+      "step": 9849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988270705510644e-05,
+      "loss": 1.7215,
+      "step": 9850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998826829956634e-05,
+      "loss": 1.6377,
+      "step": 9851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988265893375354e-05,
+      "loss": 1.4358,
+      "step": 9852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99882634869377e-05,
+      "loss": 1.1824,
+      "step": 9853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988261080253374e-05,
+      "loss": 1.102,
+      "step": 9854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988258673322376e-05,
+      "loss": 1.1893,
+      "step": 9855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99882562661447e-05,
+      "loss": 0.9691,
+      "step": 9856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998825385872035e-05,
+      "loss": 0.7626,
+      "step": 9857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998825145104933e-05,
+      "loss": 0.5945,
+      "step": 9858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988249043131634e-05,
+      "loss": 1.1725,
+      "step": 9859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998824663496727e-05,
+      "loss": 1.2295,
+      "step": 9860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988244226556225e-05,
+      "loss": 1.0714,
+      "step": 9861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988241817898506e-05,
+      "loss": 0.6421,
+      "step": 9862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998823940899412e-05,
+      "loss": 1.0365,
+      "step": 9863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998823699984306e-05,
+      "loss": 1.0579,
+      "step": 9864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988234590445326e-05,
+      "loss": 0.9681,
+      "step": 9865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988232180800915e-05,
+      "loss": 1.1143,
+      "step": 9866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988229770909835e-05,
+      "loss": 1.2837,
+      "step": 9867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998822736077208e-05,
+      "loss": 1.1725,
+      "step": 9868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998822495038765e-05,
+      "loss": 1.0829,
+      "step": 9869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988222539756544e-05,
+      "loss": 1.0975,
+      "step": 9870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998822012887877e-05,
+      "loss": 1.1761,
+      "step": 9871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988217717754324e-05,
+      "loss": 1.1875,
+      "step": 9872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99882153063832e-05,
+      "loss": 1.2365,
+      "step": 9873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988212894765405e-05,
+      "loss": 1.3565,
+      "step": 9874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988210482900935e-05,
+      "loss": 1.0789,
+      "step": 9875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99882080707898e-05,
+      "loss": 1.159,
+      "step": 9876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988205658431985e-05,
+      "loss": 1.3166,
+      "step": 9877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99882032458275e-05,
+      "loss": 0.9486,
+      "step": 9878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998820083297634e-05,
+      "loss": 0.9855,
+      "step": 9879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988198419878506e-05,
+      "loss": 0.9638,
+      "step": 9880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988196006534e-05,
+      "loss": 1.309,
+      "step": 9881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998819359294282e-05,
+      "loss": 1.0043,
+      "step": 9882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998819117910497e-05,
+      "loss": 1.1235,
+      "step": 9883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998818876502044e-05,
+      "loss": 1.2024,
+      "step": 9884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998818635068925e-05,
+      "loss": 0.9608,
+      "step": 9885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988183936111374e-05,
+      "loss": 0.9681,
+      "step": 9886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998818152128683e-05,
+      "loss": 1.2117,
+      "step": 9887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988179106215614e-05,
+      "loss": 1.1707,
+      "step": 9888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998817669089773e-05,
+      "loss": 0.6553,
+      "step": 9889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998817427533317e-05,
+      "loss": 1.047,
+      "step": 9890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998817185952193e-05,
+      "loss": 0.9099,
+      "step": 9891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998816944346403e-05,
+      "loss": 1.2208,
+      "step": 9892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998816702715944e-05,
+      "loss": 1.1677,
+      "step": 9893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998816461060819e-05,
+      "loss": 1.0616,
+      "step": 9894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988162193810266e-05,
+      "loss": 1.1913,
+      "step": 9895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998815977676567e-05,
+      "loss": 1.8925,
+      "step": 9896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99881573594744e-05,
+      "loss": 1.3356,
+      "step": 9897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998815494193646e-05,
+      "loss": 1.1028,
+      "step": 9898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998815252415184e-05,
+      "loss": 1.2409,
+      "step": 9899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998815010612056e-05,
+      "loss": 1.1552,
+      "step": 9900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998814768784259e-05,
+      "loss": 1.0364,
+      "step": 9901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998814526931796e-05,
+      "loss": 0.6646,
+      "step": 9902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998814285054666e-05,
+      "loss": 0.9307,
+      "step": 9903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998814043152868e-05,
+      "loss": 0.902,
+      "step": 9904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998813801226403e-05,
+      "loss": 1.3225,
+      "step": 9905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998813559275271e-05,
+      "loss": 1.0705,
+      "step": 9906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998813317299471e-05,
+      "loss": 1.1976,
+      "step": 9907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998813075299005e-05,
+      "loss": 0.8782,
+      "step": 9908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988128332738706e-05,
+      "loss": 1.282,
+      "step": 9909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998812591224069e-05,
+      "loss": 1.3072,
+      "step": 9910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998812349149601e-05,
+      "loss": 1.2509,
+      "step": 9911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998812107050466e-05,
+      "loss": 1.2104,
+      "step": 9912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998811864926662e-05,
+      "loss": 1.1429,
+      "step": 9913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988116227781925e-05,
+      "loss": 1.34,
+      "step": 9914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998811380605055e-05,
+      "loss": 0.9676,
+      "step": 9915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988111384072514e-05,
+      "loss": 0.9763,
+      "step": 9916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998810896184779e-05,
+      "loss": 1.2044,
+      "step": 9917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99881065393764e-05,
+      "loss": 1.0246,
+      "step": 9918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998810411665834e-05,
+      "loss": 1.0786,
+      "step": 9919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998810169369361e-05,
+      "loss": 1.0917,
+      "step": 9920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988099270482205e-05,
+      "loss": 1.1904,
+      "step": 9921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988096847024124e-05,
+      "loss": 1.2247,
+      "step": 9922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998809442331937e-05,
+      "loss": 1.1823,
+      "step": 9923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998809199936796e-05,
+      "loss": 1.0782,
+      "step": 9924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998808957516986e-05,
+      "loss": 0.9249,
+      "step": 9925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988087150725096e-05,
+      "loss": 1.1004,
+      "step": 9926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998808472603366e-05,
+      "loss": 1.1934,
+      "step": 9927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998808230109555e-05,
+      "loss": 1.3222,
+      "step": 9928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998807987591077e-05,
+      "loss": 1.2111,
+      "step": 9929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998807745047932e-05,
+      "loss": 1.0947,
+      "step": 9930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988075024801194e-05,
+      "loss": 1.2274,
+      "step": 9931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988072598876404e-05,
+      "loss": 1.1896,
+      "step": 9932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998807017270494e-05,
+      "loss": 1.0705,
+      "step": 9933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99880677462868e-05,
+      "loss": 1.2374,
+      "step": 9934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998806531962199e-05,
+      "loss": 1.466,
+      "step": 9935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988062892710506e-05,
+      "loss": 0.999,
+      "step": 9936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998806046555235e-05,
+      "loss": 1.0264,
+      "step": 9937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998805803814752e-05,
+      "loss": 1.4109,
+      "step": 9938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998805561049603e-05,
+      "loss": 1.1204,
+      "step": 9939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998805318259786e-05,
+      "loss": 1.11,
+      "step": 9940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998805075445302e-05,
+      "loss": 0.5099,
+      "step": 9941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998804832606151e-05,
+      "loss": 0.4688,
+      "step": 9942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998804589742332e-05,
+      "loss": 0.3469,
+      "step": 9943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998804346853847e-05,
+      "loss": 0.1816,
+      "step": 9944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998804103940694e-05,
+      "loss": 0.1119,
+      "step": 9945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988038610028745e-05,
+      "loss": 0.1391,
+      "step": 9946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998803618040388e-05,
+      "loss": 0.7113,
+      "step": 9947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998803375053234e-05,
+      "loss": 1.4228,
+      "step": 9948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998803132041413e-05,
+      "loss": 1.482,
+      "step": 9949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988028890049244e-05,
+      "loss": 1.5651,
+      "step": 9950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988026459437685e-05,
+      "loss": 1.1674,
+      "step": 9951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998802402857946e-05,
+      "loss": 1.149,
+      "step": 9952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988021597474565e-05,
+      "loss": 1.245,
+      "step": 9953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988019166123004e-05,
+      "loss": 1.153,
+      "step": 9954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998801673452476e-05,
+      "loss": 0.9997,
+      "step": 9955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998801430267985e-05,
+      "loss": 1.0358,
+      "step": 9956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998801187058827e-05,
+      "loss": 1.0735,
+      "step": 9957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988009438250017e-05,
+      "loss": 1.0173,
+      "step": 9958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988007005665094e-05,
+      "loss": 1.2286,
+      "step": 9959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988004572833496e-05,
+      "loss": 1.1883,
+      "step": 9960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998800213975523e-05,
+      "loss": 1.0479,
+      "step": 9961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987999706430296e-05,
+      "loss": 1.4511,
+      "step": 9962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998799727285869e-05,
+      "loss": 1.6647,
+      "step": 9963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987994839040405e-05,
+      "loss": 1.3232,
+      "step": 9964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998799240497546e-05,
+      "loss": 1.1381,
+      "step": 9965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998798997066384e-05,
+      "loss": 0.5195,
+      "step": 9966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987987536105544e-05,
+      "loss": 1.1474,
+      "step": 9967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998798510130058e-05,
+      "loss": 0.6412,
+      "step": 9968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998798266624895e-05,
+      "loss": 0.6699,
+      "step": 9969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998798023095065e-05,
+      "loss": 1.0617,
+      "step": 9970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998797779540567e-05,
+      "loss": 1.1165,
+      "step": 9971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987975359614026e-05,
+      "loss": 1.2524,
+      "step": 9972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998797292357571e-05,
+      "loss": 1.064,
+      "step": 9973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987970487290724e-05,
+      "loss": 1.1155,
+      "step": 9974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998796805075907e-05,
+      "loss": 1.4537,
+      "step": 9975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987965613980737e-05,
+      "loss": 1.5731,
+      "step": 9976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987963176955735e-05,
+      "loss": 1.3807,
+      "step": 9977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998796073968407e-05,
+      "loss": 1.4099,
+      "step": 9978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987958302165724e-05,
+      "loss": 1.3611,
+      "step": 9979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987955864400714e-05,
+      "loss": 1.2911,
+      "step": 9980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987953426389035e-05,
+      "loss": 1.2902,
+      "step": 9981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987950988130686e-05,
+      "loss": 1.2613,
+      "step": 9982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998794854962566e-05,
+      "loss": 1.1833,
+      "step": 9983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987946110873965e-05,
+      "loss": 1.0884,
+      "step": 9984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99879436718756e-05,
+      "loss": 1.4898,
+      "step": 9985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998794123263057e-05,
+      "loss": 1.5612,
+      "step": 9986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987938793138863e-05,
+      "loss": 1.1664,
+      "step": 9987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998793635340049e-05,
+      "loss": 1.3252,
+      "step": 9988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998793391341544e-05,
+      "loss": 1.0936,
+      "step": 9989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998793147318372e-05,
+      "loss": 1.1245,
+      "step": 9990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998792903270534e-05,
+      "loss": 1.123,
+      "step": 9991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998792659198028e-05,
+      "loss": 1.253,
+      "step": 9992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998792415100856e-05,
+      "loss": 1.044,
+      "step": 9993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998792170979016e-05,
+      "loss": 1.0668,
+      "step": 9994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998791926832509e-05,
+      "loss": 1.3681,
+      "step": 9995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987916826613355e-05,
+      "loss": 1.0164,
+      "step": 9996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998791438465494e-05,
+      "loss": 1.3209,
+      "step": 9997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987911942449874e-05,
+      "loss": 0.9155,
+      "step": 9998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998790949999812e-05,
+      "loss": 1.1548,
+      "step": 9999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998790705729971e-05,
+      "loss": 1.1474,
+      "step": 10000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0545191764831543,
+      "eval_runtime": 85.9243,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 4.038,
+      "step": 10000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998790461435462e-05,
+      "loss": 0.9587,
+      "step": 10001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987902171162856e-05,
+      "loss": 1.1316,
+      "step": 10002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987899727724427e-05,
+      "loss": 1.2127,
+      "step": 10003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987897284039334e-05,
+      "loss": 1.0664,
+      "step": 10004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987894840107565e-05,
+      "loss": 1.0854,
+      "step": 10005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987892395929127e-05,
+      "loss": 1.1176,
+      "step": 10006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998788995150402e-05,
+      "loss": 1.1545,
+      "step": 10007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998788750683224e-05,
+      "loss": 1.1105,
+      "step": 10008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987885061913794e-05,
+      "loss": 1.8349,
+      "step": 10009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998788261674867e-05,
+      "loss": 1.2364,
+      "step": 10010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987880171336884e-05,
+      "loss": 1.3148,
+      "step": 10011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998787772567843e-05,
+      "loss": 0.9995,
+      "step": 10012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987875279773296e-05,
+      "loss": 1.2341,
+      "step": 10013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998787283362151e-05,
+      "loss": 1.3008,
+      "step": 10014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998787038722304e-05,
+      "loss": 0.7189,
+      "step": 10015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99878679405779e-05,
+      "loss": 1.0264,
+      "step": 10016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998786549368609e-05,
+      "loss": 1.1821,
+      "step": 10017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998786304654762e-05,
+      "loss": 0.9963,
+      "step": 10018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998786059916247e-05,
+      "loss": 1.0399,
+      "step": 10019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998785815153066e-05,
+      "loss": 1.0536,
+      "step": 10020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998785570365218e-05,
+      "loss": 1.2832,
+      "step": 10021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998785325552702e-05,
+      "loss": 1.0504,
+      "step": 10022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99878508071552e-05,
+      "loss": 0.9803,
+      "step": 10023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99878483585367e-05,
+      "loss": 0.9373,
+      "step": 10024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998784590967154e-05,
+      "loss": 1.2109,
+      "step": 10025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987843460559705e-05,
+      "loss": 1.3152,
+      "step": 10026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987841011201206e-05,
+      "loss": 1.4545,
+      "step": 10027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998783856159604e-05,
+      "loss": 1.1669,
+      "step": 10028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998783611174419e-05,
+      "loss": 1.2884,
+      "step": 10029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998783366164568e-05,
+      "loss": 2.1015,
+      "step": 10030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99878312113005e-05,
+      "loss": 1.3097,
+      "step": 10031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987828760708654e-05,
+      "loss": 0.4655,
+      "step": 10032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998782630987013e-05,
+      "loss": 0.5985,
+      "step": 10033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987823858784946e-05,
+      "loss": 1.4386,
+      "step": 10034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998782140745309e-05,
+      "loss": 1.3263,
+      "step": 10035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987818955874565e-05,
+      "loss": 1.1649,
+      "step": 10036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987816504049364e-05,
+      "loss": 1.2357,
+      "step": 10037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998781405197751e-05,
+      "loss": 1.011,
+      "step": 10038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987811599658973e-05,
+      "loss": 1.3485,
+      "step": 10039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998780914709377e-05,
+      "loss": 1.03,
+      "step": 10040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99878066942819e-05,
+      "loss": 0.361,
+      "step": 10041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987804241223356e-05,
+      "loss": 1.4018,
+      "step": 10042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987801787918144e-05,
+      "loss": 1.2243,
+      "step": 10043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998779933436627e-05,
+      "loss": 0.8755,
+      "step": 10044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998779688056772e-05,
+      "loss": 1.2429,
+      "step": 10045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987794426522505e-05,
+      "loss": 1.2466,
+      "step": 10046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987791972230615e-05,
+      "loss": 1.1774,
+      "step": 10047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998778951769206e-05,
+      "loss": 1.3135,
+      "step": 10048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998778706290684e-05,
+      "loss": 1.2678,
+      "step": 10049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998778460787494e-05,
+      "loss": 1.1939,
+      "step": 10050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998778215259639e-05,
+      "loss": 1.1722,
+      "step": 10051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998777969707115e-05,
+      "loss": 1.1416,
+      "step": 10052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998777724129925e-05,
+      "loss": 1.092,
+      "step": 10053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987774785280686e-05,
+      "loss": 1.1459,
+      "step": 10054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987772329015447e-05,
+      "loss": 1.047,
+      "step": 10055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987769872503544e-05,
+      "loss": 1.2231,
+      "step": 10056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987767415744966e-05,
+      "loss": 1.2234,
+      "step": 10057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987764958739724e-05,
+      "loss": 0.9417,
+      "step": 10058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987762501487813e-05,
+      "loss": 0.9673,
+      "step": 10059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987760043989226e-05,
+      "loss": 1.2353,
+      "step": 10060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998775758624398e-05,
+      "loss": 1.1185,
+      "step": 10061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998775512825207e-05,
+      "loss": 1.0758,
+      "step": 10062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987752670013474e-05,
+      "loss": 1.0797,
+      "step": 10063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998775021152822e-05,
+      "loss": 1.1433,
+      "step": 10064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99877477527963e-05,
+      "loss": 0.9668,
+      "step": 10065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987745293817703e-05,
+      "loss": 0.7362,
+      "step": 10066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998774283459244e-05,
+      "loss": 0.916,
+      "step": 10067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998774037512052e-05,
+      "loss": 1.0283,
+      "step": 10068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998773791540192e-05,
+      "loss": 1.2069,
+      "step": 10069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998773545543666e-05,
+      "loss": 0.7493,
+      "step": 10070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987732995224726e-05,
+      "loss": 0.9055,
+      "step": 10071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998773053476612e-05,
+      "loss": 0.9794,
+      "step": 10072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987728074060847e-05,
+      "loss": 1.4686,
+      "step": 10073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987725613108906e-05,
+      "loss": 1.1496,
+      "step": 10074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987723151910296e-05,
+      "loss": 1.0109,
+      "step": 10075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998772069046502e-05,
+      "loss": 0.953,
+      "step": 10076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998771822877308e-05,
+      "loss": 1.1296,
+      "step": 10077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998771576683446e-05,
+      "loss": 0.6741,
+      "step": 10078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998771330464919e-05,
+      "loss": 0.9119,
+      "step": 10079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987710842217236e-05,
+      "loss": 0.5591,
+      "step": 10080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998770837953862e-05,
+      "loss": 0.4912,
+      "step": 10081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998770591661334e-05,
+      "loss": 0.3447,
+      "step": 10082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998770345344138e-05,
+      "loss": 0.2822,
+      "step": 10083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987700990022764e-05,
+      "loss": 0.1761,
+      "step": 10084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998769852635747e-05,
+      "loss": 0.1722,
+      "step": 10085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998769606244552e-05,
+      "loss": 0.4746,
+      "step": 10086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987693598286886e-05,
+      "loss": 1.0607,
+      "step": 10087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99876911338816e-05,
+      "loss": 1.2912,
+      "step": 10088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987688669229635e-05,
+      "loss": 1.1969,
+      "step": 10089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998768620433101e-05,
+      "loss": 1.169,
+      "step": 10090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987683739185706e-05,
+      "loss": 1.293,
+      "step": 10091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998768127379374e-05,
+      "loss": 1.08,
+      "step": 10092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998767880815511e-05,
+      "loss": 0.8878,
+      "step": 10093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987676342269816e-05,
+      "loss": 1.5998,
+      "step": 10094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998767387613784e-05,
+      "loss": 1.1856,
+      "step": 10095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987671409759205e-05,
+      "loss": 1.3133,
+      "step": 10096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987668943133905e-05,
+      "loss": 1.43,
+      "step": 10097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998766647626193e-05,
+      "loss": 0.6464,
+      "step": 10098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998766400914329e-05,
+      "loss": 1.3154,
+      "step": 10099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998766154177798e-05,
+      "loss": 1.3452,
+      "step": 10100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998765907416601e-05,
+      "loss": 0.9236,
+      "step": 10101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987656606307365e-05,
+      "loss": 0.8636,
+      "step": 10102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987654138202055e-05,
+      "loss": 1.3781,
+      "step": 10103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987651669850075e-05,
+      "loss": 1.2948,
+      "step": 10104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998764920125143e-05,
+      "loss": 1.1604,
+      "step": 10105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998764673240612e-05,
+      "loss": 1.1219,
+      "step": 10106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998764426331414e-05,
+      "loss": 1.3069,
+      "step": 10107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987641793975496e-05,
+      "loss": 1.5225,
+      "step": 10108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987639324390175e-05,
+      "loss": 1.0722,
+      "step": 10109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998763685455819e-05,
+      "loss": 0.8805,
+      "step": 10110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998763438447954e-05,
+      "loss": 1.0155,
+      "step": 10111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998763191415423e-05,
+      "loss": 1.0353,
+      "step": 10112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998762944358224e-05,
+      "loss": 1.3936,
+      "step": 10113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998762697276359e-05,
+      "loss": 1.2141,
+      "step": 10114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998762450169827e-05,
+      "loss": 1.3697,
+      "step": 10115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987622030386286e-05,
+      "loss": 1.1035,
+      "step": 10116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998761955882763e-05,
+      "loss": 1.2115,
+      "step": 10117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998761708702231e-05,
+      "loss": 0.8148,
+      "step": 10118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998761461497032e-05,
+      "loss": 0.94,
+      "step": 10119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998761214267166e-05,
+      "loss": 1.0566,
+      "step": 10120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998760967012634e-05,
+      "loss": 1.2239,
+      "step": 10121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998760719733435e-05,
+      "loss": 0.9975,
+      "step": 10122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99876047242957e-05,
+      "loss": 1.152,
+      "step": 10123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998760225101037e-05,
+      "loss": 1.0156,
+      "step": 10124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987599777478376e-05,
+      "loss": 1.3767,
+      "step": 10125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998759730369972e-05,
+      "loss": 0.971,
+      "step": 10126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987594829674395e-05,
+      "loss": 1.2178,
+      "step": 10127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99875923554024e-05,
+      "loss": 1.136,
+      "step": 10128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987589880883744e-05,
+      "loss": 1.0516,
+      "step": 10129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998758740611842e-05,
+      "loss": 1.1713,
+      "step": 10130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998758493110642e-05,
+      "loss": 1.0194,
+      "step": 10131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998758245584776e-05,
+      "loss": 1.0846,
+      "step": 10132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998757998034243e-05,
+      "loss": 0.9837,
+      "step": 10133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998757750459044e-05,
+      "loss": 0.9976,
+      "step": 10134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998757502859178e-05,
+      "loss": 1.0837,
+      "step": 10135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998757255234645e-05,
+      "loss": 1.0913,
+      "step": 10136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998757007585446e-05,
+      "loss": 0.5574,
+      "step": 10137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987567599115794e-05,
+      "loss": 0.5928,
+      "step": 10138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998756512213046e-05,
+      "loss": 0.7991,
+      "step": 10139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987562644898473e-05,
+      "loss": 1.1505,
+      "step": 10140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998756016741981e-05,
+      "loss": 1.046,
+      "step": 10141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998755768969449e-05,
+      "loss": 1.133,
+      "step": 10142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998755521172249e-05,
+      "loss": 1.1574,
+      "step": 10143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998755273350383e-05,
+      "loss": 1.1993,
+      "step": 10144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99875502550385e-05,
+      "loss": 1.1066,
+      "step": 10145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99875477763265e-05,
+      "loss": 1.1578,
+      "step": 10146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998754529736784e-05,
+      "loss": 0.9662,
+      "step": 10147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998754281816251e-05,
+      "loss": 1.1373,
+      "step": 10148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987540338710523e-05,
+      "loss": 1.1941,
+      "step": 10149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998753785901186e-05,
+      "loss": 0.9031,
+      "step": 10150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987535379066534e-05,
+      "loss": 0.9422,
+      "step": 10151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987532898874545e-05,
+      "loss": 1.0914,
+      "step": 10152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998753041843588e-05,
+      "loss": 1.0855,
+      "step": 10153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998752793775056e-05,
+      "loss": 1.1082,
+      "step": 10154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998752545681856e-05,
+      "loss": 1.0152,
+      "step": 10155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998752297563991e-05,
+      "loss": 1.1159,
+      "step": 10156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998752049421458e-05,
+      "loss": 1.4043,
+      "step": 10157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998751801254259e-05,
+      "loss": 1.177,
+      "step": 10158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998751553062393e-05,
+      "loss": 1.1901,
+      "step": 10159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987513048458603e-05,
+      "loss": 0.8716,
+      "step": 10160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987510566046616e-05,
+      "loss": 0.9739,
+      "step": 10161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998750808338796e-05,
+      "loss": 0.9237,
+      "step": 10162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998750560048264e-05,
+      "loss": 0.8929,
+      "step": 10163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998750311733065e-05,
+      "loss": 0.9751,
+      "step": 10164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987500633931996e-05,
+      "loss": 1.2216,
+      "step": 10165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987498150286674e-05,
+      "loss": 1.19,
+      "step": 10166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998749566639469e-05,
+      "loss": 1.2974,
+      "step": 10167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987493182256035e-05,
+      "loss": 1.2664,
+      "step": 10168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998749069787072e-05,
+      "loss": 0.8888,
+      "step": 10169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998748821323873e-05,
+      "loss": 0.9302,
+      "step": 10170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998748572836008e-05,
+      "loss": 1.3394,
+      "step": 10171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998748324323477e-05,
+      "loss": 0.7946,
+      "step": 10172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998748075786278e-05,
+      "loss": 0.5738,
+      "step": 10173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987478272244136e-05,
+      "loss": 0.8554,
+      "step": 10174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987475786378816e-05,
+      "loss": 1.3842,
+      "step": 10175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998747330026684e-05,
+      "loss": 1.4482,
+      "step": 10176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998747081390819e-05,
+      "loss": 1.4754,
+      "step": 10177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987468327302876e-05,
+      "loss": 1.29,
+      "step": 10178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987465840450905e-05,
+      "loss": 1.1728,
+      "step": 10179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987463353352256e-05,
+      "loss": 1.3689,
+      "step": 10180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998746086600695e-05,
+      "loss": 1.229,
+      "step": 10181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998745837841498e-05,
+      "loss": 1.9158,
+      "step": 10182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998745589057634e-05,
+      "loss": 1.06,
+      "step": 10183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998745340249103e-05,
+      "loss": 1.0289,
+      "step": 10184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998745091415906e-05,
+      "loss": 1.1272,
+      "step": 10185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998744842558042e-05,
+      "loss": 1.09,
+      "step": 10186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998744593675512e-05,
+      "loss": 1.127,
+      "step": 10187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998744344768315e-05,
+      "loss": 1.1904,
+      "step": 10188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987440958364516e-05,
+      "loss": 0.9387,
+      "step": 10189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998743846879922e-05,
+      "loss": 0.9157,
+      "step": 10190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987435978987254e-05,
+      "loss": 1.0315,
+      "step": 10191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987433488928626e-05,
+      "loss": 1.5297,
+      "step": 10192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998743099862333e-05,
+      "loss": 1.3291,
+      "step": 10193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987428508071367e-05,
+      "loss": 1.2469,
+      "step": 10194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998742601727274e-05,
+      "loss": 0.5735,
+      "step": 10195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998742352622745e-05,
+      "loss": 1.1063,
+      "step": 10196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987421034935494e-05,
+      "loss": 1.3326,
+      "step": 10197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998741854339687e-05,
+      "loss": 1.0274,
+      "step": 10198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998741605161158e-05,
+      "loss": 1.5276,
+      "step": 10199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987413559579636e-05,
+      "loss": 1.4601,
+      "step": 10200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987411067301015e-05,
+      "loss": 0.9156,
+      "step": 10201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998740857477573e-05,
+      "loss": 1.0569,
+      "step": 10202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987406082003785e-05,
+      "loss": 0.9294,
+      "step": 10203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998740358898517e-05,
+      "loss": 1.1716,
+      "step": 10204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998740109571989e-05,
+      "loss": 0.6784,
+      "step": 10205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998739860220795e-05,
+      "loss": 0.723,
+      "step": 10206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998739610844934e-05,
+      "loss": 0.7166,
+      "step": 10207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987393614444066e-05,
+      "loss": 0.6935,
+      "step": 10208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998739112019213e-05,
+      "loss": 1.0554,
+      "step": 10209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998738862569352e-05,
+      "loss": 1.0829,
+      "step": 10210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987386130948255e-05,
+      "loss": 0.9208,
+      "step": 10211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987383635956317e-05,
+      "loss": 1.1552,
+      "step": 10212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998738114071772e-05,
+      "loss": 1.3641,
+      "step": 10213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998737864523246e-05,
+      "loss": 0.9493,
+      "step": 10214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998737614950053e-05,
+      "loss": 1.0372,
+      "step": 10215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987373653521936e-05,
+      "loss": 1.0349,
+      "step": 10216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987371157296684e-05,
+      "loss": 1.3293,
+      "step": 10217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998736866082475e-05,
+      "loss": 1.1373,
+      "step": 10218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987366164106164e-05,
+      "loss": 1.2073,
+      "step": 10219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998736366714091e-05,
+      "loss": 1.0581,
+      "step": 10220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987361169929e-05,
+      "loss": 1.7155,
+      "step": 10221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998735867247042e-05,
+      "loss": 1.326,
+      "step": 10222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987356174765165e-05,
+      "loss": 1.0172,
+      "step": 10223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998735367681326e-05,
+      "loss": 1.2729,
+      "step": 10224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987351178614686e-05,
+      "loss": 0.8779,
+      "step": 10225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998734868016944e-05,
+      "loss": 1.1188,
+      "step": 10226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987346181477535e-05,
+      "loss": 1.2608,
+      "step": 10227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987343682538965e-05,
+      "loss": 1.058,
+      "step": 10228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998734118335373e-05,
+      "loss": 1.0546,
+      "step": 10229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998733868392184e-05,
+      "loss": 1.0555,
+      "step": 10230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998733618424327e-05,
+      "loss": 1.0756,
+      "step": 10231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987333684318045e-05,
+      "loss": 2.1973,
+      "step": 10232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987331184146155e-05,
+      "loss": 1.1023,
+      "step": 10233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987328683727595e-05,
+      "loss": 1.2884,
+      "step": 10234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998732618306238e-05,
+      "loss": 0.9701,
+      "step": 10235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998732368215049e-05,
+      "loss": 1.4618,
+      "step": 10236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998732118099194e-05,
+      "loss": 1.9832,
+      "step": 10237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998731867958673e-05,
+      "loss": 1.0656,
+      "step": 10238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998731617793485e-05,
+      "loss": 0.9133,
+      "step": 10239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987313676036305e-05,
+      "loss": 0.7601,
+      "step": 10240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987311173891105e-05,
+      "loss": 0.8458,
+      "step": 10241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998730867149924e-05,
+      "loss": 1.1284,
+      "step": 10242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998730616886069e-05,
+      "loss": 0.9589,
+      "step": 10243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998730366597549e-05,
+      "loss": 0.9388,
+      "step": 10244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987301162843634e-05,
+      "loss": 0.2874,
+      "step": 10245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998729865946511e-05,
+      "loss": 0.2173,
+      "step": 10246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998729615583991e-05,
+      "loss": 0.2438,
+      "step": 10247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998729365196806e-05,
+      "loss": 0.173,
+      "step": 10248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987291147849544e-05,
+      "loss": 0.2369,
+      "step": 10249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998728864348435e-05,
+      "loss": 0.2054,
+      "step": 10250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998728613887251e-05,
+      "loss": 0.205,
+      "step": 10251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987283634013995e-05,
+      "loss": 0.5671,
+      "step": 10252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998728112890882e-05,
+      "loss": 0.6497,
+      "step": 10253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998727862355698e-05,
+      "loss": 0.658,
+      "step": 10254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998727611795848e-05,
+      "loss": 0.645,
+      "step": 10255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987273612113314e-05,
+      "loss": 0.2322,
+      "step": 10256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998727110602148e-05,
+      "loss": 0.1367,
+      "step": 10257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998726859968299e-05,
+      "loss": 0.352,
+      "step": 10258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998726609309783e-05,
+      "loss": 1.0021,
+      "step": 10259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998726358626601e-05,
+      "loss": 1.5276,
+      "step": 10260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998726107918752e-05,
+      "loss": 1.2101,
+      "step": 10261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987258571862375e-05,
+      "loss": 1.2582,
+      "step": 10262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987256064290554e-05,
+      "loss": 1.0663,
+      "step": 10263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998725355647208e-05,
+      "loss": 1.2152,
+      "step": 10264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987251048406945e-05,
+      "loss": 1.0573,
+      "step": 10265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987248540095135e-05,
+      "loss": 1.2546,
+      "step": 10266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998724603153667e-05,
+      "loss": 1.1061,
+      "step": 10267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998724352273154e-05,
+      "loss": 1.0008,
+      "step": 10268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987241013679745e-05,
+      "loss": 1.1719,
+      "step": 10269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987238504381285e-05,
+      "loss": 0.9274,
+      "step": 10270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998723599483616e-05,
+      "loss": 1.1606,
+      "step": 10271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998723348504438e-05,
+      "loss": 1.1728,
+      "step": 10272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998723097500593e-05,
+      "loss": 1.9997,
+      "step": 10273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987228464720816e-05,
+      "loss": 1.7191,
+      "step": 10274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998722595418904e-05,
+      "loss": 1.1818,
+      "step": 10275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987223443410605e-05,
+      "loss": 1.0861,
+      "step": 10276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987220932385506e-05,
+      "loss": 0.8447,
+      "step": 10277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987218421113737e-05,
+      "loss": 0.4378,
+      "step": 10278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998721590959531e-05,
+      "loss": 0.3998,
+      "step": 10279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998721339783021e-05,
+      "loss": 0.6704,
+      "step": 10280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998721088581846e-05,
+      "loss": 1.2137,
+      "step": 10281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998720837356004e-05,
+      "loss": 0.7144,
+      "step": 10282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987205861054956e-05,
+      "loss": 1.0599,
+      "step": 10283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998720334830321e-05,
+      "loss": 0.772,
+      "step": 10284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99872008353048e-05,
+      "loss": 0.9947,
+      "step": 10285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998719832205974e-05,
+      "loss": 1.2363,
+      "step": 10286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987195808567996e-05,
+      "loss": 0.9215,
+      "step": 10287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987193294829606e-05,
+      "loss": 1.3128,
+      "step": 10288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998719078084454e-05,
+      "loss": 1.0097,
+      "step": 10289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998718826661282e-05,
+      "loss": 1.1504,
+      "step": 10290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998718575213443e-05,
+      "loss": 1.2552,
+      "step": 10291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987183237409384e-05,
+      "loss": 1.0095,
+      "step": 10292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987180722437674e-05,
+      "loss": 1.0155,
+      "step": 10293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99871782072193e-05,
+      "loss": 1.1331,
+      "step": 10294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998717569175426e-05,
+      "loss": 1.3524,
+      "step": 10295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998717317604256e-05,
+      "loss": 1.3273,
+      "step": 10296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998717066008419e-05,
+      "loss": 1.216,
+      "step": 10297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987168143879165e-05,
+      "loss": 1.551,
+      "step": 10298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998716562742748e-05,
+      "loss": 0.6469,
+      "step": 10299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998716311072912e-05,
+      "loss": 0.8778,
+      "step": 10300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998716059378411e-05,
+      "loss": 1.2854,
+      "step": 10301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998715807659243e-05,
+      "loss": 1.2548,
+      "step": 10302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998715555915409e-05,
+      "loss": 0.8318,
+      "step": 10303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987153041469086e-05,
+      "loss": 0.7398,
+      "step": 10304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987150523537415e-05,
+      "loss": 1.1692,
+      "step": 10305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987148005359095e-05,
+      "loss": 0.6863,
+      "step": 10306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987145486934106e-05,
+      "loss": 1.3325,
+      "step": 10307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998714296826245e-05,
+      "loss": 1.3931,
+      "step": 10308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998714044934413e-05,
+      "loss": 1.1937,
+      "step": 10309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987137930179154e-05,
+      "loss": 1.1837,
+      "step": 10310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998713541076751e-05,
+      "loss": 1.5333,
+      "step": 10311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998713289110921e-05,
+      "loss": 0.6901,
+      "step": 10312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987130371204244e-05,
+      "loss": 1.7999,
+      "step": 10313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987127851052615e-05,
+      "loss": 1.1222,
+      "step": 10314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998712533065433e-05,
+      "loss": 1.2129,
+      "step": 10315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998712281000937e-05,
+      "loss": 1.3231,
+      "step": 10316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998712028911776e-05,
+      "loss": 1.2286,
+      "step": 10317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998711776797947e-05,
+      "loss": 1.1108,
+      "step": 10318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998711524659454e-05,
+      "loss": 1.0662,
+      "step": 10319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998711272496294e-05,
+      "loss": 0.9464,
+      "step": 10320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998711020308467e-05,
+      "loss": 0.9488,
+      "step": 10321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987107680959745e-05,
+      "loss": 1.0501,
+      "step": 10322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998710515858816e-05,
+      "loss": 1.2066,
+      "step": 10323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998710263596991e-05,
+      "loss": 1.1524,
+      "step": 10324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998710011310499e-05,
+      "loss": 1.0147,
+      "step": 10325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998709758999341e-05,
+      "loss": 1.0825,
+      "step": 10326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987095066635175e-05,
+      "loss": 1.1649,
+      "step": 10327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998709254303028e-05,
+      "loss": 1.1194,
+      "step": 10328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998709001917872e-05,
+      "loss": 1.1093,
+      "step": 10329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987087495080484e-05,
+      "loss": 1.2593,
+      "step": 10330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998708497073561e-05,
+      "loss": 1.1496,
+      "step": 10331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987082446144056e-05,
+      "loss": 1.1416,
+      "step": 10332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998707992130585e-05,
+      "loss": 1.196,
+      "step": 10333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998707739622098e-05,
+      "loss": 0.9892,
+      "step": 10334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998707487088944e-05,
+      "loss": 1.2026,
+      "step": 10335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998707234531125e-05,
+      "loss": 0.8014,
+      "step": 10336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998706981948639e-05,
+      "loss": 2.2308,
+      "step": 10337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998706729341487e-05,
+      "loss": 1.0938,
+      "step": 10338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998706476709669e-05,
+      "loss": 0.7971,
+      "step": 10339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998706224053185e-05,
+      "loss": 1.2027,
+      "step": 10340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998705971372034e-05,
+      "loss": 1.139,
+      "step": 10341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998705718666218e-05,
+      "loss": 0.6991,
+      "step": 10342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998705465935734e-05,
+      "loss": 1.3724,
+      "step": 10343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998705213180586e-05,
+      "loss": 1.1193,
+      "step": 10344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998704960400771e-05,
+      "loss": 1.005,
+      "step": 10345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987047075962893e-05,
+      "loss": 1.098,
+      "step": 10346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998704454767143e-05,
+      "loss": 1.116,
+      "step": 10347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987042019133284e-05,
+      "loss": 1.0914,
+      "step": 10348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987039490348485e-05,
+      "loss": 1.8293,
+      "step": 10349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998703696131703e-05,
+      "loss": 1.2251,
+      "step": 10350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987034432038906e-05,
+      "loss": 1.2625,
+      "step": 10351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987031902514126e-05,
+      "loss": 1.3011,
+      "step": 10352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998702937274268e-05,
+      "loss": 0.5868,
+      "step": 10353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998702684272458e-05,
+      "loss": 0.522,
+      "step": 10354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987024312459815e-05,
+      "loss": 0.915,
+      "step": 10355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987021781948384e-05,
+      "loss": 1.1415,
+      "step": 10356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998701925119029e-05,
+      "loss": 1.3289,
+      "step": 10357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998701672018554e-05,
+      "loss": 1.1288,
+      "step": 10358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987014188934134e-05,
+      "loss": 1.0588,
+      "step": 10359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998701165743606e-05,
+      "loss": 1.0426,
+      "step": 10360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987009125691326e-05,
+      "loss": 1.0733,
+      "step": 10361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998700659369993e-05,
+      "loss": 0.9528,
+      "step": 10362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987004061461875e-05,
+      "loss": 1.1307,
+      "step": 10363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987001528977155e-05,
+      "loss": 1.1637,
+      "step": 10364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998699899624578e-05,
+      "loss": 1.2464,
+      "step": 10365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998699646326774e-05,
+      "loss": 0.776,
+      "step": 10366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998699393004304e-05,
+      "loss": 0.906,
+      "step": 10367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986991396571674e-05,
+      "loss": 0.9508,
+      "step": 10368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986988862853654e-05,
+      "loss": 1.0688,
+      "step": 10369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986986328888964e-05,
+      "loss": 1.1141,
+      "step": 10370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986983794677625e-05,
+      "loss": 0.9474,
+      "step": 10371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998698126021961e-05,
+      "loss": 0.6913,
+      "step": 10372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986978725514945e-05,
+      "loss": 0.4359,
+      "step": 10373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998697619056362e-05,
+      "loss": 1.1969,
+      "step": 10374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986973655365634e-05,
+      "loss": 1.2375,
+      "step": 10375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986971119920975e-05,
+      "loss": 1.0497,
+      "step": 10376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998696858422967e-05,
+      "loss": 0.9796,
+      "step": 10377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986966048291694e-05,
+      "loss": 1.0029,
+      "step": 10378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986963512107067e-05,
+      "loss": 1.1816,
+      "step": 10379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986960975675776e-05,
+      "loss": 1.1951,
+      "step": 10380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998695843899782e-05,
+      "loss": 1.143,
+      "step": 10381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998695590207321e-05,
+      "loss": 1.0644,
+      "step": 10382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998695336490193e-05,
+      "loss": 1.1941,
+      "step": 10383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986950827483994e-05,
+      "loss": 1.1132,
+      "step": 10384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986948289819396e-05,
+      "loss": 0.549,
+      "step": 10385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998694575190814e-05,
+      "loss": 1.0158,
+      "step": 10386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998694321375022e-05,
+      "loss": 1.3903,
+      "step": 10387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998694067534564e-05,
+      "loss": 1.4583,
+      "step": 10388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986938136694405e-05,
+      "loss": 1.1652,
+      "step": 10389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99869355977965e-05,
+      "loss": 1.2458,
+      "step": 10390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998693305865194e-05,
+      "loss": 1.2432,
+      "step": 10391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986930519260724e-05,
+      "loss": 1.3848,
+      "step": 10392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986927979622844e-05,
+      "loss": 1.0191,
+      "step": 10393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99869254397383e-05,
+      "loss": 0.9536,
+      "step": 10394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986922899607104e-05,
+      "loss": 1.1373,
+      "step": 10395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986920359229237e-05,
+      "loss": 1.2518,
+      "step": 10396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998691781860472e-05,
+      "loss": 1.3977,
+      "step": 10397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998691527773353e-05,
+      "loss": 1.0759,
+      "step": 10398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998691273661569e-05,
+      "loss": 1.6515,
+      "step": 10399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986910195251186e-05,
+      "loss": 1.5402,
+      "step": 10400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986907653640025e-05,
+      "loss": 1.5662,
+      "step": 10401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99869051117822e-05,
+      "loss": 1.4326,
+      "step": 10402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986902569677714e-05,
+      "loss": 1.3334,
+      "step": 10403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986900027326564e-05,
+      "loss": 1.2097,
+      "step": 10404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986897484728766e-05,
+      "loss": 1.2671,
+      "step": 10405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99868949418843e-05,
+      "loss": 1.1214,
+      "step": 10406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998689239879317e-05,
+      "loss": 1.349,
+      "step": 10407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986889855455386e-05,
+      "loss": 1.3452,
+      "step": 10408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998688731187094e-05,
+      "loss": 1.3616,
+      "step": 10409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986884768039844e-05,
+      "loss": 1.3294,
+      "step": 10410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986882223962076e-05,
+      "loss": 1.2834,
+      "step": 10411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998687967963766e-05,
+      "loss": 1.1846,
+      "step": 10412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998687713506657e-05,
+      "loss": 1.2795,
+      "step": 10413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998687459024882e-05,
+      "loss": 1.1784,
+      "step": 10414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998687204518442e-05,
+      "loss": 1.3204,
+      "step": 10415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998686949987336e-05,
+      "loss": 1.1975,
+      "step": 10416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986866954315634e-05,
+      "loss": 1.1086,
+      "step": 10417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998686440851125e-05,
+      "loss": 1.285,
+      "step": 10418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986861862460196e-05,
+      "loss": 1.3251,
+      "step": 10419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99868593161625e-05,
+      "loss": 0.9713,
+      "step": 10420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986856769618134e-05,
+      "loss": 1.2093,
+      "step": 10421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998685422282711e-05,
+      "loss": 0.9838,
+      "step": 10422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998685167578943e-05,
+      "loss": 1.0117,
+      "step": 10423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986849128505084e-05,
+      "loss": 1.0491,
+      "step": 10424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998684658097409e-05,
+      "loss": 1.4273,
+      "step": 10425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998684403319642e-05,
+      "loss": 1.3291,
+      "step": 10426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99868414851721e-05,
+      "loss": 1.0863,
+      "step": 10427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998683893690112e-05,
+      "loss": 0.884,
+      "step": 10428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986836388383484e-05,
+      "loss": 1.0212,
+      "step": 10429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986833839619184e-05,
+      "loss": 1.1829,
+      "step": 10430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998683129060822e-05,
+      "loss": 0.7684,
+      "step": 10431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99868287413506e-05,
+      "loss": 1.2292,
+      "step": 10432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998682619184633e-05,
+      "loss": 1.5942,
+      "step": 10433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998682364209539e-05,
+      "loss": 3.0241,
+      "step": 10434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998682109209779e-05,
+      "loss": 1.0854,
+      "step": 10435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998681854185353e-05,
+      "loss": 1.2607,
+      "step": 10436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998681599136262e-05,
+      "loss": 1.3714,
+      "step": 10437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998681344062504e-05,
+      "loss": 1.0285,
+      "step": 10438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998681088964081e-05,
+      "loss": 1.0756,
+      "step": 10439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998680833840991e-05,
+      "loss": 1.2608,
+      "step": 10440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998680578693236e-05,
+      "loss": 1.3557,
+      "step": 10441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998680323520814e-05,
+      "loss": 1.3043,
+      "step": 10442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998680068323728e-05,
+      "loss": 0.7495,
+      "step": 10443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998679813101975e-05,
+      "loss": 1.3086,
+      "step": 10444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986795578555555e-05,
+      "loss": 1.6532,
+      "step": 10445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998679302584471e-05,
+      "loss": 1.1917,
+      "step": 10446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99867904728872e-05,
+      "loss": 1.2263,
+      "step": 10447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998678791968303e-05,
+      "loss": 1.052,
+      "step": 10448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998678536623221e-05,
+      "loss": 1.1604,
+      "step": 10449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998678281253472e-05,
+      "loss": 1.1795,
+      "step": 10450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998678025859058e-05,
+      "loss": 1.3203,
+      "step": 10451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998677770439978e-05,
+      "loss": 1.0137,
+      "step": 10452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998677514996232e-05,
+      "loss": 1.2603,
+      "step": 10453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998677259527819e-05,
+      "loss": 1.1167,
+      "step": 10454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998677004034742e-05,
+      "loss": 1.2954,
+      "step": 10455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998676748516997e-05,
+      "loss": 1.1379,
+      "step": 10456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998676492974588e-05,
+      "loss": 1.1313,
+      "step": 10457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998676237407512e-05,
+      "loss": 1.1246,
+      "step": 10458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986759818157704e-05,
+      "loss": 0.5331,
+      "step": 10459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986757261993636e-05,
+      "loss": 0.9279,
+      "step": 10460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99867547055829e-05,
+      "loss": 1.4327,
+      "step": 10461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998675214892551e-05,
+      "loss": 1.1658,
+      "step": 10462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998674959202146e-05,
+      "loss": 0.872,
+      "step": 10463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998674703487075e-05,
+      "loss": 2.5678,
+      "step": 10464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998674447747338e-05,
+      "loss": 1.6675,
+      "step": 10465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986741919829363e-05,
+      "loss": 1.2018,
+      "step": 10466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998673936193867e-05,
+      "loss": 1.1596,
+      "step": 10467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986736803801326e-05,
+      "loss": 0.9724,
+      "step": 10468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998673424541733e-05,
+      "loss": 1.0377,
+      "step": 10469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998673168678667e-05,
+      "loss": 0.9609,
+      "step": 10470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998672912790935e-05,
+      "loss": 1.1886,
+      "step": 10471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986726568785367e-05,
+      "loss": 1.1853,
+      "step": 10472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998672400941474e-05,
+      "loss": 0.8812,
+      "step": 10473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986721449797444e-05,
+      "loss": 0.8519,
+      "step": 10474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986718889933495e-05,
+      "loss": 1.2709,
+      "step": 10475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998671632982288e-05,
+      "loss": 1.3796,
+      "step": 10476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998671376946561e-05,
+      "loss": 1.5551,
+      "step": 10477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998671120886168e-05,
+      "loss": 1.0944,
+      "step": 10478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99867086480111e-05,
+      "loss": 1.1696,
+      "step": 10479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998670608691385e-05,
+      "loss": 1.1202,
+      "step": 10480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998670352556995e-05,
+      "loss": 1.2514,
+      "step": 10481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998670096397939e-05,
+      "loss": 0.9168,
+      "step": 10482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986698402142175e-05,
+      "loss": 1.1231,
+      "step": 10483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986695840058295e-05,
+      "loss": 0.9077,
+      "step": 10484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998669327772776e-05,
+      "loss": 1.162,
+      "step": 10485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998669071515057e-05,
+      "loss": 1.3504,
+      "step": 10486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998668815232672e-05,
+      "loss": 1.1128,
+      "step": 10487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998668558925621e-05,
+      "loss": 0.9506,
+      "step": 10488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998668302593904e-05,
+      "loss": 0.9646,
+      "step": 10489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998668046237522e-05,
+      "loss": 0.9416,
+      "step": 10490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986677898564735e-05,
+      "loss": 0.8455,
+      "step": 10491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986675334507594e-05,
+      "loss": 0.8593,
+      "step": 10492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986672770203797e-05,
+      "loss": 0.7396,
+      "step": 10493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998667020565334e-05,
+      "loss": 0.7632,
+      "step": 10494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998666764085622e-05,
+      "loss": 1.1413,
+      "step": 10495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986665075812455e-05,
+      "loss": 0.995,
+      "step": 10496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998666251052202e-05,
+      "loss": 1.4334,
+      "step": 10497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986659944984937e-05,
+      "loss": 0.8874,
+      "step": 10498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998665737920118e-05,
+      "loss": 1.0983,
+      "step": 10499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998665481317079e-05,
+      "loss": 1.3465,
+      "step": 10500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0590955018997192,
+      "eval_runtime": 83.8028,
+      "eval_samples_per_second": 16.527,
+      "eval_steps_per_second": 4.141,
+      "step": 10500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998665224689372e-05,
+      "loss": 0.8434,
+      "step": 10501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998664968037e-05,
+      "loss": 0.758,
+      "step": 10502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998664711359963e-05,
+      "loss": 0.8072,
+      "step": 10503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986644546582596e-05,
+      "loss": 1.0505,
+      "step": 10504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99866419793189e-05,
+      "loss": 1.1189,
+      "step": 10505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998663941180855e-05,
+      "loss": 1.2439,
+      "step": 10506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986636844051546e-05,
+      "loss": 1.1577,
+      "step": 10507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986634276047875e-05,
+      "loss": 1.0675,
+      "step": 10508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998663170779756e-05,
+      "loss": 1.2374,
+      "step": 10509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998662913930058e-05,
+      "loss": 1.16,
+      "step": 10510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998662657055694e-05,
+      "loss": 1.1783,
+      "step": 10511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986624001566644e-05,
+      "loss": 1.1148,
+      "step": 10512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998662143232969e-05,
+      "loss": 1.207,
+      "step": 10513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998661886284608e-05,
+      "loss": 1.1054,
+      "step": 10514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986616293115816e-05,
+      "loss": 1.2763,
+      "step": 10515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998661372313889e-05,
+      "loss": 0.8169,
+      "step": 10516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998661115291531e-05,
+      "loss": 1.0664,
+      "step": 10517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998660858244507e-05,
+      "loss": 1.1742,
+      "step": 10518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998660601172818e-05,
+      "loss": 0.9612,
+      "step": 10519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998660344076462e-05,
+      "loss": 1.2785,
+      "step": 10520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986600869554415e-05,
+      "loss": 1.2946,
+      "step": 10521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998659829809754e-05,
+      "loss": 1.2146,
+      "step": 10522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998659572639402e-05,
+      "loss": 0.9512,
+      "step": 10523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998659315444384e-05,
+      "loss": 1.1009,
+      "step": 10524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986590582247e-05,
+      "loss": 1.1699,
+      "step": 10525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986588009803506e-05,
+      "loss": 1.2749,
+      "step": 10526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998658543711335e-05,
+      "loss": 1.187,
+      "step": 10527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998658286417653e-05,
+      "loss": 1.0756,
+      "step": 10528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986580290993065e-05,
+      "loss": 1.1111,
+      "step": 10529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998657771756294e-05,
+      "loss": 1.0458,
+      "step": 10530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998657514388616e-05,
+      "loss": 0.9649,
+      "step": 10531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998657256996273e-05,
+      "loss": 0.9339,
+      "step": 10532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998656999579263e-05,
+      "loss": 0.6619,
+      "step": 10533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998656742137588e-05,
+      "loss": 0.5385,
+      "step": 10534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998656484671247e-05,
+      "loss": 1.0374,
+      "step": 10535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986562271802406e-05,
+      "loss": 0.5005,
+      "step": 10536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998655969664568e-05,
+      "loss": 0.3531,
+      "step": 10537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99865571212423e-05,
+      "loss": 0.5739,
+      "step": 10538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986554545592266e-05,
+      "loss": 0.689,
+      "step": 10539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998655196969557e-05,
+      "loss": 1.1641,
+      "step": 10540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998654939355222e-05,
+      "loss": 0.9358,
+      "step": 10541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986546817162214e-05,
+      "loss": 1.1991,
+      "step": 10542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986544240525556e-05,
+      "loss": 1.6141,
+      "step": 10543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986541663642236e-05,
+      "loss": 1.8724,
+      "step": 10544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998653908651226e-05,
+      "loss": 1.3541,
+      "step": 10545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986536509135626e-05,
+      "loss": 1.0405,
+      "step": 10546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998653393151234e-05,
+      "loss": 0.9787,
+      "step": 10547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986531353642386e-05,
+      "loss": 0.9666,
+      "step": 10548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986528775525785e-05,
+      "loss": 0.9589,
+      "step": 10549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998652619716253e-05,
+      "loss": 1.0897,
+      "step": 10550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998652361855261e-05,
+      "loss": 1.1917,
+      "step": 10551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998652103969604e-05,
+      "loss": 1.1123,
+      "step": 10552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986518460592816e-05,
+      "loss": 0.8103,
+      "step": 10553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998651588124292e-05,
+      "loss": 1.1595,
+      "step": 10554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986513301646385e-05,
+      "loss": 1.1804,
+      "step": 10555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998651072180319e-05,
+      "loss": 1.0873,
+      "step": 10556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998650814171334e-05,
+      "loss": 1.1571,
+      "step": 10557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986505561376826e-05,
+      "loss": 1.1775,
+      "step": 10558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998650298079366e-05,
+      "loss": 1.4187,
+      "step": 10559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986500399963835e-05,
+      "loss": 1.3483,
+      "step": 10560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986497818887355e-05,
+      "loss": 1.0625,
+      "step": 10561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998649523756423e-05,
+      "loss": 1.1701,
+      "step": 10562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986492655994435e-05,
+      "loss": 1.0178,
+      "step": 10563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998649007417798e-05,
+      "loss": 1.0857,
+      "step": 10564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986487492114885e-05,
+      "loss": 0.7984,
+      "step": 10565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998648490980512e-05,
+      "loss": 1.3078,
+      "step": 10566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998648232724871e-05,
+      "loss": 1.1336,
+      "step": 10567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998647974444564e-05,
+      "loss": 0.9312,
+      "step": 10568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986477161395904e-05,
+      "loss": 1.1834,
+      "step": 10569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998647457809953e-05,
+      "loss": 0.9114,
+      "step": 10570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998647199455649e-05,
+      "loss": 1.1802,
+      "step": 10571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986469410766786e-05,
+      "loss": 1.0653,
+      "step": 10572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986466826730435e-05,
+      "loss": 0.5615,
+      "step": 10573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986464242447435e-05,
+      "loss": 0.6149,
+      "step": 10574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998646165791777e-05,
+      "loss": 0.8447,
+      "step": 10575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998645907314146e-05,
+      "loss": 0.4378,
+      "step": 10576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986456488118483e-05,
+      "loss": 1.224,
+      "step": 10577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998645390284885e-05,
+      "loss": 1.3924,
+      "step": 10578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998645131733256e-05,
+      "loss": 1.2941,
+      "step": 10579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998644873156962e-05,
+      "loss": 1.149,
+      "step": 10580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998644614556002e-05,
+      "loss": 1.003,
+      "step": 10581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986443559303774e-05,
+      "loss": 1.0581,
+      "step": 10582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998644097280086e-05,
+      "loss": 0.9061,
+      "step": 10583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986438386051295e-05,
+      "loss": 1.0999,
+      "step": 10584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998643579905508e-05,
+      "loss": 0.9629,
+      "step": 10585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99864332118122e-05,
+      "loss": 1.1814,
+      "step": 10586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998643062432267e-05,
+      "loss": 1.1246,
+      "step": 10587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998642803658649e-05,
+      "loss": 1.0821,
+      "step": 10588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986425448603646e-05,
+      "loss": 1.1333,
+      "step": 10589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998642286037415e-05,
+      "loss": 0.9511,
+      "step": 10590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998642027189799e-05,
+      "loss": 1.1535,
+      "step": 10591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998641768317518e-05,
+      "loss": 1.1484,
+      "step": 10592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998641509420572e-05,
+      "loss": 0.636,
+      "step": 10593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99864125049896e-05,
+      "loss": 1.128,
+      "step": 10594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986409915526834e-05,
+      "loss": 0.9554,
+      "step": 10595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986407325817395e-05,
+      "loss": 0.5884,
+      "step": 10596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986404735861314e-05,
+      "loss": 0.849,
+      "step": 10597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986402145658576e-05,
+      "loss": 1.3073,
+      "step": 10598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998639955520918e-05,
+      "loss": 1.1573,
+      "step": 10599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986396964513126e-05,
+      "loss": 1.026,
+      "step": 10600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998639437357042e-05,
+      "loss": 1.1495,
+      "step": 10601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998639178238106e-05,
+      "loss": 0.9721,
+      "step": 10602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998638919094505e-05,
+      "loss": 1.2164,
+      "step": 10603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986386599262374e-05,
+      "loss": 1.0202,
+      "step": 10604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986384007333045e-05,
+      "loss": 0.9831,
+      "step": 10605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986381415157066e-05,
+      "loss": 1.0955,
+      "step": 10606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998637882273443e-05,
+      "loss": 1.1857,
+      "step": 10607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986376230065134e-05,
+      "loss": 1.3291,
+      "step": 10608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986373637149193e-05,
+      "loss": 1.1299,
+      "step": 10609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998637104398659e-05,
+      "loss": 1.1586,
+      "step": 10610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998636845057734e-05,
+      "loss": 1.114,
+      "step": 10611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986365856921424e-05,
+      "loss": 1.36,
+      "step": 10612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998636326301886e-05,
+      "loss": 1.0706,
+      "step": 10613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998636066886963e-05,
+      "loss": 1.2458,
+      "step": 10614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998635807447376e-05,
+      "loss": 1.0786,
+      "step": 10615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998635547983123e-05,
+      "loss": 1.108,
+      "step": 10616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986352884942044e-05,
+      "loss": 1.1663,
+      "step": 10617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998635028980621e-05,
+      "loss": 1.181,
+      "step": 10618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986347694423714e-05,
+      "loss": 0.9882,
+      "step": 10619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986345098794565e-05,
+      "loss": 1.0248,
+      "step": 10620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998634250291876e-05,
+      "loss": 0.8304,
+      "step": 10621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99863399067963e-05,
+      "loss": 1.1176,
+      "step": 10622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998633731042719e-05,
+      "loss": 1.0107,
+      "step": 10623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998633471381142e-05,
+      "loss": 1.4044,
+      "step": 10624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986332116949e-05,
+      "loss": 4.4344,
+      "step": 10625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998632951983992e-05,
+      "loss": 1.1785,
+      "step": 10626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986326922484186e-05,
+      "loss": 1.0471,
+      "step": 10627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99863243248818e-05,
+      "loss": 1.2066,
+      "step": 10628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986321727032756e-05,
+      "loss": 0.9483,
+      "step": 10629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998631912893707e-05,
+      "loss": 1.0307,
+      "step": 10630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986316530594716e-05,
+      "loss": 1.0138,
+      "step": 10631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986313932005715e-05,
+      "loss": 0.9078,
+      "step": 10632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998631133317005e-05,
+      "loss": 1.1706,
+      "step": 10633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998630873408774e-05,
+      "loss": 1.1575,
+      "step": 10634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998630613475878e-05,
+      "loss": 1.1129,
+      "step": 10635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998630353518315e-05,
+      "loss": 1.542,
+      "step": 10636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998630093536088e-05,
+      "loss": 0.982,
+      "step": 10637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998629833529195e-05,
+      "loss": 1.1706,
+      "step": 10638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998629573497636e-05,
+      "loss": 0.9877,
+      "step": 10639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998629313441413e-05,
+      "loss": 1.0951,
+      "step": 10640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998629053360524e-05,
+      "loss": 1.1293,
+      "step": 10641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998628793254969e-05,
+      "loss": 1.1675,
+      "step": 10642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998628533124749e-05,
+      "loss": 1.1676,
+      "step": 10643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986282729698633e-05,
+      "loss": 0.8128,
+      "step": 10644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998628012790313e-05,
+      "loss": 1.0571,
+      "step": 10645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998627752586096e-05,
+      "loss": 1.369,
+      "step": 10646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986274923572154e-05,
+      "loss": 1.4563,
+      "step": 10647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998627232103668e-05,
+      "loss": 1.2891,
+      "step": 10648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998626971825455e-05,
+      "loss": 0.8835,
+      "step": 10649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998626711522577e-05,
+      "loss": 0.9248,
+      "step": 10650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998626451195034e-05,
+      "loss": 0.9569,
+      "step": 10651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998626190842825e-05,
+      "loss": 1.1694,
+      "step": 10652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998625930465952e-05,
+      "loss": 1.1396,
+      "step": 10653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986256700644125e-05,
+      "loss": 1.131,
+      "step": 10654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998625409638208e-05,
+      "loss": 1.0525,
+      "step": 10655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998625149187337e-05,
+      "loss": 1.4169,
+      "step": 10656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998624888711802e-05,
+      "loss": 1.29,
+      "step": 10657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998624628211601e-05,
+      "loss": 0.8785,
+      "step": 10658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998624367686735e-05,
+      "loss": 1.1464,
+      "step": 10659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998624107137203e-05,
+      "loss": 1.2562,
+      "step": 10660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998623846563007e-05,
+      "loss": 1.3613,
+      "step": 10661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998623585964144e-05,
+      "loss": 1.4084,
+      "step": 10662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998623325340616e-05,
+      "loss": 0.99,
+      "step": 10663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986230646924234e-05,
+      "loss": 1.432,
+      "step": 10664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986228040195646e-05,
+      "loss": 1.4405,
+      "step": 10665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986225433220416e-05,
+      "loss": 1.2277,
+      "step": 10666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986222825998517e-05,
+      "loss": 1.0819,
+      "step": 10667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986220218529974e-05,
+      "loss": 1.0059,
+      "step": 10668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986217610814776e-05,
+      "loss": 0.8576,
+      "step": 10669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998621500285292e-05,
+      "loss": 0.6349,
+      "step": 10670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998621239464442e-05,
+      "loss": 0.9552,
+      "step": 10671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998620978618927e-05,
+      "loss": 1.2087,
+      "step": 10672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998620717748745e-05,
+      "loss": 5.6711,
+      "step": 10673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998620456853899e-05,
+      "loss": 3.5529,
+      "step": 10674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986201959343873e-05,
+      "loss": 0.8959,
+      "step": 10675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99861993499021e-05,
+      "loss": 1.1657,
+      "step": 10676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998619674021367e-05,
+      "loss": 1.0985,
+      "step": 10677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986194130278597e-05,
+      "loss": 1.3274,
+      "step": 10678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998619152009686e-05,
+      "loss": 1.1938,
+      "step": 10679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998618890966847e-05,
+      "loss": 1.4198,
+      "step": 10680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998618629899344e-05,
+      "loss": 1.1039,
+      "step": 10681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998618368807175e-05,
+      "loss": 1.1376,
+      "step": 10682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99861810769034e-05,
+      "loss": 1.0047,
+      "step": 10683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998617846548841e-05,
+      "loss": 5.4369,
+      "step": 10684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998617585382676e-05,
+      "loss": 4.1474,
+      "step": 10685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986173241918456e-05,
+      "loss": 1.1917,
+      "step": 10686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986170629763505e-05,
+      "loss": 1.261,
+      "step": 10687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99861680173619e-05,
+      "loss": 1.0599,
+      "step": 10688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998616540471363e-05,
+      "loss": 1.2908,
+      "step": 10689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998616279181872e-05,
+      "loss": 0.9679,
+      "step": 10690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998616017867716e-05,
+      "loss": 1.0874,
+      "step": 10691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998615756528893e-05,
+      "loss": 1.1285,
+      "step": 10692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986154951654066e-05,
+      "loss": 0.9024,
+      "step": 10693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986152337772536e-05,
+      "loss": 0.9533,
+      "step": 10694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986149723644357e-05,
+      "loss": 1.5688,
+      "step": 10695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998614710926952e-05,
+      "loss": 1.0835,
+      "step": 10696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986144494648043e-05,
+      "loss": 1.0618,
+      "step": 10697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99861418797799e-05,
+      "loss": 0.9249,
+      "step": 10698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998613926466511e-05,
+      "loss": 0.9606,
+      "step": 10699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986136649303674e-05,
+      "loss": 1.0427,
+      "step": 10700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998613403369557e-05,
+      "loss": 0.5878,
+      "step": 10701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998613141784083e-05,
+      "loss": 0.7082,
+      "step": 10702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998612880173943e-05,
+      "loss": 0.9104,
+      "step": 10703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998612618539138e-05,
+      "loss": 0.8221,
+      "step": 10704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998612356879667e-05,
+      "loss": 0.7588,
+      "step": 10705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998612095195532e-05,
+      "loss": 0.9248,
+      "step": 10706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99861183348673e-05,
+      "loss": 0.8445,
+      "step": 10707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998611571753264e-05,
+      "loss": 0.8719,
+      "step": 10708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998611309995133e-05,
+      "loss": 0.6463,
+      "step": 10709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998611048212337e-05,
+      "loss": 0.558,
+      "step": 10710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986107864048746e-05,
+      "loss": 0.5944,
+      "step": 10711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986105245727476e-05,
+      "loss": 1.0485,
+      "step": 10712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998610262715955e-05,
+      "loss": 1.4276,
+      "step": 10713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986100008344974e-05,
+      "loss": 1.1766,
+      "step": 10714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998609738928375e-05,
+      "loss": 1.0757,
+      "step": 10715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998609476997587e-05,
+      "loss": 1.0755,
+      "step": 10716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998609215042134e-05,
+      "loss": 1.106,
+      "step": 10717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998608953062015e-05,
+      "loss": 0.9057,
+      "step": 10718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998608691057232e-05,
+      "loss": 0.8137,
+      "step": 10719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998608429027783e-05,
+      "loss": 1.2992,
+      "step": 10720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998608166973669e-05,
+      "loss": 1.1847,
+      "step": 10721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986079048948884e-05,
+      "loss": 1.3237,
+      "step": 10722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998607642791444e-05,
+      "loss": 1.5411,
+      "step": 10723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998607380663335e-05,
+      "loss": 1.2056,
+      "step": 10724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99860711851056e-05,
+      "loss": 1.0274,
+      "step": 10725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998606856333119e-05,
+      "loss": 1.3021,
+      "step": 10726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986065941310144e-05,
+      "loss": 1.1686,
+      "step": 10727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998606331904244e-05,
+      "loss": 1.3316,
+      "step": 10728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986060696528084e-05,
+      "loss": 1.3269,
+      "step": 10729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986058073767074e-05,
+      "loss": 0.9971,
+      "step": 10730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998605545075941e-05,
+      "loss": 0.9264,
+      "step": 10731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99860528275051e-05,
+      "loss": 0.9454,
+      "step": 10732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986050204004134e-05,
+      "loss": 1.2553,
+      "step": 10733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998604758025652e-05,
+      "loss": 1.1812,
+      "step": 10734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998604495626224e-05,
+      "loss": 1.0018,
+      "step": 10735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998604233202133e-05,
+      "loss": 0.8275,
+      "step": 10736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998603970753376e-05,
+      "loss": 1.2006,
+      "step": 10737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998603708279954e-05,
+      "loss": 1.1241,
+      "step": 10738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986034457818664e-05,
+      "loss": 1.2169,
+      "step": 10739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986031832591134e-05,
+      "loss": 1.0624,
+      "step": 10740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986029207116955e-05,
+      "loss": 1.0672,
+      "step": 10741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998602658139613e-05,
+      "loss": 0.9789,
+      "step": 10742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998602395542864e-05,
+      "loss": 1.0947,
+      "step": 10743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986021329214516e-05,
+      "loss": 1.75,
+      "step": 10744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986018702753734e-05,
+      "loss": 1.4034,
+      "step": 10745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986016076046295e-05,
+      "loss": 0.8919,
+      "step": 10746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998601344909221e-05,
+      "loss": 1.254,
+      "step": 10747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998601082189147e-05,
+      "loss": 1.0796,
+      "step": 10748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998600819444408e-05,
+      "loss": 1.2729,
+      "step": 10749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986005566750036e-05,
+      "loss": 1.1744,
+      "step": 10750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998600293880935e-05,
+      "loss": 1.2834,
+      "step": 10751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986000310622004e-05,
+      "loss": 1.1745,
+      "step": 10752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985997682188014e-05,
+      "loss": 1.195,
+      "step": 10753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998599505350736e-05,
+      "loss": 1.1958,
+      "step": 10754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985992424580066e-05,
+      "loss": 1.4716,
+      "step": 10755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998598979540612e-05,
+      "loss": 1.2591,
+      "step": 10756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998598716598552e-05,
+      "loss": 1.2078,
+      "step": 10757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998598453631827e-05,
+      "loss": 1.0542,
+      "step": 10758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998598190640437e-05,
+      "loss": 1.0827,
+      "step": 10759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998597927624382e-05,
+      "loss": 1.4116,
+      "step": 10760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985976645836615e-05,
+      "loss": 1.0502,
+      "step": 10761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985974015182755e-05,
+      "loss": 0.6948,
+      "step": 10762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985971384282246e-05,
+      "loss": 0.5402,
+      "step": 10763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985968753135094e-05,
+      "loss": 0.5075,
+      "step": 10764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998596612174129e-05,
+      "loss": 0.4851,
+      "step": 10765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998596349010083e-05,
+      "loss": 0.3154,
+      "step": 10766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985960858213724e-05,
+      "loss": 0.3732,
+      "step": 10767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985958226079955e-05,
+      "loss": 0.402,
+      "step": 10768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985955593699544e-05,
+      "loss": 0.3886,
+      "step": 10769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985952961072484e-05,
+      "loss": 1.1539,
+      "step": 10770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985950328198774e-05,
+      "loss": 1.3236,
+      "step": 10771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998594769507841e-05,
+      "loss": 1.0445,
+      "step": 10772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985945061711393e-05,
+      "loss": 0.8028,
+      "step": 10773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998594242809773e-05,
+      "loss": 0.4059,
+      "step": 10774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985939794237416e-05,
+      "loss": 1.045,
+      "step": 10775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998593716013045e-05,
+      "loss": 1.4564,
+      "step": 10776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998593452577683e-05,
+      "loss": 1.2097,
+      "step": 10777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998593189117656e-05,
+      "loss": 1.0419,
+      "step": 10778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985929256329644e-05,
+      "loss": 1.1569,
+      "step": 10779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998592662123607e-05,
+      "loss": 1.1554,
+      "step": 10780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985923985895855e-05,
+      "loss": 1.0237,
+      "step": 10781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985921350308984e-05,
+      "loss": 0.9649,
+      "step": 10782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985918714475463e-05,
+      "loss": 1.0963,
+      "step": 10783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998591607839529e-05,
+      "loss": 1.0937,
+      "step": 10784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985913442068475e-05,
+      "loss": 1.3142,
+      "step": 10785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998591080549499e-05,
+      "loss": 0.7393,
+      "step": 10786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985908168674875e-05,
+      "loss": 1.1603,
+      "step": 10787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99859055316081e-05,
+      "loss": 1.3147,
+      "step": 10788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998590289429468e-05,
+      "loss": 1.0646,
+      "step": 10789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998590025673461e-05,
+      "loss": 0.9826,
+      "step": 10790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985897618927886e-05,
+      "loss": 0.9658,
+      "step": 10791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985894980874516e-05,
+      "loss": 1.1047,
+      "step": 10792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998589234257448e-05,
+      "loss": 0.9545,
+      "step": 10793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998588970402781e-05,
+      "loss": 1.2241,
+      "step": 10794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998588706523449e-05,
+      "loss": 0.6729,
+      "step": 10795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985884426194516e-05,
+      "loss": 0.9699,
+      "step": 10796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998588178690789e-05,
+      "loss": 0.7535,
+      "step": 10797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998587914737461e-05,
+      "loss": 1.4064,
+      "step": 10798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985876507594685e-05,
+      "loss": 1.1086,
+      "step": 10799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985873867568114e-05,
+      "loss": 1.1563,
+      "step": 10800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998587122729489e-05,
+      "loss": 1.0799,
+      "step": 10801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998586858677501e-05,
+      "loss": 0.8134,
+      "step": 10802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998586594600849e-05,
+      "loss": 1.1417,
+      "step": 10803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998586330499532e-05,
+      "loss": 1.1602,
+      "step": 10804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998586066373549e-05,
+      "loss": 1.1683,
+      "step": 10805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985858022229015e-05,
+      "loss": 1.2593,
+      "step": 10806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998585538047589e-05,
+      "loss": 1.0802,
+      "step": 10807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998585273847612e-05,
+      "loss": 0.6904,
+      "step": 10808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985850096229694e-05,
+      "loss": 0.9784,
+      "step": 10809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985847453736624e-05,
+      "loss": 1.2396,
+      "step": 10810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99858448109969e-05,
+      "loss": 1.1279,
+      "step": 10811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998584216801052e-05,
+      "loss": 1.1777,
+      "step": 10812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99858395247775e-05,
+      "loss": 1.4183,
+      "step": 10813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985836881297826e-05,
+      "loss": 1.5293,
+      "step": 10814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99858342375715e-05,
+      "loss": 1.2572,
+      "step": 10815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998583159359853e-05,
+      "loss": 1.444,
+      "step": 10816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998582894937891e-05,
+      "loss": 1.1602,
+      "step": 10817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998582630491264e-05,
+      "loss": 0.947,
+      "step": 10818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998582366019972e-05,
+      "loss": 1.0774,
+      "step": 10819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998582101524015e-05,
+      "loss": 1.0768,
+      "step": 10820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985818370033933e-05,
+      "loss": 1.3893,
+      "step": 10821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985815724581066e-05,
+      "loss": 1.2026,
+      "step": 10822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985813078881536e-05,
+      "loss": 1.0682,
+      "step": 10823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998581043293538e-05,
+      "loss": 1.1521,
+      "step": 10824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985807786742556e-05,
+      "loss": 1.0433,
+      "step": 10825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985805140303085e-05,
+      "loss": 0.8613,
+      "step": 10826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998580249361697e-05,
+      "loss": 1.178,
+      "step": 10827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998579984668421e-05,
+      "loss": 0.9097,
+      "step": 10828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998579719950479e-05,
+      "loss": 1.1461,
+      "step": 10829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998579455207873e-05,
+      "loss": 1.397,
+      "step": 10830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998579190440601e-05,
+      "loss": 1.2818,
+      "step": 10831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985789256486654e-05,
+      "loss": 1.2913,
+      "step": 10832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998578660832064e-05,
+      "loss": 1.5145,
+      "step": 10833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998578395990798e-05,
+      "loss": 1.4816,
+      "step": 10834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985781311248667e-05,
+      "loss": 0.6521,
+      "step": 10835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998577866234271e-05,
+      "loss": 0.9915,
+      "step": 10836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985776013190105e-05,
+      "loss": 1.2507,
+      "step": 10837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985773363790843e-05,
+      "loss": 1.0145,
+      "step": 10838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998577071414494e-05,
+      "loss": 1.146,
+      "step": 10839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985768064252386e-05,
+      "loss": 0.9321,
+      "step": 10840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985765414113184e-05,
+      "loss": 1.0879,
+      "step": 10841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985762763727326e-05,
+      "loss": 1.134,
+      "step": 10842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985760113094825e-05,
+      "loss": 0.8009,
+      "step": 10843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985757462215675e-05,
+      "loss": 1.0375,
+      "step": 10844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985754811089876e-05,
+      "loss": 1.517,
+      "step": 10845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998575215971743e-05,
+      "loss": 1.6569,
+      "step": 10846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998574950809833e-05,
+      "loss": 1.3168,
+      "step": 10847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998574685623258e-05,
+      "loss": 1.1746,
+      "step": 10848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998574420412019e-05,
+      "loss": 1.3589,
+      "step": 10849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998574155176114e-05,
+      "loss": 1.181,
+      "step": 10850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985738899155454e-05,
+      "loss": 1.0479,
+      "step": 10851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998573624630311e-05,
+      "loss": 1.0858,
+      "step": 10852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998573359320412e-05,
+      "loss": 1.0836,
+      "step": 10853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998573093985848e-05,
+      "loss": 1.3111,
+      "step": 10854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99857282862662e-05,
+      "loss": 1.0305,
+      "step": 10855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985725632427263e-05,
+      "loss": 1.1478,
+      "step": 10856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985722978341674e-05,
+      "loss": 0.5696,
+      "step": 10857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998572032400944e-05,
+      "loss": 0.5121,
+      "step": 10858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998571766943057e-05,
+      "loss": 0.7915,
+      "step": 10859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998571501460503e-05,
+      "loss": 1.0314,
+      "step": 10860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985712359532856e-05,
+      "loss": 0.9899,
+      "step": 10861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998570970421402e-05,
+      "loss": 1.0597,
+      "step": 10862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985707048648555e-05,
+      "loss": 1.1329,
+      "step": 10863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998570439283643e-05,
+      "loss": 0.8242,
+      "step": 10864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998570173677766e-05,
+      "loss": 1.0128,
+      "step": 10865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998569908047224e-05,
+      "loss": 1.2129,
+      "step": 10866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998569642392017e-05,
+      "loss": 1.1209,
+      "step": 10867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998569376712145e-05,
+      "loss": 0.6446,
+      "step": 10868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985691110076086e-05,
+      "loss": 0.5986,
+      "step": 10869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985688452784076e-05,
+      "loss": 0.905,
+      "step": 10870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998568579524541e-05,
+      "loss": 1.1289,
+      "step": 10871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99856831374601e-05,
+      "loss": 0.9742,
+      "step": 10872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985680479428146e-05,
+      "loss": 1.108,
+      "step": 10873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998567782114954e-05,
+      "loss": 1.1507,
+      "step": 10874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998567516262429e-05,
+      "loss": 1.1444,
+      "step": 10875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998567250385239e-05,
+      "loss": 1.1622,
+      "step": 10876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998566984483384e-05,
+      "loss": 1.0635,
+      "step": 10877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985667185568643e-05,
+      "loss": 1.2776,
+      "step": 10878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998566452605679e-05,
+      "loss": 1.2305,
+      "step": 10879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985661866298296e-05,
+      "loss": 1.2408,
+      "step": 10880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998565920629316e-05,
+      "loss": 1.2548,
+      "step": 10881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998565654604137e-05,
+      "loss": 1.0354,
+      "step": 10882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985653885542924e-05,
+      "loss": 1.0019,
+      "step": 10883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985651224797846e-05,
+      "loss": 0.9741,
+      "step": 10884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998564856380611e-05,
+      "loss": 1.2451,
+      "step": 10885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985645902567735e-05,
+      "loss": 1.3416,
+      "step": 10886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99856432410827e-05,
+      "loss": 1.1328,
+      "step": 10887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998564057935102e-05,
+      "loss": 1.2498,
+      "step": 10888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985637917372704e-05,
+      "loss": 1.1604,
+      "step": 10889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998563525514773e-05,
+      "loss": 1.0942,
+      "step": 10890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985632592676115e-05,
+      "loss": 1.0734,
+      "step": 10891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998562992995784e-05,
+      "loss": 1.1091,
+      "step": 10892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998562726699293e-05,
+      "loss": 1.1214,
+      "step": 10893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985624603781365e-05,
+      "loss": 0.9871,
+      "step": 10894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998562194032316e-05,
+      "loss": 1.0969,
+      "step": 10895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985619276618304e-05,
+      "loss": 1.3956,
+      "step": 10896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99856166126668e-05,
+      "loss": 0.9659,
+      "step": 10897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985613948468646e-05,
+      "loss": 1.0178,
+      "step": 10898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985611284023843e-05,
+      "loss": 1.0859,
+      "step": 10899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998560861933239e-05,
+      "loss": 1.0746,
+      "step": 10900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99856059543943e-05,
+      "loss": 1.1855,
+      "step": 10901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998560328920956e-05,
+      "loss": 1.2518,
+      "step": 10902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985600623778174e-05,
+      "loss": 1.0322,
+      "step": 10903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998559795810013e-05,
+      "loss": 1.1024,
+      "step": 10904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998559529217545e-05,
+      "loss": 1.1613,
+      "step": 10905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998559262600412e-05,
+      "loss": 1.0327,
+      "step": 10906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998558995958614e-05,
+      "loss": 1.1118,
+      "step": 10907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998558729292151e-05,
+      "loss": 0.9098,
+      "step": 10908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998558462601024e-05,
+      "loss": 1.2527,
+      "step": 10909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998558195885232e-05,
+      "loss": 1.0318,
+      "step": 10910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985579291447754e-05,
+      "loss": 1.4081,
+      "step": 10911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998557662379654e-05,
+      "loss": 1.3483,
+      "step": 10912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998557395589868e-05,
+      "loss": 1.1597,
+      "step": 10913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998557128775417e-05,
+      "loss": 1.2172,
+      "step": 10914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998556861936301e-05,
+      "loss": 1.0755,
+      "step": 10915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998556595072521e-05,
+      "loss": 0.8813,
+      "step": 10916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985563281840764e-05,
+      "loss": 1.2554,
+      "step": 10917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985560612709667e-05,
+      "loss": 0.6499,
+      "step": 10918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998555794333192e-05,
+      "loss": 0.3899,
+      "step": 10919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998555527370753e-05,
+      "loss": 0.3813,
+      "step": 10920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998555260383649e-05,
+      "loss": 0.4494,
+      "step": 10921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998554993371881e-05,
+      "loss": 0.447,
+      "step": 10922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998554726335448e-05,
+      "loss": 1.1859,
+      "step": 10923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985544592743504e-05,
+      "loss": 0.9167,
+      "step": 10924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998554192188588e-05,
+      "loss": 0.8268,
+      "step": 10925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998553925078161e-05,
+      "loss": 0.9506,
+      "step": 10926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998553657943069e-05,
+      "loss": 1.0431,
+      "step": 10927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998553390783313e-05,
+      "loss": 0.9283,
+      "step": 10928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998553123598891e-05,
+      "loss": 1.0289,
+      "step": 10929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998552856389806e-05,
+      "loss": 1.1909,
+      "step": 10930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998552589156055e-05,
+      "loss": 1.4488,
+      "step": 10931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99855232189764e-05,
+      "loss": 1.6222,
+      "step": 10932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998552054614561e-05,
+      "loss": 1.1969,
+      "step": 10933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985517873068156e-05,
+      "loss": 0.8274,
+      "step": 10934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998551519974407e-05,
+      "loss": 1.3114,
+      "step": 10935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998551252617333e-05,
+      "loss": 0.9761,
+      "step": 10936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998550985235595e-05,
+      "loss": 1.1793,
+      "step": 10937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998550717829192e-05,
+      "loss": 0.4267,
+      "step": 10938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998550450398124e-05,
+      "loss": 0.2122,
+      "step": 10939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998550182942392e-05,
+      "loss": 0.2233,
+      "step": 10940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998549915461995e-05,
+      "loss": 0.1119,
+      "step": 10941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998549647956934e-05,
+      "loss": 0.0817,
+      "step": 10942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998549380427208e-05,
+      "loss": 0.1277,
+      "step": 10943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998549112872817e-05,
+      "loss": 0.0484,
+      "step": 10944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998548845293761e-05,
+      "loss": 0.0294,
+      "step": 10945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998548577690041e-05,
+      "loss": 0.2159,
+      "step": 10946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998548310061656e-05,
+      "loss": 1.1548,
+      "step": 10947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985480424086076e-05,
+      "loss": 1.1134,
+      "step": 10948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998547774730893e-05,
+      "loss": 0.8281,
+      "step": 10949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998547507028515e-05,
+      "loss": 1.2531,
+      "step": 10950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998547239301472e-05,
+      "loss": 0.9875,
+      "step": 10951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998546971549764e-05,
+      "loss": 0.9606,
+      "step": 10952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985467037733916e-05,
+      "loss": 0.5359,
+      "step": 10953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998546435972354e-05,
+      "loss": 0.8507,
+      "step": 10954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985461681466525e-05,
+      "loss": 1.152,
+      "step": 10955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998545900296287e-05,
+      "loss": 1.2994,
+      "step": 10956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985456324212564e-05,
+      "loss": 1.015,
+      "step": 10957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998545364521561e-05,
+      "loss": 1.0323,
+      "step": 10958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985450965972014e-05,
+      "loss": 1.0276,
+      "step": 10959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985448286481764e-05,
+      "loss": 0.9741,
+      "step": 10960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998544560674487e-05,
+      "loss": 0.5701,
+      "step": 10961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998544292676134e-05,
+      "loss": 1.1095,
+      "step": 10962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985440246531155e-05,
+      "loss": 1.0364,
+      "step": 10963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998543756605433e-05,
+      "loss": 1.1118,
+      "step": 10964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998543488533086e-05,
+      "loss": 1.0536,
+      "step": 10965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998543220436074e-05,
+      "loss": 1.1477,
+      "step": 10966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998542952314398e-05,
+      "loss": 1.2684,
+      "step": 10967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998542684168056e-05,
+      "loss": 1.2744,
+      "step": 10968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99854241599705e-05,
+      "loss": 1.2041,
+      "step": 10969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99854214780138e-05,
+      "loss": 1.0253,
+      "step": 10970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998541879581046e-05,
+      "loss": 0.9462,
+      "step": 10971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998541611336047e-05,
+      "loss": 1.1174,
+      "step": 10972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998541343066383e-05,
+      "loss": 0.7509,
+      "step": 10973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998541074772055e-05,
+      "loss": 1.1098,
+      "step": 10974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985408064530615e-05,
+      "loss": 1.0753,
+      "step": 10975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998540538109404e-05,
+      "loss": 0.9133,
+      "step": 10976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998540269741082e-05,
+      "loss": 1.1574,
+      "step": 10977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985400013480954e-05,
+      "loss": 1.0544,
+      "step": 10978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985397329304444e-05,
+      "loss": 0.5128,
+      "step": 10979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998539464488129e-05,
+      "loss": 0.4893,
+      "step": 10980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998539196021149e-05,
+      "loss": 1.188,
+      "step": 10981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998538927529504e-05,
+      "loss": 1.3035,
+      "step": 10982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985386590131946e-05,
+      "loss": 1.0466,
+      "step": 10983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998538390472222e-05,
+      "loss": 1.345,
+      "step": 10984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998538121906583e-05,
+      "loss": 0.963,
+      "step": 10985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985378533162806e-05,
+      "loss": 1.0332,
+      "step": 10986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985375847013136e-05,
+      "loss": 0.6341,
+      "step": 10987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998537316061681e-05,
+      "loss": 1.2144,
+      "step": 10988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985370473973856e-05,
+      "loss": 1.1795,
+      "step": 10989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985367787084245e-05,
+      "loss": 0.9995,
+      "step": 10990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998536509994799e-05,
+      "loss": 1.219,
+      "step": 10991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99853624125651e-05,
+      "loss": 1.206,
+      "step": 10992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998535972493555e-05,
+      "loss": 1.2541,
+      "step": 10993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985357037059365e-05,
+      "loss": 2.1262,
+      "step": 10994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998535434893653e-05,
+      "loss": 1.2823,
+      "step": 10995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998535166056706e-05,
+      "loss": 0.9986,
+      "step": 10996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998534897195093e-05,
+      "loss": 1.1551,
+      "step": 10997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985346283088165e-05,
+      "loss": 1.1561,
+      "step": 10998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998534359397875e-05,
+      "loss": 1.1731,
+      "step": 10999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99853409046227e-05,
+      "loss": 1.0143,
+      "step": 11000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0610970258712769,
+      "eval_runtime": 86.5342,
+      "eval_samples_per_second": 16.005,
+      "eval_steps_per_second": 4.01,
+      "step": 11000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998533821501999e-05,
+      "loss": 1.0851,
+      "step": 11001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998533552517065e-05,
+      "loss": 1.0072,
+      "step": 11002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998533283507466e-05,
+      "loss": 1.0418,
+      "step": 11003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985330144732024e-05,
+      "loss": 0.8995,
+      "step": 11004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998532745414274e-05,
+      "loss": 0.4762,
+      "step": 11005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998532476330682e-05,
+      "loss": 1.2354,
+      "step": 11006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998532207222425e-05,
+      "loss": 1.5269,
+      "step": 11007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998531938089503e-05,
+      "loss": 1.3113,
+      "step": 11008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998531668931918e-05,
+      "loss": 1.17,
+      "step": 11009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985313997496674e-05,
+      "loss": 1.1982,
+      "step": 11010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998531130542753e-05,
+      "loss": 0.9628,
+      "step": 11011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985308613111733e-05,
+      "loss": 1.1221,
+      "step": 11012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99853059205493e-05,
+      "loss": 1.0852,
+      "step": 11013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998530322774022e-05,
+      "loss": 1.0638,
+      "step": 11014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998530053468449e-05,
+      "loss": 1.4711,
+      "step": 11015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998529784138212e-05,
+      "loss": 1.0756,
+      "step": 11016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998529514783311e-05,
+      "loss": 1.2878,
+      "step": 11017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985292454037445e-05,
+      "loss": 1.5491,
+      "step": 11018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998528975999515e-05,
+      "loss": 1.0756,
+      "step": 11019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99852870657062e-05,
+      "loss": 1.0738,
+      "step": 11020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998528437117061e-05,
+      "loss": 0.9079,
+      "step": 11021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985281676388374e-05,
+      "loss": 1.3341,
+      "step": 11022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985278981359495e-05,
+      "loss": 1.7209,
+      "step": 11023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985276286083974e-05,
+      "loss": 1.4991,
+      "step": 11024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99852735905618e-05,
+      "loss": 0.9747,
+      "step": 11025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998527089479299e-05,
+      "loss": 1.0822,
+      "step": 11026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985268198777535e-05,
+      "loss": 1.0989,
+      "step": 11027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998526550251544e-05,
+      "loss": 0.8828,
+      "step": 11028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985262806006696e-05,
+      "loss": 0.5451,
+      "step": 11029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998526010925131e-05,
+      "loss": 1.0593,
+      "step": 11030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985257412249275e-05,
+      "loss": 1.2572,
+      "step": 11031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985254715000594e-05,
+      "loss": 1.1468,
+      "step": 11032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998525201750528e-05,
+      "loss": 1.3098,
+      "step": 11033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998524931976332e-05,
+      "loss": 1.1122,
+      "step": 11034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998524662177471e-05,
+      "loss": 0.8548,
+      "step": 11035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998524392353946e-05,
+      "loss": 0.9925,
+      "step": 11036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985241225057565e-05,
+      "loss": 1.0917,
+      "step": 11037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998523852632903e-05,
+      "loss": 1.0632,
+      "step": 11038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985235827353844e-05,
+      "loss": 1.0606,
+      "step": 11039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985233128132016e-05,
+      "loss": 0.9986,
+      "step": 11040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985230428663546e-05,
+      "loss": 1.0103,
+      "step": 11041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985227728948434e-05,
+      "loss": 0.67,
+      "step": 11042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998522502898668e-05,
+      "loss": 1.176,
+      "step": 11043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985222328778275e-05,
+      "loss": 1.2349,
+      "step": 11044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985219628323235e-05,
+      "loss": 1.3533,
+      "step": 11045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998521692762154e-05,
+      "loss": 1.4718,
+      "step": 11046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998521422667321e-05,
+      "loss": 0.7743,
+      "step": 11047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985211525478235e-05,
+      "loss": 1.2536,
+      "step": 11048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998520882403662e-05,
+      "loss": 1.1083,
+      "step": 11049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998520612234835e-05,
+      "loss": 1.0355,
+      "step": 11050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998520342041345e-05,
+      "loss": 1.0358,
+      "step": 11051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998520071823191e-05,
+      "loss": 0.9751,
+      "step": 11052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985198015803715e-05,
+      "loss": 0.9187,
+      "step": 11053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998519531312887e-05,
+      "loss": 1.0517,
+      "step": 11054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985192610207396e-05,
+      "loss": 1.2088,
+      "step": 11055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985189907039276e-05,
+      "loss": 1.1475,
+      "step": 11056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985187203624506e-05,
+      "loss": 0.7752,
+      "step": 11057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99851844999631e-05,
+      "loss": 0.8341,
+      "step": 11058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998518179605505e-05,
+      "loss": 0.8604,
+      "step": 11059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998517909190035e-05,
+      "loss": 0.9871,
+      "step": 11060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998517638749901e-05,
+      "loss": 0.514,
+      "step": 11061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998517368285103e-05,
+      "loss": 1.3165,
+      "step": 11062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985170977956406e-05,
+      "loss": 1.1155,
+      "step": 11063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998516827281514e-05,
+      "loss": 0.7402,
+      "step": 11064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998516556742723e-05,
+      "loss": 0.7353,
+      "step": 11065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998516286179268e-05,
+      "loss": 0.6377,
+      "step": 11066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998516015591148e-05,
+      "loss": 0.6585,
+      "step": 11067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998515744978364e-05,
+      "loss": 0.6662,
+      "step": 11068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998515474340916e-05,
+      "loss": 0.6446,
+      "step": 11069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985152036788036e-05,
+      "loss": 0.6433,
+      "step": 11070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985149329920266e-05,
+      "loss": 0.6352,
+      "step": 11071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998514662280585e-05,
+      "loss": 0.6851,
+      "step": 11072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99851439154448e-05,
+      "loss": 0.6676,
+      "step": 11073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998514120783711e-05,
+      "loss": 0.6724,
+      "step": 11074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998513849998277e-05,
+      "loss": 0.6671,
+      "step": 11075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985135791881784e-05,
+      "loss": 0.7058,
+      "step": 11076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998513308353416e-05,
+      "loss": 0.7303,
+      "step": 11077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998513037493989e-05,
+      "loss": 0.7453,
+      "step": 11078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998512766609898e-05,
+      "loss": 0.8886,
+      "step": 11079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998512495701142e-05,
+      "loss": 1.077,
+      "step": 11080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998512224767723e-05,
+      "loss": 1.3381,
+      "step": 11081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998511953809639e-05,
+      "loss": 1.021,
+      "step": 11082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985116828268905e-05,
+      "loss": 1.1992,
+      "step": 11083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998511411819479e-05,
+      "loss": 1.2632,
+      "step": 11084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998511140787402e-05,
+      "loss": 0.8558,
+      "step": 11085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998510869730662e-05,
+      "loss": 0.8319,
+      "step": 11086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998510598649256e-05,
+      "loss": 0.8487,
+      "step": 11087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985103275431863e-05,
+      "loss": 1.1844,
+      "step": 11088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998510056412453e-05,
+      "loss": 1.1179,
+      "step": 11089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998509785257056e-05,
+      "loss": 1.1867,
+      "step": 11090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985095140769933e-05,
+      "loss": 1.2414,
+      "step": 11091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998509242872267e-05,
+      "loss": 1.1789,
+      "step": 11092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998508971642877e-05,
+      "loss": 1.3805,
+      "step": 11093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998508700388822e-05,
+      "loss": 0.9683,
+      "step": 11094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998508429110103e-05,
+      "loss": 1.2478,
+      "step": 11095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985081578067194e-05,
+      "loss": 1.7467,
+      "step": 11096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998507886478673e-05,
+      "loss": 1.1114,
+      "step": 11097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985076151259604e-05,
+      "loss": 0.9845,
+      "step": 11098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998507343748585e-05,
+      "loss": 0.4566,
+      "step": 11099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998507072346545e-05,
+      "loss": 0.8858,
+      "step": 11100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998506800919841e-05,
+      "loss": 1.0575,
+      "step": 11101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998506529468472e-05,
+      "loss": 1.1572,
+      "step": 11102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985062579924393e-05,
+      "loss": 1.1657,
+      "step": 11103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998505986491743e-05,
+      "loss": 0.8033,
+      "step": 11104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985057149663816e-05,
+      "loss": 0.7907,
+      "step": 11105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998505443416357e-05,
+      "loss": 1.0515,
+      "step": 11106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998505171841667e-05,
+      "loss": 1.2635,
+      "step": 11107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985049002423135e-05,
+      "loss": 1.2101,
+      "step": 11108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998504628618296e-05,
+      "loss": 1.2433,
+      "step": 11109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998504356969613e-05,
+      "loss": 1.0422,
+      "step": 11110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998504085296267e-05,
+      "loss": 1.072,
+      "step": 11111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998503813598256e-05,
+      "loss": 1.0027,
+      "step": 11112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998503541875582e-05,
+      "loss": 1.4205,
+      "step": 11113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985032701282434e-05,
+      "loss": 1.4059,
+      "step": 11114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99850299835624e-05,
+      "loss": 1.1584,
+      "step": 11115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985027265595736e-05,
+      "loss": 1.2866,
+      "step": 11116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998502454738242e-05,
+      "loss": 1.1668,
+      "step": 11117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998502182892247e-05,
+      "loss": 1.2536,
+      "step": 11118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985019110215875e-05,
+      "loss": 1.6954,
+      "step": 11119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998501639126264e-05,
+      "loss": 1.1413,
+      "step": 11120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998501367206276e-05,
+      "loss": 0.9921,
+      "step": 11121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998501095261624e-05,
+      "loss": 0.9011,
+      "step": 11122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985008232923075e-05,
+      "loss": 1.3716,
+      "step": 11123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985005512983275e-05,
+      "loss": 1.0429,
+      "step": 11124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985002792796834e-05,
+      "loss": 3.123,
+      "step": 11125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998500007236374e-05,
+      "loss": 2.4726,
+      "step": 11126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984997351684016e-05,
+      "loss": 0.9741,
+      "step": 11127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998499463075764e-05,
+      "loss": 0.3929,
+      "step": 11128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984991909584635e-05,
+      "loss": 1.1812,
+      "step": 11129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998498918816499e-05,
+      "loss": 1.3151,
+      "step": 11130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998498646649869e-05,
+      "loss": 1.2314,
+      "step": 11131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998498374458576e-05,
+      "loss": 1.1442,
+      "step": 11132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998498102242618e-05,
+      "loss": 1.1083,
+      "step": 11133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998497830001997e-05,
+      "loss": 0.9796,
+      "step": 11134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998497557736711e-05,
+      "loss": 1.0991,
+      "step": 11135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998497285446761e-05,
+      "loss": 1.0666,
+      "step": 11136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998497013132147e-05,
+      "loss": 1.2087,
+      "step": 11137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998496740792868e-05,
+      "loss": 1.1696,
+      "step": 11138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984964684289266e-05,
+      "loss": 1.259,
+      "step": 11139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99849619604032e-05,
+      "loss": 1.1692,
+      "step": 11140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99849592362705e-05,
+      "loss": 1.1599,
+      "step": 11141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998495651189116e-05,
+      "loss": 1.1001,
+      "step": 11142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984953787265166e-05,
+      "loss": 1.228,
+      "step": 11143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998495106239254e-05,
+      "loss": 0.9668,
+      "step": 11144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984948337273274e-05,
+      "loss": 0.8809,
+      "step": 11145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998494561190736e-05,
+      "loss": 1.2722,
+      "step": 11146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998494288629482e-05,
+      "loss": 1.1455,
+      "step": 11147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998494016043562e-05,
+      "loss": 0.994,
+      "step": 11148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984937434329796e-05,
+      "loss": 1.2156,
+      "step": 11149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998493470797731e-05,
+      "loss": 1.0701,
+      "step": 11150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99849319813782e-05,
+      "loss": 1.0988,
+      "step": 11151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984929254532454e-05,
+      "loss": 0.8265,
+      "step": 11152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998492652744006e-05,
+      "loss": 1.4351,
+      "step": 11153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984923800101026e-05,
+      "loss": 0.8209,
+      "step": 11154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984921072515345e-05,
+      "loss": 1.0785,
+      "step": 11155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998491834468303e-05,
+      "loss": 1.0431,
+      "step": 11156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984915616604076e-05,
+      "loss": 1.0766,
+      "step": 11157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998491288827848e-05,
+      "loss": 1.1995,
+      "step": 11158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998491015970624e-05,
+      "loss": 0.9651,
+      "step": 11159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984907430887363e-05,
+      "loss": 0.7261,
+      "step": 11160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998490470182184e-05,
+      "loss": 1.0632,
+      "step": 11161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998490197250968e-05,
+      "loss": 1.1548,
+      "step": 11162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998489924295088e-05,
+      "loss": 1.144,
+      "step": 11163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998489651314544e-05,
+      "loss": 1.1877,
+      "step": 11164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998489378309337e-05,
+      "loss": 0.7849,
+      "step": 11165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998489105279464e-05,
+      "loss": 1.0305,
+      "step": 11166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998488832224928e-05,
+      "loss": 1.0139,
+      "step": 11167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998488559145728e-05,
+      "loss": 1.0266,
+      "step": 11168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984882860418637e-05,
+      "loss": 1.2371,
+      "step": 11169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998488012913336e-05,
+      "loss": 1.2292,
+      "step": 11170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998487739760144e-05,
+      "loss": 1.25,
+      "step": 11171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998487466582287e-05,
+      "loss": 1.1316,
+      "step": 11172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998487193379767e-05,
+      "loss": 1.1257,
+      "step": 11173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998486920152583e-05,
+      "loss": 0.9874,
+      "step": 11174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984866469007344e-05,
+      "loss": 0.9263,
+      "step": 11175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998486373624222e-05,
+      "loss": 1.2612,
+      "step": 11176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998486100323045e-05,
+      "loss": 1.3142,
+      "step": 11177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984858269972055e-05,
+      "loss": 1.019,
+      "step": 11178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998485553646701e-05,
+      "loss": 0.9364,
+      "step": 11179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984852802715335e-05,
+      "loss": 0.7494,
+      "step": 11180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984850068717004e-05,
+      "loss": 0.9415,
+      "step": 11181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984847334472045e-05,
+      "loss": 1.0966,
+      "step": 11182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998484459998044e-05,
+      "loss": 1.0315,
+      "step": 11183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99848418652422e-05,
+      "loss": 1.1631,
+      "step": 11184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998483913025732e-05,
+      "loss": 1.348,
+      "step": 11185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998483639502579e-05,
+      "loss": 1.2972,
+      "step": 11186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998483365954764e-05,
+      "loss": 0.8458,
+      "step": 11187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998483092382283e-05,
+      "loss": 0.9921,
+      "step": 11188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998482818785139e-05,
+      "loss": 1.1262,
+      "step": 11189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998482545163331e-05,
+      "loss": 0.9357,
+      "step": 11190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998482271516859e-05,
+      "loss": 1.1249,
+      "step": 11191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998481997845723e-05,
+      "loss": 0.9263,
+      "step": 11192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984817241499227e-05,
+      "loss": 1.1577,
+      "step": 11193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984814504294585e-05,
+      "loss": 1.2202,
+      "step": 11194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998481176684331e-05,
+      "loss": 1.1899,
+      "step": 11195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998480902914539e-05,
+      "loss": 1.3262,
+      "step": 11196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998480629120084e-05,
+      "loss": 1.3787,
+      "step": 11197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984803553009635e-05,
+      "loss": 1.1495,
+      "step": 11198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984800814571795e-05,
+      "loss": 1.1057,
+      "step": 11199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984798075887326e-05,
+      "loss": 1.1455,
+      "step": 11200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99847953369562e-05,
+      "loss": 1.1896,
+      "step": 11201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984792597778453e-05,
+      "loss": 1.0,
+      "step": 11202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998478985835405e-05,
+      "loss": 1.0723,
+      "step": 11203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984787118683025e-05,
+      "loss": 1.1118,
+      "step": 11204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998478437876535e-05,
+      "loss": 1.128,
+      "step": 11205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998478163860104e-05,
+      "loss": 1.2559,
+      "step": 11206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998477889819009e-05,
+      "loss": 1.1666,
+      "step": 11207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99847761575325e-05,
+      "loss": 1.0831,
+      "step": 11208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998477341662826e-05,
+      "loss": 1.1202,
+      "step": 11209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99847706754774e-05,
+      "loss": 0.7362,
+      "step": 11210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998476793407989e-05,
+      "loss": 0.8456,
+      "step": 11211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998476519243574e-05,
+      "loss": 0.4883,
+      "step": 11212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984762450544955e-05,
+      "loss": 1.166,
+      "step": 11213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984759708407526e-05,
+      "loss": 1.0985,
+      "step": 11214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998475696602346e-05,
+      "loss": 0.7811,
+      "step": 11215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998475422339276e-05,
+      "loss": 1.166,
+      "step": 11216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998475148051542e-05,
+      "loss": 1.0113,
+      "step": 11217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998474873739144e-05,
+      "loss": 1.2124,
+      "step": 11218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998474599402082e-05,
+      "loss": 1.0753,
+      "step": 11219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998474325040355e-05,
+      "loss": 1.2656,
+      "step": 11220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984740506539654e-05,
+      "loss": 1.1407,
+      "step": 11221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984737762429125e-05,
+      "loss": 0.7786,
+      "step": 11222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998473501807195e-05,
+      "loss": 0.963,
+      "step": 11223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998473227346813e-05,
+      "loss": 1.0386,
+      "step": 11224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998472952861768e-05,
+      "loss": 1.1661,
+      "step": 11225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998472678352059e-05,
+      "loss": 1.217,
+      "step": 11226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984724038176854e-05,
+      "loss": 0.7382,
+      "step": 11227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984721292586485e-05,
+      "loss": 1.2549,
+      "step": 11228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998471854674947e-05,
+      "loss": 0.9207,
+      "step": 11229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984715800665825e-05,
+      "loss": 1.0814,
+      "step": 11230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998471305433554e-05,
+      "loss": 1.1673,
+      "step": 11231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998471030775862e-05,
+      "loss": 1.0175,
+      "step": 11232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998470756093506e-05,
+      "loss": 1.1739,
+      "step": 11233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998470481386486e-05,
+      "loss": 0.8953,
+      "step": 11234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998470206654802e-05,
+      "loss": 1.0875,
+      "step": 11235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998469931898454e-05,
+      "loss": 1.0619,
+      "step": 11236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998469657117443e-05,
+      "loss": 0.7886,
+      "step": 11237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998469382311767e-05,
+      "loss": 0.75,
+      "step": 11238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998469107481428e-05,
+      "loss": 1.1156,
+      "step": 11239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998468832626425e-05,
+      "loss": 1.859,
+      "step": 11240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998468557746758e-05,
+      "loss": 1.4037,
+      "step": 11241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998468282842427e-05,
+      "loss": 1.1006,
+      "step": 11242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998468007913432e-05,
+      "loss": 0.8764,
+      "step": 11243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998467732959774e-05,
+      "loss": 1.0423,
+      "step": 11244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984674579814515e-05,
+      "loss": 0.9886,
+      "step": 11245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984671829784656e-05,
+      "loss": 0.8288,
+      "step": 11246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984669079508155e-05,
+      "loss": 1.3285,
+      "step": 11247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998466632898502e-05,
+      "loss": 1.2376,
+      "step": 11248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998466357821524e-05,
+      "loss": 1.1547,
+      "step": 11249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998466082719883e-05,
+      "loss": 1.1883,
+      "step": 11250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998465807593578e-05,
+      "loss": 0.918,
+      "step": 11251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998465532442609e-05,
+      "loss": 1.1977,
+      "step": 11252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984652572669765e-05,
+      "loss": 1.2453,
+      "step": 11253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984649820666794e-05,
+      "loss": 1.1108,
+      "step": 11254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984647068417193e-05,
+      "loss": 1.2474,
+      "step": 11255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998464431592095e-05,
+      "loss": 1.0577,
+      "step": 11256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998464156317807e-05,
+      "loss": 1.0733,
+      "step": 11257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998463881018856e-05,
+      "loss": 0.9653,
+      "step": 11258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99846360569524e-05,
+      "loss": 1.0213,
+      "step": 11259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998463330346961e-05,
+      "loss": 0.9565,
+      "step": 11260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998463054974018e-05,
+      "loss": 1.2118,
+      "step": 11261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998462779576412e-05,
+      "loss": 1.1782,
+      "step": 11262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984625041541414e-05,
+      "loss": 1.181,
+      "step": 11263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984622287072066e-05,
+      "loss": 1.0004,
+      "step": 11264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998461953235608e-05,
+      "loss": 1.2491,
+      "step": 11265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998461677739347e-05,
+      "loss": 1.4297,
+      "step": 11266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998461402218421e-05,
+      "loss": 1.1271,
+      "step": 11267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984611266728316e-05,
+      "loss": 1.1888,
+      "step": 11268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998460851102578e-05,
+      "loss": 1.2104,
+      "step": 11269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984605755076614e-05,
+      "loss": 1.0469,
+      "step": 11270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998460299888081e-05,
+      "loss": 1.071,
+      "step": 11271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998460024243836e-05,
+      "loss": 1.1764,
+      "step": 11272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984597485749286e-05,
+      "loss": 1.5474,
+      "step": 11273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998459472881356e-05,
+      "loss": 1.2111,
+      "step": 11274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998459197163121e-05,
+      "loss": 1.0373,
+      "step": 11275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998458921420222e-05,
+      "loss": 0.9903,
+      "step": 11276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984586456526585e-05,
+      "loss": 0.9765,
+      "step": 11277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998458369860432e-05,
+      "loss": 1.1196,
+      "step": 11278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984580940435414e-05,
+      "loss": 1.2614,
+      "step": 11279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998457818201987e-05,
+      "loss": 0.6967,
+      "step": 11280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984575423357686e-05,
+      "loss": 0.5994,
+      "step": 11281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998457266444887e-05,
+      "loss": 0.6136,
+      "step": 11282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984569905293416e-05,
+      "loss": 0.5445,
+      "step": 11283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998456714589133e-05,
+      "loss": 0.7206,
+      "step": 11284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984564386242596e-05,
+      "loss": 0.2923,
+      "step": 11285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998456162634722e-05,
+      "loss": 0.2527,
+      "step": 11286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998455886620523e-05,
+      "loss": 0.1374,
+      "step": 11287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998455610581659e-05,
+      "loss": 0.099,
+      "step": 11288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998455334518131e-05,
+      "loss": 0.1067,
+      "step": 11289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998455058429939e-05,
+      "loss": 0.0849,
+      "step": 11290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998454782317085e-05,
+      "loss": 0.0381,
+      "step": 11291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998454506179566e-05,
+      "loss": 0.0622,
+      "step": 11292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998454230017383e-05,
+      "loss": 0.0453,
+      "step": 11293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998453953830538e-05,
+      "loss": 0.2759,
+      "step": 11294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998453677619027e-05,
+      "loss": 0.3704,
+      "step": 11295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998453401382855e-05,
+      "loss": 0.3534,
+      "step": 11296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998453125122017e-05,
+      "loss": 0.3021,
+      "step": 11297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998452848836516e-05,
+      "loss": 0.2176,
+      "step": 11298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984525725263515e-05,
+      "loss": 0.8484,
+      "step": 11299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998452296191523e-05,
+      "loss": 0.9705,
+      "step": 11300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984520198320314e-05,
+      "loss": 2.57,
+      "step": 11301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984517434478756e-05,
+      "loss": 4.6057,
+      "step": 11302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998451467039056e-05,
+      "loss": 4.4689,
+      "step": 11303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984511906055733e-05,
+      "loss": 4.3654,
+      "step": 11304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984509141474276e-05,
+      "loss": 4.1739,
+      "step": 11305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998450637664617e-05,
+      "loss": 3.8854,
+      "step": 11306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998450361157143e-05,
+      "loss": 3.6601,
+      "step": 11307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998450084625006e-05,
+      "loss": 3.5853,
+      "step": 11308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984498080682047e-05,
+      "loss": 3.4867,
+      "step": 11309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99844953148674e-05,
+      "loss": 3.2333,
+      "step": 11310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998449254880612e-05,
+      "loss": 3.3127,
+      "step": 11311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998448978249819e-05,
+      "loss": 3.4235,
+      "step": 11312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998448701594364e-05,
+      "loss": 3.0145,
+      "step": 11313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998448424914244e-05,
+      "loss": 3.0155,
+      "step": 11314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998448148209461e-05,
+      "loss": 2.968,
+      "step": 11315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998447871480015e-05,
+      "loss": 2.445,
+      "step": 11316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998447594725905e-05,
+      "loss": 0.4463,
+      "step": 11317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998447317947131e-05,
+      "loss": 1.2936,
+      "step": 11318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998447041143694e-05,
+      "loss": 1.6815,
+      "step": 11319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984467643155924e-05,
+      "loss": 1.3543,
+      "step": 11320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998446487462828e-05,
+      "loss": 0.8619,
+      "step": 11321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998446210585399e-05,
+      "loss": 1.2308,
+      "step": 11322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998445933683308e-05,
+      "loss": 1.9087,
+      "step": 11323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998445656756552e-05,
+      "loss": 1.0822,
+      "step": 11324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998445379805133e-05,
+      "loss": 0.9731,
+      "step": 11325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99844510282905e-05,
+      "loss": 0.9626,
+      "step": 11326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998444825828304e-05,
+      "loss": 1.0638,
+      "step": 11327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998444548802894e-05,
+      "loss": 1.3116,
+      "step": 11328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99844427175282e-05,
+      "loss": 0.529,
+      "step": 11329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998443994678084e-05,
+      "loss": 1.1222,
+      "step": 11330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984437175786824e-05,
+      "loss": 1.2004,
+      "step": 11331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998443440454619e-05,
+      "loss": 1.0878,
+      "step": 11332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984431633058906e-05,
+      "loss": 1.2296,
+      "step": 11333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984428861324994e-05,
+      "loss": 1.2536,
+      "step": 11334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984426089344446e-05,
+      "loss": 1.0364,
+      "step": 11335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984423317117255e-05,
+      "loss": 1.6685,
+      "step": 11336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998442054464344e-05,
+      "loss": 1.0711,
+      "step": 11337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998441777192298e-05,
+      "loss": 0.8518,
+      "step": 11338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998441499895589e-05,
+      "loss": 0.7944,
+      "step": 11339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998441222574216e-05,
+      "loss": 1.0822,
+      "step": 11340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99844094522818e-05,
+      "loss": 0.9353,
+      "step": 11341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984406678574796e-05,
+      "loss": 1.3678,
+      "step": 11342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998440390462116e-05,
+      "loss": 1.8005,
+      "step": 11343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984401130420886e-05,
+      "loss": 1.3246,
+      "step": 11344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998439835597398e-05,
+      "loss": 1.0563,
+      "step": 11345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998439558128045e-05,
+      "loss": 1.4994,
+      "step": 11346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998439280634027e-05,
+      "loss": 3.3592,
+      "step": 11347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998439003115346e-05,
+      "loss": 5.7064,
+      "step": 11348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998438725572001e-05,
+      "loss": 5.6699,
+      "step": 11349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998438448003992e-05,
+      "loss": 5.6864,
+      "step": 11350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998438170411321e-05,
+      "loss": 5.5852,
+      "step": 11351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984378927939856e-05,
+      "loss": 5.5625,
+      "step": 11352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984376151519866e-05,
+      "loss": 5.5607,
+      "step": 11353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984373374853246e-05,
+      "loss": 5.4607,
+      "step": 11354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984370597939985e-05,
+      "loss": 5.2944,
+      "step": 11355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998436782078009e-05,
+      "loss": 5.3528,
+      "step": 11356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998436504337356e-05,
+      "loss": 1.5605,
+      "step": 11357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99843622657204e-05,
+      "loss": 0.9664,
+      "step": 11358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984359487820596e-05,
+      "loss": 0.8769,
+      "step": 11359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998435670967416e-05,
+      "loss": 0.7703,
+      "step": 11360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984353931281094e-05,
+      "loss": 0.6933,
+      "step": 11361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998435115264138e-05,
+      "loss": 0.6304,
+      "step": 11362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998434837375505e-05,
+      "loss": 0.599,
+      "step": 11363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984345594622074e-05,
+      "loss": 0.5209,
+      "step": 11364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998434281524246e-05,
+      "loss": 0.5048,
+      "step": 11365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984340035616215e-05,
+      "loss": 0.45,
+      "step": 11366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998433725574334e-05,
+      "loss": 0.3872,
+      "step": 11367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998433447562383e-05,
+      "loss": 0.3557,
+      "step": 11368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998433169525768e-05,
+      "loss": 0.327,
+      "step": 11369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99843289146449e-05,
+      "loss": 0.2999,
+      "step": 11370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998432613378547e-05,
+      "loss": 0.2661,
+      "step": 11371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998432335267942e-05,
+      "loss": 0.2799,
+      "step": 11372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984320571326735e-05,
+      "loss": 0.2492,
+      "step": 11373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998431778972741e-05,
+      "loss": 0.2166,
+      "step": 11374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984315007881455e-05,
+      "loss": 0.2357,
+      "step": 11375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998431222578886e-05,
+      "loss": 0.217,
+      "step": 11376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998430944344963e-05,
+      "loss": 0.2398,
+      "step": 11377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998430666086377e-05,
+      "loss": 0.205,
+      "step": 11378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998430387803128e-05,
+      "loss": 0.2642,
+      "step": 11379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998430109495215e-05,
+      "loss": 0.2823,
+      "step": 11380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998429831162638e-05,
+      "loss": 0.7373,
+      "step": 11381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998429552805398e-05,
+      "loss": 1.2914,
+      "step": 11382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998429274423495e-05,
+      "loss": 0.6993,
+      "step": 11383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984289960169286e-05,
+      "loss": 1.2087,
+      "step": 11384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998428717585698e-05,
+      "loss": 1.1759,
+      "step": 11385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998428439129804e-05,
+      "loss": 0.8273,
+      "step": 11386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984281606492466e-05,
+      "loss": 1.3036,
+      "step": 11387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998427882144027e-05,
+      "loss": 1.8611,
+      "step": 11388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998427603614143e-05,
+      "loss": 1.269,
+      "step": 11389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984273250595946e-05,
+      "loss": 1.1731,
+      "step": 11390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998427046480384e-05,
+      "loss": 1.1321,
+      "step": 11391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99842676787651e-05,
+      "loss": 1.0621,
+      "step": 11392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984264892479724e-05,
+      "loss": 1.1258,
+      "step": 11393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998426210594771e-05,
+      "loss": 1.191,
+      "step": 11394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998425931916907e-05,
+      "loss": 1.1869,
+      "step": 11395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998425653214379e-05,
+      "loss": 1.1527,
+      "step": 11396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984253744871877e-05,
+      "loss": 0.9867,
+      "step": 11397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998425095735333e-05,
+      "loss": 0.5327,
+      "step": 11398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998424816958815e-05,
+      "loss": 0.9447,
+      "step": 11399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998424538157633e-05,
+      "loss": 1.1777,
+      "step": 11400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984242593317885e-05,
+      "loss": 1.1002,
+      "step": 11401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99842398048128e-05,
+      "loss": 1.1605,
+      "step": 11402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998423701606109e-05,
+      "loss": 1.0985,
+      "step": 11403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984234227062734e-05,
+      "loss": 0.8734,
+      "step": 11404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998423143781775e-05,
+      "loss": 1.2233,
+      "step": 11405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998422864832614e-05,
+      "loss": 1.0246,
+      "step": 11406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998422585858788e-05,
+      "loss": 1.3118,
+      "step": 11407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984223068603e-05,
+      "loss": 1.359,
+      "step": 11408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998422027837147e-05,
+      "loss": 1.284,
+      "step": 11409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984217487893324e-05,
+      "loss": 1.2624,
+      "step": 11410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998421469716853e-05,
+      "loss": 0.9914,
+      "step": 11411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998421190619711e-05,
+      "loss": 0.9909,
+      "step": 11412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984209114979064e-05,
+      "loss": 1.0387,
+      "step": 11413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998420632351437e-05,
+      "loss": 1.1785,
+      "step": 11414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998420353180305e-05,
+      "loss": 0.9459,
+      "step": 11415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99842007398451e-05,
+      "loss": 1.1308,
+      "step": 11416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998419794764051e-05,
+      "loss": 1.2018,
+      "step": 11417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998419515518929e-05,
+      "loss": 1.0973,
+      "step": 11418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984192362491436e-05,
+      "loss": 1.1917,
+      "step": 11419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984189569546944e-05,
+      "loss": 0.8545,
+      "step": 11420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984186776355824e-05,
+      "loss": 0.8211,
+      "step": 11421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998418398291807e-05,
+      "loss": 1.113,
+      "step": 11422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998418118923368e-05,
+      "loss": 1.032,
+      "step": 11423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998417839530266e-05,
+      "loss": 1.2295,
+      "step": 11424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984175601125005e-05,
+      "loss": 1.0615,
+      "step": 11425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998417280670071e-05,
+      "loss": 0.175,
+      "step": 11426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998417001202979e-05,
+      "loss": 0.3851,
+      "step": 11427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984167217112235e-05,
+      "loss": 1.0646,
+      "step": 11428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998416442194805e-05,
+      "loss": 1.359,
+      "step": 11429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998416162653723e-05,
+      "loss": 1.206,
+      "step": 11430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998415883087977e-05,
+      "loss": 1.2987,
+      "step": 11431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998415603497568e-05,
+      "loss": 0.8914,
+      "step": 11432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984153238824964e-05,
+      "loss": 1.0199,
+      "step": 11433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984150442427614e-05,
+      "loss": 0.9027,
+      "step": 11434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998414764578362e-05,
+      "loss": 0.9879,
+      "step": 11435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984144848893006e-05,
+      "loss": 1.1858,
+      "step": 11436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984142051755756e-05,
+      "loss": 1.6896,
+      "step": 11437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998413925437186e-05,
+      "loss": 1.8751,
+      "step": 11438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998413645674135e-05,
+      "loss": 1.1724,
+      "step": 11439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99841336588642e-05,
+      "loss": 1.2096,
+      "step": 11440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998413086074042e-05,
+      "loss": 0.9712,
+      "step": 11441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984128062369995e-05,
+      "loss": 1.1346,
+      "step": 11442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984125263752944e-05,
+      "loss": 1.4867,
+      "step": 11443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984122464889264e-05,
+      "loss": 1.1887,
+      "step": 11444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998411966577895e-05,
+      "loss": 1.2453,
+      "step": 11445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984116866422004e-05,
+      "loss": 1.0821,
+      "step": 11446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984114066818424e-05,
+      "loss": 0.858,
+      "step": 11447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984111266968215e-05,
+      "loss": 1.3127,
+      "step": 11448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984108466871364e-05,
+      "loss": 1.443,
+      "step": 11449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998410566652788e-05,
+      "loss": 1.2599,
+      "step": 11450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984102865937774e-05,
+      "loss": 0.9952,
+      "step": 11451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984100065101036e-05,
+      "loss": 1.2394,
+      "step": 11452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998409726401766e-05,
+      "loss": 1.1049,
+      "step": 11453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984094462687646e-05,
+      "loss": 0.6838,
+      "step": 11454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998409166111101e-05,
+      "loss": 1.3454,
+      "step": 11455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984088859287734e-05,
+      "loss": 1.0764,
+      "step": 11456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984086057217824e-05,
+      "loss": 1.18,
+      "step": 11457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998408325490129e-05,
+      "loss": 0.9963,
+      "step": 11458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984080452338125e-05,
+      "loss": 1.1228,
+      "step": 11459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984077649528315e-05,
+      "loss": 0.8998,
+      "step": 11460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998407484647188e-05,
+      "loss": 0.9379,
+      "step": 11461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984072043168815e-05,
+      "loss": 1.5251,
+      "step": 11462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998406923961911e-05,
+      "loss": 1.5672,
+      "step": 11463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998406643582279e-05,
+      "loss": 1.1538,
+      "step": 11464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998406363177982e-05,
+      "loss": 1.1419,
+      "step": 11465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998406082749023e-05,
+      "loss": 1.306,
+      "step": 11466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984058022953996e-05,
+      "loss": 1.0548,
+      "step": 11467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984055218171135e-05,
+      "loss": 1.177,
+      "step": 11468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984052413141645e-05,
+      "loss": 1.223,
+      "step": 11469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998404960786552e-05,
+      "loss": 1.4315,
+      "step": 11470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984046802342764e-05,
+      "loss": 1.4485,
+      "step": 11471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984043996573374e-05,
+      "loss": 0.9677,
+      "step": 11472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984041190557355e-05,
+      "loss": 1.0573,
+      "step": 11473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984038384294706e-05,
+      "loss": 1.4296,
+      "step": 11474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984035577785416e-05,
+      "loss": 3.2775,
+      "step": 11475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99840327710295e-05,
+      "loss": 1.2259,
+      "step": 11476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984029964026954e-05,
+      "loss": 1.2345,
+      "step": 11477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998402715677778e-05,
+      "loss": 1.1811,
+      "step": 11478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984024349281964e-05,
+      "loss": 1.2459,
+      "step": 11479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998402154153952e-05,
+      "loss": 1.0402,
+      "step": 11480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984018733550445e-05,
+      "loss": 1.2996,
+      "step": 11481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984015925314745e-05,
+      "loss": 1.4455,
+      "step": 11482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99840131168324e-05,
+      "loss": 1.0468,
+      "step": 11483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998401030810343e-05,
+      "loss": 0.9074,
+      "step": 11484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984007499127825e-05,
+      "loss": 1.6606,
+      "step": 11485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99840046899056e-05,
+      "loss": 1.3244,
+      "step": 11486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998400188043673e-05,
+      "loss": 0.9966,
+      "step": 11487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998399907072123e-05,
+      "loss": 0.7921,
+      "step": 11488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998399626075911e-05,
+      "loss": 0.9289,
+      "step": 11489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983993450550345e-05,
+      "loss": 1.4705,
+      "step": 11490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998399064009496e-05,
+      "loss": 1.2707,
+      "step": 11491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983987829392936e-05,
+      "loss": 1.1782,
+      "step": 11492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998398501844428e-05,
+      "loss": 1.0509,
+      "step": 11493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983982207249e-05,
+      "loss": 1.2076,
+      "step": 11494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998397939580708e-05,
+      "loss": 1.2198,
+      "step": 11495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998397658411854e-05,
+      "loss": 1.2317,
+      "step": 11496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998397377218336e-05,
+      "loss": 1.0691,
+      "step": 11497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998397096000155e-05,
+      "loss": 1.4243,
+      "step": 11498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99839681475731e-05,
+      "loss": 1.2581,
+      "step": 11499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998396533489803e-05,
+      "loss": 1.1266,
+      "step": 11500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0503673553466797,
+      "eval_runtime": 83.5269,
+      "eval_samples_per_second": 16.581,
+      "eval_steps_per_second": 4.154,
+      "step": 11500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983962521976326e-05,
+      "loss": 0.9645,
+      "step": 11501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983959708807994e-05,
+      "loss": 1.2468,
+      "step": 11502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998395689539303e-05,
+      "loss": 1.0432,
+      "step": 11503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998395408173143e-05,
+      "loss": 1.2302,
+      "step": 11504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983951267823206e-05,
+      "loss": 1.081,
+      "step": 11505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983948453668345e-05,
+      "loss": 0.9665,
+      "step": 11506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998394563926685e-05,
+      "loss": 0.8637,
+      "step": 11507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998394282461874e-05,
+      "loss": 1.2088,
+      "step": 11508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983940009723976e-05,
+      "loss": 1.144,
+      "step": 11509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99839371945826e-05,
+      "loss": 1.9551,
+      "step": 11510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998393437919459e-05,
+      "loss": 1.3644,
+      "step": 11511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998393156355994e-05,
+      "loss": 1.1624,
+      "step": 11512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983928747678663e-05,
+      "loss": 1.0378,
+      "step": 11513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998392593155076e-05,
+      "loss": 1.1334,
+      "step": 11514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998392311517622e-05,
+      "loss": 1.2368,
+      "step": 11515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983920298555054e-05,
+      "loss": 1.1113,
+      "step": 11516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998391748168725e-05,
+      "loss": 1.0115,
+      "step": 11517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998391466457282e-05,
+      "loss": 1.3807,
+      "step": 11518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983911847211764e-05,
+      "loss": 1.6326,
+      "step": 11519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998390902960407e-05,
+      "loss": 1.2276,
+      "step": 11520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998390621174975e-05,
+      "loss": 1.0294,
+      "step": 11521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99839033936488e-05,
+      "loss": 0.6542,
+      "step": 11522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983900575301214e-05,
+      "loss": 0.7583,
+      "step": 11523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983897756707e-05,
+      "loss": 0.9786,
+      "step": 11524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998389493786616e-05,
+      "loss": 1.024,
+      "step": 11525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983892118778684e-05,
+      "loss": 1.0794,
+      "step": 11526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998388929944458e-05,
+      "loss": 1.2192,
+      "step": 11527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983886479863846e-05,
+      "loss": 1.3275,
+      "step": 11528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998388366003648e-05,
+      "loss": 0.9568,
+      "step": 11529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983880839962485e-05,
+      "loss": 1.1889,
+      "step": 11530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998387801964186e-05,
+      "loss": 0.9586,
+      "step": 11531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99838751990746e-05,
+      "loss": 1.2669,
+      "step": 11532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983872378260714e-05,
+      "loss": 1.2115,
+      "step": 11533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99838695572002e-05,
+      "loss": 1.479,
+      "step": 11534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998386673589305e-05,
+      "loss": 1.3306,
+      "step": 11535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983863914339276e-05,
+      "loss": 1.3134,
+      "step": 11536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983861092538865e-05,
+      "loss": 0.8815,
+      "step": 11537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983858270491826e-05,
+      "loss": 1.1442,
+      "step": 11538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998385544819816e-05,
+      "loss": 1.404,
+      "step": 11539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998385262565786e-05,
+      "loss": 0.7192,
+      "step": 11540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998384980287093e-05,
+      "loss": 1.7457,
+      "step": 11541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998384697983738e-05,
+      "loss": 1.2463,
+      "step": 11542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998384415655719e-05,
+      "loss": 1.0283,
+      "step": 11543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998384133303037e-05,
+      "loss": 1.0004,
+      "step": 11544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998383850925692e-05,
+      "loss": 0.9757,
+      "step": 11545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983835685236844e-05,
+      "loss": 1.1777,
+      "step": 11546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998383286097014e-05,
+      "loss": 0.9577,
+      "step": 11547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99838300364568e-05,
+      "loss": 1.2794,
+      "step": 11548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998382721169683e-05,
+      "loss": 1.1624,
+      "step": 11549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998382438669024e-05,
+      "loss": 1.4319,
+      "step": 11550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983821561437003e-05,
+      "loss": 1.134,
+      "step": 11551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983818735937146e-05,
+      "loss": 1.0575,
+      "step": 11552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998381591019066e-05,
+      "loss": 0.9679,
+      "step": 11553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998381308419755e-05,
+      "loss": 1.13,
+      "step": 11554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99838102579578e-05,
+      "loss": 1.1842,
+      "step": 11555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998380743147143e-05,
+      "loss": 1.129,
+      "step": 11556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998380460473842e-05,
+      "loss": 0.9742,
+      "step": 11557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998380177775879e-05,
+      "loss": 0.9647,
+      "step": 11558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998379895053252e-05,
+      "loss": 1.1006,
+      "step": 11559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983796123059626e-05,
+      "loss": 1.0966,
+      "step": 11560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99837932953401e-05,
+      "loss": 1.3214,
+      "step": 11561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998379046737395e-05,
+      "loss": 0.9388,
+      "step": 11562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998378763916117e-05,
+      "loss": 1.0182,
+      "step": 11563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983784810701756e-05,
+      "loss": 1.1197,
+      "step": 11564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983781981995717e-05,
+      "loss": 0.9419,
+      "step": 11565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998377915304304e-05,
+      "loss": 0.6806,
+      "step": 11566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983776323843744e-05,
+      "loss": 0.8233,
+      "step": 11567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998377349439781e-05,
+      "loss": 0.8634,
+      "step": 11568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983770664705255e-05,
+      "loss": 0.5917,
+      "step": 11569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998376783476607e-05,
+      "loss": 1.2584,
+      "step": 11570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998376500458025e-05,
+      "loss": 1.2925,
+      "step": 11571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998376217414781e-05,
+      "loss": 1.3929,
+      "step": 11572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983759343468725e-05,
+      "loss": 1.0965,
+      "step": 11573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983756512543025e-05,
+      "loss": 1.5753,
+      "step": 11574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983753681370696e-05,
+      "loss": 1.0878,
+      "step": 11575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983750849951725e-05,
+      "loss": 0.8005,
+      "step": 11576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998374801828614e-05,
+      "loss": 0.1209,
+      "step": 11577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983745186373923e-05,
+      "loss": 0.0624,
+      "step": 11578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983742354215066e-05,
+      "loss": 0.0626,
+      "step": 11579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998373952180959e-05,
+      "loss": 0.1138,
+      "step": 11580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998373668915748e-05,
+      "loss": 0.0397,
+      "step": 11581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998373385625874e-05,
+      "loss": 0.0496,
+      "step": 11582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998373102311338e-05,
+      "loss": 0.0782,
+      "step": 11583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983728189721384e-05,
+      "loss": 0.4105,
+      "step": 11584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998372535608276e-05,
+      "loss": 1.2428,
+      "step": 11585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998372252219751e-05,
+      "loss": 0.8774,
+      "step": 11586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983719688065625e-05,
+      "loss": 1.1977,
+      "step": 11587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998371685368712e-05,
+      "loss": 1.2278,
+      "step": 11588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998371401906198e-05,
+      "loss": 0.7271,
+      "step": 11589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998371118419021e-05,
+      "loss": 0.562,
+      "step": 11590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983708349071814e-05,
+      "loss": 1.1148,
+      "step": 11591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983705513706794e-05,
+      "loss": 1.038,
+      "step": 11592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998370267809514e-05,
+      "loss": 1.3651,
+      "step": 11593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998369984223686e-05,
+      "loss": 1.2701,
+      "step": 11594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998369700613195e-05,
+      "loss": 0.9801,
+      "step": 11595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983694169780406e-05,
+      "loss": 1.1135,
+      "step": 11596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998369133318224e-05,
+      "loss": 1.1544,
+      "step": 11597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998368849633744e-05,
+      "loss": 1.1129,
+      "step": 11598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998368565924603e-05,
+      "loss": 1.0495,
+      "step": 11599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983682821907976e-05,
+      "loss": 0.9521,
+      "step": 11600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998367998432329e-05,
+      "loss": 1.2354,
+      "step": 11601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998367714649198e-05,
+      "loss": 0.9108,
+      "step": 11602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998367430841404e-05,
+      "loss": 1.4155,
+      "step": 11603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998367147008948e-05,
+      "loss": 1.2151,
+      "step": 11604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998366863151828e-05,
+      "loss": 1.066,
+      "step": 11605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983665792700464e-05,
+      "loss": 0.9034,
+      "step": 11606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998366295363601e-05,
+      "loss": 1.1208,
+      "step": 11607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998366011432493e-05,
+      "loss": 1.1024,
+      "step": 11608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983657274767225e-05,
+      "loss": 1.1969,
+      "step": 11609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998365443496289e-05,
+      "loss": 1.2588,
+      "step": 11610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998365159491193e-05,
+      "loss": 1.196,
+      "step": 11611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998364875461433e-05,
+      "loss": 0.9279,
+      "step": 11612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998364591407011e-05,
+      "loss": 1.0814,
+      "step": 11613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998364307327927e-05,
+      "loss": 1.2077,
+      "step": 11614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983640232241794e-05,
+      "loss": 0.9932,
+      "step": 11615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998363739095769e-05,
+      "loss": 1.1417,
+      "step": 11616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998363454942696e-05,
+      "loss": 1.229,
+      "step": 11617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983631707649595e-05,
+      "loss": 1.2227,
+      "step": 11618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998362886562561e-05,
+      "loss": 1.1244,
+      "step": 11619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983626023354994e-05,
+      "loss": 1.1602,
+      "step": 11620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983623180837756e-05,
+      "loss": 1.2657,
+      "step": 11621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983620338073883e-05,
+      "loss": 1.2006,
+      "step": 11622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998361749506338e-05,
+      "loss": 1.1533,
+      "step": 11623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998361465180626e-05,
+      "loss": 1.0951,
+      "step": 11624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983611808302505e-05,
+      "loss": 1.0147,
+      "step": 11625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998360896455212e-05,
+      "loss": 1.3346,
+      "step": 11626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998360612055511e-05,
+      "loss": 1.5097,
+      "step": 11627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983603276311473e-05,
+      "loss": 0.9217,
+      "step": 11628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998360043182121e-05,
+      "loss": 1.0167,
+      "step": 11629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998359758708432e-05,
+      "loss": 1.0593,
+      "step": 11630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99835947421008e-05,
+      "loss": 0.9394,
+      "step": 11631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983591896870654e-05,
+      "loss": 0.9581,
+      "step": 11632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983589051393884e-05,
+      "loss": 1.3864,
+      "step": 11633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998358620567048e-05,
+      "loss": 0.5484,
+      "step": 11634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998358335970045e-05,
+      "loss": 0.3537,
+      "step": 11635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99835805134838e-05,
+      "loss": 1.1215,
+      "step": 11636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983577667020507e-05,
+      "loss": 1.1501,
+      "step": 11637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99835748203106e-05,
+      "loss": 1.1637,
+      "step": 11638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998357197335406e-05,
+      "loss": 0.8523,
+      "step": 11639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998356912615089e-05,
+      "loss": 1.2117,
+      "step": 11640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99835662787011e-05,
+      "loss": 1.1864,
+      "step": 11641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998356343100468e-05,
+      "loss": 0.9856,
+      "step": 11642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983560583061625e-05,
+      "loss": 1.0273,
+      "step": 11643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998355773487196e-05,
+      "loss": 1.1164,
+      "step": 11644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983554886435654e-05,
+      "loss": 1.209,
+      "step": 11645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998355203775272e-05,
+      "loss": 1.2607,
+      "step": 11646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998354918882317e-05,
+      "loss": 0.9828,
+      "step": 11647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983546339646984e-05,
+      "loss": 1.1073,
+      "step": 11648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998354349022417e-05,
+      "loss": 0.8334,
+      "step": 11649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998354064055474e-05,
+      "loss": 1.2648,
+      "step": 11650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998353779063867e-05,
+      "loss": 1.2619,
+      "step": 11651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998353494047598e-05,
+      "loss": 1.405,
+      "step": 11652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998353209006666e-05,
+      "loss": 1.4111,
+      "step": 11653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998352923941072e-05,
+      "loss": 0.9003,
+      "step": 11654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998352638850814e-05,
+      "loss": 1.1074,
+      "step": 11655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983523537358945e-05,
+      "loss": 0.9703,
+      "step": 11656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983520685963115e-05,
+      "loss": 1.0727,
+      "step": 11657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998351783432066e-05,
+      "loss": 1.1307,
+      "step": 11658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998351498243159e-05,
+      "loss": 0.7871,
+      "step": 11659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998351213029588e-05,
+      "loss": 0.7997,
+      "step": 11660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998350927791355e-05,
+      "loss": 0.7349,
+      "step": 11661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998350642528459e-05,
+      "loss": 0.934,
+      "step": 11662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983503572409004e-05,
+      "loss": 0.9828,
+      "step": 11663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998350071928679e-05,
+      "loss": 1.241,
+      "step": 11664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998349786591795e-05,
+      "loss": 1.3398,
+      "step": 11665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998349501230248e-05,
+      "loss": 1.0326,
+      "step": 11666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998349215844039e-05,
+      "loss": 1.2806,
+      "step": 11667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998348930433167e-05,
+      "loss": 1.0945,
+      "step": 11668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998348644997633e-05,
+      "loss": 1.1345,
+      "step": 11669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983483595374356e-05,
+      "loss": 1.2521,
+      "step": 11670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983480740525756e-05,
+      "loss": 1.3518,
+      "step": 11671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998347788543053e-05,
+      "loss": 1.1941,
+      "step": 11672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983475030088675e-05,
+      "loss": 1.2435,
+      "step": 11673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99834721745002e-05,
+      "loss": 1.1946,
+      "step": 11674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99834693186651e-05,
+      "loss": 0.9514,
+      "step": 11675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983466462583375e-05,
+      "loss": 0.9711,
+      "step": 11676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998346360625501e-05,
+      "loss": 1.1204,
+      "step": 11677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998346074968003e-05,
+      "loss": 0.6449,
+      "step": 11678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998345789285842e-05,
+      "loss": 0.932,
+      "step": 11679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998345503579018e-05,
+      "loss": 1.0239,
+      "step": 11680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983452178475326e-05,
+      "loss": 1.2577,
+      "step": 11681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998344932091384e-05,
+      "loss": 1.2697,
+      "step": 11682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998344646310573e-05,
+      "loss": 1.0062,
+      "step": 11683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998344360505099e-05,
+      "loss": 0.96,
+      "step": 11684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998344074674962e-05,
+      "loss": 1.4697,
+      "step": 11685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998343788820163e-05,
+      "loss": 1.0183,
+      "step": 11686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998343502940701e-05,
+      "loss": 0.9672,
+      "step": 11687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998343217036577e-05,
+      "loss": 0.8766,
+      "step": 11688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99834293110779e-05,
+      "loss": 1.0352,
+      "step": 11689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99834264515434e-05,
+      "loss": 1.0626,
+      "step": 11690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983423591762285e-05,
+      "loss": 0.9342,
+      "step": 11691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998342073173454e-05,
+      "loss": 0.9113,
+      "step": 11692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983417871460164e-05,
+      "loss": 1.1392,
+      "step": 11693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998341501093917e-05,
+      "loss": 1.147,
+      "step": 11694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998341215017154e-05,
+      "loss": 1.2468,
+      "step": 11695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998340928915729e-05,
+      "loss": 1.2733,
+      "step": 11696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983406427896416e-05,
+      "loss": 0.771,
+      "step": 11697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998340356638892e-05,
+      "loss": 1.2064,
+      "step": 11698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998340070463479e-05,
+      "loss": 1.2075,
+      "step": 11699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998339784263404e-05,
+      "loss": 1.1184,
+      "step": 11700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983394980386653e-05,
+      "loss": 1.0077,
+      "step": 11701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998339211789266e-05,
+      "loss": 1.3163,
+      "step": 11702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998338925515202e-05,
+      "loss": 1.1615,
+      "step": 11703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998338639216477e-05,
+      "loss": 1.0435,
+      "step": 11704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983383528930896e-05,
+      "loss": 1.4037,
+      "step": 11705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983380665450386e-05,
+      "loss": 1.2341,
+      "step": 11706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983377801723254e-05,
+      "loss": 0.9345,
+      "step": 11707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99833749377495e-05,
+      "loss": 0.874,
+      "step": 11708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983372073529124e-05,
+      "loss": 1.1688,
+      "step": 11709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998336920906211e-05,
+      "loss": 1.1019,
+      "step": 11710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998336634434848e-05,
+      "loss": 1.3786,
+      "step": 11711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998336347938822e-05,
+      "loss": 1.439,
+      "step": 11712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998336061418134e-05,
+      "loss": 1.339,
+      "step": 11713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998335774872783e-05,
+      "loss": 1.1371,
+      "step": 11714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99833548830277e-05,
+      "loss": 0.8706,
+      "step": 11715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998335201708094e-05,
+      "loss": 0.7482,
+      "step": 11716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998334915088756e-05,
+      "loss": 0.9907,
+      "step": 11717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983346284447554e-05,
+      "loss": 0.9305,
+      "step": 11718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998334341776092e-05,
+      "loss": 1.0183,
+      "step": 11719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983340550827656e-05,
+      "loss": 1.0315,
+      "step": 11720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983337683647774e-05,
+      "loss": 0.9853,
+      "step": 11721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998333481622127e-05,
+      "loss": 0.8974,
+      "step": 11722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998333194854813e-05,
+      "loss": 0.9707,
+      "step": 11723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983329080628374e-05,
+      "loss": 1.1358,
+      "step": 11724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998332621246199e-05,
+      "loss": 1.0645,
+      "step": 11725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983323344048984e-05,
+      "loss": 1.1729,
+      "step": 11726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998332047538935e-05,
+      "loss": 0.9998,
+      "step": 11727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998331760648309e-05,
+      "loss": 0.8516,
+      "step": 11728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998331473733021e-05,
+      "loss": 0.4824,
+      "step": 11729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99833118679307e-05,
+      "loss": 0.5791,
+      "step": 11730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998330899828457e-05,
+      "loss": 0.4122,
+      "step": 11731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983306128391814e-05,
+      "loss": 0.4782,
+      "step": 11732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998330325825243e-05,
+      "loss": 0.4057,
+      "step": 11733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998330038786643e-05,
+      "loss": 0.9686,
+      "step": 11734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983297517233796e-05,
+      "loss": 1.0921,
+      "step": 11735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998329464635454e-05,
+      "loss": 0.9476,
+      "step": 11736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998329177522866e-05,
+      "loss": 1.1128,
+      "step": 11737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998328890385616e-05,
+      "loss": 1.2486,
+      "step": 11738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983286032237025e-05,
+      "loss": 1.1701,
+      "step": 11739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983283160371275e-05,
+      "loss": 1.0324,
+      "step": 11740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99832802882589e-05,
+      "loss": 1.2524,
+      "step": 11741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983277415899896e-05,
+      "loss": 0.9121,
+      "step": 11742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983274543294266e-05,
+      "loss": 0.7149,
+      "step": 11743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983271670442014e-05,
+      "loss": 0.7126,
+      "step": 11744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998326879734314e-05,
+      "loss": 0.7163,
+      "step": 11745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998326592399764e-05,
+      "loss": 0.6952,
+      "step": 11746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998326305040551e-05,
+      "loss": 0.6841,
+      "step": 11747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998326017656677e-05,
+      "loss": 0.6924,
+      "step": 11748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983257302481404e-05,
+      "loss": 0.7164,
+      "step": 11749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99832544281494e-05,
+      "loss": 0.6867,
+      "step": 11750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998325155357078e-05,
+      "loss": 0.6912,
+      "step": 11751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983248678745536e-05,
+      "loss": 0.6771,
+      "step": 11752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983245803673665e-05,
+      "loss": 0.6655,
+      "step": 11753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998324292835517e-05,
+      "loss": 0.534,
+      "step": 11754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998324005279006e-05,
+      "loss": 0.4779,
+      "step": 11755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998323717697831e-05,
+      "loss": 0.6134,
+      "step": 11756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998323430091995e-05,
+      "loss": 0.6505,
+      "step": 11757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983231424614964e-05,
+      "loss": 0.658,
+      "step": 11758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998322854806335e-05,
+      "loss": 0.6802,
+      "step": 11759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998322567126511e-05,
+      "loss": 0.6745,
+      "step": 11760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998322279422025e-05,
+      "loss": 0.6616,
+      "step": 11761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998321991692877e-05,
+      "loss": 0.6549,
+      "step": 11762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983217039390663e-05,
+      "loss": 0.6667,
+      "step": 11763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998321416160593e-05,
+      "loss": 0.6433,
+      "step": 11764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998321128357457e-05,
+      "loss": 0.6033,
+      "step": 11765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983208405296586e-05,
+      "loss": 0.6027,
+      "step": 11766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983205526771985e-05,
+      "loss": 0.5814,
+      "step": 11767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998320264800076e-05,
+      "loss": 0.5885,
+      "step": 11768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998319976898292e-05,
+      "loss": 0.5924,
+      "step": 11769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983196889718436e-05,
+      "loss": 0.5976,
+      "step": 11770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998319401020733e-05,
+      "loss": 0.584,
+      "step": 11771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983191130449614e-05,
+      "loss": 0.567,
+      "step": 11772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998318825044527e-05,
+      "loss": 0.5799,
+      "step": 11773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983185370194305e-05,
+      "loss": 0.6011,
+      "step": 11774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998318248969671e-05,
+      "loss": 0.6608,
+      "step": 11775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99831796089525e-05,
+      "loss": 0.6653,
+      "step": 11776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998317672796166e-05,
+      "loss": 0.6688,
+      "step": 11777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99831738467242e-05,
+      "loss": 0.6603,
+      "step": 11778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998317096524011e-05,
+      "loss": 0.6453,
+      "step": 11779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983168083509404e-05,
+      "loss": 0.6551,
+      "step": 11780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983165201532066e-05,
+      "loss": 0.6575,
+      "step": 11781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998316231930811e-05,
+      "loss": 0.6216,
+      "step": 11782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998315943683754e-05,
+      "loss": 0.6309,
+      "step": 11783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983156554120334e-05,
+      "loss": 0.6641,
+      "step": 11784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998315367115651e-05,
+      "loss": 0.9049,
+      "step": 11785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998315078794606e-05,
+      "loss": 1.3519,
+      "step": 11786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998314790448899e-05,
+      "loss": 1.1333,
+      "step": 11787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983145020785296e-05,
+      "loss": 0.67,
+      "step": 11788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998314213683497e-05,
+      "loss": 1.1663,
+      "step": 11789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998313925263803e-05,
+      "loss": 1.0568,
+      "step": 11790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998313636819447e-05,
+      "loss": 1.3779,
+      "step": 11791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983133483504285e-05,
+      "loss": 1.1582,
+      "step": 11792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983130598567475e-05,
+      "loss": 0.9026,
+      "step": 11793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983127713384036e-05,
+      "loss": 1.1377,
+      "step": 11794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998312482795399e-05,
+      "loss": 1.2038,
+      "step": 11795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983121942277306e-05,
+      "loss": 1.0308,
+      "step": 11796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998311905635401e-05,
+      "loss": 0.9214,
+      "step": 11797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998311617018409e-05,
+      "loss": 1.11,
+      "step": 11798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998311328376754e-05,
+      "loss": 1.3039,
+      "step": 11799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998311039710437e-05,
+      "loss": 1.0181,
+      "step": 11800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998310751019457e-05,
+      "loss": 0.7254,
+      "step": 11801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998310462303816e-05,
+      "loss": 0.9878,
+      "step": 11802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998310173563512e-05,
+      "loss": 1.1497,
+      "step": 11803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998309884798546e-05,
+      "loss": 1.106,
+      "step": 11804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998309596008918e-05,
+      "loss": 0.331,
+      "step": 11805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983093071946274e-05,
+      "loss": 0.6411,
+      "step": 11806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998309018355674e-05,
+      "loss": 1.171,
+      "step": 11807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998308729492059e-05,
+      "loss": 1.2667,
+      "step": 11808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983084406037815e-05,
+      "loss": 0.9939,
+      "step": 11809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998308151690842e-05,
+      "loss": 1.0861,
+      "step": 11810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99830786275324e-05,
+      "loss": 1.1063,
+      "step": 11811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983075737909765e-05,
+      "loss": 1.1264,
+      "step": 11812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99830728480405e-05,
+      "loss": 0.9674,
+      "step": 11813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998306995792462e-05,
+      "loss": 1.1379,
+      "step": 11814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998306706756211e-05,
+      "loss": 0.9703,
+      "step": 11815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983064176952976e-05,
+      "loss": 1.3968,
+      "step": 11816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998306128609722e-05,
+      "loss": 1.0956,
+      "step": 11817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998305839499485e-05,
+      "loss": 2.2113,
+      "step": 11818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983055503645854e-05,
+      "loss": 2.0404,
+      "step": 11819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983052612050234e-05,
+      "loss": 1.96,
+      "step": 11820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983049720208e-05,
+      "loss": 1.9934,
+      "step": 11821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983046828119127e-05,
+      "loss": 1.0033,
+      "step": 11822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983043935783647e-05,
+      "loss": 1.083,
+      "step": 11823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998304104320154e-05,
+      "loss": 0.9911,
+      "step": 11824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998303815037281e-05,
+      "loss": 1.1688,
+      "step": 11825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983035257297454e-05,
+      "loss": 1.3553,
+      "step": 11826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998303236397548e-05,
+      "loss": 0.9616,
+      "step": 11827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998302947040688e-05,
+      "loss": 1.381,
+      "step": 11828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998302657659167e-05,
+      "loss": 1.2144,
+      "step": 11829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998302368252983e-05,
+      "loss": 0.9461,
+      "step": 11830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998302078822137e-05,
+      "loss": 0.5215,
+      "step": 11831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983017893666284e-05,
+      "loss": 0.9689,
+      "step": 11832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998301499886458e-05,
+      "loss": 1.4347,
+      "step": 11833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998301210381625e-05,
+      "loss": 0.9473,
+      "step": 11834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99830092085213e-05,
+      "loss": 1.5753,
+      "step": 11835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998300631297973e-05,
+      "loss": 1.5506,
+      "step": 11836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983003417191544e-05,
+      "loss": 1.1735,
+      "step": 11837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998300052115673e-05,
+      "loss": 1.0264,
+      "step": 11838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998299762487529e-05,
+      "loss": 1.2732,
+      "step": 11839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982994728347235e-05,
+      "loss": 1.0026,
+      "step": 11840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998299183157256e-05,
+      "loss": 0.9684,
+      "step": 11841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998298893455126e-05,
+      "loss": 1.1025,
+      "step": 11842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982986037283335e-05,
+      "loss": 0.8175,
+      "step": 11843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998298313976879e-05,
+      "loss": 0.6195,
+      "step": 11844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982980242007624e-05,
+      "loss": 1.047,
+      "step": 11845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998297734399984e-05,
+      "loss": 0.8917,
+      "step": 11846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998297444574543e-05,
+      "loss": 1.0824,
+      "step": 11847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982971547244406e-05,
+      "loss": 1.0504,
+      "step": 11848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998296864849675e-05,
+      "loss": 0.9859,
+      "step": 11849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998296574950249e-05,
+      "loss": 1.2751,
+      "step": 11850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982962850261595e-05,
+      "loss": 1.1871,
+      "step": 11851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998295995077408e-05,
+      "loss": 1.3634,
+      "step": 11852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982957051039944e-05,
+      "loss": 1.5256,
+      "step": 11853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982954151059185e-05,
+      "loss": 1.6309,
+      "step": 11854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998295125083181e-05,
+      "loss": 1.72,
+      "step": 11855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998294835035781e-05,
+      "loss": 2.9109,
+      "step": 11856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998294544963719e-05,
+      "loss": 0.969,
+      "step": 11857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998294254866995e-05,
+      "loss": 0.7834,
+      "step": 11858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982939647456087e-05,
+      "loss": 1.0156,
+      "step": 11859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99829367459956e-05,
+      "loss": 1.1561,
+      "step": 11860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982933844288495e-05,
+      "loss": 1.1978,
+      "step": 11861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998293094233477e-05,
+      "loss": 1.1596,
+      "step": 11862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998292804013442e-05,
+      "loss": 1.118,
+      "step": 11863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982925137687456e-05,
+      "loss": 1.2925,
+      "step": 11864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998292223499387e-05,
+      "loss": 1.0552,
+      "step": 11865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998291933205366e-05,
+      "loss": 1.5542,
+      "step": 11866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998291642886683e-05,
+      "loss": 1.1159,
+      "step": 11867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982913525433376e-05,
+      "loss": 0.9709,
+      "step": 11868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982910621753306e-05,
+      "loss": 0.993,
+      "step": 11869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982907717826614e-05,
+      "loss": 1.3587,
+      "step": 11870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99829048136533e-05,
+      "loss": 1.2833,
+      "step": 11871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998290190923337e-05,
+      "loss": 1.1247,
+      "step": 11872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998289900456681e-05,
+      "loss": 0.836,
+      "step": 11873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982896099653644e-05,
+      "loss": 0.9815,
+      "step": 11874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998289319449384e-05,
+      "loss": 0.985,
+      "step": 11875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982890289087423e-05,
+      "loss": 0.7515,
+      "step": 11876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998288738343439e-05,
+      "loss": 1.1292,
+      "step": 11877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998288447753473e-05,
+      "loss": 1.2825,
+      "step": 11878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998288157138845e-05,
+      "loss": 0.9876,
+      "step": 11879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998287866499556e-05,
+      "loss": 1.0972,
+      "step": 11880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998287575835604e-05,
+      "loss": 1.5533,
+      "step": 11881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99828728514699e-05,
+      "loss": 1.0838,
+      "step": 11882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998286994433714e-05,
+      "loss": 0.9633,
+      "step": 11883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998286703695776e-05,
+      "loss": 1.0257,
+      "step": 11884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998286412933176e-05,
+      "loss": 1.0479,
+      "step": 11885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982861221459135e-05,
+      "loss": 1.1619,
+      "step": 11886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99828583133399e-05,
+      "loss": 1.0674,
+      "step": 11887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998285540497403e-05,
+      "loss": 1.2695,
+      "step": 11888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998285249636155e-05,
+      "loss": 1.2531,
+      "step": 11889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998284958750245e-05,
+      "loss": 0.8401,
+      "step": 11890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998284667839673e-05,
+      "loss": 0.7986,
+      "step": 11891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982843769044384e-05,
+      "loss": 1.2023,
+      "step": 11892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982840859445424e-05,
+      "loss": 0.6585,
+      "step": 11893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998283794959984e-05,
+      "loss": 1.1371,
+      "step": 11894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998283503950764e-05,
+      "loss": 1.1359,
+      "step": 11895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982832129168824e-05,
+      "loss": 0.848,
+      "step": 11896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998282921858337e-05,
+      "loss": 1.4398,
+      "step": 11897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998282630775131e-05,
+      "loss": 1.3705,
+      "step": 11898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998282339667263e-05,
+      "loss": 1.0344,
+      "step": 11899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982820485347324e-05,
+      "loss": 1.21,
+      "step": 11900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998281757377541e-05,
+      "loss": 1.0661,
+      "step": 11901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998281466195687e-05,
+      "loss": 0.636,
+      "step": 11902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99828117498917e-05,
+      "loss": 0.464,
+      "step": 11903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998280883757992e-05,
+      "loss": 0.4931,
+      "step": 11904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982805925021526e-05,
+      "loss": 0.4886,
+      "step": 11905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99828030122165e-05,
+      "loss": 0.4493,
+      "step": 11906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998280009916486e-05,
+      "loss": 0.4238,
+      "step": 11907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982797185866605e-05,
+      "loss": 0.3304,
+      "step": 11908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998279427232172e-05,
+      "loss": 0.2891,
+      "step": 11909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998279135853022e-05,
+      "loss": 0.5425,
+      "step": 11910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99827884444921e-05,
+      "loss": 1.1136,
+      "step": 11911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982785530207366e-05,
+      "loss": 1.098,
+      "step": 11912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982782615676004e-05,
+      "loss": 1.3658,
+      "step": 11913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982779700898034e-05,
+      "loss": 1.0967,
+      "step": 11914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998277678587343e-05,
+      "loss": 0.8413,
+      "step": 11915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998277387060221e-05,
+      "loss": 1.1046,
+      "step": 11916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982770955084376e-05,
+      "loss": 0.5204,
+      "step": 11917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982768039319924e-05,
+      "loss": 0.5353,
+      "step": 11918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998276512330885e-05,
+      "loss": 0.7756,
+      "step": 11919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982762207051154e-05,
+      "loss": 1.1151,
+      "step": 11920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982759290546836e-05,
+      "loss": 1.0285,
+      "step": 11921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99827563737959e-05,
+      "loss": 0.9638,
+      "step": 11922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998275345679835e-05,
+      "loss": 1.0292,
+      "step": 11923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998275053955418e-05,
+      "loss": 1.0397,
+      "step": 11924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998274762206339e-05,
+      "loss": 0.2139,
+      "step": 11925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998274470432599e-05,
+      "loss": 0.1712,
+      "step": 11926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982741786341956e-05,
+      "loss": 0.0851,
+      "step": 11927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982738868111304e-05,
+      "loss": 0.0965,
+      "step": 11928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998273594963404e-05,
+      "loss": 0.0471,
+      "step": 11929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998273303091016e-05,
+      "loss": 0.172,
+      "step": 11930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982730111939646e-05,
+      "loss": 0.3087,
+      "step": 11931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982727192722525e-05,
+      "loss": 1.0428,
+      "step": 11932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998272427325878e-05,
+      "loss": 1.0664,
+      "step": 11933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982721353548424e-05,
+      "loss": 1.4479,
+      "step": 11934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982718433591444e-05,
+      "loss": 0.9686,
+      "step": 11935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998271551338784e-05,
+      "loss": 0.7969,
+      "step": 11936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982712592937616e-05,
+      "loss": 0.9167,
+      "step": 11937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998270967224078e-05,
+      "loss": 1.133,
+      "step": 11938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982706751297334e-05,
+      "loss": 1.3677,
+      "step": 11939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998270383010726e-05,
+      "loss": 0.8491,
+      "step": 11940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982700908670564e-05,
+      "loss": 1.178,
+      "step": 11941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998269798698725e-05,
+      "loss": 1.2003,
+      "step": 11942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998269506505732e-05,
+      "loss": 1.2625,
+      "step": 11943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998269214288077e-05,
+      "loss": 1.0666,
+      "step": 11944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99826892204576e-05,
+      "loss": 0.9202,
+      "step": 11945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982686297787816e-05,
+      "loss": 1.2332,
+      "step": 11946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998268337487141e-05,
+      "loss": 0.8747,
+      "step": 11947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998268045170838e-05,
+      "loss": 0.788,
+      "step": 11948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998267752829874e-05,
+      "loss": 0.9752,
+      "step": 11949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998267460464249e-05,
+      "loss": 0.8442,
+      "step": 11950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99826716807396e-05,
+      "loss": 1.1171,
+      "step": 11951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998266875659011e-05,
+      "loss": 0.8666,
+      "step": 11952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998266583219399e-05,
+      "loss": 1.3084,
+      "step": 11953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998266290755126e-05,
+      "loss": 1.5812,
+      "step": 11954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998265998266191e-05,
+      "loss": 1.4263,
+      "step": 11955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998265705752593e-05,
+      "loss": 1.3035,
+      "step": 11956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982654132143346e-05,
+      "loss": 1.618,
+      "step": 11957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998265120651414e-05,
+      "loss": 1.44,
+      "step": 11958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998264828063831e-05,
+      "loss": 0.9215,
+      "step": 11959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998264535451587e-05,
+      "loss": 0.9074,
+      "step": 11960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998264242814681e-05,
+      "loss": 1.1559,
+      "step": 11961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998263950153113e-05,
+      "loss": 0.8698,
+      "step": 11962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998263657466883e-05,
+      "loss": 1.4071,
+      "step": 11963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982633647559915e-05,
+      "loss": 1.1193,
+      "step": 11964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998263072020438e-05,
+      "loss": 0.8669,
+      "step": 11965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998262779260223e-05,
+      "loss": 1.0289,
+      "step": 11966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982624864753456e-05,
+      "loss": 1.0742,
+      "step": 11967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998262193665807e-05,
+      "loss": 0.9148,
+      "step": 11968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982619008316065e-05,
+      "loss": 1.1374,
+      "step": 11969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998261607972744e-05,
+      "loss": 1.2256,
+      "step": 11970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99826131508922e-05,
+      "loss": 0.9827,
+      "step": 11971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982610221810335e-05,
+      "loss": 0.8465,
+      "step": 11972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982607292481856e-05,
+      "loss": 0.9577,
+      "step": 11973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998260436290676e-05,
+      "loss": 1.0681,
+      "step": 11974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998260143308505e-05,
+      "loss": 1.0878,
+      "step": 11975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998259850301672e-05,
+      "loss": 1.1793,
+      "step": 11976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982595572701776e-05,
+      "loss": 1.0593,
+      "step": 11977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998259264214021e-05,
+      "loss": 1.1046,
+      "step": 11978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998258971133202e-05,
+      "loss": 1.1013,
+      "step": 11979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998258678027722e-05,
+      "loss": 1.0016,
+      "step": 11980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998258384897581e-05,
+      "loss": 1.2534,
+      "step": 11981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982580917427765e-05,
+      "loss": 1.0986,
+      "step": 11982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998257798563312e-05,
+      "loss": 1.0771,
+      "step": 11983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998257505359184e-05,
+      "loss": 1.2647,
+      "step": 11984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998257212130395e-05,
+      "loss": 0.9088,
+      "step": 11985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982569188769444e-05,
+      "loss": 1.4022,
+      "step": 11986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998256625598833e-05,
+      "loss": 1.1231,
+      "step": 11987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998256332296059e-05,
+      "loss": 1.1074,
+      "step": 11988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998256038968623e-05,
+      "loss": 1.1353,
+      "step": 11989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982557456165244e-05,
+      "loss": 1.07,
+      "step": 11990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982554522397656e-05,
+      "loss": 0.887,
+      "step": 11991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982551588383445e-05,
+      "loss": 0.6104,
+      "step": 11992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998254865412262e-05,
+      "loss": 1.2055,
+      "step": 11993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998254571961518e-05,
+      "loss": 1.4669,
+      "step": 11994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982542784861115e-05,
+      "loss": 1.1528,
+      "step": 11995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998253984986043e-05,
+      "loss": 1.1111,
+      "step": 11996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982536914613135e-05,
+      "loss": 1.1834,
+      "step": 11997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982533979119226e-05,
+      "loss": 0.9809,
+      "step": 11998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982531043378695e-05,
+      "loss": 1.3561,
+      "step": 11999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998252810739155e-05,
+      "loss": 1.0595,
+      "step": 12000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0549529790878296,
+      "eval_runtime": 95.3857,
+      "eval_samples_per_second": 14.52,
+      "eval_steps_per_second": 3.638,
+      "step": 12000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982525171157786e-05,
+      "loss": 1.407,
+      "step": 12001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99825222346774e-05,
+      "loss": 1.2803,
+      "step": 12002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99825192979504e-05,
+      "loss": 1.2961,
+      "step": 12003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982516360976794e-05,
+      "loss": 1.1177,
+      "step": 12004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982513423756564e-05,
+      "loss": 1.2042,
+      "step": 12005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998251048628971e-05,
+      "loss": 1.0742,
+      "step": 12006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982507548576244e-05,
+      "loss": 1.3132,
+      "step": 12007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998250461061617e-05,
+      "loss": 1.3119,
+      "step": 12008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998250167240946e-05,
+      "loss": 1.468,
+      "step": 12009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998249873395615e-05,
+      "loss": 1.5029,
+      "step": 12010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998249579525621e-05,
+      "loss": 0.4397,
+      "step": 12011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998249285630967e-05,
+      "loss": 0.9234,
+      "step": 12012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99824899171165e-05,
+      "loss": 1.1719,
+      "step": 12013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998248697767672e-05,
+      "loss": 1.0474,
+      "step": 12014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982484037990316e-05,
+      "loss": 1.0646,
+      "step": 12015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982481098057304e-05,
+      "loss": 1.647,
+      "step": 12016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982478157877676e-05,
+      "loss": 1.5274,
+      "step": 12017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982475217451426e-05,
+      "loss": 1.4571,
+      "step": 12018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998247227677856e-05,
+      "loss": 1.371,
+      "step": 12019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998246933585908e-05,
+      "loss": 0.9796,
+      "step": 12020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982466394692984e-05,
+      "loss": 0.8978,
+      "step": 12021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998246345328027e-05,
+      "loss": 1.0147,
+      "step": 12022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998246051162093e-05,
+      "loss": 1.2353,
+      "step": 12023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998245756971499e-05,
+      "loss": 1.0887,
+      "step": 12024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998245462756243e-05,
+      "loss": 0.9905,
+      "step": 12025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998245168516325e-05,
+      "loss": 1.1723,
+      "step": 12026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998244874251745e-05,
+      "loss": 1.1858,
+      "step": 12027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982445799625036e-05,
+      "loss": 1.1936,
+      "step": 12028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998244285648601e-05,
+      "loss": 1.1744,
+      "step": 12029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998243991310036e-05,
+      "loss": 1.1702,
+      "step": 12030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998243696946811e-05,
+      "loss": 1.2338,
+      "step": 12031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998243402558923e-05,
+      "loss": 1.0368,
+      "step": 12032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982431081463735e-05,
+      "loss": 1.0198,
+      "step": 12033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998242813709163e-05,
+      "loss": 0.8942,
+      "step": 12034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99824251924729e-05,
+      "loss": 1.2525,
+      "step": 12035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998242224760756e-05,
+      "loss": 1.1418,
+      "step": 12036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998241930249561e-05,
+      "loss": 1.1844,
+      "step": 12037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998241635713704e-05,
+      "loss": 1.4223,
+      "step": 12038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982413411531845e-05,
+      "loss": 1.1155,
+      "step": 12039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998241046568004e-05,
+      "loss": 1.1506,
+      "step": 12040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998240751958163e-05,
+      "loss": 1.1888,
+      "step": 12041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998240457323658e-05,
+      "loss": 1.2224,
+      "step": 12042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982401626644936e-05,
+      "loss": 1.2334,
+      "step": 12043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998239867980667e-05,
+      "loss": 0.8968,
+      "step": 12044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998239573272179e-05,
+      "loss": 1.0604,
+      "step": 12045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998239278539028e-05,
+      "loss": 0.7093,
+      "step": 12046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982389837812175e-05,
+      "loss": 1.4507,
+      "step": 12047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982386889987444e-05,
+      "loss": 1.3586,
+      "step": 12048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99823839419161e-05,
+      "loss": 0.9533,
+      "step": 12049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998238099359814e-05,
+      "loss": 0.9154,
+      "step": 12050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998237804503356e-05,
+      "loss": 0.9695,
+      "step": 12051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998237509622237e-05,
+      "loss": 1.1887,
+      "step": 12052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998237214716456e-05,
+      "loss": 1.1882,
+      "step": 12053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998236919786014e-05,
+      "loss": 1.0273,
+      "step": 12054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982366248309096e-05,
+      "loss": 1.229,
+      "step": 12055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998236329851145e-05,
+      "loss": 1.2101,
+      "step": 12056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998236034846718e-05,
+      "loss": 0.9866,
+      "step": 12057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998235739817629e-05,
+      "loss": 1.1917,
+      "step": 12058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998235444763879e-05,
+      "loss": 0.853,
+      "step": 12059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998235149685468e-05,
+      "loss": 1.0745,
+      "step": 12060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998234854582395e-05,
+      "loss": 1.1735,
+      "step": 12061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99823455945466e-05,
+      "loss": 1.3232,
+      "step": 12062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998234264302264e-05,
+      "loss": 1.137,
+      "step": 12063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982339691252066e-05,
+      "loss": 1.0724,
+      "step": 12064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982336739234875e-05,
+      "loss": 1.1275,
+      "step": 12065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998233378697107e-05,
+      "loss": 1.189,
+      "step": 12066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998233083446065e-05,
+      "loss": 1.0839,
+      "step": 12067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998232788170361e-05,
+      "loss": 1.1865,
+      "step": 12068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982324928699964e-05,
+      "loss": 1.2996,
+      "step": 12069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982321975449696e-05,
+      "loss": 1.0822,
+      "step": 12070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998231902195282e-05,
+      "loss": 0.9162,
+      "step": 12071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982316068209315e-05,
+      "loss": 1.1151,
+      "step": 12072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99823131142192e-05,
+      "loss": 0.9708,
+      "step": 12073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998231015998248e-05,
+      "loss": 0.8525,
+      "step": 12074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998230720549914e-05,
+      "loss": 1.1532,
+      "step": 12075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998230425076919e-05,
+      "loss": 1.061,
+      "step": 12076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998230129579261e-05,
+      "loss": 1.1644,
+      "step": 12077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998229834056943e-05,
+      "loss": 1.3484,
+      "step": 12078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998229538509963e-05,
+      "loss": 1.3026,
+      "step": 12079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982292429383216e-05,
+      "loss": 1.1669,
+      "step": 12080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982289473420186e-05,
+      "loss": 1.1338,
+      "step": 12081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998228651721054e-05,
+      "loss": 0.9417,
+      "step": 12082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998228356075428e-05,
+      "loss": 1.1192,
+      "step": 12083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998228060405141e-05,
+      "loss": 1.1765,
+      "step": 12084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998227764710192e-05,
+      "loss": 1.162,
+      "step": 12085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982274689905814e-05,
+      "loss": 1.1069,
+      "step": 12086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982271732463106e-05,
+      "loss": 1.2318,
+      "step": 12087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982268774773776e-05,
+      "loss": 1.0605,
+      "step": 12088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982265816837824e-05,
+      "loss": 0.5878,
+      "step": 12089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998226285865527e-05,
+      "loss": 0.6539,
+      "step": 12090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998225990022609e-05,
+      "loss": 0.9472,
+      "step": 12091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998225694155031e-05,
+      "loss": 1.1295,
+      "step": 12092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99822539826279e-05,
+      "loss": 1.0472,
+      "step": 12093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998225102345888e-05,
+      "loss": 1.2961,
+      "step": 12094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982248064043255e-05,
+      "loss": 1.1211,
+      "step": 12095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998224510438101e-05,
+      "loss": 0.8796,
+      "step": 12096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998224214447215e-05,
+      "loss": 0.8588,
+      "step": 12097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998223918431668e-05,
+      "loss": 0.6565,
+      "step": 12098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982236223914585e-05,
+      "loss": 1.061,
+      "step": 12099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982233263265884e-05,
+      "loss": 1.2461,
+      "step": 12100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998223030237057e-05,
+      "loss": 1.4045,
+      "step": 12101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998222734122864e-05,
+      "loss": 1.216,
+      "step": 12102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998222437984009e-05,
+      "loss": 1.0652,
+      "step": 12103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982221418204935e-05,
+      "loss": 1.2943,
+      "step": 12104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998221845632316e-05,
+      "loss": 1.3192,
+      "step": 12105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998221549419477e-05,
+      "loss": 1.3339,
+      "step": 12106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998221253181978e-05,
+      "loss": 1.5881,
+      "step": 12107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998220956919816e-05,
+      "loss": 1.3818,
+      "step": 12108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982206606329935e-05,
+      "loss": 1.2337,
+      "step": 12109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982203643215094e-05,
+      "loss": 1.2094,
+      "step": 12110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998220067985364e-05,
+      "loss": 1.3412,
+      "step": 12111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982197716245574e-05,
+      "loss": 1.1303,
+      "step": 12112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998219475239089e-05,
+      "loss": 0.9769,
+      "step": 12113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998219178828959e-05,
+      "loss": 1.0795,
+      "step": 12114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982188823941675e-05,
+      "loss": 1.0715,
+      "step": 12115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982185859347156e-05,
+      "loss": 1.1885,
+      "step": 12116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982182894506015e-05,
+      "loss": 0.9226,
+      "step": 12117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982179929418265e-05,
+      "loss": 0.9401,
+      "step": 12118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998217696408391e-05,
+      "loss": 0.7499,
+      "step": 12119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982173998502926e-05,
+      "loss": 0.364,
+      "step": 12120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998217103267533e-05,
+      "loss": 1.0029,
+      "step": 12121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982168066601126e-05,
+      "loss": 0.9788,
+      "step": 12122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998216510028031e-05,
+      "loss": 0.8128,
+      "step": 12123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998216213371287e-05,
+      "loss": 0.8819,
+      "step": 12124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998215916689883e-05,
+      "loss": 1.0108,
+      "step": 12125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998215619983817e-05,
+      "loss": 0.9376,
+      "step": 12126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982153232530895e-05,
+      "loss": 0.7809,
+      "step": 12127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998215026497701e-05,
+      "loss": 0.7568,
+      "step": 12128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998214729717652e-05,
+      "loss": 1.0232,
+      "step": 12129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998214432912941e-05,
+      "loss": 1.3491,
+      "step": 12130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998214136083568e-05,
+      "loss": 1.2105,
+      "step": 12131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998213839229534e-05,
+      "loss": 1.145,
+      "step": 12132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998213542350839e-05,
+      "loss": 1.1217,
+      "step": 12133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998213245447483e-05,
+      "loss": 1.166,
+      "step": 12134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982129485194644e-05,
+      "loss": 1.2469,
+      "step": 12135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998212651566786e-05,
+      "loss": 1.2689,
+      "step": 12136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998212354589445e-05,
+      "loss": 1.0761,
+      "step": 12137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982120575874436e-05,
+      "loss": 0.984,
+      "step": 12138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982117605607805e-05,
+      "loss": 1.2007,
+      "step": 12139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982114635094566e-05,
+      "loss": 1.1728,
+      "step": 12140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998211166433471e-05,
+      "loss": 1.186,
+      "step": 12141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998210869332824e-05,
+      "loss": 1.2243,
+      "step": 12142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982105722075154e-05,
+      "loss": 1.4051,
+      "step": 12143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998210275057547e-05,
+      "loss": 1.0489,
+      "step": 12144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982099778829164e-05,
+      "loss": 0.8759,
+      "step": 12145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998209680683624e-05,
+      "loss": 1.2142,
+      "step": 12146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982093834596706e-05,
+      "loss": 1.3355,
+      "step": 12147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982090862110564e-05,
+      "loss": 0.9924,
+      "step": 12148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982087889377806e-05,
+      "loss": 0.9815,
+      "step": 12149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982084916398434e-05,
+      "loss": 0.6959,
+      "step": 12150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998208194317245e-05,
+      "loss": 0.4738,
+      "step": 12151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982078969699856e-05,
+      "loss": 0.3914,
+      "step": 12152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998207599598065e-05,
+      "loss": 0.6012,
+      "step": 12153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998207302201482e-05,
+      "loss": 0.3795,
+      "step": 12154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982070047802394e-05,
+      "loss": 0.3312,
+      "step": 12155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998206707334335e-05,
+      "loss": 0.4454,
+      "step": 12156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998206409863769e-05,
+      "loss": 0.9173,
+      "step": 12157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998206112368542e-05,
+      "loss": 1.0747,
+      "step": 12158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998205814848653e-05,
+      "loss": 1.2573,
+      "step": 12159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998205517304104e-05,
+      "loss": 1.3131,
+      "step": 12160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998205219734893e-05,
+      "loss": 0.9893,
+      "step": 12161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998204922141021e-05,
+      "loss": 0.5699,
+      "step": 12162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998204624522488e-05,
+      "loss": 0.7396,
+      "step": 12163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998204326879294e-05,
+      "loss": 0.9701,
+      "step": 12164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998204029211438e-05,
+      "loss": 1.189,
+      "step": 12165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998203731518921e-05,
+      "loss": 1.275,
+      "step": 12166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998203433801743e-05,
+      "loss": 1.1439,
+      "step": 12167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998203136059904e-05,
+      "loss": 0.8104,
+      "step": 12168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982028382934034e-05,
+      "loss": 1.1535,
+      "step": 12169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982025405022416e-05,
+      "loss": 1.0484,
+      "step": 12170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998202242686418e-05,
+      "loss": 0.9965,
+      "step": 12171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982019448459346e-05,
+      "loss": 0.8001,
+      "step": 12172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998201646980789e-05,
+      "loss": 0.9192,
+      "step": 12173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998201349090983e-05,
+      "loss": 1.187,
+      "step": 12174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982010511765154e-05,
+      "loss": 1.1789,
+      "step": 12175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982007532373864e-05,
+      "loss": 0.832,
+      "step": 12176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998200455273596e-05,
+      "loss": 1.1056,
+      "step": 12177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998200157285145e-05,
+      "loss": 1.0418,
+      "step": 12178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998199859272032e-05,
+      "loss": 0.8702,
+      "step": 12179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998199561234259e-05,
+      "loss": 1.1537,
+      "step": 12180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981992631718244e-05,
+      "loss": 1.0839,
+      "step": 12181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998198965084728e-05,
+      "loss": 0.9476,
+      "step": 12182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981986669729705e-05,
+      "loss": 0.5703,
+      "step": 12183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998198368836553e-05,
+      "loss": 0.6009,
+      "step": 12184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998198070675473e-05,
+      "loss": 1.3527,
+      "step": 12185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998197772489733e-05,
+      "loss": 1.3717,
+      "step": 12186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998197474279331e-05,
+      "loss": 1.1882,
+      "step": 12187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998197176044268e-05,
+      "loss": 0.9134,
+      "step": 12188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998196877784544e-05,
+      "loss": 0.7254,
+      "step": 12189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998196579500159e-05,
+      "loss": 0.9603,
+      "step": 12190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981962811911125e-05,
+      "loss": 1.0766,
+      "step": 12191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998195982857405e-05,
+      "loss": 1.2593,
+      "step": 12192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998195684499036e-05,
+      "loss": 1.2716,
+      "step": 12193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998195386116007e-05,
+      "loss": 1.1075,
+      "step": 12194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998195087708316e-05,
+      "loss": 1.1545,
+      "step": 12195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998194789275964e-05,
+      "loss": 1.0028,
+      "step": 12196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99819449081895e-05,
+      "loss": 1.1678,
+      "step": 12197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998194192337276e-05,
+      "loss": 1.1907,
+      "step": 12198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998193893830941e-05,
+      "loss": 1.3121,
+      "step": 12199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981935952999445e-05,
+      "loss": 0.7019,
+      "step": 12200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981932967442866e-05,
+      "loss": 1.1035,
+      "step": 12201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998192998163968e-05,
+      "loss": 1.6263,
+      "step": 12202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981926995589876e-05,
+      "loss": 0.9957,
+      "step": 12203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998192400929347e-05,
+      "loss": 0.9056,
+      "step": 12204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998192102275045e-05,
+      "loss": 0.8507,
+      "step": 12205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998191803596082e-05,
+      "loss": 0.9602,
+      "step": 12206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998191504892458e-05,
+      "loss": 0.8681,
+      "step": 12207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998191206164172e-05,
+      "loss": 0.7919,
+      "step": 12208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998190907411226e-05,
+      "loss": 0.784,
+      "step": 12209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981906086336184e-05,
+      "loss": 0.739,
+      "step": 12210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99819030983135e-05,
+      "loss": 0.7427,
+      "step": 12211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99819001100442e-05,
+      "loss": 0.7395,
+      "step": 12212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99818971215283e-05,
+      "loss": 0.6996,
+      "step": 12213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998189413276578e-05,
+      "loss": 0.7456,
+      "step": 12214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998189114375665e-05,
+      "loss": 0.7073,
+      "step": 12215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981888154500914e-05,
+      "loss": 0.7032,
+      "step": 12216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998188516499855e-05,
+      "loss": 0.6384,
+      "step": 12217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998188217524959e-05,
+      "loss": 0.6803,
+      "step": 12218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998187918525402e-05,
+      "loss": 0.6206,
+      "step": 12219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981876195011844e-05,
+      "loss": 0.6704,
+      "step": 12220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998187320452305e-05,
+      "loss": 1.6482,
+      "step": 12221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998187021378765e-05,
+      "loss": 0.9323,
+      "step": 12222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998186722280563e-05,
+      "loss": 1.0517,
+      "step": 12223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998186423157701e-05,
+      "loss": 0.9098,
+      "step": 12224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981861240101765e-05,
+      "loss": 1.2377,
+      "step": 12225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981858248379925e-05,
+      "loss": 1.1893,
+      "step": 12226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998185525641147e-05,
+      "loss": 0.9471,
+      "step": 12227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981852264196405e-05,
+      "loss": 1.2663,
+      "step": 12228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981849271734726e-05,
+      "loss": 1.4243,
+      "step": 12229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981846279026445e-05,
+      "loss": 0.8721,
+      "step": 12230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998184328607154e-05,
+      "loss": 1.2458,
+      "step": 12231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998184029287004e-05,
+      "loss": 1.2852,
+      "step": 12232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981837299421916e-05,
+      "loss": 1.1038,
+      "step": 12233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981834305727194e-05,
+      "loss": 1.0695,
+      "step": 12234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981831311785856e-05,
+      "loss": 1.1814,
+      "step": 12235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998182831759791e-05,
+      "loss": 1.1959,
+      "step": 12236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998182532316335e-05,
+      "loss": 0.9837,
+      "step": 12237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981822328482186e-05,
+      "loss": 1.0547,
+      "step": 12238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99818193335544e-05,
+      "loss": 1.3387,
+      "step": 12239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998181633838001e-05,
+      "loss": 1.2434,
+      "step": 12240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998181334295902e-05,
+      "loss": 1.1832,
+      "step": 12241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998181034729141e-05,
+      "loss": 2.241,
+      "step": 12242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981807351377194e-05,
+      "loss": 1.0094,
+      "step": 12243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981804355216366e-05,
+      "loss": 0.8128,
+      "step": 12244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998180135880893e-05,
+      "loss": 1.1457,
+      "step": 12245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998179836215488e-05,
+      "loss": 1.1867,
+      "step": 12246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998179536525422e-05,
+      "loss": 1.2103,
+      "step": 12247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998179236810696e-05,
+      "loss": 1.129,
+      "step": 12248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998178937071308e-05,
+      "loss": 1.0006,
+      "step": 12249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998178637307259e-05,
+      "loss": 1.1473,
+      "step": 12250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981783375185495e-05,
+      "loss": 1.2892,
+      "step": 12251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998178037705179e-05,
+      "loss": 1.1412,
+      "step": 12252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998177737867148e-05,
+      "loss": 0.8711,
+      "step": 12253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998177438004455e-05,
+      "loss": 0.9427,
+      "step": 12254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998177138117102e-05,
+      "loss": 1.1894,
+      "step": 12255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998176838205087e-05,
+      "loss": 1.3187,
+      "step": 12256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998176538268412e-05,
+      "loss": 1.5896,
+      "step": 12257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981762383070765e-05,
+      "loss": 1.7242,
+      "step": 12258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998175938321079e-05,
+      "loss": 1.3832,
+      "step": 12259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99817563831042e-05,
+      "loss": 1.3111,
+      "step": 12260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981753382751015e-05,
+      "loss": 1.1332,
+      "step": 12261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998175038215121e-05,
+      "loss": 0.7401,
+      "step": 12262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998174738130481e-05,
+      "loss": 1.1572,
+      "step": 12263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998174438021179e-05,
+      "loss": 0.963,
+      "step": 12264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998174137887216e-05,
+      "loss": 0.9892,
+      "step": 12265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981738377285924e-05,
+      "loss": 2.0799,
+      "step": 12266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998173537545308e-05,
+      "loss": 1.0048,
+      "step": 12267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998173237337362e-05,
+      "loss": 1.4656,
+      "step": 12268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981729371047555e-05,
+      "loss": 0.895,
+      "step": 12269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998172636847488e-05,
+      "loss": 1.1404,
+      "step": 12270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99817233656556e-05,
+      "loss": 1.145,
+      "step": 12271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998172036258971e-05,
+      "loss": 1.2568,
+      "step": 12272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99817173592772e-05,
+      "loss": 1.4198,
+      "step": 12273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981714355718094e-05,
+      "loss": 1.2216,
+      "step": 12274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981711351912374e-05,
+      "loss": 1.1723,
+      "step": 12275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981708347860044e-05,
+      "loss": 0.9818,
+      "step": 12276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981705343561114e-05,
+      "loss": 1.1023,
+      "step": 12277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998170233901557e-05,
+      "loss": 0.602,
+      "step": 12278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981699334223406e-05,
+      "loss": 0.888,
+      "step": 12279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998169632918465e-05,
+      "loss": 0.8708,
+      "step": 12280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998169332389928e-05,
+      "loss": 1.2637,
+      "step": 12281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998169031836729e-05,
+      "loss": 1.0338,
+      "step": 12282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998168731258871e-05,
+      "loss": 0.7036,
+      "step": 12283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99816843065635e-05,
+      "loss": 0.7508,
+      "step": 12284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99816813002917e-05,
+      "loss": 0.584,
+      "step": 12285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998167829377328e-05,
+      "loss": 0.4087,
+      "step": 12286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998167528700826e-05,
+      "loss": 1.1411,
+      "step": 12287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998167227999663e-05,
+      "loss": 1.0434,
+      "step": 12288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981669272738386e-05,
+      "loss": 0.7202,
+      "step": 12289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998166626523354e-05,
+      "loss": 0.831,
+      "step": 12290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998166325748208e-05,
+      "loss": 0.7361,
+      "step": 12291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998166024948401e-05,
+      "loss": 0.954,
+      "step": 12292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998165724123933e-05,
+      "loss": 1.6453,
+      "step": 12293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981654232748056e-05,
+      "loss": 1.0373,
+      "step": 12294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981651224010164e-05,
+      "loss": 0.9059,
+      "step": 12295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998164821502566e-05,
+      "loss": 1.0726,
+      "step": 12296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981645205794555e-05,
+      "loss": 1.3252,
+      "step": 12297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998164219631684e-05,
+      "loss": 1.1061,
+      "step": 12298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998163918659251e-05,
+      "loss": 0.9898,
+      "step": 12299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981636176621577e-05,
+      "loss": 0.9904,
+      "step": 12300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981633166404033e-05,
+      "loss": 1.2752,
+      "step": 12301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998163015593988e-05,
+      "loss": 0.8749,
+      "step": 12302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998162714522913e-05,
+      "loss": 1.0561,
+      "step": 12303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981624134271766e-05,
+      "loss": 1.0469,
+      "step": 12304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998162112306779e-05,
+      "loss": 1.0939,
+      "step": 12305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998161811161721e-05,
+      "loss": 0.8517,
+      "step": 12306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981615099920015e-05,
+      "loss": 1.1539,
+      "step": 12307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998161208797621e-05,
+      "loss": 1.1281,
+      "step": 12308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981609075785814e-05,
+      "loss": 0.976,
+      "step": 12309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981606063348794e-05,
+      "loss": 0.6655,
+      "step": 12310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998160305066518e-05,
+      "loss": 1.7917,
+      "step": 12311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998160003773494e-05,
+      "loss": 1.8101,
+      "step": 12312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981597024558115e-05,
+      "loss": 1.9419,
+      "step": 12313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998159401113466e-05,
+      "loss": 1.9558,
+      "step": 12314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998159099746461e-05,
+      "loss": 1.7066,
+      "step": 12315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981587983547954e-05,
+      "loss": 1.8039,
+      "step": 12316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998158496938468e-05,
+      "loss": 1.7127,
+      "step": 12317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998158195497481e-05,
+      "loss": 1.2072,
+      "step": 12318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998157894031832e-05,
+      "loss": 0.3727,
+      "step": 12319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998157592541523e-05,
+      "loss": 0.2639,
+      "step": 12320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998157291026553e-05,
+      "loss": 0.5425,
+      "step": 12321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981569894869226e-05,
+      "loss": 0.5329,
+      "step": 12322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981566879226314e-05,
+      "loss": 0.7372,
+      "step": 12323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998156386333679e-05,
+      "loss": 1.1371,
+      "step": 12324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998156084720066e-05,
+      "loss": 1.0917,
+      "step": 12325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998155783081793e-05,
+      "loss": 1.0032,
+      "step": 12326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998155481418858e-05,
+      "loss": 1.1574,
+      "step": 12327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998155179731263e-05,
+      "loss": 1.24,
+      "step": 12328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998154878019007e-05,
+      "loss": 1.3952,
+      "step": 12329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981545762820903e-05,
+      "loss": 1.1931,
+      "step": 12330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998154274520514e-05,
+      "loss": 0.7442,
+      "step": 12331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998153972734275e-05,
+      "loss": 1.1324,
+      "step": 12332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981536709233766e-05,
+      "loss": 0.605,
+      "step": 12333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998153369087817e-05,
+      "loss": 0.5132,
+      "step": 12334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998153067227597e-05,
+      "loss": 0.4633,
+      "step": 12335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998152765342716e-05,
+      "loss": 0.4099,
+      "step": 12336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998152463433175e-05,
+      "loss": 0.445,
+      "step": 12337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998152161498972e-05,
+      "loss": 0.5687,
+      "step": 12338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998151859540109e-05,
+      "loss": 1.0694,
+      "step": 12339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998151557556585e-05,
+      "loss": 1.2817,
+      "step": 12340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981512555484005e-05,
+      "loss": 1.1015,
+      "step": 12341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998150953515556e-05,
+      "loss": 1.2148,
+      "step": 12342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99815065145805e-05,
+      "loss": 1.1055,
+      "step": 12343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998150349375883e-05,
+      "loss": 0.8014,
+      "step": 12344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981500472690554e-05,
+      "loss": 1.1865,
+      "step": 12345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998149745137568e-05,
+      "loss": 0.8474,
+      "step": 12346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998149442981419e-05,
+      "loss": 1.171,
+      "step": 12347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99814914080061e-05,
+      "loss": 1.1856,
+      "step": 12348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981488385951406e-05,
+      "loss": 1.1312,
+      "step": 12349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998148536365009e-05,
+      "loss": 1.0987,
+      "step": 12350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998148234110218e-05,
+      "loss": 1.1335,
+      "step": 12351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998147931830767e-05,
+      "loss": 1.0597,
+      "step": 12352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998147629526654e-05,
+      "loss": 1.0757,
+      "step": 12353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981473271978805e-05,
+      "loss": 0.9938,
+      "step": 12354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998147024844446e-05,
+      "loss": 1.2549,
+      "step": 12355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998146722466352e-05,
+      "loss": 1.1512,
+      "step": 12356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981464200635966e-05,
+      "loss": 1.1168,
+      "step": 12357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998146117636181e-05,
+      "loss": 0.9795,
+      "step": 12358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981458151841045e-05,
+      "loss": 1.1687,
+      "step": 12359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998145512707367e-05,
+      "loss": 1.1541,
+      "step": 12360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998145210205969e-05,
+      "loss": 1.0797,
+      "step": 12361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998144907679911e-05,
+      "loss": 1.126,
+      "step": 12362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998144605129191e-05,
+      "loss": 1.1401,
+      "step": 12363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981443025538125e-05,
+      "loss": 0.9566,
+      "step": 12364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981439999537715e-05,
+      "loss": 1.0134,
+      "step": 12365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998143697329071e-05,
+      "loss": 1.2251,
+      "step": 12366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998143394679709e-05,
+      "loss": 1.0263,
+      "step": 12367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998143092005687e-05,
+      "loss": 1.1975,
+      "step": 12368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981427893070045e-05,
+      "loss": 1.7828,
+      "step": 12369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981424865836606e-05,
+      "loss": 0.8206,
+      "step": 12370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981421838356565e-05,
+      "loss": 0.7618,
+      "step": 12371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998141881062992e-05,
+      "loss": 0.5565,
+      "step": 12372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981415782656664e-05,
+      "loss": 0.4305,
+      "step": 12373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981412754436805e-05,
+      "loss": 0.3556,
+      "step": 12374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998140972597034e-05,
+      "loss": 0.7321,
+      "step": 12375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998140669725727e-05,
+      "loss": 0.7104,
+      "step": 12376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99814036682976e-05,
+      "loss": 0.4816,
+      "step": 12377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981400639091314e-05,
+      "loss": 0.211,
+      "step": 12378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981397609638426e-05,
+      "loss": 0.319,
+      "step": 12379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981394579938935e-05,
+      "loss": 0.5376,
+      "step": 12380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998139154999283e-05,
+      "loss": 0.5669,
+      "step": 12381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998138851980013e-05,
+      "loss": 0.6685,
+      "step": 12382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998138548936082e-05,
+      "loss": 0.3601,
+      "step": 12383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99813824586749e-05,
+      "loss": 0.6418,
+      "step": 12384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998137942774237e-05,
+      "loss": 0.9678,
+      "step": 12385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998137639656325e-05,
+      "loss": 0.7382,
+      "step": 12386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981373365137515e-05,
+      "loss": 0.8277,
+      "step": 12387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998137033346517e-05,
+      "loss": 0.4406,
+      "step": 12388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998136730154623e-05,
+      "loss": 0.6967,
+      "step": 12389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998136426938068e-05,
+      "loss": 0.4694,
+      "step": 12390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998136123696852e-05,
+      "loss": 0.5338,
+      "step": 12391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981358204309755e-05,
+      "loss": 0.6247,
+      "step": 12392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981355171404394e-05,
+      "loss": 0.3445,
+      "step": 12393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981352138252424e-05,
+      "loss": 0.297,
+      "step": 12394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998134910485384e-05,
+      "loss": 0.9048,
+      "step": 12395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998134607120866e-05,
+      "loss": 1.2166,
+      "step": 12396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981343037316876e-05,
+      "loss": 1.181,
+      "step": 12397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998134000317848e-05,
+      "loss": 1.2389,
+      "step": 12398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998133696879348e-05,
+      "loss": 1.0067,
+      "step": 12399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998133393416187e-05,
+      "loss": 1.1272,
+      "step": 12400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998133089928366e-05,
+      "loss": 1.2163,
+      "step": 12401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998132786415885e-05,
+      "loss": 0.8942,
+      "step": 12402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998132482878743e-05,
+      "loss": 1.1119,
+      "step": 12403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99813217931694e-05,
+      "loss": 1.3623,
+      "step": 12404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998131875730477e-05,
+      "loss": 0.9595,
+      "step": 12405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998131572119354e-05,
+      "loss": 0.7142,
+      "step": 12406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99813126848357e-05,
+      "loss": 1.2325,
+      "step": 12407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998130964823126e-05,
+      "loss": 1.4484,
+      "step": 12408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998130661138021e-05,
+      "loss": 1.345,
+      "step": 12409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998130357428255e-05,
+      "loss": 1.3685,
+      "step": 12410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981300536938294e-05,
+      "loss": 1.1315,
+      "step": 12411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981297499347424e-05,
+      "loss": 0.7519,
+      "step": 12412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998129446150995e-05,
+      "loss": 1.1386,
+      "step": 12413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998129142342588e-05,
+      "loss": 1.145,
+      "step": 12414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981288385095204e-05,
+      "loss": 1.0816,
+      "step": 12415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981285346517914e-05,
+      "loss": 1.1919,
+      "step": 12416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998128230769403e-05,
+      "loss": 1.1038,
+      "step": 12417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998127926862354e-05,
+      "loss": 1.1895,
+      "step": 12418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998127622930644e-05,
+      "loss": 1.1287,
+      "step": 12419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981273189742734e-05,
+      "loss": 1.1634,
+      "step": 12420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998127014993242e-05,
+      "loss": 0.6406,
+      "step": 12421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998126710987552e-05,
+      "loss": 0.8229,
+      "step": 12422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981264069572e-05,
+      "loss": 1.1455,
+      "step": 12423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998126102902188e-05,
+      "loss": 1.0545,
+      "step": 12424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998125798822515e-05,
+      "loss": 1.0934,
+      "step": 12425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998125494718182e-05,
+      "loss": 1.0143,
+      "step": 12426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998125190589189e-05,
+      "loss": 1.1201,
+      "step": 12427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998124886435535e-05,
+      "loss": 0.9573,
+      "step": 12428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981245822572206e-05,
+      "loss": 0.8915,
+      "step": 12429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998124278054246e-05,
+      "loss": 1.3156,
+      "step": 12430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981239738266106e-05,
+      "loss": 1.2664,
+      "step": 12431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998123669574315e-05,
+      "loss": 1.0817,
+      "step": 12432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998123365297359e-05,
+      "loss": 0.702,
+      "step": 12433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998123060995743e-05,
+      "loss": 1.0154,
+      "step": 12434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981227566694656e-05,
+      "loss": 1.2546,
+      "step": 12435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998122452318529e-05,
+      "loss": 1.1136,
+      "step": 12436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998122147942931e-05,
+      "loss": 0.9993,
+      "step": 12437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981218435426724e-05,
+      "loss": 1.0578,
+      "step": 12438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998121539117754e-05,
+      "loss": 0.7509,
+      "step": 12439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998121234668175e-05,
+      "loss": 1.0293,
+      "step": 12440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998120930193936e-05,
+      "loss": 1.2657,
+      "step": 12441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998120625695036e-05,
+      "loss": 1.0948,
+      "step": 12442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981203211714764e-05,
+      "loss": 1.2965,
+      "step": 12443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981200166232556e-05,
+      "loss": 1.1268,
+      "step": 12444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981197120503746e-05,
+      "loss": 1.1302,
+      "step": 12445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981194074528334e-05,
+      "loss": 1.464,
+      "step": 12446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998119102830632e-05,
+      "loss": 1.2115,
+      "step": 12447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99811879818377e-05,
+      "loss": 1.1439,
+      "step": 12448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981184935122474e-05,
+      "loss": 1.3663,
+      "step": 12449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998118188816064e-05,
+      "loss": 1.1248,
+      "step": 12450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981178840952215e-05,
+      "loss": 1.0585,
+      "step": 12451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998117579349718e-05,
+      "loss": 1.1164,
+      "step": 12452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998117274579555e-05,
+      "loss": 1.0238,
+      "step": 12453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981169697847294e-05,
+      "loss": 0.7298,
+      "step": 12454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998116664965245e-05,
+      "loss": 1.1957,
+      "step": 12455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981163601211e-05,
+      "loss": 1.012,
+      "step": 12456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998116055252295e-05,
+      "loss": 1.1416,
+      "step": 12457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998115750358829e-05,
+      "loss": 1.1676,
+      "step": 12458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981154454407036e-05,
+      "loss": 1.1385,
+      "step": 12459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998115140497918e-05,
+      "loss": 1.0225,
+      "step": 12460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998114835530471e-05,
+      "loss": 0.7041,
+      "step": 12461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998114530538364e-05,
+      "loss": 0.8163,
+      "step": 12462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998114225521596e-05,
+      "loss": 0.5854,
+      "step": 12463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981139204801686e-05,
+      "loss": 0.4704,
+      "step": 12464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981136154140806e-05,
+      "loss": 0.5542,
+      "step": 12465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981133103233325e-05,
+      "loss": 0.5285,
+      "step": 12466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998113005207924e-05,
+      "loss": 0.6402,
+      "step": 12467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998112700067854e-05,
+      "loss": 0.7608,
+      "step": 12468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998112394903125e-05,
+      "loss": 0.8731,
+      "step": 12469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981120897137355e-05,
+      "loss": 0.7127,
+      "step": 12470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998111784499686e-05,
+      "loss": 1.0558,
+      "step": 12471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998111479260976e-05,
+      "loss": 1.4552,
+      "step": 12472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998111173997605e-05,
+      "loss": 0.5671,
+      "step": 12473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998110868709574e-05,
+      "loss": 1.531,
+      "step": 12474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981105633968825e-05,
+      "loss": 0.9289,
+      "step": 12475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998110258059532e-05,
+      "loss": 0.9232,
+      "step": 12476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99810995269752e-05,
+      "loss": 1.111,
+      "step": 12477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998109647310848e-05,
+      "loss": 1.0808,
+      "step": 12478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981093418995157e-05,
+      "loss": 1.274,
+      "step": 12479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998109036463523e-05,
+      "loss": 1.0235,
+      "step": 12480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998108731002871e-05,
+      "loss": 1.1525,
+      "step": 12481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998108425517557e-05,
+      "loss": 1.1978,
+      "step": 12482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998108120007584e-05,
+      "loss": 1.1724,
+      "step": 12483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99810781447295e-05,
+      "loss": 1.2258,
+      "step": 12484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981075089136563e-05,
+      "loss": 1.1161,
+      "step": 12485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998107203329702e-05,
+      "loss": 0.6646,
+      "step": 12486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981068977210873e-05,
+      "loss": 0.4699,
+      "step": 12487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981065920878126e-05,
+      "loss": 0.4123,
+      "step": 12488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981062864298776e-05,
+      "loss": 0.3083,
+      "step": 12489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998105980747283e-05,
+      "loss": 0.8927,
+      "step": 12490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981056750400265e-05,
+      "loss": 1.4359,
+      "step": 12491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998105369308112e-05,
+      "loss": 0.9881,
+      "step": 12492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998105063551536e-05,
+      "loss": 1.3489,
+      "step": 12493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998104757770299e-05,
+      "loss": 1.2331,
+      "step": 12494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998104451964403e-05,
+      "loss": 0.9847,
+      "step": 12495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998104146133846e-05,
+      "loss": 1.0687,
+      "step": 12496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981038402786296e-05,
+      "loss": 1.0977,
+      "step": 12497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998103534398752e-05,
+      "loss": 1.1294,
+      "step": 12498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998103228494215e-05,
+      "loss": 0.8384,
+      "step": 12499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981029225650174e-05,
+      "loss": 0.8916,
+      "step": 12500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0588072538375854,
+      "eval_runtime": 85.3376,
+      "eval_samples_per_second": 16.23,
+      "eval_steps_per_second": 4.066,
+      "step": 12500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981026166111597e-05,
+      "loss": 0.9107,
+      "step": 12501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998102310632641e-05,
+      "loss": 1.0434,
+      "step": 12502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998102004629463e-05,
+      "loss": 0.9114,
+      "step": 12503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981016986016246e-05,
+      "loss": 1.0933,
+      "step": 12504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998101392549126e-05,
+      "loss": 1.0656,
+      "step": 12505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998101086471967e-05,
+      "loss": 0.8722,
+      "step": 12506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998100780370148e-05,
+      "loss": 0.9555,
+      "step": 12507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998100474243669e-05,
+      "loss": 1.1079,
+      "step": 12508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998100168092529e-05,
+      "loss": 1.1997,
+      "step": 12509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99809986191673e-05,
+      "loss": 0.8179,
+      "step": 12510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99809955571627e-05,
+      "loss": 0.8281,
+      "step": 12511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99809924949115e-05,
+      "loss": 1.0142,
+      "step": 12512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998098943241369e-05,
+      "loss": 1.0984,
+      "step": 12513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998098636966929e-05,
+      "loss": 1.2993,
+      "step": 12514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998098330667829e-05,
+      "loss": 0.802,
+      "step": 12515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998098024344068e-05,
+      "loss": 1.1062,
+      "step": 12516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980977179956465e-05,
+      "loss": 1.3741,
+      "step": 12517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980974116225657e-05,
+      "loss": 1.2558,
+      "step": 12518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998097105224825e-05,
+      "loss": 0.8686,
+      "step": 12519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998096798802423e-05,
+      "loss": 0.5407,
+      "step": 12520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998096492355362e-05,
+      "loss": 0.9241,
+      "step": 12521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99809618588364e-05,
+      "loss": 1.1364,
+      "step": 12522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980958793872577e-05,
+      "loss": 1.2324,
+      "step": 12523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998095572866216e-05,
+      "loss": 0.91,
+      "step": 12524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998095266320514e-05,
+      "loss": 1.3726,
+      "step": 12525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980949597501515e-05,
+      "loss": 1.0983,
+      "step": 12526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998094653155129e-05,
+      "loss": 1.1695,
+      "step": 12527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980943465354465e-05,
+      "loss": 1.1617,
+      "step": 12528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980940398911045e-05,
+      "loss": 1.1853,
+      "step": 12529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998093733222101e-05,
+      "loss": 1.0265,
+      "step": 12530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980934265284384e-05,
+      "loss": 0.9487,
+      "step": 12531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998093119810115e-05,
+      "loss": 0.7968,
+      "step": 12532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980928130671316e-05,
+      "loss": 1.2177,
+      "step": 12533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998092506299489e-05,
+      "loss": 0.8995,
+      "step": 12534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998092199507185e-05,
+      "loss": 1.1374,
+      "step": 12535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980918926902215e-05,
+      "loss": 1.1554,
+      "step": 12536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998091585848598e-05,
+      "loss": 1.1725,
+      "step": 12537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998091278982314e-05,
+      "loss": 1.2441,
+      "step": 12538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980909720913705e-05,
+      "loss": 0.9656,
+      "step": 12539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980906651757664e-05,
+      "loss": 1.0499,
+      "step": 12540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998090358235502e-05,
+      "loss": 1.3264,
+      "step": 12541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998090051270578e-05,
+      "loss": 0.9362,
+      "step": 12542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998089744280994e-05,
+      "loss": 1.0823,
+      "step": 12543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998089437266749e-05,
+      "loss": 1.3716,
+      "step": 12544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998089130227844e-05,
+      "loss": 0.5029,
+      "step": 12545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99808882316428e-05,
+      "loss": 1.1508,
+      "step": 12546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998088516076055e-05,
+      "loss": 1.5744,
+      "step": 12547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998088208963171e-05,
+      "loss": 1.0503,
+      "step": 12548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998087901825626e-05,
+      "loss": 1.1751,
+      "step": 12549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998087594663421e-05,
+      "loss": 1.1362,
+      "step": 12550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998087287476556e-05,
+      "loss": 1.1305,
+      "step": 12551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998086980265031e-05,
+      "loss": 1.2603,
+      "step": 12552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980866730288464e-05,
+      "loss": 1.2407,
+      "step": 12553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998086365768001e-05,
+      "loss": 1.2634,
+      "step": 12554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998086058482496e-05,
+      "loss": 0.9507,
+      "step": 12555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980857511723303e-05,
+      "loss": 0.9109,
+      "step": 12556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998085443837506e-05,
+      "loss": 0.8525,
+      "step": 12557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99808513647802e-05,
+      "loss": 1.0178,
+      "step": 12558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998084829093875e-05,
+      "loss": 1.0699,
+      "step": 12559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998084521685069e-05,
+      "loss": 1.0353,
+      "step": 12560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998084214251604e-05,
+      "loss": 1.2007,
+      "step": 12561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998083906793478e-05,
+      "loss": 1.2147,
+      "step": 12562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998083599310693e-05,
+      "loss": 1.0938,
+      "step": 12563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998083291803247e-05,
+      "loss": 0.8749,
+      "step": 12564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998082984271142e-05,
+      "loss": 1.1915,
+      "step": 12565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998082676714376e-05,
+      "loss": 1.2303,
+      "step": 12566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980823691329505e-05,
+      "loss": 1.3084,
+      "step": 12567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980820615268655e-05,
+      "loss": 1.021,
+      "step": 12568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99808175389612e-05,
+      "loss": 0.8386,
+      "step": 12569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998081446240714e-05,
+      "loss": 0.5628,
+      "step": 12570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998081138560649e-05,
+      "loss": 0.394,
+      "step": 12571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980808308559225e-05,
+      "loss": 0.3166,
+      "step": 12572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998080523126537e-05,
+      "loss": 0.3661,
+      "step": 12573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998080215372491e-05,
+      "loss": 0.2504,
+      "step": 12574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998079907593786e-05,
+      "loss": 0.1394,
+      "step": 12575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99807959979042e-05,
+      "loss": 0.1276,
+      "step": 12576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998079291962394e-05,
+      "loss": 0.0873,
+      "step": 12577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998078984109709e-05,
+      "loss": 0.1252,
+      "step": 12578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998078676232363e-05,
+      "loss": 0.1228,
+      "step": 12579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998078368330358e-05,
+      "loss": 0.1049,
+      "step": 12580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998078060403692e-05,
+      "loss": 0.2173,
+      "step": 12581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980777524523667e-05,
+      "loss": 0.1332,
+      "step": 12582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998077444476381e-05,
+      "loss": 0.0962,
+      "step": 12583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998077136475736e-05,
+      "loss": 0.0435,
+      "step": 12584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99807682845043e-05,
+      "loss": 0.024,
+      "step": 12585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980765204004645e-05,
+      "loss": 0.0397,
+      "step": 12586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980762123258395e-05,
+      "loss": 0.0451,
+      "step": 12587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980759042265544e-05,
+      "loss": 0.0358,
+      "step": 12588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998075596102609e-05,
+      "loss": 0.0299,
+      "step": 12589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980752879540035e-05,
+      "loss": 0.0337,
+      "step": 12590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980749797807385e-05,
+      "loss": 0.0337,
+      "step": 12591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998074671582813e-05,
+      "loss": 0.0328,
+      "step": 12592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980743633602286e-05,
+      "loss": 0.0385,
+      "step": 12593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998074055112984e-05,
+      "loss": 0.0282,
+      "step": 12594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998073746841079e-05,
+      "loss": 0.123,
+      "step": 12595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980734385445135e-05,
+      "loss": 1.1476,
+      "step": 12596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998073130223289e-05,
+      "loss": 1.3075,
+      "step": 12597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980728218774045e-05,
+      "loss": 1.1733,
+      "step": 12598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980725135068595e-05,
+      "loss": 1.0228,
+      "step": 12599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998072205111655e-05,
+      "loss": 1.1655,
+      "step": 12600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998071896691791e-05,
+      "loss": 1.2506,
+      "step": 12601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998071588247266e-05,
+      "loss": 1.0348,
+      "step": 12602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980712797780814e-05,
+      "loss": 0.8461,
+      "step": 12603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980709712842375e-05,
+      "loss": 0.7891,
+      "step": 12604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980706627657333e-05,
+      "loss": 0.9184,
+      "step": 12605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998070354222569e-05,
+      "loss": 1.1337,
+      "step": 12606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998070045654746e-05,
+      "loss": 1.1779,
+      "step": 12607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998069737062262e-05,
+      "loss": 1.1481,
+      "step": 12608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998069428445118e-05,
+      "loss": 1.2349,
+      "step": 12609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998069119803315e-05,
+      "loss": 1.1853,
+      "step": 12610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998068811136851e-05,
+      "loss": 1.2069,
+      "step": 12611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998068502445728e-05,
+      "loss": 0.9469,
+      "step": 12612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998068193729945e-05,
+      "loss": 1.2307,
+      "step": 12613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998067884989502e-05,
+      "loss": 0.8482,
+      "step": 12614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980675762243986e-05,
+      "loss": 1.2056,
+      "step": 12615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980672674346366e-05,
+      "loss": 0.9474,
+      "step": 12616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998066958620213e-05,
+      "loss": 1.1598,
+      "step": 12617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980666497811307e-05,
+      "loss": 1.2106,
+      "step": 12618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998066340917388e-05,
+      "loss": 1.0984,
+      "step": 12619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998066032028986e-05,
+      "loss": 1.2588,
+      "step": 12620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998065723115924e-05,
+      "loss": 1.129,
+      "step": 12621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998065414178202e-05,
+      "loss": 0.9005,
+      "step": 12622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99806510521582e-05,
+      "loss": 1.1697,
+      "step": 12623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998064796228779e-05,
+      "loss": 1.2553,
+      "step": 12624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980644872170765e-05,
+      "loss": 1.0182,
+      "step": 12625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998064178180716e-05,
+      "loss": 1.2193,
+      "step": 12626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998063869119695e-05,
+      "loss": 1.0012,
+      "step": 12627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980635600340134e-05,
+      "loss": 1.1328,
+      "step": 12628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980632509236724e-05,
+      "loss": 1.2161,
+      "step": 12629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998062941788671e-05,
+      "loss": 0.9976,
+      "step": 12630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980626326290114e-05,
+      "loss": 1.162,
+      "step": 12631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998062323444691e-05,
+      "loss": 1.1436,
+      "step": 12632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99806201423571e-05,
+      "loss": 1.2007,
+      "step": 12633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980617050020704e-05,
+      "loss": 1.162,
+      "step": 12634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980613957437704e-05,
+      "loss": 1.2691,
+      "step": 12635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998061086460811e-05,
+      "loss": 1.2555,
+      "step": 12636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998060777153192e-05,
+      "loss": 0.7833,
+      "step": 12637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998060467820913e-05,
+      "loss": 0.6888,
+      "step": 12638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980601584639735e-05,
+      "loss": 1.0737,
+      "step": 12639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980598490823746e-05,
+      "loss": 1.1638,
+      "step": 12640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998059539676116e-05,
+      "loss": 1.257,
+      "step": 12641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998059230245198e-05,
+      "loss": 0.9882,
+      "step": 12642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99805892078962e-05,
+      "loss": 1.2478,
+      "step": 12643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998058611309382e-05,
+      "loss": 1.2171,
+      "step": 12644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998058301804484e-05,
+      "loss": 0.894,
+      "step": 12645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980579922749266e-05,
+      "loss": 1.1506,
+      "step": 12646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998057682720709e-05,
+      "loss": 1.0116,
+      "step": 12647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980573731418324e-05,
+      "loss": 1.2914,
+      "step": 12648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998057063538295e-05,
+      "loss": 1.2192,
+      "step": 12649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998056753910099e-05,
+      "loss": 1.235,
+      "step": 12650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998056444257242e-05,
+      "loss": 1.0878,
+      "step": 12651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980561345797264e-05,
+      "loss": 1.2564,
+      "step": 12652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99805582487755e-05,
+      "loss": 1.0418,
+      "step": 12653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998055515150715e-05,
+      "loss": 1.008,
+      "step": 12654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998055205399219e-05,
+      "loss": 1.3084,
+      "step": 12655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980548956230644e-05,
+      "loss": 1.2171,
+      "step": 12656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998054585822249e-05,
+      "loss": 0.9538,
+      "step": 12657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998054275996774e-05,
+      "loss": 1.1178,
+      "step": 12658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99805396614664e-05,
+      "loss": 1.0045,
+      "step": 12659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998053656271846e-05,
+      "loss": 1.2685,
+      "step": 12660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998053346372392e-05,
+      "loss": 1.0377,
+      "step": 12661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980530364482786e-05,
+      "loss": 0.9865,
+      "step": 12662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998052726499505e-05,
+      "loss": 1.273,
+      "step": 12663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998052416526072e-05,
+      "loss": 0.9645,
+      "step": 12664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99805210652798e-05,
+      "loss": 1.2464,
+      "step": 12665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998051796505227e-05,
+      "loss": 1.0677,
+      "step": 12666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998051486457815e-05,
+      "loss": 0.9309,
+      "step": 12667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998051176385743e-05,
+      "loss": 1.0929,
+      "step": 12668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980508662890114e-05,
+      "loss": 0.8478,
+      "step": 12669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998050556167621e-05,
+      "loss": 0.6965,
+      "step": 12670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998050246021569e-05,
+      "loss": 0.7587,
+      "step": 12671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998049935850858e-05,
+      "loss": 0.8848,
+      "step": 12672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980496256554885e-05,
+      "loss": 0.9169,
+      "step": 12673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998049315435458e-05,
+      "loss": 1.0632,
+      "step": 12674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998049005190768e-05,
+      "loss": 1.4337,
+      "step": 12675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980486949214186e-05,
+      "loss": 0.9088,
+      "step": 12676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980483846274094e-05,
+      "loss": 1.1794,
+      "step": 12677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998048074308741e-05,
+      "loss": 1.1089,
+      "step": 12678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998047763965412e-05,
+      "loss": 1.0994,
+      "step": 12679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998047453597424e-05,
+      "loss": 1.035,
+      "step": 12680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980471432047757e-05,
+      "loss": 0.5826,
+      "step": 12681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998046832787468e-05,
+      "loss": 0.378,
+      "step": 12682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980465223455014e-05,
+      "loss": 0.582,
+      "step": 12683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980462118788744e-05,
+      "loss": 0.6371,
+      "step": 12684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998045901387588e-05,
+      "loss": 1.3622,
+      "step": 12685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998045590871642e-05,
+      "loss": 0.9852,
+      "step": 12686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998045280331036e-05,
+      "loss": 1.2683,
+      "step": 12687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980449697657705e-05,
+      "loss": 1.1576,
+      "step": 12688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998044659175845e-05,
+      "loss": 1.333,
+      "step": 12689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980443485612606e-05,
+      "loss": 1.3333,
+      "step": 12690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980440379220163e-05,
+      "loss": 1.0612,
+      "step": 12691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980437272581126e-05,
+      "loss": 1.1116,
+      "step": 12692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998043416569549e-05,
+      "loss": 1.3357,
+      "step": 12693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980431058563246e-05,
+      "loss": 0.7662,
+      "step": 12694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980427951184424e-05,
+      "loss": 0.2322,
+      "step": 12695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980424843559e-05,
+      "loss": 0.4027,
+      "step": 12696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998042173568698e-05,
+      "loss": 0.4733,
+      "step": 12697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998041862756836e-05,
+      "loss": 0.6448,
+      "step": 12698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980415519203144e-05,
+      "loss": 0.1055,
+      "step": 12699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998041241059133e-05,
+      "loss": 0.7113,
+      "step": 12700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998040930173292e-05,
+      "loss": 0.0967,
+      "step": 12701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980406192627926e-05,
+      "loss": 0.5632,
+      "step": 12702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998040308327633e-05,
+      "loss": 0.9085,
+      "step": 12703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980399973678126e-05,
+      "loss": 1.2519,
+      "step": 12704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998039686383334e-05,
+      "loss": 1.1727,
+      "step": 12705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998039375374195e-05,
+      "loss": 1.2409,
+      "step": 12706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980390643403976e-05,
+      "loss": 1.1744,
+      "step": 12707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998038753281939e-05,
+      "loss": 1.0515,
+      "step": 12708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998038442198822e-05,
+      "loss": 0.8253,
+      "step": 12709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998038131091045e-05,
+      "loss": 0.9936,
+      "step": 12710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998037819958608e-05,
+      "loss": 1.1891,
+      "step": 12711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998037508801512e-05,
+      "loss": 1.3771,
+      "step": 12712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998037197619756e-05,
+      "loss": 1.0651,
+      "step": 12713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998036886413341e-05,
+      "loss": 1.07,
+      "step": 12714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998036575182266e-05,
+      "loss": 1.0145,
+      "step": 12715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980362639265316e-05,
+      "loss": 1.0349,
+      "step": 12716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980359526461376e-05,
+      "loss": 1.2096,
+      "step": 12717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998035641341084e-05,
+      "loss": 0.9221,
+      "step": 12718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998035330011371e-05,
+      "loss": 1.2303,
+      "step": 12719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998035018656998e-05,
+      "loss": 0.8573,
+      "step": 12720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998034707277966e-05,
+      "loss": 1.3344,
+      "step": 12721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998034395874274e-05,
+      "loss": 0.8321,
+      "step": 12722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998034084445923e-05,
+      "loss": 0.8328,
+      "step": 12723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998033772992912e-05,
+      "loss": 0.8378,
+      "step": 12724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998033461515242e-05,
+      "loss": 0.7233,
+      "step": 12725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998033150012912e-05,
+      "loss": 0.5925,
+      "step": 12726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998032838485922e-05,
+      "loss": 0.6885,
+      "step": 12727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998032526934273e-05,
+      "loss": 0.9141,
+      "step": 12728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998032215357965e-05,
+      "loss": 2.9479,
+      "step": 12729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998031903756997e-05,
+      "loss": 1.1008,
+      "step": 12730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998031592131369e-05,
+      "loss": 0.9494,
+      "step": 12731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998031280481083e-05,
+      "loss": 1.0106,
+      "step": 12732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980309688061354e-05,
+      "loss": 1.2855,
+      "step": 12733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998030657106529e-05,
+      "loss": 1.1777,
+      "step": 12734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998030345382264e-05,
+      "loss": 1.3056,
+      "step": 12735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998030033633339e-05,
+      "loss": 1.0951,
+      "step": 12736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980297218597545e-05,
+      "loss": 1.4119,
+      "step": 12737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980294100615104e-05,
+      "loss": 1.2088,
+      "step": 12738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998029098238607e-05,
+      "loss": 1.381,
+      "step": 12739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980287863910435e-05,
+      "loss": 1.1912,
+      "step": 12740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998028474518821e-05,
+      "loss": 0.9648,
+      "step": 12741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980281626219394e-05,
+      "loss": 1.0769,
+      "step": 12742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998027850700398e-05,
+      "loss": 1.1291,
+      "step": 12743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980275387541966e-05,
+      "loss": 1.2115,
+      "step": 12744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998027226783336e-05,
+      "loss": 1.1611,
+      "step": 12745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980269147878165e-05,
+      "loss": 1.1745,
+      "step": 12746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998026602767637e-05,
+      "loss": 0.7889,
+      "step": 12747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998026290722798e-05,
+      "loss": 0.973,
+      "step": 12748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980259786533e-05,
+      "loss": 0.8657,
+      "step": 12749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980256665591415e-05,
+      "loss": 1.1629,
+      "step": 12750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980253544403245e-05,
+      "loss": 0.835,
+      "step": 12751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998025042296848e-05,
+      "loss": 1.2045,
+      "step": 12752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998024730128712e-05,
+      "loss": 0.9768,
+      "step": 12753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998024417935916e-05,
+      "loss": 0.8842,
+      "step": 12754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998024105718461e-05,
+      "loss": 1.3598,
+      "step": 12755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980237934763464e-05,
+      "loss": 1.1881,
+      "step": 12756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980234812095724e-05,
+      "loss": 1.1587,
+      "step": 12757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998023168918139e-05,
+      "loss": 1.0493,
+      "step": 12758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980228566020466e-05,
+      "loss": 0.9941,
+      "step": 12759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998022544261294e-05,
+      "loss": 1.0987,
+      "step": 12760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998022231895883e-05,
+      "loss": 0.8508,
+      "step": 12761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998021919505811e-05,
+      "loss": 1.0839,
+      "step": 12762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998021607091081e-05,
+      "loss": 1.209,
+      "step": 12763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998021294651691e-05,
+      "loss": 1.2368,
+      "step": 12764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998020982187642e-05,
+      "loss": 0.9204,
+      "step": 12765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998020669698933e-05,
+      "loss": 1.0901,
+      "step": 12766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998020357185564e-05,
+      "loss": 1.3258,
+      "step": 12767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980200446475376e-05,
+      "loss": 1.1116,
+      "step": 12768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980197320848496e-05,
+      "loss": 1.1504,
+      "step": 12769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998019419497504e-05,
+      "loss": 0.928,
+      "step": 12770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998019106885498e-05,
+      "loss": 1.0945,
+      "step": 12771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998018794248833e-05,
+      "loss": 1.0755,
+      "step": 12772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998018481587509e-05,
+      "loss": 0.9464,
+      "step": 12773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980181689015245e-05,
+      "loss": 0.9836,
+      "step": 12774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980178561908816e-05,
+      "loss": 0.7677,
+      "step": 12775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980175434555784e-05,
+      "loss": 0.9265,
+      "step": 12776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980172306956164e-05,
+      "loss": 0.9324,
+      "step": 12777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998016917910995e-05,
+      "loss": 1.0801,
+      "step": 12778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998016605101714e-05,
+      "loss": 1.2038,
+      "step": 12779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998016292267774e-05,
+      "loss": 0.5574,
+      "step": 12780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998015979409175e-05,
+      "loss": 1.1233,
+      "step": 12781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998015666525916e-05,
+      "loss": 0.9915,
+      "step": 12782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998015353617997e-05,
+      "loss": 1.1283,
+      "step": 12783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99801504068542e-05,
+      "loss": 1.2538,
+      "step": 12784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998014727728183e-05,
+      "loss": 1.0479,
+      "step": 12785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980144147462866e-05,
+      "loss": 1.1323,
+      "step": 12786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998014101739731e-05,
+      "loss": 1.2417,
+      "step": 12787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998013788708516e-05,
+      "loss": 1.0424,
+      "step": 12788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998013475652642e-05,
+      "loss": 1.068,
+      "step": 12789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980131625721084e-05,
+      "loss": 1.3921,
+      "step": 12790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980128494669155e-05,
+      "loss": 1.0747,
+      "step": 12791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998012536337063e-05,
+      "loss": 1.2593,
+      "step": 12792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998012223182551e-05,
+      "loss": 1.1864,
+      "step": 12793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99801191000338e-05,
+      "loss": 1.0663,
+      "step": 12794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99801159679955e-05,
+      "loss": 1.13,
+      "step": 12795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99801128357106e-05,
+      "loss": 1.0854,
+      "step": 12796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998010970317911e-05,
+      "loss": 1.0491,
+      "step": 12797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998010657040103e-05,
+      "loss": 1.0693,
+      "step": 12798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998010343737636e-05,
+      "loss": 1.0836,
+      "step": 12799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998010030410508e-05,
+      "loss": 1.1157,
+      "step": 12800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998009717058722e-05,
+      "loss": 1.3284,
+      "step": 12801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998009403682277e-05,
+      "loss": 0.8892,
+      "step": 12802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998009090281172e-05,
+      "loss": 1.1005,
+      "step": 12803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998008776855408e-05,
+      "loss": 0.7738,
+      "step": 12804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998008463404985e-05,
+      "loss": 0.8511,
+      "step": 12805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998008149929903e-05,
+      "loss": 0.9628,
+      "step": 12806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980078364301606e-05,
+      "loss": 1.1077,
+      "step": 12807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998007522905759e-05,
+      "loss": 0.8475,
+      "step": 12808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998007209356699e-05,
+      "loss": 0.328,
+      "step": 12809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998006895782978e-05,
+      "loss": 0.5646,
+      "step": 12810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980065821846e-05,
+      "loss": 1.0582,
+      "step": 12811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980062685615615e-05,
+      "loss": 1.2618,
+      "step": 12812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980059549138636e-05,
+      "loss": 1.1225,
+      "step": 12813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998005641241507e-05,
+      "loss": 0.8237,
+      "step": 12814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998005327544491e-05,
+      "loss": 0.933,
+      "step": 12815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980050138228156e-05,
+      "loss": 0.808,
+      "step": 12816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998004700076481e-05,
+      "loss": 0.7977,
+      "step": 12817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998004386305487e-05,
+      "loss": 1.1029,
+      "step": 12818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998004072509834e-05,
+      "loss": 1.0664,
+      "step": 12819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980037586895225e-05,
+      "loss": 0.919,
+      "step": 12820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99800344484455e-05,
+      "loss": 0.8339,
+      "step": 12821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99800313097492e-05,
+      "loss": 0.8931,
+      "step": 12822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998002817080629e-05,
+      "loss": 1.1265,
+      "step": 12823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998002503161681e-05,
+      "loss": 1.0405,
+      "step": 12824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980021892180715e-05,
+      "loss": 1.2704,
+      "step": 12825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980018752498035e-05,
+      "loss": 0.983,
+      "step": 12826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998001561256877e-05,
+      "loss": 1.1701,
+      "step": 12827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998001247239291e-05,
+      "loss": 0.9842,
+      "step": 12828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998000933197046e-05,
+      "loss": 0.9698,
+      "step": 12829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998000619130141e-05,
+      "loss": 0.813,
+      "step": 12830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998000305038577e-05,
+      "loss": 0.8723,
+      "step": 12831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997999990922354e-05,
+      "loss": 1.0961,
+      "step": 12832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997999676781472e-05,
+      "loss": 1.2978,
+      "step": 12833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997999362615931e-05,
+      "loss": 1.3828,
+      "step": 12834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99799904842573e-05,
+      "loss": 1.0502,
+      "step": 12835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99799873421087e-05,
+      "loss": 1.2325,
+      "step": 12836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997998419971351e-05,
+      "loss": 1.2105,
+      "step": 12837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979981057071725e-05,
+      "loss": 1.4,
+      "step": 12838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997997791418336e-05,
+      "loss": 0.9596,
+      "step": 12839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997997477104839e-05,
+      "loss": 0.9279,
+      "step": 12840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997997162766683e-05,
+      "loss": 1.3523,
+      "step": 12841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979968484038684e-05,
+      "loss": 1.2956,
+      "step": 12842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997996534016395e-05,
+      "loss": 1.2156,
+      "step": 12843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997996219604261e-05,
+      "loss": 1.2605,
+      "step": 12844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997995905167468e-05,
+      "loss": 1.11,
+      "step": 12845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997995590706017e-05,
+      "loss": 1.1698,
+      "step": 12846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997995276219907e-05,
+      "loss": 1.0783,
+      "step": 12847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979949617091366e-05,
+      "loss": 1.4209,
+      "step": 12848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979946471737075e-05,
+      "loss": 1.0297,
+      "step": 12849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997994332613619e-05,
+      "loss": 1.1634,
+      "step": 12850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997994018028872e-05,
+      "loss": 1.3217,
+      "step": 12851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997993703419466e-05,
+      "loss": 0.9882,
+      "step": 12852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979933887854e-05,
+      "loss": 1.0151,
+      "step": 12853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997993074126675e-05,
+      "loss": 1.0105,
+      "step": 12854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979927594432916e-05,
+      "loss": 1.0664,
+      "step": 12855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997992444735248e-05,
+      "loss": 1.1174,
+      "step": 12856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997992130002546e-05,
+      "loss": 1.2691,
+      "step": 12857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997991815245185e-05,
+      "loss": 1.1248,
+      "step": 12858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979915004631644e-05,
+      "loss": 1.1102,
+      "step": 12859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997991185656485e-05,
+      "loss": 1.1124,
+      "step": 12860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979908708251465e-05,
+      "loss": 1.1699,
+      "step": 12861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979905559691487e-05,
+      "loss": 1.2475,
+      "step": 12862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997990241088491e-05,
+      "loss": 1.0424,
+      "step": 12863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997989926183175e-05,
+      "loss": 1.1789,
+      "step": 12864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997989611253201e-05,
+      "loss": 0.8412,
+      "step": 12865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997989296298566e-05,
+      "loss": 0.9935,
+      "step": 12866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997988981319273e-05,
+      "loss": 1.3858,
+      "step": 12867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997988666315321e-05,
+      "loss": 1.2902,
+      "step": 12868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997988351286709e-05,
+      "loss": 1.7678,
+      "step": 12869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997988036233439e-05,
+      "loss": 1.1934,
+      "step": 12870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979877211555094e-05,
+      "loss": 1.1657,
+      "step": 12871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979874060529206e-05,
+      "loss": 0.9137,
+      "step": 12872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997987090925672e-05,
+      "loss": 1.0796,
+      "step": 12873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997986775773766e-05,
+      "loss": 1.1675,
+      "step": 12874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979864605972e-05,
+      "loss": 1.1648,
+      "step": 12875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979861453959745e-05,
+      "loss": 1.8118,
+      "step": 12876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997985830170091e-05,
+      "loss": 0.8699,
+      "step": 12877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997985514919548e-05,
+      "loss": 1.2232,
+      "step": 12878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997985199644345e-05,
+      "loss": 1.1903,
+      "step": 12879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979848843444845e-05,
+      "loss": 1.1403,
+      "step": 12880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997984569019964e-05,
+      "loss": 1.2959,
+      "step": 12881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979842536707845e-05,
+      "loss": 1.0923,
+      "step": 12882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997983938296946e-05,
+      "loss": 0.8126,
+      "step": 12883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997983622898449e-05,
+      "loss": 1.1587,
+      "step": 12884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997983307475292e-05,
+      "loss": 0.941,
+      "step": 12885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979829920274766e-05,
+      "loss": 1.2729,
+      "step": 12886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997982676555002e-05,
+      "loss": 1.2418,
+      "step": 12887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997982361057868e-05,
+      "loss": 1.1064,
+      "step": 12888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997982045536076e-05,
+      "loss": 1.249,
+      "step": 12889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997981729989624e-05,
+      "loss": 1.2328,
+      "step": 12890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997981414418513e-05,
+      "loss": 1.0787,
+      "step": 12891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997981098822744e-05,
+      "loss": 1.0769,
+      "step": 12892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997980783202315e-05,
+      "loss": 0.9737,
+      "step": 12893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997980467557227e-05,
+      "loss": 0.965,
+      "step": 12894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99798015188748e-05,
+      "loss": 0.9079,
+      "step": 12895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997979836193075e-05,
+      "loss": 0.7914,
+      "step": 12896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99797952047401e-05,
+      "loss": 0.9449,
+      "step": 12897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979792047302863e-05,
+      "loss": 0.9801,
+      "step": 12898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997978888961904e-05,
+      "loss": 1.3363,
+      "step": 12899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997978573168862e-05,
+      "loss": 1.0618,
+      "step": 12900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997978257351161e-05,
+      "loss": 0.2838,
+      "step": 12901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979779415088016e-05,
+      "loss": 0.3652,
+      "step": 12902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979776256417825e-05,
+      "loss": 0.9881,
+      "step": 12903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979773097501046e-05,
+      "loss": 1.7913,
+      "step": 12904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979769938337685e-05,
+      "loss": 1.7617,
+      "step": 12905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997976677892773e-05,
+      "loss": 1.3546,
+      "step": 12906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997976361927118e-05,
+      "loss": 1.2825,
+      "step": 12907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997976045936804e-05,
+      "loss": 1.0407,
+      "step": 12908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997975729921831e-05,
+      "loss": 0.9894,
+      "step": 12909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979754138822e-05,
+      "loss": 1.2142,
+      "step": 12910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979750978179094e-05,
+      "loss": 1.1767,
+      "step": 12911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99797478172896e-05,
+      "loss": 0.8767,
+      "step": 12912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979744656153516e-05,
+      "loss": 1.2397,
+      "step": 12913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997974149477085e-05,
+      "loss": 1.2256,
+      "step": 12914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997973833314158e-05,
+      "loss": 0.8299,
+      "step": 12915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979735171265726e-05,
+      "loss": 1.1286,
+      "step": 12916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979732009143286e-05,
+      "loss": 0.9414,
+      "step": 12917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997972884677425e-05,
+      "loss": 1.035,
+      "step": 12918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979725684158636e-05,
+      "loss": 1.084,
+      "step": 12919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979722521296424e-05,
+      "loss": 0.8994,
+      "step": 12920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979719358187625e-05,
+      "loss": 1.2614,
+      "step": 12921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997971619483224e-05,
+      "loss": 1.0149,
+      "step": 12922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979713031230255e-05,
+      "loss": 1.0084,
+      "step": 12923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99797098673817e-05,
+      "loss": 1.2544,
+      "step": 12924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997970670328654e-05,
+      "loss": 0.9306,
+      "step": 12925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997970353894479e-05,
+      "loss": 1.2647,
+      "step": 12926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979700374356455e-05,
+      "loss": 1.2281,
+      "step": 12927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997969720952154e-05,
+      "loss": 0.5042,
+      "step": 12928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979694044440026e-05,
+      "loss": 0.6964,
+      "step": 12929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997969087911192e-05,
+      "loss": 1.3595,
+      "step": 12930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979687713537224e-05,
+      "loss": 1.0151,
+      "step": 12931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997968454771595e-05,
+      "loss": 1.1689,
+      "step": 12932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997968138164808e-05,
+      "loss": 1.1825,
+      "step": 12933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997967821533362e-05,
+      "loss": 0.9164,
+      "step": 12934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997967504877258e-05,
+      "loss": 1.018,
+      "step": 12935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997967188196494e-05,
+      "loss": 0.7397,
+      "step": 12936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997966871491072e-05,
+      "loss": 1.758,
+      "step": 12937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979665547609914e-05,
+      "loss": 1.2953,
+      "step": 12938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979662380062506e-05,
+      "loss": 1.0949,
+      "step": 12939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979659212268516e-05,
+      "loss": 1.389,
+      "step": 12940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997965604422794e-05,
+      "loss": 1.1682,
+      "step": 12941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997965287594077e-05,
+      "loss": 1.4026,
+      "step": 12942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997964970740702e-05,
+      "loss": 1.5013,
+      "step": 12943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997964653862667e-05,
+      "loss": 1.0865,
+      "step": 12944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997964336959974e-05,
+      "loss": 1.0767,
+      "step": 12945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979640200326214e-05,
+      "loss": 1.4635,
+      "step": 12946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997963703080611e-05,
+      "loss": 1.2607,
+      "step": 12947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997963386103941e-05,
+      "loss": 1.0479,
+      "step": 12948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997963069102613e-05,
+      "loss": 1.2824,
+      "step": 12949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997962752076625e-05,
+      "loss": 1.3158,
+      "step": 12950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979624350259776e-05,
+      "loss": 1.0888,
+      "step": 12951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997962117950673e-05,
+      "loss": 0.8503,
+      "step": 12952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997961800850709e-05,
+      "loss": 0.7097,
+      "step": 12953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997961483726086e-05,
+      "loss": 0.8083,
+      "step": 12954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997961166576805e-05,
+      "loss": 1.0741,
+      "step": 12955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997960849402864e-05,
+      "loss": 1.1947,
+      "step": 12956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979605322042644e-05,
+      "loss": 0.9949,
+      "step": 12957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979602149810066e-05,
+      "loss": 1.1383,
+      "step": 12958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99795989773309e-05,
+      "loss": 0.9662,
+      "step": 12959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997959580460514e-05,
+      "loss": 0.824,
+      "step": 12960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997959263163279e-05,
+      "loss": 1.1551,
+      "step": 12961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997958945841386e-05,
+      "loss": 1.2205,
+      "step": 12962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997958628494834e-05,
+      "loss": 0.4988,
+      "step": 12963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997958311123623e-05,
+      "loss": 0.2457,
+      "step": 12964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997957993727753e-05,
+      "loss": 0.2279,
+      "step": 12965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997957676307224e-05,
+      "loss": 0.1507,
+      "step": 12966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997957358862038e-05,
+      "loss": 0.39,
+      "step": 12967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997957041392192e-05,
+      "loss": 0.773,
+      "step": 12968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997956723897686e-05,
+      "loss": 0.7261,
+      "step": 12969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979564063785226e-05,
+      "loss": 0.7207,
+      "step": 12970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979560888347e-05,
+      "loss": 0.6963,
+      "step": 12971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979557712662196e-05,
+      "loss": 0.5992,
+      "step": 12972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979554536730795e-05,
+      "loss": 0.7173,
+      "step": 12973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979551360552806e-05,
+      "loss": 0.7709,
+      "step": 12974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979548184128235e-05,
+      "loss": 0.6878,
+      "step": 12975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997954500745707e-05,
+      "loss": 0.5592,
+      "step": 12976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997954183053932e-05,
+      "loss": 0.645,
+      "step": 12977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997953865337498e-05,
+      "loss": 0.5575,
+      "step": 12978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979535475964056e-05,
+      "loss": 0.5624,
+      "step": 12979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979532298306544e-05,
+      "loss": 0.562,
+      "step": 12980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979529120402444e-05,
+      "loss": 0.5335,
+      "step": 12981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979525942251756e-05,
+      "loss": 0.4988,
+      "step": 12982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997952276385448e-05,
+      "loss": 0.502,
+      "step": 12983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997951958521062e-05,
+      "loss": 0.4721,
+      "step": 12984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997951640632017e-05,
+      "loss": 0.4373,
+      "step": 12985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997951322718314e-05,
+      "loss": 0.4362,
+      "step": 12986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997951004779952e-05,
+      "loss": 0.4396,
+      "step": 12987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99795068681693e-05,
+      "loss": 0.4634,
+      "step": 12988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997950368829251e-05,
+      "loss": 0.5829,
+      "step": 12989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997950050816912e-05,
+      "loss": 0.5246,
+      "step": 12990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979497327799145e-05,
+      "loss": 0.4869,
+      "step": 12991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997949414718259e-05,
+      "loss": 0.498,
+      "step": 12992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997949096631944e-05,
+      "loss": 0.4482,
+      "step": 12993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997948778520971e-05,
+      "loss": 0.4193,
+      "step": 12994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997948460385339e-05,
+      "loss": 0.4703,
+      "step": 12995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997948142225048e-05,
+      "loss": 0.5105,
+      "step": 12996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997947824040099e-05,
+      "loss": 0.4898,
+      "step": 12997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979475058304906e-05,
+      "loss": 0.4599,
+      "step": 12998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997947187596224e-05,
+      "loss": 0.4422,
+      "step": 12999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997946869337298e-05,
+      "loss": 0.4958,
+      "step": 13000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0994994640350342,
+      "eval_runtime": 84.1793,
+      "eval_samples_per_second": 16.453,
+      "eval_steps_per_second": 4.122,
+      "step": 13000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997946551053714e-05,
+      "loss": 0.5183,
+      "step": 13001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997946232745471e-05,
+      "loss": 0.4278,
+      "step": 13002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99794591441257e-05,
+      "loss": 0.4494,
+      "step": 13003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979455960550096e-05,
+      "loss": 0.4726,
+      "step": 13004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997945277672791e-05,
+      "loss": 0.47,
+      "step": 13005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997944959265913e-05,
+      "loss": 0.5201,
+      "step": 13006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997944640834377e-05,
+      "loss": 0.5025,
+      "step": 13007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979443223781824e-05,
+      "loss": 0.525,
+      "step": 13008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997944003897329e-05,
+      "loss": 0.4876,
+      "step": 13009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997943685391817e-05,
+      "loss": 0.8739,
+      "step": 13010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997943366861646e-05,
+      "loss": 1.0577,
+      "step": 13011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997943048306817e-05,
+      "loss": 1.1702,
+      "step": 13012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979427297273286e-05,
+      "loss": 1.2193,
+      "step": 13013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997942411123182e-05,
+      "loss": 1.2447,
+      "step": 13014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997942092494376e-05,
+      "loss": 1.5422,
+      "step": 13015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997941773840913e-05,
+      "loss": 1.0676,
+      "step": 13016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99794145516279e-05,
+      "loss": 1.3999,
+      "step": 13017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979411364600095e-05,
+      "loss": 0.8781,
+      "step": 13018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997940817732569e-05,
+      "loss": 1.171,
+      "step": 13019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997940498980471e-05,
+      "loss": 1.1286,
+      "step": 13020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979401802037135e-05,
+      "loss": 1.2802,
+      "step": 13021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997939861402298e-05,
+      "loss": 0.9592,
+      "step": 13022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997939542576223e-05,
+      "loss": 1.0764,
+      "step": 13023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997939223725491e-05,
+      "loss": 1.2927,
+      "step": 13024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997938904850099e-05,
+      "loss": 1.04,
+      "step": 13025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979385859500497e-05,
+      "loss": 1.0126,
+      "step": 13026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979382670253406e-05,
+      "loss": 1.3136,
+      "step": 13027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997937948075973e-05,
+      "loss": 1.3859,
+      "step": 13028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997937629101947e-05,
+      "loss": 1.697,
+      "step": 13029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997937310103263e-05,
+      "loss": 1.0814,
+      "step": 13030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99793699107992e-05,
+      "loss": 0.9613,
+      "step": 13031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997936672031918e-05,
+      "loss": 1.0867,
+      "step": 13032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997936352959258e-05,
+      "loss": 1.217,
+      "step": 13033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997936033861939e-05,
+      "loss": 1.0016,
+      "step": 13034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997935714739962e-05,
+      "loss": 0.9677,
+      "step": 13035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997935395593326e-05,
+      "loss": 1.2878,
+      "step": 13036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997935076422031e-05,
+      "loss": 1.1637,
+      "step": 13037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979347572260785e-05,
+      "loss": 0.9675,
+      "step": 13038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997934438005467e-05,
+      "loss": 1.9872,
+      "step": 13039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979341187601965e-05,
+      "loss": 0.9205,
+      "step": 13040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997933799490267e-05,
+      "loss": 0.8962,
+      "step": 13041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99793348019568e-05,
+      "loss": 1.0186,
+      "step": 13042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979331608764344e-05,
+      "loss": 1.146,
+      "step": 13043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99793284153253e-05,
+      "loss": 1.4336,
+      "step": 13044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979325221639675e-05,
+      "loss": 1.1118,
+      "step": 13045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997932202770746e-05,
+      "loss": 0.8323,
+      "step": 13046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997931883352866e-05,
+      "loss": 2.7557,
+      "step": 13047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997931563910328e-05,
+      "loss": 0.3497,
+      "step": 13048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99793124444313e-05,
+      "loss": 1.1898,
+      "step": 13049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979309249512754e-05,
+      "loss": 1.3632,
+      "step": 13050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979306054347607e-05,
+      "loss": 1.1675,
+      "step": 13051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979302858935884e-05,
+      "loss": 1.0783,
+      "step": 13052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997929966327757e-05,
+      "loss": 1.0931,
+      "step": 13053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997929646737267e-05,
+      "loss": 1.1867,
+      "step": 13054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979293271221195e-05,
+      "loss": 1.4483,
+      "step": 13055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997929007482313e-05,
+      "loss": 1.4049,
+      "step": 13056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997928687817848e-05,
+      "loss": 1.5023,
+      "step": 13057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997928368128724e-05,
+      "loss": 0.9887,
+      "step": 13058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997928048414941e-05,
+      "loss": 1.3957,
+      "step": 13059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997927728676501e-05,
+      "loss": 1.084,
+      "step": 13060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997927408913402e-05,
+      "loss": 1.2527,
+      "step": 13061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979270891256445e-05,
+      "loss": 1.4548,
+      "step": 13062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997926769313228e-05,
+      "loss": 1.1246,
+      "step": 13063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997926449476153e-05,
+      "loss": 0.9704,
+      "step": 13064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997926129614421e-05,
+      "loss": 1.164,
+      "step": 13065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979258097280294e-05,
+      "loss": 1.577,
+      "step": 13066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997925489816979e-05,
+      "loss": 1.0228,
+      "step": 13067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997925169881271e-05,
+      "loss": 0.8215,
+      "step": 13068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997924849920904e-05,
+      "loss": 0.9098,
+      "step": 13069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979245299358785e-05,
+      "loss": 1.2712,
+      "step": 13070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997924209926195e-05,
+      "loss": 1.2067,
+      "step": 13071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997923889891852e-05,
+      "loss": 1.1048,
+      "step": 13072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997923569832852e-05,
+      "loss": 1.1711,
+      "step": 13073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979232497491925e-05,
+      "loss": 1.35,
+      "step": 13074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997922929640875e-05,
+      "loss": 0.7622,
+      "step": 13075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997922609507898e-05,
+      "loss": 1.3351,
+      "step": 13076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997922289350264e-05,
+      "loss": 0.9535,
+      "step": 13077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979219691679715e-05,
+      "loss": 0.9206,
+      "step": 13078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99792164896102e-05,
+      "loss": 1.1329,
+      "step": 13079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99792132872941e-05,
+      "loss": 0.8848,
+      "step": 13080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979210084731414e-05,
+      "loss": 0.7544,
+      "step": 13081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997920688192215e-05,
+      "loss": 0.8494,
+      "step": 13082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99792036788663e-05,
+      "loss": 1.131,
+      "step": 13083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997920047556387e-05,
+      "loss": 0.7298,
+      "step": 13084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997919727201484e-05,
+      "loss": 0.1745,
+      "step": 13085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979194068219235e-05,
+      "loss": 0.1724,
+      "step": 13086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979190864177053e-05,
+      "loss": 0.1669,
+      "step": 13087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979187659888284e-05,
+      "loss": 0.1473,
+      "step": 13088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979184455352926e-05,
+      "loss": 0.5057,
+      "step": 13089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979181250570986e-05,
+      "loss": 0.536,
+      "step": 13090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979178045542465e-05,
+      "loss": 0.6683,
+      "step": 13091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997917484026735e-05,
+      "loss": 0.6792,
+      "step": 13092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979171634745665e-05,
+      "loss": 0.6064,
+      "step": 13093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979168428977386e-05,
+      "loss": 0.5759,
+      "step": 13094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997916522296253e-05,
+      "loss": 0.5447,
+      "step": 13095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979162016701084e-05,
+      "loss": 0.4281,
+      "step": 13096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997915881019306e-05,
+      "loss": 0.5687,
+      "step": 13097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997915560343845e-05,
+      "loss": 0.5711,
+      "step": 13098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997915239643726e-05,
+      "loss": 0.532,
+      "step": 13099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979149189189476e-05,
+      "loss": 0.4495,
+      "step": 13100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997914598169512e-05,
+      "loss": 0.5383,
+      "step": 13101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997914277395417e-05,
+      "loss": 0.6257,
+      "step": 13102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997913956596664e-05,
+      "loss": 0.4699,
+      "step": 13103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997913635773253e-05,
+      "loss": 0.4719,
+      "step": 13104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979133149251834e-05,
+      "loss": 0.4495,
+      "step": 13105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997912994052456e-05,
+      "loss": 0.4682,
+      "step": 13106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997912673155069e-05,
+      "loss": 0.403,
+      "step": 13107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997912352233025e-05,
+      "loss": 0.3921,
+      "step": 13108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997912031286322e-05,
+      "loss": 0.3934,
+      "step": 13109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997911710314961e-05,
+      "loss": 0.4463,
+      "step": 13110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997911389318941e-05,
+      "loss": 0.5515,
+      "step": 13111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997911068298263e-05,
+      "loss": 0.5984,
+      "step": 13112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979107472529266e-05,
+      "loss": 0.4758,
+      "step": 13113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979104261829324e-05,
+      "loss": 0.462,
+      "step": 13114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979101050882794e-05,
+      "loss": 0.4119,
+      "step": 13115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997909783968968e-05,
+      "loss": 0.4225,
+      "step": 13116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997909462824999e-05,
+      "loss": 0.4138,
+      "step": 13117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997909141656371e-05,
+      "loss": 0.5493,
+      "step": 13118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979088204630845e-05,
+      "loss": 0.5572,
+      "step": 13119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997908499245141e-05,
+      "loss": 0.4677,
+      "step": 13120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979081780025376e-05,
+      "loss": 0.4902,
+      "step": 13121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997907856735277e-05,
+      "loss": 0.4652,
+      "step": 13122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997907535443357e-05,
+      "loss": 0.5826,
+      "step": 13123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979072141267796e-05,
+      "loss": 0.4451,
+      "step": 13124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997906892785544e-05,
+      "loss": 0.4576,
+      "step": 13125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997906571419649e-05,
+      "loss": 0.4687,
+      "step": 13126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997906250029097e-05,
+      "loss": 0.4176,
+      "step": 13127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997905928613886e-05,
+      "loss": 0.4297,
+      "step": 13128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997905607174017e-05,
+      "loss": 0.4745,
+      "step": 13129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99790528570949e-05,
+      "loss": 0.5335,
+      "step": 13130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979049642203045e-05,
+      "loss": 0.482,
+      "step": 13131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997904642706461e-05,
+      "loss": 0.5525,
+      "step": 13132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979043211679586e-05,
+      "loss": 0.8318,
+      "step": 13133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997903999604798e-05,
+      "loss": 1.1823,
+      "step": 13134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979036780169795e-05,
+      "loss": 1.3099,
+      "step": 13135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997903356404503e-05,
+      "loss": 0.9838,
+      "step": 13136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997903034767368e-05,
+      "loss": 1.0567,
+      "step": 13137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997902713105575e-05,
+      "loss": 1.2956,
+      "step": 13138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997902391419123e-05,
+      "loss": 1.2132,
+      "step": 13139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997902069708013e-05,
+      "loss": 1.2639,
+      "step": 13140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979017479722446e-05,
+      "loss": 1.2568,
+      "step": 13141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997901426211818e-05,
+      "loss": 1.2874,
+      "step": 13142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979011044267345e-05,
+      "loss": 1.105,
+      "step": 13143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997900782616991e-05,
+      "loss": 1.1131,
+      "step": 13144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979004607825904e-05,
+      "loss": 1.1554,
+      "step": 13145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979001389235315e-05,
+      "loss": 0.9998,
+      "step": 13146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997899817039814e-05,
+      "loss": 1.0174,
+      "step": 13147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978994951314386e-05,
+      "loss": 1.1904,
+      "step": 13148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978991731984046e-05,
+      "loss": 0.8066,
+      "step": 13149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978988512407124e-05,
+      "loss": 0.5018,
+      "step": 13150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997898529258362e-05,
+      "loss": 0.3624,
+      "step": 13151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978982072513536e-05,
+      "loss": 0.3332,
+      "step": 13152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997897885219687e-05,
+      "loss": 0.2909,
+      "step": 13153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978975631633616e-05,
+      "loss": 0.2502,
+      "step": 13154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997897241082379e-05,
+      "loss": 1.0549,
+      "step": 13155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978969189767377e-05,
+      "loss": 1.093,
+      "step": 13156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997896596846438e-05,
+      "loss": 1.2906,
+      "step": 13157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978962746914805e-05,
+      "loss": 0.7933,
+      "step": 13158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997895952511865e-05,
+      "loss": 2.054,
+      "step": 13159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997895630307591e-05,
+      "loss": 1.0205,
+      "step": 13160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997895308078658e-05,
+      "loss": 1.2437,
+      "step": 13161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997894985825068e-05,
+      "loss": 1.0688,
+      "step": 13162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978946635468196e-05,
+      "loss": 0.919,
+      "step": 13163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978943412439134e-05,
+      "loss": 0.8673,
+      "step": 13164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978940189163483e-05,
+      "loss": 0.4756,
+      "step": 13165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997893696564125e-05,
+      "loss": 0.872,
+      "step": 13166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978933741872445e-05,
+      "loss": 1.2752,
+      "step": 13167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997893051785705e-05,
+      "loss": 1.132,
+      "step": 13168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997892729359508e-05,
+      "loss": 0.8541,
+      "step": 13169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997892406908652e-05,
+      "loss": 1.2013,
+      "step": 13170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997892084433139e-05,
+      "loss": 1.0825,
+      "step": 13171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997891761932967e-05,
+      "loss": 1.1343,
+      "step": 13172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997891439408137e-05,
+      "loss": 1.1043,
+      "step": 13173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978911168586486e-05,
+      "loss": 1.1395,
+      "step": 13174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997890794284502e-05,
+      "loss": 1.1256,
+      "step": 13175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997890471685698e-05,
+      "loss": 1.09,
+      "step": 13176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978901490622354e-05,
+      "loss": 1.0381,
+      "step": 13177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997889826414115e-05,
+      "loss": 0.9623,
+      "step": 13178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997889503741337e-05,
+      "loss": 0.575,
+      "step": 13179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978891810438995e-05,
+      "loss": 0.7448,
+      "step": 13180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997888858321804e-05,
+      "loss": 0.6096,
+      "step": 13181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997888535575051e-05,
+      "loss": 1.2684,
+      "step": 13182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978882128036404e-05,
+      "loss": 1.1718,
+      "step": 13183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997887890007571e-05,
+      "loss": 1.1802,
+      "step": 13184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997887567186844e-05,
+      "loss": 0.7917,
+      "step": 13185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997887244341458e-05,
+      "loss": 1.0472,
+      "step": 13186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997886921471415e-05,
+      "loss": 1.371,
+      "step": 13187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997886598576713e-05,
+      "loss": 1.1039,
+      "step": 13188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978862756573535e-05,
+      "loss": 1.1748,
+      "step": 13189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997885952713336e-05,
+      "loss": 1.0792,
+      "step": 13190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99788562974466e-05,
+      "loss": 1.1257,
+      "step": 13191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997885306751326e-05,
+      "loss": 1.1188,
+      "step": 13192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997884983733334e-05,
+      "loss": 1.0585,
+      "step": 13193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978846606906834e-05,
+      "loss": 0.8592,
+      "step": 13194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997884337623376e-05,
+      "loss": 1.3862,
+      "step": 13195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99788401453141e-05,
+      "loss": 1.1329,
+      "step": 13196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978836914147855e-05,
+      "loss": 1.3172,
+      "step": 13197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997883368273503e-05,
+      "loss": 1.1406,
+      "step": 13198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997883045107563e-05,
+      "loss": 1.0109,
+      "step": 13199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978827219169643e-05,
+      "loss": 1.282,
+      "step": 13200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978823987017074e-05,
+      "loss": 0.9855,
+      "step": 13201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997882075461793e-05,
+      "loss": 0.7936,
+      "step": 13202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997881752197221e-05,
+      "loss": 1.0041,
+      "step": 13203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99788142890799e-05,
+      "loss": 1.1821,
+      "step": 13204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997881105594102e-05,
+      "loss": 1.3851,
+      "step": 13205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978807822555554e-05,
+      "loss": 0.9859,
+      "step": 13206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99788045889235e-05,
+      "loss": 0.9828,
+      "step": 13207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997880135504488e-05,
+      "loss": 0.9934,
+      "step": 13208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997879812091967e-05,
+      "loss": 0.9731,
+      "step": 13209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997879488654789e-05,
+      "loss": 1.0061,
+      "step": 13210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997879165192951e-05,
+      "loss": 1.2834,
+      "step": 13211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997878841706457e-05,
+      "loss": 1.2319,
+      "step": 13212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997878518195304e-05,
+      "loss": 1.1768,
+      "step": 13213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997878194659493e-05,
+      "loss": 1.0656,
+      "step": 13214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997877871099025e-05,
+      "loss": 1.0897,
+      "step": 13215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997877547513898e-05,
+      "loss": 1.3611,
+      "step": 13216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978772239041134e-05,
+      "loss": 0.9368,
+      "step": 13217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99787690026967e-05,
+      "loss": 1.487,
+      "step": 13218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997876576610569e-05,
+      "loss": 1.0307,
+      "step": 13219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997876252926811e-05,
+      "loss": 1.1302,
+      "step": 13220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997875929218394e-05,
+      "loss": 1.1502,
+      "step": 13221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997875605485319e-05,
+      "loss": 1.1711,
+      "step": 13222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997875281727587e-05,
+      "loss": 0.9221,
+      "step": 13223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997874957945196e-05,
+      "loss": 0.903,
+      "step": 13224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997874634138148e-05,
+      "loss": 1.0384,
+      "step": 13225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978743103064415e-05,
+      "loss": 1.1008,
+      "step": 13226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997873986450077e-05,
+      "loss": 1.1846,
+      "step": 13227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997873662569054e-05,
+      "loss": 1.3003,
+      "step": 13228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997873338663374e-05,
+      "loss": 1.1651,
+      "step": 13229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978730147330355e-05,
+      "loss": 0.9041,
+      "step": 13230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978726907780396e-05,
+      "loss": 1.1599,
+      "step": 13231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997872366798385e-05,
+      "loss": 1.2287,
+      "step": 13232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997872042794073e-05,
+      "loss": 1.3725,
+      "step": 13233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978717187651025e-05,
+      "loss": 1.2067,
+      "step": 13234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997871394711474e-05,
+      "loss": 1.1195,
+      "step": 13235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997871070633188e-05,
+      "loss": 1.0614,
+      "step": 13236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997870746530244e-05,
+      "loss": 0.9316,
+      "step": 13237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978704224026426e-05,
+      "loss": 0.9898,
+      "step": 13238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997870098250382e-05,
+      "loss": 0.7764,
+      "step": 13239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997869774073465e-05,
+      "loss": 0.5434,
+      "step": 13240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997869449871889e-05,
+      "loss": 1.0947,
+      "step": 13241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997869125645656e-05,
+      "loss": 1.0978,
+      "step": 13242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997868801394764e-05,
+      "loss": 1.242,
+      "step": 13243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978684771192144e-05,
+      "loss": 1.0641,
+      "step": 13244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997868152819007e-05,
+      "loss": 1.096,
+      "step": 13245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978678284941426e-05,
+      "loss": 1.0832,
+      "step": 13246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997867504144619e-05,
+      "loss": 1.04,
+      "step": 13247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997867179770438e-05,
+      "loss": 1.0938,
+      "step": 13248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997866855371599e-05,
+      "loss": 1.0092,
+      "step": 13249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978665309481025e-05,
+      "loss": 1.2841,
+      "step": 13250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978662064999485e-05,
+      "loss": 1.3628,
+      "step": 13251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997865882027135e-05,
+      "loss": 1.3454,
+      "step": 13252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997865557529665e-05,
+      "loss": 1.1328,
+      "step": 13253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997865233007537e-05,
+      "loss": 1.2953,
+      "step": 13254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99786490846075e-05,
+      "loss": 1.0408,
+      "step": 13255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997864583889307e-05,
+      "loss": 1.0113,
+      "step": 13256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978642592932046e-05,
+      "loss": 0.99,
+      "step": 13257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997863934672445e-05,
+      "loss": 1.3442,
+      "step": 13258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997863610027027e-05,
+      "loss": 1.1026,
+      "step": 13259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997863285356952e-05,
+      "loss": 1.1804,
+      "step": 13260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997862960662218e-05,
+      "loss": 1.118,
+      "step": 13261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978626359428274e-05,
+      "loss": 0.8905,
+      "step": 13262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997862311198778e-05,
+      "loss": 0.859,
+      "step": 13263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997861986430071e-05,
+      "loss": 0.9503,
+      "step": 13264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978616616367065e-05,
+      "loss": 0.837,
+      "step": 13265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978613368186837e-05,
+      "loss": 0.8207,
+      "step": 13266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997861011976004e-05,
+      "loss": 1.2311,
+      "step": 13267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978606871086656e-05,
+      "loss": 1.1697,
+      "step": 13268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997860362216669e-05,
+      "loss": 0.8688,
+      "step": 13269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997860037300015e-05,
+      "loss": 0.7533,
+      "step": 13270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978597123587035e-05,
+      "loss": 0.8438,
+      "step": 13271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997859387392734e-05,
+      "loss": 1.1314,
+      "step": 13272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978590624021074e-05,
+      "loss": 1.059,
+      "step": 13273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978587373868214e-05,
+      "loss": 1.2934,
+      "step": 13274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997858412346879e-05,
+      "loss": 1.2194,
+      "step": 13275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997858087282278e-05,
+      "loss": 1.388,
+      "step": 13276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997857762193019e-05,
+      "loss": 1.0481,
+      "step": 13277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997857437079103e-05,
+      "loss": 0.8995,
+      "step": 13278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997857111940529e-05,
+      "loss": 1.1896,
+      "step": 13279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997856786777297e-05,
+      "loss": 1.0073,
+      "step": 13280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997856461589407e-05,
+      "loss": 0.8425,
+      "step": 13281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978561363768595e-05,
+      "loss": 1.5723,
+      "step": 13282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997855811139654e-05,
+      "loss": 1.0289,
+      "step": 13283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997855485877792e-05,
+      "loss": 0.913,
+      "step": 13284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99785516059127e-05,
+      "loss": 1.1144,
+      "step": 13285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997854835280091e-05,
+      "loss": 1.0904,
+      "step": 13286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978545099442555e-05,
+      "loss": 0.9672,
+      "step": 13287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997854184583761e-05,
+      "loss": 1.1621,
+      "step": 13288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997853859198609e-05,
+      "loss": 0.9643,
+      "step": 13289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978535337887994e-05,
+      "loss": 1.0162,
+      "step": 13290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997853208354332e-05,
+      "loss": 1.1104,
+      "step": 13291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978528828952066e-05,
+      "loss": 1.0895,
+      "step": 13292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997852557411424e-05,
+      "loss": 0.9818,
+      "step": 13293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997852231902983e-05,
+      "loss": 1.1447,
+      "step": 13294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997851906369885e-05,
+      "loss": 1.1971,
+      "step": 13295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997851580812129e-05,
+      "loss": 0.8869,
+      "step": 13296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997851255229714e-05,
+      "loss": 1.4195,
+      "step": 13297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997850929622643e-05,
+      "loss": 1.0647,
+      "step": 13298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978506039909135e-05,
+      "loss": 0.9029,
+      "step": 13299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997850278334526e-05,
+      "loss": 1.104,
+      "step": 13300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978499526534816e-05,
+      "loss": 1.2008,
+      "step": 13301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978496269477784e-05,
+      "loss": 0.8578,
+      "step": 13302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997849301217419e-05,
+      "loss": 1.1522,
+      "step": 13303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978489754624e-05,
+      "loss": 1.3556,
+      "step": 13304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997848649682725e-05,
+      "loss": 1.2666,
+      "step": 13305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997848323878391e-05,
+      "loss": 1.2673,
+      "step": 13306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978479980494005e-05,
+      "loss": 1.0229,
+      "step": 13307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997847672195751e-05,
+      "loss": 1.771,
+      "step": 13308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997847346317445e-05,
+      "loss": 0.9576,
+      "step": 13309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997847020414481e-05,
+      "loss": 1.0482,
+      "step": 13310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997846694486858e-05,
+      "loss": 0.9268,
+      "step": 13311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997846368534579e-05,
+      "loss": 1.0142,
+      "step": 13312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997846042557642e-05,
+      "loss": 0.9943,
+      "step": 13313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997845716556047e-05,
+      "loss": 0.9842,
+      "step": 13314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997845390529794e-05,
+      "loss": 1.0571,
+      "step": 13315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978450644788834e-05,
+      "loss": 1.0647,
+      "step": 13316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978447384033155e-05,
+      "loss": 1.3043,
+      "step": 13317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99784441230309e-05,
+      "loss": 1.1047,
+      "step": 13318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997844086178206e-05,
+      "loss": 1.0826,
+      "step": 13319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978437600286656e-05,
+      "loss": 0.7416,
+      "step": 13320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978434338544665e-05,
+      "loss": 0.9795,
+      "step": 13321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978431076556106e-05,
+      "loss": 1.1324,
+      "step": 13322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997842781432096e-05,
+      "loss": 0.931,
+      "step": 13323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997842455183925e-05,
+      "loss": 0.9355,
+      "step": 13324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997842128911095e-05,
+      "loss": 1.0136,
+      "step": 13325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997841802613609e-05,
+      "loss": 1.1904,
+      "step": 13326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997841476291464e-05,
+      "loss": 0.8844,
+      "step": 13327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997841149944662e-05,
+      "loss": 1.1935,
+      "step": 13328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978408235732025e-05,
+      "loss": 1.3043,
+      "step": 13329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997840497177084e-05,
+      "loss": 1.4427,
+      "step": 13330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978401707563096e-05,
+      "loss": 1.121,
+      "step": 13331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997839844310876e-05,
+      "loss": 1.0611,
+      "step": 13332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997839517840786e-05,
+      "loss": 1.2253,
+      "step": 13333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997839191346038e-05,
+      "loss": 1.2705,
+      "step": 13334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997838864826633e-05,
+      "loss": 0.8734,
+      "step": 13335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997838538282569e-05,
+      "loss": 1.1791,
+      "step": 13336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997838211713848e-05,
+      "loss": 1.1229,
+      "step": 13337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99783788512047e-05,
+      "loss": 1.0637,
+      "step": 13338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997837558502434e-05,
+      "loss": 1.0856,
+      "step": 13339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99783723185974e-05,
+      "loss": 1.2392,
+      "step": 13340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997836905192389e-05,
+      "loss": 0.953,
+      "step": 13341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99783657850038e-05,
+      "loss": 1.0463,
+      "step": 13342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997836251783714e-05,
+      "loss": 1.5255,
+      "step": 13343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99783592504239e-05,
+      "loss": 1.0337,
+      "step": 13344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997835598276408e-05,
+      "loss": 1.0521,
+      "step": 13345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997835271485769e-05,
+      "loss": 1.051,
+      "step": 13346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997834944670472e-05,
+      "loss": 1.206,
+      "step": 13347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997834617830518e-05,
+      "loss": 0.962,
+      "step": 13348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997834290965906e-05,
+      "loss": 0.9511,
+      "step": 13349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997833964076637e-05,
+      "loss": 1.265,
+      "step": 13350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99783363716271e-05,
+      "loss": 1.032,
+      "step": 13351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978333102241247e-05,
+      "loss": 1.1215,
+      "step": 13352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978329832608825e-05,
+      "loss": 0.7762,
+      "step": 13353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978326562729835e-05,
+      "loss": 1.1542,
+      "step": 13354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997832329260426e-05,
+      "loss": 1.1663,
+      "step": 13355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997832002223211e-05,
+      "loss": 1.0352,
+      "step": 13356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978316751613384e-05,
+      "loss": 0.9754,
+      "step": 13357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997831348074808e-05,
+      "loss": 1.048,
+      "step": 13358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997831020963621e-05,
+      "loss": 1.1207,
+      "step": 13359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997830693827776e-05,
+      "loss": 0.8837,
+      "step": 13360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997830366667274e-05,
+      "loss": 1.3253,
+      "step": 13361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997830039482114e-05,
+      "loss": 1.2562,
+      "step": 13362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997829712272296e-05,
+      "loss": 1.3778,
+      "step": 13363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997829385037821e-05,
+      "loss": 0.9268,
+      "step": 13364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997829057778689e-05,
+      "loss": 1.0259,
+      "step": 13365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997828730494899e-05,
+      "loss": 1.2288,
+      "step": 13366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997828403186451e-05,
+      "loss": 1.1526,
+      "step": 13367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978280758533455e-05,
+      "loss": 1.0788,
+      "step": 13368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997827748495583e-05,
+      "loss": 1.1302,
+      "step": 13369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978274211131633e-05,
+      "loss": 1.4603,
+      "step": 13370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978270937060854e-05,
+      "loss": 1.3785,
+      "step": 13371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978267662743506e-05,
+      "loss": 1.2116,
+      "step": 13372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997826438817958e-05,
+      "loss": 1.2782,
+      "step": 13373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978261113369074e-05,
+      "loss": 1.1429,
+      "step": 13374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978257838312e-05,
+      "loss": 1.1785,
+      "step": 13375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978254563008356e-05,
+      "loss": 0.774,
+      "step": 13376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997825128745812e-05,
+      "loss": 1.0027,
+      "step": 13377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997824801166132e-05,
+      "loss": 0.9855,
+      "step": 13378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997824473561795e-05,
+      "loss": 1.1095,
+      "step": 13379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978241459328004e-05,
+      "loss": 1.1271,
+      "step": 13380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997823818279147e-05,
+      "loss": 1.0207,
+      "step": 13381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997823490600838e-05,
+      "loss": 1.0327,
+      "step": 13382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99782316289787e-05,
+      "loss": 1.0698,
+      "step": 13383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997822835170245e-05,
+      "loss": 1.2344,
+      "step": 13384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997822507417963e-05,
+      "loss": 1.2526,
+      "step": 13385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997822179641023e-05,
+      "loss": 1.2137,
+      "step": 13386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997821851839426e-05,
+      "loss": 0.9287,
+      "step": 13387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997821524013171e-05,
+      "loss": 1.1097,
+      "step": 13388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997821196162259e-05,
+      "loss": 1.0661,
+      "step": 13389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99782086828669e-05,
+      "loss": 0.9219,
+      "step": 13390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978205403864634e-05,
+      "loss": 1.1103,
+      "step": 13391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997820212461578e-05,
+      "loss": 0.8699,
+      "step": 13392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978198845120364e-05,
+      "loss": 0.889,
+      "step": 13393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997819556537837e-05,
+      "loss": 1.1113,
+      "step": 13394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997819228538981e-05,
+      "loss": 1.1604,
+      "step": 13395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997818900515466e-05,
+      "loss": 1.141,
+      "step": 13396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978185724672954e-05,
+      "loss": 0.5784,
+      "step": 13397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978182443944665e-05,
+      "loss": 1.2137,
+      "step": 13398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978179162969794e-05,
+      "loss": 1.1008,
+      "step": 13399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997817588174836e-05,
+      "loss": 0.967,
+      "step": 13400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997817260028035e-05,
+      "loss": 1.0273,
+      "step": 13401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997816931856577e-05,
+      "loss": 1.1577,
+      "step": 13402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978166036604606e-05,
+      "loss": 1.2207,
+      "step": 13403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978162754396875e-05,
+      "loss": 1.4783,
+      "step": 13404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978159471942563e-05,
+      "loss": 1.1987,
+      "step": 13405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997815618924169e-05,
+      "loss": 1.4815,
+      "step": 13406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997815290629423e-05,
+      "loss": 1.1812,
+      "step": 13407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99781496231002e-05,
+      "loss": 0.9092,
+      "step": 13408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978146339659596e-05,
+      "loss": 1.0899,
+      "step": 13409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978143055972424e-05,
+      "loss": 1.3144,
+      "step": 13410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997813977203867e-05,
+      "loss": 1.1962,
+      "step": 13411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997813648785835e-05,
+      "loss": 1.2084,
+      "step": 13412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978133203431454e-05,
+      "loss": 1.0493,
+      "step": 13413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978129918757984e-05,
+      "loss": 1.1011,
+      "step": 13414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997812663383794e-05,
+      "loss": 1.3777,
+      "step": 13415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978123348671326e-05,
+      "loss": 1.651,
+      "step": 13416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978120063258125e-05,
+      "loss": 1.5813,
+      "step": 13417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997811677759836e-05,
+      "loss": 1.5543,
+      "step": 13418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978113491692025e-05,
+      "loss": 1.2562,
+      "step": 13419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997811020553911e-05,
+      "loss": 1.6136,
+      "step": 13420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978106919139634e-05,
+      "loss": 1.7078,
+      "step": 13421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997810363249357e-05,
+      "loss": 1.173,
+      "step": 13422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978100345600944e-05,
+      "loss": 1.5011,
+      "step": 13423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997809705846174e-05,
+      "loss": 1.6022,
+      "step": 13424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978093771075955e-05,
+      "loss": 1.2602,
+      "step": 13425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997809048344361e-05,
+      "loss": 1.3651,
+      "step": 13426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997808719556468e-05,
+      "loss": 1.3125,
+      "step": 13427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978083907439185e-05,
+      "loss": 1.3091,
+      "step": 13428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978080619067115e-05,
+      "loss": 0.9429,
+      "step": 13429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997807733044847e-05,
+      "loss": 1.1973,
+      "step": 13430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997807404158325e-05,
+      "loss": 1.0719,
+      "step": 13431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997807075247146e-05,
+      "loss": 1.0395,
+      "step": 13432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99780674631131e-05,
+      "loss": 1.1366,
+      "step": 13433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978064173508164e-05,
+      "loss": 1.019,
+      "step": 13434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997806088365665e-05,
+      "loss": 1.0591,
+      "step": 13435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997805759355857e-05,
+      "loss": 1.0975,
+      "step": 13436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997805430321392e-05,
+      "loss": 1.2038,
+      "step": 13437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997805101262269e-05,
+      "loss": 1.2305,
+      "step": 13438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997804772178489e-05,
+      "loss": 1.1124,
+      "step": 13439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978044430700514e-05,
+      "loss": 1.1629,
+      "step": 13440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997804113936957e-05,
+      "loss": 1.0923,
+      "step": 13441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978037847792054e-05,
+      "loss": 0.9277,
+      "step": 13442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997803455596796e-05,
+      "loss": 0.981,
+      "step": 13443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978031263897295e-05,
+      "loss": 1.1139,
+      "step": 13444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997802797158006e-05,
+      "loss": 1.2224,
+      "step": 13445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997802467901626e-05,
+      "loss": 0.944,
+      "step": 13446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997802138620587e-05,
+      "loss": 1.0513,
+      "step": 13447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978018093148915e-05,
+      "loss": 1.0209,
+      "step": 13448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997801479984539e-05,
+      "loss": 1.0556,
+      "step": 13449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978011506295294e-05,
+      "loss": 1.188,
+      "step": 13450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978008212498615e-05,
+      "loss": 1.4315,
+      "step": 13451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997800491845538e-05,
+      "loss": 1.6668,
+      "step": 13452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997800162416556e-05,
+      "loss": 1.1929,
+      "step": 13453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997799832962917e-05,
+      "loss": 1.0631,
+      "step": 13454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997799503484621e-05,
+      "loss": 1.1428,
+      "step": 13455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997799173981668e-05,
+      "loss": 0.8503,
+      "step": 13456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997798844454057e-05,
+      "loss": 0.9871,
+      "step": 13457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99779851490179e-05,
+      "loss": 1.2522,
+      "step": 13458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997798185324865e-05,
+      "loss": 1.2566,
+      "step": 13459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997797855723283e-05,
+      "loss": 0.9995,
+      "step": 13460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997797526097043e-05,
+      "loss": 0.896,
+      "step": 13461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997797196446147e-05,
+      "loss": 1.0634,
+      "step": 13462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997796866770593e-05,
+      "loss": 0.6953,
+      "step": 13463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997796537070382e-05,
+      "loss": 1.043,
+      "step": 13464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997796207345514e-05,
+      "loss": 1.3509,
+      "step": 13465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997795877595989e-05,
+      "loss": 1.317,
+      "step": 13466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997795547821806e-05,
+      "loss": 1.2163,
+      "step": 13467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997795218022966e-05,
+      "loss": 1.1202,
+      "step": 13468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99779488819947e-05,
+      "loss": 1.0508,
+      "step": 13469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977945583513155e-05,
+      "loss": 1.203,
+      "step": 13470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997794228478504e-05,
+      "loss": 1.082,
+      "step": 13471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977938985810356e-05,
+      "loss": 0.8471,
+      "step": 13472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99779356865891e-05,
+      "loss": 0.4605,
+      "step": 13473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997793238712127e-05,
+      "loss": 0.2344,
+      "step": 13474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997792908740687e-05,
+      "loss": 0.3221,
+      "step": 13475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977925787445904e-05,
+      "loss": 1.3215,
+      "step": 13476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997792248723836e-05,
+      "loss": 1.2194,
+      "step": 13477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997791918678425e-05,
+      "loss": 0.6998,
+      "step": 13478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997791588608356e-05,
+      "loss": 1.1459,
+      "step": 13479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977912585136304e-05,
+      "loss": 0.9691,
+      "step": 13480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997790928394248e-05,
+      "loss": 0.8361,
+      "step": 13481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997790598250208e-05,
+      "loss": 1.3147,
+      "step": 13482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977902680815105e-05,
+      "loss": 0.9627,
+      "step": 13483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997789937888156e-05,
+      "loss": 0.7577,
+      "step": 13484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977896076701446e-05,
+      "loss": 0.5085,
+      "step": 13485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997789277427477e-05,
+      "loss": 0.6398,
+      "step": 13486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997788947160151e-05,
+      "loss": 1.1449,
+      "step": 13487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997788616868168e-05,
+      "loss": 1.0784,
+      "step": 13488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997788286551528e-05,
+      "loss": 0.9939,
+      "step": 13489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977879562102316e-05,
+      "loss": 1.1397,
+      "step": 13490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997787625844277e-05,
+      "loss": 1.3888,
+      "step": 13491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977872954536666e-05,
+      "loss": 1.2933,
+      "step": 13492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997786965038398e-05,
+      "loss": 1.1572,
+      "step": 13493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977866345984724e-05,
+      "loss": 1.1605,
+      "step": 13494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99778630413389e-05,
+      "loss": 1.1386,
+      "step": 13495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977859736446503e-05,
+      "loss": 1.1907,
+      "step": 13496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997785643130754e-05,
+      "loss": 1.1242,
+      "step": 13497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977853125922004e-05,
+      "loss": 0.7879,
+      "step": 13498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997784982028989e-05,
+      "loss": 0.6311,
+      "step": 13499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997784651441121e-05,
+      "loss": 0.883,
+      "step": 13500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0440155267715454,
+      "eval_runtime": 86.2254,
+      "eval_samples_per_second": 16.063,
+      "eval_steps_per_second": 4.024,
+      "step": 13500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997784320828596e-05,
+      "loss": 0.7417,
+      "step": 13501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977839901914144e-05,
+      "loss": 0.8161,
+      "step": 13502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977836595295744e-05,
+      "loss": 0.7489,
+      "step": 13503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997783328843079e-05,
+      "loss": 0.5493,
+      "step": 13504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997782998131926e-05,
+      "loss": 0.4742,
+      "step": 13505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997782667396115e-05,
+      "loss": 0.4656,
+      "step": 13506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997782336635648e-05,
+      "loss": 0.4238,
+      "step": 13507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997782005850523e-05,
+      "loss": 0.438,
+      "step": 13508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997781675040742e-05,
+      "loss": 0.5162,
+      "step": 13509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997781344206303e-05,
+      "loss": 0.4822,
+      "step": 13510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997781013347208e-05,
+      "loss": 0.4873,
+      "step": 13511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977806824634554e-05,
+      "loss": 0.5565,
+      "step": 13512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977803515550455e-05,
+      "loss": 0.4265,
+      "step": 13513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997780020621978e-05,
+      "loss": 0.4158,
+      "step": 13514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977796896642546e-05,
+      "loss": 0.4371,
+      "step": 13515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997779358681874e-05,
+      "loss": 0.6408,
+      "step": 13516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997779027674836e-05,
+      "loss": 0.5063,
+      "step": 13517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997778696643142e-05,
+      "loss": 0.4855,
+      "step": 13518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997778365586789e-05,
+      "loss": 0.5207,
+      "step": 13519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997778034505781e-05,
+      "loss": 0.4506,
+      "step": 13520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997777703400115e-05,
+      "loss": 0.4622,
+      "step": 13521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997777372269792e-05,
+      "loss": 0.4479,
+      "step": 13522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977770411148125e-05,
+      "loss": 0.4662,
+      "step": 13523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997776709935176e-05,
+      "loss": 0.4799,
+      "step": 13524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997776378730882e-05,
+      "loss": 0.4619,
+      "step": 13525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997776047501931e-05,
+      "loss": 0.4242,
+      "step": 13526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997775716248323e-05,
+      "loss": 0.4824,
+      "step": 13527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997775384970058e-05,
+      "loss": 0.463,
+      "step": 13528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977750536671366e-05,
+      "loss": 0.4151,
+      "step": 13529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997774722339557e-05,
+      "loss": 0.416,
+      "step": 13530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977743909873216e-05,
+      "loss": 0.4624,
+      "step": 13531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977740596104285e-05,
+      "loss": 0.4563,
+      "step": 13532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997773728208879e-05,
+      "loss": 0.4372,
+      "step": 13533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997773396782672e-05,
+      "loss": 0.5665,
+      "step": 13534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997773065331809e-05,
+      "loss": 0.4993,
+      "step": 13535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997772733856288e-05,
+      "loss": 0.4868,
+      "step": 13536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99777240235611e-05,
+      "loss": 0.5029,
+      "step": 13537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997772070831276e-05,
+      "loss": 0.51,
+      "step": 13538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997771739281784e-05,
+      "loss": 0.5196,
+      "step": 13539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997771407707635e-05,
+      "loss": 0.4814,
+      "step": 13540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977710761088304e-05,
+      "loss": 0.4599,
+      "step": 13541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997770744485368e-05,
+      "loss": 0.4118,
+      "step": 13542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977704128372483e-05,
+      "loss": 0.4171,
+      "step": 13543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977700811644725e-05,
+      "loss": 0.4187,
+      "step": 13544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997769749467039e-05,
+      "loss": 0.4279,
+      "step": 13545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997769417744949e-05,
+      "loss": 0.4406,
+      "step": 13546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997769085998202e-05,
+      "loss": 0.4483,
+      "step": 13547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977687542267984e-05,
+      "loss": 0.4418,
+      "step": 13548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997768422430737e-05,
+      "loss": 0.4313,
+      "step": 13549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997768090610019e-05,
+      "loss": 0.4078,
+      "step": 13550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977677587646446e-05,
+      "loss": 0.4018,
+      "step": 13551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997767426894613e-05,
+      "loss": 0.5116,
+      "step": 13552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997767094999925e-05,
+      "loss": 0.7394,
+      "step": 13553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997766763080579e-05,
+      "loss": 0.5174,
+      "step": 13554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997766431136577e-05,
+      "loss": 0.4871,
+      "step": 13555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977660991679176e-05,
+      "loss": 0.7955,
+      "step": 13556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977657671746014e-05,
+      "loss": 1.3388,
+      "step": 13557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977654351566285e-05,
+      "loss": 0.621,
+      "step": 13558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997765103113998e-05,
+      "loss": 0.3241,
+      "step": 13559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977647710467115e-05,
+      "loss": 0.6977,
+      "step": 13560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997764438954768e-05,
+      "loss": 1.4052,
+      "step": 13561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977641068381674e-05,
+      "loss": 1.2271,
+      "step": 13562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99776377469691e-05,
+      "loss": 1.379,
+      "step": 13563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997763442530996e-05,
+      "loss": 1.7043,
+      "step": 13564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997763110340424e-05,
+      "loss": 1.728,
+      "step": 13565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997762778125197e-05,
+      "loss": 1.3264,
+      "step": 13566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977624458853116e-05,
+      "loss": 1.2016,
+      "step": 13567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99776211362077e-05,
+      "loss": 0.9734,
+      "step": 13568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997761781331571e-05,
+      "loss": 1.1336,
+      "step": 13569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997761449017716e-05,
+      "loss": 1.1087,
+      "step": 13570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997761116679204e-05,
+      "loss": 1.0512,
+      "step": 13571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977607843160346e-05,
+      "loss": 1.1937,
+      "step": 13572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977604519282085e-05,
+      "loss": 1.1724,
+      "step": 13573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977601195157255e-05,
+      "loss": 1.1913,
+      "step": 13574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977597870785864e-05,
+      "loss": 1.0414,
+      "step": 13575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997759454616789e-05,
+      "loss": 1.2591,
+      "step": 13576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977591221303366e-05,
+      "loss": 0.7551,
+      "step": 13577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997758789619226e-05,
+      "loss": 1.1312,
+      "step": 13578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997758457083459e-05,
+      "loss": 0.9839,
+      "step": 13579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997758124523035e-05,
+      "loss": 1.0888,
+      "step": 13580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997757791937955e-05,
+      "loss": 1.0604,
+      "step": 13581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977574593282174e-05,
+      "loss": 1.04,
+      "step": 13582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997757126693823e-05,
+      "loss": 1.0527,
+      "step": 13583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997756794034772e-05,
+      "loss": 0.9416,
+      "step": 13584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997756461351064e-05,
+      "loss": 1.3833,
+      "step": 13585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977561286427e-05,
+      "loss": 1.0416,
+      "step": 13586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977557959096785e-05,
+      "loss": 0.8595,
+      "step": 13587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997755463152e-05,
+      "loss": 0.4039,
+      "step": 13588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997755130369664e-05,
+      "loss": 0.312,
+      "step": 13589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977547975626734e-05,
+      "loss": 0.8537,
+      "step": 13590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997754464731025e-05,
+      "loss": 0.9302,
+      "step": 13591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997754131874719e-05,
+      "loss": 1.1255,
+      "step": 13592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997753798993757e-05,
+      "loss": 1.3073,
+      "step": 13593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977534660881384e-05,
+      "loss": 1.0837,
+      "step": 13594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997753133157863e-05,
+      "loss": 0.9998,
+      "step": 13595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99775280020293e-05,
+      "loss": 1.1223,
+      "step": 13596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977524672233414e-05,
+      "loss": 0.9514,
+      "step": 13597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997752134219095e-05,
+      "loss": 1.4154,
+      "step": 13598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997751801190193e-05,
+      "loss": 1.2093,
+      "step": 13599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997751468136632e-05,
+      "loss": 1.1492,
+      "step": 13600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997751135058416e-05,
+      "loss": 1.0651,
+      "step": 13601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997750801955543e-05,
+      "loss": 1.1727,
+      "step": 13602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997750468828013e-05,
+      "loss": 1.0718,
+      "step": 13603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997750135675827e-05,
+      "loss": 1.1734,
+      "step": 13604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977498024989845e-05,
+      "loss": 1.1093,
+      "step": 13605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997749469297484e-05,
+      "loss": 0.9622,
+      "step": 13606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997749136071327e-05,
+      "loss": 1.1411,
+      "step": 13607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997748802820514e-05,
+      "loss": 1.2939,
+      "step": 13608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997748469545044e-05,
+      "loss": 1.1855,
+      "step": 13609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977481362449174e-05,
+      "loss": 1.1944,
+      "step": 13610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997747802920133e-05,
+      "loss": 1.3632,
+      "step": 13611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977474695706936e-05,
+      "loss": 0.9715,
+      "step": 13612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977471361965965e-05,
+      "loss": 0.9561,
+      "step": 13613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977468027978426e-05,
+      "loss": 1.3744,
+      "step": 13614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977464693744326e-05,
+      "loss": 1.0924,
+      "step": 13615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997746135926366e-05,
+      "loss": 1.1654,
+      "step": 13616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977458024536416e-05,
+      "loss": 1.1529,
+      "step": 13617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977454689562605e-05,
+      "loss": 0.9529,
+      "step": 13618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977451354342233e-05,
+      "loss": 1.2269,
+      "step": 13619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99774480188753e-05,
+      "loss": 1.3335,
+      "step": 13620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997744468316179e-05,
+      "loss": 1.3038,
+      "step": 13621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977441347201725e-05,
+      "loss": 1.1051,
+      "step": 13622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997743801099508e-05,
+      "loss": 1.0032,
+      "step": 13623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997743467454188e-05,
+      "loss": 1.327,
+      "step": 13624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977431337842105e-05,
+      "loss": 1.2184,
+      "step": 13625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977428000895765e-05,
+      "loss": 1.1238,
+      "step": 13626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977424663702857e-05,
+      "loss": 1.5121,
+      "step": 13627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997742132626339e-05,
+      "loss": 1.0176,
+      "step": 13628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977417988577343e-05,
+      "loss": 0.9828,
+      "step": 13629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977414650644745e-05,
+      "loss": 0.8862,
+      "step": 13630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997741131246557e-05,
+      "loss": 0.8507,
+      "step": 13631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997740797403984e-05,
+      "loss": 0.8304,
+      "step": 13632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997740463536753e-05,
+      "loss": 0.7821,
+      "step": 13633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997740129644865e-05,
+      "loss": 0.7177,
+      "step": 13634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977397957283215e-05,
+      "loss": 0.8141,
+      "step": 13635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977394617871216e-05,
+      "loss": 0.684,
+      "step": 13636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997739127821264e-05,
+      "loss": 0.7407,
+      "step": 13637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997738793830751e-05,
+      "loss": 0.973,
+      "step": 13638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977384598155805e-05,
+      "loss": 1.2165,
+      "step": 13639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997738125775754e-05,
+      "loss": 1.1518,
+      "step": 13640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977377917112697e-05,
+      "loss": 1.25,
+      "step": 13641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99773745762213e-05,
+      "loss": 0.8807,
+      "step": 13642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997737123508333e-05,
+      "loss": 1.2195,
+      "step": 13643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99773678936988e-05,
+      "loss": 1.1316,
+      "step": 13644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99773645520677e-05,
+      "loss": 1.1911,
+      "step": 13645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977361210190033e-05,
+      "loss": 0.9835,
+      "step": 13646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99773578680658e-05,
+      "loss": 0.8384,
+      "step": 13647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977354525695e-05,
+      "loss": 1.0689,
+      "step": 13648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997735118307764e-05,
+      "loss": 1.3812,
+      "step": 13649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977347840213704e-05,
+      "loss": 1.0354,
+      "step": 13650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997734449710322e-05,
+      "loss": 1.2625,
+      "step": 13651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977341153746155e-05,
+      "loss": 1.1085,
+      "step": 13652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997733781014253e-05,
+      "loss": 0.6286,
+      "step": 13653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977334466292335e-05,
+      "loss": 1.4704,
+      "step": 13654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997733112219557e-05,
+      "loss": 0.4904,
+      "step": 13655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997732777785225e-05,
+      "loss": 0.2927,
+      "step": 13656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997732443326236e-05,
+      "loss": 0.2228,
+      "step": 13657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997732108842591e-05,
+      "loss": 0.7885,
+      "step": 13658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997731774334289e-05,
+      "loss": 1.2204,
+      "step": 13659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99773143980133e-05,
+      "loss": 1.2167,
+      "step": 13660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997731105243715e-05,
+      "loss": 1.1974,
+      "step": 13661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977307706614427e-05,
+      "loss": 1.1804,
+      "step": 13662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997730436054515e-05,
+      "loss": 1.1207,
+      "step": 13663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977301014229296e-05,
+      "loss": 0.9005,
+      "step": 13664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997729766766688e-05,
+      "loss": 0.9722,
+      "step": 13665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997729432085791e-05,
+      "loss": 0.9487,
+      "step": 13666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997729097380236e-05,
+      "loss": 1.0496,
+      "step": 13667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977287626500246e-05,
+      "loss": 1.1762,
+      "step": 13668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977284278951575e-05,
+      "loss": 0.9811,
+      "step": 13669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977280931156335e-05,
+      "loss": 0.8552,
+      "step": 13670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997727758311453e-05,
+      "loss": 1.0674,
+      "step": 13671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997727423482616e-05,
+      "loss": 1.0416,
+      "step": 13672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997727088629123e-05,
+      "loss": 1.0856,
+      "step": 13673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997726753750972e-05,
+      "loss": 1.3287,
+      "step": 13674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997726418848165e-05,
+      "loss": 1.2644,
+      "step": 13675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997726083920703e-05,
+      "loss": 0.824,
+      "step": 13676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997725748968583e-05,
+      "loss": 1.0309,
+      "step": 13677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977254139918073e-05,
+      "loss": 1.0639,
+      "step": 13678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997725078990374e-05,
+      "loss": 1.3336,
+      "step": 13679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997724743964285e-05,
+      "loss": 1.3766,
+      "step": 13680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99772440891354e-05,
+      "loss": 0.897,
+      "step": 13681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997724073838138e-05,
+      "loss": 1.1617,
+      "step": 13682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977237387380796e-05,
+      "loss": 1.2439,
+      "step": 13683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997723403613364e-05,
+      "loss": 1.3029,
+      "step": 13684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997723068463993e-05,
+      "loss": 1.1788,
+      "step": 13685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997722733289964e-05,
+      "loss": 1.1932,
+      "step": 13686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99772239809128e-05,
+      "loss": 1.3462,
+      "step": 13687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977220628679395e-05,
+      "loss": 1.4004,
+      "step": 13688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977217276199426e-05,
+      "loss": 1.3485,
+      "step": 13689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997721392347289e-05,
+      "loss": 1.3562,
+      "step": 13690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977210570499784e-05,
+      "loss": 1.2233,
+      "step": 13691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997720721728011e-05,
+      "loss": 1.0757,
+      "step": 13692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977203863813884e-05,
+      "loss": 0.9933,
+      "step": 13693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997720051010108e-05,
+      "loss": 1.1821,
+      "step": 13694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997719715614173e-05,
+      "loss": 1.3159,
+      "step": 13695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997719380193581e-05,
+      "loss": 1.2139,
+      "step": 13696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997719044748331e-05,
+      "loss": 2.1006,
+      "step": 13697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977187092784264e-05,
+      "loss": 1.1615,
+      "step": 13698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997718373783864e-05,
+      "loss": 1.2347,
+      "step": 13699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997718038264646e-05,
+      "loss": 1.1515,
+      "step": 13700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997717702720772e-05,
+      "loss": 0.917,
+      "step": 13701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99771736715224e-05,
+      "loss": 0.9968,
+      "step": 13702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977170315590535e-05,
+      "loss": 1.0793,
+      "step": 13703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977166959412095e-05,
+      "loss": 0.9374,
+      "step": 13704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977163602987094e-05,
+      "loss": 1.0369,
+      "step": 13705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977160246315525e-05,
+      "loss": 1.1964,
+      "step": 13706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977156889397394e-05,
+      "loss": 0.9708,
+      "step": 13707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99771535322327e-05,
+      "loss": 1.1762,
+      "step": 13708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977150174821444e-05,
+      "loss": 1.1125,
+      "step": 13709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977146817163623e-05,
+      "loss": 1.1945,
+      "step": 13710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977143459259235e-05,
+      "loss": 0.8692,
+      "step": 13711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977140101108286e-05,
+      "loss": 0.5989,
+      "step": 13712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977136742710775e-05,
+      "loss": 0.9482,
+      "step": 13713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99771333840667e-05,
+      "loss": 0.6659,
+      "step": 13714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997713002517606e-05,
+      "loss": 1.0247,
+      "step": 13715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997712666603885e-05,
+      "loss": 1.2016,
+      "step": 13716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997712330665508e-05,
+      "loss": 0.8925,
+      "step": 13717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977119947024744e-05,
+      "loss": 1.1947,
+      "step": 13718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977116587147854e-05,
+      "loss": 1.1535,
+      "step": 13719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997711322702439e-05,
+      "loss": 1.2114,
+      "step": 13720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997710986665437e-05,
+      "loss": 0.9377,
+      "step": 13721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997710650603779e-05,
+      "loss": 1.2785,
+      "step": 13722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997710314517463e-05,
+      "loss": 1.0253,
+      "step": 13723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997709978406492e-05,
+      "loss": 1.0243,
+      "step": 13724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977096422708643e-05,
+      "loss": 1.2286,
+      "step": 13725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977093061105804e-05,
+      "loss": 1.0534,
+      "step": 13726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99770896992564e-05,
+      "loss": 1.121,
+      "step": 13727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977086337160436e-05,
+      "loss": 1.4994,
+      "step": 13728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99770829748179e-05,
+      "loss": 0.7856,
+      "step": 13729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99770796122288e-05,
+      "loss": 0.9881,
+      "step": 13730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997707624939315e-05,
+      "loss": 0.8364,
+      "step": 13731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997707288631093e-05,
+      "loss": 1.0286,
+      "step": 13732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997706952298215e-05,
+      "loss": 1.2626,
+      "step": 13733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99770661594068e-05,
+      "loss": 1.2507,
+      "step": 13734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977062795584893e-05,
+      "loss": 1.246,
+      "step": 13735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997705943151643e-05,
+      "loss": 1.1015,
+      "step": 13736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997705606720139e-05,
+      "loss": 0.9253,
+      "step": 13737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997705270263979e-05,
+      "loss": 1.2107,
+      "step": 13738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997704933783163e-05,
+      "loss": 1.1612,
+      "step": 13739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99770459727769e-05,
+      "loss": 1.1212,
+      "step": 13740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997704260747562e-05,
+      "loss": 1.1298,
+      "step": 13741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997703924192777e-05,
+      "loss": 1.1805,
+      "step": 13742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997703587613336e-05,
+      "loss": 0.9882,
+      "step": 13743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997703251009238e-05,
+      "loss": 0.9589,
+      "step": 13744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997702914380484e-05,
+      "loss": 1.1904,
+      "step": 13745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977025777270746e-05,
+      "loss": 1.1239,
+      "step": 13746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997702241049008e-05,
+      "loss": 1.209,
+      "step": 13747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997701904346286e-05,
+      "loss": 0.9033,
+      "step": 13748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977015676189065e-05,
+      "loss": 0.9875,
+      "step": 13749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997701230866871e-05,
+      "loss": 1.0884,
+      "step": 13750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977008940901804e-05,
+      "loss": 1.3568,
+      "step": 13751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997700557288832e-05,
+      "loss": 1.1509,
+      "step": 13752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977002204628284e-05,
+      "loss": 1.2721,
+      "step": 13753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997699883612169e-05,
+      "loss": 1.1617,
+      "step": 13754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997699546736853e-05,
+      "loss": 1.172,
+      "step": 13755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99769920983688e-05,
+      "loss": 1.154,
+      "step": 13756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997698872912251e-05,
+      "loss": 1.2378,
+      "step": 13757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976985359629656e-05,
+      "loss": 1.0012,
+      "step": 13758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976981989890245e-05,
+      "loss": 1.0612,
+      "step": 13759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997697861990427e-05,
+      "loss": 1.1235,
+      "step": 13760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976975249671734e-05,
+      "loss": 1.1316,
+      "step": 13761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976971879192633e-05,
+      "loss": 0.9852,
+      "step": 13762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997696850846697e-05,
+      "loss": 1.066,
+      "step": 13763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976965137494756e-05,
+      "loss": 1.265,
+      "step": 13764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976961766275965e-05,
+      "loss": 1.0337,
+      "step": 13765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997695839481062e-05,
+      "loss": 1.1193,
+      "step": 13766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997695502309871e-05,
+      "loss": 1.2403,
+      "step": 13767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997695165114023e-05,
+      "loss": 1.0786,
+      "step": 13768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976948278935204e-05,
+      "loss": 0.9102,
+      "step": 13769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997694490648361e-05,
+      "loss": 1.4613,
+      "step": 13770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997694153378545e-05,
+      "loss": 1.3508,
+      "step": 13771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997693816084073e-05,
+      "loss": 1.226,
+      "step": 13772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997693478764945e-05,
+      "loss": 1.1583,
+      "step": 13773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976931414211604e-05,
+      "loss": 0.9677,
+      "step": 13774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99769280405272e-05,
+      "loss": 1.0938,
+      "step": 13775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997692466659624e-05,
+      "loss": 1.1531,
+      "step": 13776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997692129241871e-05,
+      "loss": 1.0068,
+      "step": 13777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976917917994615e-05,
+      "loss": 1.1943,
+      "step": 13778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997691454332397e-05,
+      "loss": 1.0498,
+      "step": 13779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976911168406754e-05,
+      "loss": 1.2163,
+      "step": 13780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997690779324298e-05,
+      "loss": 1.2518,
+      "step": 13781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997690441783265e-05,
+      "loss": 1.278,
+      "step": 13782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997690104217575e-05,
+      "loss": 1.1028,
+      "step": 13783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976897666272285e-05,
+      "loss": 0.6773,
+      "step": 13784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976894290122265e-05,
+      "loss": 1.1019,
+      "step": 13785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997689091372569e-05,
+      "loss": 1.1219,
+      "step": 13786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997688753708255e-05,
+      "loss": 1.1479,
+      "step": 13787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997688416019284e-05,
+      "loss": 1.0981,
+      "step": 13788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997688078305658e-05,
+      "loss": 0.9778,
+      "step": 13789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997687740567375e-05,
+      "loss": 1.0922,
+      "step": 13790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997687402804436e-05,
+      "loss": 1.0741,
+      "step": 13791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997687065016841e-05,
+      "loss": 0.9276,
+      "step": 13792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976867272045904e-05,
+      "loss": 1.2373,
+      "step": 13793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997686389367683e-05,
+      "loss": 0.9606,
+      "step": 13794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976860515061196e-05,
+      "loss": 1.0542,
+      "step": 13795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997685713619901e-05,
+      "loss": 1.1851,
+      "step": 13796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997685375709025e-05,
+      "loss": 1.0827,
+      "step": 13797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976850377734935e-05,
+      "loss": 0.9717,
+      "step": 13798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997684699813306e-05,
+      "loss": 1.2958,
+      "step": 13799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997684361828462e-05,
+      "loss": 0.897,
+      "step": 13800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976840238189624e-05,
+      "loss": 1.0937,
+      "step": 13801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976836857848065e-05,
+      "loss": 0.9241,
+      "step": 13802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976833477259944e-05,
+      "loss": 0.94,
+      "step": 13803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997683009642526e-05,
+      "loss": 1.1254,
+      "step": 13804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997682671534402e-05,
+      "loss": 0.9907,
+      "step": 13805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997682333401622e-05,
+      "loss": 1.2535,
+      "step": 13806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997681995244186e-05,
+      "loss": 1.0171,
+      "step": 13807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997681657062093e-05,
+      "loss": 0.9388,
+      "step": 13808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997681318855345e-05,
+      "loss": 0.8717,
+      "step": 13809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976809806239404e-05,
+      "loss": 1.3064,
+      "step": 13810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99768064236788e-05,
+      "loss": 1.0957,
+      "step": 13811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997680304087163e-05,
+      "loss": 1.0988,
+      "step": 13812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976799657817906e-05,
+      "loss": 1.0316,
+      "step": 13813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976796274517627e-05,
+      "loss": 0.9699,
+      "step": 13814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997679289097077e-05,
+      "loss": 0.864,
+      "step": 13815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997678950717737e-05,
+      "loss": 0.9019,
+      "step": 13816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976786123137395e-05,
+      "loss": 1.2512,
+      "step": 13817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997678273885087e-05,
+      "loss": 1.2744,
+      "step": 13818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997677935431778e-05,
+      "loss": 1.0352,
+      "step": 13819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976775969538134e-05,
+      "loss": 1.2294,
+      "step": 13820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997677258451193e-05,
+      "loss": 1.4391,
+      "step": 13821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997676919923916e-05,
+      "loss": 0.9864,
+      "step": 13822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997676581371983e-05,
+      "loss": 1.2285,
+      "step": 13823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997676242795394e-05,
+      "loss": 1.2572,
+      "step": 13824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997675904194149e-05,
+      "loss": 1.0138,
+      "step": 13825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997675565568248e-05,
+      "loss": 0.3226,
+      "step": 13826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976752269176914e-05,
+      "loss": 1.0268,
+      "step": 13827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976748882424786e-05,
+      "loss": 1.1284,
+      "step": 13828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997674549542609e-05,
+      "loss": 1.2994,
+      "step": 13829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976742108180846e-05,
+      "loss": 0.5558,
+      "step": 13830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976738720689034e-05,
+      "loss": 0.984,
+      "step": 13831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997673533295067e-05,
+      "loss": 1.2748,
+      "step": 13832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976731944965735e-05,
+      "loss": 1.2244,
+      "step": 13833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997672855673425e-05,
+      "loss": 1.4405,
+      "step": 13834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99767251682562e-05,
+      "loss": 1.2221,
+      "step": 13835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99767217795316e-05,
+      "loss": 1.0759,
+      "step": 13836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997671839056043e-05,
+      "loss": 0.7912,
+      "step": 13837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99767150013427e-05,
+      "loss": 0.9559,
+      "step": 13838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976711611878416e-05,
+      "loss": 1.0427,
+      "step": 13839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997670822216757e-05,
+      "loss": 1.1443,
+      "step": 13840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976704832210165e-05,
+      "loss": 0.9395,
+      "step": 13841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976701442006195e-05,
+      "loss": 1.1513,
+      "step": 13842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997669805155567e-05,
+      "loss": 1.2022,
+      "step": 13843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997669466085859e-05,
+      "loss": 1.0524,
+      "step": 13844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976691269914944e-05,
+      "loss": 1.107,
+      "step": 13845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997668787872474e-05,
+      "loss": 0.8735,
+      "step": 13846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997668448728799e-05,
+      "loss": 1.0215,
+      "step": 13847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997668109560466e-05,
+      "loss": 1.1086,
+      "step": 13848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997667770367478e-05,
+      "loss": 1.0567,
+      "step": 13849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997667431149834e-05,
+      "loss": 1.3272,
+      "step": 13850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997667091907534e-05,
+      "loss": 1.423,
+      "step": 13851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997666752640578e-05,
+      "loss": 0.9504,
+      "step": 13852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997666413348966e-05,
+      "loss": 1.1183,
+      "step": 13853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976660740326985e-05,
+      "loss": 0.8809,
+      "step": 13854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997665734691775e-05,
+      "loss": 1.2235,
+      "step": 13855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976653953261955e-05,
+      "loss": 1.0271,
+      "step": 13856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99766505593596e-05,
+      "loss": 1.1365,
+      "step": 13857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997664716521069e-05,
+      "loss": 1.0548,
+      "step": 13858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997664377081521e-05,
+      "loss": 1.1949,
+      "step": 13859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997664037617319e-05,
+      "loss": 0.6866,
+      "step": 13860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99766369812846e-05,
+      "loss": 0.8584,
+      "step": 13861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976633586149444e-05,
+      "loss": 1.0817,
+      "step": 13862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997663019076774e-05,
+      "loss": 1.0412,
+      "step": 13863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997662679513948e-05,
+      "loss": 0.7251,
+      "step": 13864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997662339926465e-05,
+      "loss": 1.0131,
+      "step": 13865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976620003143265e-05,
+      "loss": 1.0863,
+      "step": 13866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997661660677533e-05,
+      "loss": 1.1812,
+      "step": 13867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997661321016083e-05,
+      "loss": 0.8374,
+      "step": 13868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997660981329977e-05,
+      "loss": 1.0997,
+      "step": 13869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976606416192154e-05,
+      "loss": 1.0978,
+      "step": 13870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997660301883797e-05,
+      "loss": 1.1241,
+      "step": 13871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976599621237236e-05,
+      "loss": 0.9462,
+      "step": 13872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976596223389945e-05,
+      "loss": 1.0272,
+      "step": 13873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976592825296086e-05,
+      "loss": 1.0391,
+      "step": 13874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997658942695568e-05,
+      "loss": 1.1377,
+      "step": 13875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997658602836871e-05,
+      "loss": 0.995,
+      "step": 13876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976582629535184e-05,
+      "loss": 1.0341,
+      "step": 13877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99765792304551e-05,
+      "loss": 1.0076,
+      "step": 13878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997657583112846e-05,
+      "loss": 0.9456,
+      "step": 13879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997657243155526e-05,
+      "loss": 1.1291,
+      "step": 13880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99765690317355e-05,
+      "loss": 0.8903,
+      "step": 13881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997656563166918e-05,
+      "loss": 1.1877,
+      "step": 13882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99765622313563e-05,
+      "loss": 1.0815,
+      "step": 13883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997655883079687e-05,
+      "loss": 0.9905,
+      "step": 13884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997655542999087e-05,
+      "loss": 1.0751,
+      "step": 13885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976552028938326e-05,
+      "loss": 0.8339,
+      "step": 13886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997654862763922e-05,
+      "loss": 1.188,
+      "step": 13887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997654522609355e-05,
+      "loss": 1.1361,
+      "step": 13888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976541824301324e-05,
+      "loss": 1.0927,
+      "step": 13889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997653842226254e-05,
+      "loss": 1.1153,
+      "step": 13890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99765350199772e-05,
+      "loss": 1.0066,
+      "step": 13891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997653161744531e-05,
+      "loss": 0.7331,
+      "step": 13892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997652821466685e-05,
+      "loss": 0.8633,
+      "step": 13893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976524811641834e-05,
+      "loss": 1.037,
+      "step": 13894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997652140837027e-05,
+      "loss": 0.7927,
+      "step": 13895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997651800485214e-05,
+      "loss": 0.9617,
+      "step": 13896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976514601087454e-05,
+      "loss": 0.7226,
+      "step": 13897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99765111970762e-05,
+      "loss": 1.095,
+      "step": 13898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976507792818404e-05,
+      "loss": 1.0521,
+      "step": 13899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976504388314044e-05,
+      "loss": 1.1347,
+      "step": 13900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997650098356313e-05,
+      "loss": 1.0732,
+      "step": 13901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997649757856565e-05,
+      "loss": 1.0931,
+      "step": 13902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997649417332162e-05,
+      "loss": 1.1045,
+      "step": 13903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997649076783103e-05,
+      "loss": 1.1639,
+      "step": 13904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997648736209389e-05,
+      "loss": 0.9362,
+      "step": 13905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997648395611018e-05,
+      "loss": 1.2209,
+      "step": 13906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997648054987992e-05,
+      "loss": 1.1402,
+      "step": 13907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99764771434031e-05,
+      "loss": 1.189,
+      "step": 13908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997647373667973e-05,
+      "loss": 1.1464,
+      "step": 13909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976470329709794e-05,
+      "loss": 1.0142,
+      "step": 13910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99764669224933e-05,
+      "loss": 1.0851,
+      "step": 13911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976463515030256e-05,
+      "loss": 1.3289,
+      "step": 13912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976460107320656e-05,
+      "loss": 1.1941,
+      "step": 13913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976456699364494e-05,
+      "loss": 0.8055,
+      "step": 13914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997645329116177e-05,
+      "loss": 0.929,
+      "step": 13915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976449882712495e-05,
+      "loss": 1.5083,
+      "step": 13916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997644647401666e-05,
+      "loss": 1.3804,
+      "step": 13917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997644306507428e-05,
+      "loss": 1.2218,
+      "step": 13918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976439655885324e-05,
+      "loss": 1.1761,
+      "step": 13919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997643624644982e-05,
+      "loss": 0.9478,
+      "step": 13920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997643283676776e-05,
+      "loss": 1.2176,
+      "step": 13921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976429426839144e-05,
+      "loss": 1.4068,
+      "step": 13922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997642601666397e-05,
+      "loss": 1.0524,
+      "step": 13923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997642260624225e-05,
+      "loss": 1.092,
+      "step": 13924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976419195573955e-05,
+      "loss": 0.937,
+      "step": 13925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976415784659114e-05,
+      "loss": 1.1258,
+      "step": 13926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997641237349771e-05,
+      "loss": 1.0027,
+      "step": 13927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976408962089756e-05,
+      "loss": 1.1325,
+      "step": 13928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997640555043524e-05,
+      "loss": 1.1987,
+      "step": 13929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976402138534173e-05,
+      "loss": 1.3612,
+      "step": 13930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997639872638654e-05,
+      "loss": 1.2419,
+      "step": 13931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997639531399236e-05,
+      "loss": 1.1688,
+      "step": 13932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997639190135162e-05,
+      "loss": 1.1912,
+      "step": 13933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997638848846432e-05,
+      "loss": 1.076,
+      "step": 13934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976385075330466e-05,
+      "loss": 1.4147,
+      "step": 13935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997638166195006e-05,
+      "loss": 0.9483,
+      "step": 13936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976378248323095e-05,
+      "loss": 1.2198,
+      "step": 13937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976374834449575e-05,
+      "loss": 0.9229,
+      "step": 13938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976371420329494e-05,
+      "loss": 1.1425,
+      "step": 13939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997636800596286e-05,
+      "loss": 1.1703,
+      "step": 13940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976364591349675e-05,
+      "loss": 1.1516,
+      "step": 13941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976361176489924e-05,
+      "loss": 0.9866,
+      "step": 13942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976357761383626e-05,
+      "loss": 1.1424,
+      "step": 13943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976354346030766e-05,
+      "loss": 1.1475,
+      "step": 13944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976350930431345e-05,
+      "loss": 0.7793,
+      "step": 13945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976347514585377e-05,
+      "loss": 0.8283,
+      "step": 13946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976344098492854e-05,
+      "loss": 1.0096,
+      "step": 13947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997634068215376e-05,
+      "loss": 0.7455,
+      "step": 13948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997633726556813e-05,
+      "loss": 1.2219,
+      "step": 13949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997633384873593e-05,
+      "loss": 1.2689,
+      "step": 13950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976330431657185e-05,
+      "loss": 0.9038,
+      "step": 13951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976327014331884e-05,
+      "loss": 0.9423,
+      "step": 13952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997632359676002e-05,
+      "loss": 1.2045,
+      "step": 13953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976320178941605e-05,
+      "loss": 1.2527,
+      "step": 13954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997631676087663e-05,
+      "loss": 1.2215,
+      "step": 13955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99763133425651e-05,
+      "loss": 0.9627,
+      "step": 13956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976309924007015e-05,
+      "loss": 1.0087,
+      "step": 13957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976306505202374e-05,
+      "loss": 0.7504,
+      "step": 13958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997630308615118e-05,
+      "loss": 1.0684,
+      "step": 13959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997629966685343e-05,
+      "loss": 1.1437,
+      "step": 13960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997629624730912e-05,
+      "loss": 1.1618,
+      "step": 13961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997629282751826e-05,
+      "loss": 1.2139,
+      "step": 13962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997628940748084e-05,
+      "loss": 1.0534,
+      "step": 13963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976285987196866e-05,
+      "loss": 1.0269,
+      "step": 13964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976282566666345e-05,
+      "loss": 1.1003,
+      "step": 13965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997627914588926e-05,
+      "loss": 1.0463,
+      "step": 13966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997627572486562e-05,
+      "loss": 1.1097,
+      "step": 13967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997627230359542e-05,
+      "loss": 1.1212,
+      "step": 13968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997626888207867e-05,
+      "loss": 1.0862,
+      "step": 13969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997626546031537e-05,
+      "loss": 1.2666,
+      "step": 13970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997626203830551e-05,
+      "loss": 1.1089,
+      "step": 13971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997625861604909e-05,
+      "loss": 1.1409,
+      "step": 13972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997625519354612e-05,
+      "loss": 1.0216,
+      "step": 13973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99762517707966e-05,
+      "loss": 1.1534,
+      "step": 13974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997624834780051e-05,
+      "loss": 1.0097,
+      "step": 13975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997624492455788e-05,
+      "loss": 0.8927,
+      "step": 13976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976241501068686e-05,
+      "loss": 1.0883,
+      "step": 13977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997623807733295e-05,
+      "loss": 1.0862,
+      "step": 13978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997623465335065e-05,
+      "loss": 1.1607,
+      "step": 13979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99762312291218e-05,
+      "loss": 1.1474,
+      "step": 13980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997622780464638e-05,
+      "loss": 1.0373,
+      "step": 13981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997622437992442e-05,
+      "loss": 0.5321,
+      "step": 13982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99762209549559e-05,
+      "loss": 0.8732,
+      "step": 13983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997621752974082e-05,
+      "loss": 1.2515,
+      "step": 13984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99762141042792e-05,
+      "loss": 1.18,
+      "step": 13985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976210678571015e-05,
+      "loss": 0.95,
+      "step": 13986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997620725261627e-05,
+      "loss": 1.0746,
+      "step": 13987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997620382641498e-05,
+      "loss": 0.9309,
+      "step": 13988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997620039996713e-05,
+      "loss": 1.0832,
+      "step": 13989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997619697327273e-05,
+      "loss": 1.0483,
+      "step": 13990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997619354633177e-05,
+      "loss": 1.0112,
+      "step": 13991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997619011914426e-05,
+      "loss": 1.1311,
+      "step": 13992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976186691710204e-05,
+      "loss": 1.0354,
+      "step": 13993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997618326402958e-05,
+      "loss": 1.3114,
+      "step": 13994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997617983610241e-05,
+      "loss": 1.0396,
+      "step": 13995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976176407928674e-05,
+      "loss": 1.159,
+      "step": 13996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997617297950839e-05,
+      "loss": 0.9205,
+      "step": 13997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997616955084156e-05,
+      "loss": 1.0981,
+      "step": 13998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997616612192817e-05,
+      "loss": 1.3084,
+      "step": 13999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976162692768224e-05,
+      "loss": 1.3947,
+      "step": 14000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0414711236953735,
+      "eval_runtime": 85.5056,
+      "eval_samples_per_second": 16.198,
+      "eval_steps_per_second": 4.058,
+      "step": 14000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997615926336172e-05,
+      "loss": 1.2238,
+      "step": 14001
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997615583370867e-05,
+      "loss": 1.2037,
+      "step": 14002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997615240380906e-05,
+      "loss": 1.175,
+      "step": 14003
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99761489736629e-05,
+      "loss": 1.0471,
+      "step": 14004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997614554327019e-05,
+      "loss": 0.9915,
+      "step": 14005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997614211263092e-05,
+      "loss": 0.9342,
+      "step": 14006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976138681745094e-05,
+      "loss": 1.1484,
+      "step": 14007
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976135250612715e-05,
+      "loss": 1.0165,
+      "step": 14008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997613181923378e-05,
+      "loss": 0.9837,
+      "step": 14009
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99761283876083e-05,
+      "loss": 0.9486,
+      "step": 14010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997612495573626e-05,
+      "loss": 1.0191,
+      "step": 14011
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997612152361767e-05,
+      "loss": 1.5051,
+      "step": 14012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997611809125252e-05,
+      "loss": 1.4434,
+      "step": 14013
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997611465864082e-05,
+      "loss": 1.3862,
+      "step": 14014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997611122578256e-05,
+      "loss": 1.4054,
+      "step": 14015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997610779267776e-05,
+      "loss": 1.4269,
+      "step": 14016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99761043593264e-05,
+      "loss": 1.4767,
+      "step": 14017
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976100925728485e-05,
+      "loss": 1.5355,
+      "step": 14018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997609749188401e-05,
+      "loss": 1.3726,
+      "step": 14019
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997609405779299e-05,
+      "loss": 1.4025,
+      "step": 14020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997609062345542e-05,
+      "loss": 1.4298,
+      "step": 14021
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997608718887129e-05,
+      "loss": 1.4052,
+      "step": 14022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976083754040606e-05,
+      "loss": 1.2636,
+      "step": 14023
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997608031896337e-05,
+      "loss": 0.9864,
+      "step": 14024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997607688363958e-05,
+      "loss": 1.2486,
+      "step": 14025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997607344806924e-05,
+      "loss": 1.2348,
+      "step": 14026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997607001225235e-05,
+      "loss": 1.0257,
+      "step": 14027
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976066576188895e-05,
+      "loss": 1.235,
+      "step": 14028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976063139878895e-05,
+      "loss": 1.1075,
+      "step": 14029
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997605970332234e-05,
+      "loss": 1.0367,
+      "step": 14030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997605626651923e-05,
+      "loss": 0.9863,
+      "step": 14031
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997605282946957e-05,
+      "loss": 1.1902,
+      "step": 14032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997604939217335e-05,
+      "loss": 0.9979,
+      "step": 14033
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997604595463059e-05,
+      "loss": 1.7464,
+      "step": 14034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997604251684127e-05,
+      "loss": 1.6848,
+      "step": 14035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976039078805395e-05,
+      "loss": 1.3606,
+      "step": 14036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997603564052297e-05,
+      "loss": 0.3411,
+      "step": 14037
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997603220199399e-05,
+      "loss": 0.2637,
+      "step": 14038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976028763218455e-05,
+      "loss": 0.2964,
+      "step": 14039
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997602532419637e-05,
+      "loss": 0.2212,
+      "step": 14040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997602188492774e-05,
+      "loss": 0.3276,
+      "step": 14041
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997601844541254e-05,
+      "loss": 0.339,
+      "step": 14042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99760150056508e-05,
+      "loss": 0.2881,
+      "step": 14043
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976011565642504e-05,
+      "loss": 0.191,
+      "step": 14044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976008125387655e-05,
+      "loss": 0.1706,
+      "step": 14045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997600468488626e-05,
+      "loss": 0.2245,
+      "step": 14046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997600124413831e-05,
+      "loss": 0.27,
+      "step": 14047
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975997803143795e-05,
+      "loss": 0.2916,
+      "step": 14048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975994361902735e-05,
+      "loss": 0.2542,
+      "step": 14049
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997599092041513e-05,
+      "loss": 0.1982,
+      "step": 14050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975987478680966e-05,
+      "loss": 0.4146,
+      "step": 14051
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975984036700244e-05,
+      "loss": 0.6991,
+      "step": 14052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997598059447297e-05,
+      "loss": 0.6499,
+      "step": 14053
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975977151999155e-05,
+      "loss": 0.3083,
+      "step": 14054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997597370927878e-05,
+      "loss": 0.2072,
+      "step": 14055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975970266311856e-05,
+      "loss": 0.4152,
+      "step": 14056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997596682309838e-05,
+      "loss": 0.7276,
+      "step": 14057
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997596337963835e-05,
+      "loss": 1.0309,
+      "step": 14058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997595993593177e-05,
+      "loss": 1.3804,
+      "step": 14059
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997595649197864e-05,
+      "loss": 2.1518,
+      "step": 14060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997595304777894e-05,
+      "loss": 1.1852,
+      "step": 14061
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99759496033327e-05,
+      "loss": 0.7476,
+      "step": 14062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975946158639916e-05,
+      "loss": 0.4246,
+      "step": 14063
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975942713700575e-05,
+      "loss": 0.7485,
+      "step": 14064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997593926851468e-05,
+      "loss": 0.3086,
+      "step": 14065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997593582308223e-05,
+      "loss": 0.2025,
+      "step": 14066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997593237740323e-05,
+      "loss": 0.1456,
+      "step": 14067
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997592893147769e-05,
+      "loss": 1.0141,
+      "step": 14068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997592548530559e-05,
+      "loss": 1.0374,
+      "step": 14069
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997592203888693e-05,
+      "loss": 1.0126,
+      "step": 14070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975918592221726e-05,
+      "loss": 1.0492,
+      "step": 14071
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975915145309964e-05,
+      "loss": 1.3127,
+      "step": 14072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975911698151654e-05,
+      "loss": 1.4635,
+      "step": 14073
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975908250746796e-05,
+      "loss": 1.2857,
+      "step": 14074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975904803095384e-05,
+      "loss": 0.9974,
+      "step": 14075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997590135519742e-05,
+      "loss": 1.5194,
+      "step": 14076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975897907052904e-05,
+      "loss": 1.1631,
+      "step": 14077
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975894458661836e-05,
+      "loss": 1.0396,
+      "step": 14078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975891010024214e-05,
+      "loss": 1.4006,
+      "step": 14079
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975887561140044e-05,
+      "loss": 1.1684,
+      "step": 14080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975884112009326e-05,
+      "loss": 1.2567,
+      "step": 14081
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975880662632054e-05,
+      "loss": 1.0493,
+      "step": 14082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997587721300823e-05,
+      "loss": 1.4118,
+      "step": 14083
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997587376313785e-05,
+      "loss": 1.0582,
+      "step": 14084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975870313020926e-05,
+      "loss": 0.9975,
+      "step": 14085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997586686265745e-05,
+      "loss": 1.1511,
+      "step": 14086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997586341204742e-05,
+      "loss": 0.9706,
+      "step": 14087
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975859961190835e-05,
+      "loss": 1.076,
+      "step": 14088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99758565100877e-05,
+      "loss": 1.3168,
+      "step": 14089
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997585305873802e-05,
+      "loss": 0.8597,
+      "step": 14090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997584960714179e-05,
+      "loss": 0.7687,
+      "step": 14091
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975846155299e-05,
+      "loss": 0.7132,
+      "step": 14092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975842703209665e-05,
+      "loss": 0.6899,
+      "step": 14093
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997583925087378e-05,
+      "loss": 0.668,
+      "step": 14094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997583579829134e-05,
+      "loss": 0.7043,
+      "step": 14095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975832345462345e-05,
+      "loss": 0.6641,
+      "step": 14096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997582889238681e-05,
+      "loss": 0.5732,
+      "step": 14097
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997582543906472e-05,
+      "loss": 0.6086,
+      "step": 14098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975821985496076e-05,
+      "loss": 0.5433,
+      "step": 14099
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975818531680886e-05,
+      "loss": 0.78,
+      "step": 14100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997581507761914e-05,
+      "loss": 1.2859,
+      "step": 14101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975811623310845e-05,
+      "loss": 0.9706,
+      "step": 14102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975808168756e-05,
+      "loss": 1.0174,
+      "step": 14103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997580471395461e-05,
+      "loss": 0.9041,
+      "step": 14104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975801258906665e-05,
+      "loss": 1.0138,
+      "step": 14105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997579780361217e-05,
+      "loss": 1.1986,
+      "step": 14106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975794348071114e-05,
+      "loss": 1.0084,
+      "step": 14107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997579089228352e-05,
+      "loss": 1.2811,
+      "step": 14108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997578743624937e-05,
+      "loss": 1.0046,
+      "step": 14109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975783979968673e-05,
+      "loss": 1.2217,
+      "step": 14110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997578052344143e-05,
+      "loss": 1.2312,
+      "step": 14111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975777066667626e-05,
+      "loss": 1.3594,
+      "step": 14112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997577360964728e-05,
+      "loss": 0.9843,
+      "step": 14113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975770152380375e-05,
+      "loss": 1.0723,
+      "step": 14114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975766694866924e-05,
+      "loss": 1.1024,
+      "step": 14115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975763237106927e-05,
+      "loss": 1.0341,
+      "step": 14116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975759779100374e-05,
+      "loss": 0.527,
+      "step": 14117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975756320847274e-05,
+      "loss": 0.4008,
+      "step": 14118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997575286234763e-05,
+      "loss": 0.41,
+      "step": 14119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975749403601425e-05,
+      "loss": 0.3649,
+      "step": 14120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997574594460867e-05,
+      "loss": 0.2659,
+      "step": 14121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997574248536937e-05,
+      "loss": 0.1907,
+      "step": 14122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975739025883514e-05,
+      "loss": 0.25,
+      "step": 14123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975735566151115e-05,
+      "loss": 1.0159,
+      "step": 14124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997573210617217e-05,
+      "loss": 1.2458,
+      "step": 14125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997572864594666e-05,
+      "loss": 1.1723,
+      "step": 14126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975725185474606e-05,
+      "loss": 1.1799,
+      "step": 14127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997572172475601e-05,
+      "loss": 1.0264,
+      "step": 14128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997571826379086e-05,
+      "loss": 1.2867,
+      "step": 14129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997571480257916e-05,
+      "loss": 1.1588,
+      "step": 14130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997571134112091e-05,
+      "loss": 1.269,
+      "step": 14131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997570787941611e-05,
+      "loss": 1.105,
+      "step": 14132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997570441746476e-05,
+      "loss": 1.1432,
+      "step": 14133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997570095526686e-05,
+      "loss": 1.0189,
+      "step": 14134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997569749282241e-05,
+      "loss": 1.552,
+      "step": 14135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975694030131415e-05,
+      "loss": 1.186,
+      "step": 14136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975690567193864e-05,
+      "loss": 0.9589,
+      "step": 14137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997568710400977e-05,
+      "loss": 1.0039,
+      "step": 14138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975683640579126e-05,
+      "loss": 0.9679,
+      "step": 14139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997568017690193e-05,
+      "loss": 1.2845,
+      "step": 14140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975676712978183e-05,
+      "loss": 1.432,
+      "step": 14141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997567324880789e-05,
+      "loss": 1.1023,
+      "step": 14142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975669784391044e-05,
+      "loss": 1.1681,
+      "step": 14143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975666319727646e-05,
+      "loss": 1.1895,
+      "step": 14144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997566285481771e-05,
+      "loss": 1.0524,
+      "step": 14145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997565938966121e-05,
+      "loss": 1.0158,
+      "step": 14146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975655924258175e-05,
+      "loss": 0.8595,
+      "step": 14147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997565245860858e-05,
+      "loss": 1.0239,
+      "step": 14148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975648992712444e-05,
+      "loss": 1.0904,
+      "step": 14149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997564552656976e-05,
+      "loss": 1.1,
+      "step": 14150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975642060180517e-05,
+      "loss": 0.9147,
+      "step": 14151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997563859354473e-05,
+      "loss": 0.966,
+      "step": 14152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99756351266624e-05,
+      "loss": 1.1773,
+      "step": 14153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997563165953351e-05,
+      "loss": 0.8457,
+      "step": 14154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997562819215808e-05,
+      "loss": 0.676,
+      "step": 14155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975624724536095e-05,
+      "loss": 1.0086,
+      "step": 14156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975621256667565e-05,
+      "loss": 0.637,
+      "step": 14157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997561778855249e-05,
+      "loss": 0.7949,
+      "step": 14158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975614320190856e-05,
+      "loss": 1.2112,
+      "step": 14159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997561085158268e-05,
+      "loss": 1.006,
+      "step": 14160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997560738272795e-05,
+      "loss": 1.1773,
+      "step": 14161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975603913626675e-05,
+      "loss": 1.1284,
+      "step": 14162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975600444278853e-05,
+      "loss": 1.0709,
+      "step": 14163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997559697468448e-05,
+      "loss": 0.9037,
+      "step": 14164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997559350484356e-05,
+      "loss": 1.2146,
+      "step": 14165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975590034756095e-05,
+      "loss": 1.0601,
+      "step": 14166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997558656442207e-05,
+      "loss": 1.1863,
+      "step": 14167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99755830938415e-05,
+      "loss": 1.3709,
+      "step": 14168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997557962301439e-05,
+      "loss": 1.3823,
+      "step": 14169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997557615194073e-05,
+      "loss": 1.089,
+      "step": 14170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997557268062052e-05,
+      "loss": 1.1908,
+      "step": 14171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997556920905376e-05,
+      "loss": 1.1416,
+      "step": 14172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997556573724045e-05,
+      "loss": 0.6685,
+      "step": 14173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99755622651806e-05,
+      "loss": 1.248,
+      "step": 14174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997555879287419e-05,
+      "loss": 0.8823,
+      "step": 14175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997555532032124e-05,
+      "loss": 1.2691,
+      "step": 14176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997555184752174e-05,
+      "loss": 0.9468,
+      "step": 14177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975548374475686e-05,
+      "loss": 1.1383,
+      "step": 14178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997554490118309e-05,
+      "loss": 1.1812,
+      "step": 14179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975541427643945e-05,
+      "loss": 1.1472,
+      "step": 14180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997553795385825e-05,
+      "loss": 1.1894,
+      "step": 14181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975534479826014e-05,
+      "loss": 1.0304,
+      "step": 14182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975531005547227e-05,
+      "loss": 1.3028,
+      "step": 14183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975527531021885e-05,
+      "loss": 0.9467,
+      "step": 14184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997552405625e-05,
+      "loss": 1.227,
+      "step": 14185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997552058123157e-05,
+      "loss": 1.0007,
+      "step": 14186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997551710596658e-05,
+      "loss": 1.0721,
+      "step": 14187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975513630455064e-05,
+      "loss": 1.1782,
+      "step": 14188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997551015469698e-05,
+      "loss": 1.1244,
+      "step": 14189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975506678692365e-05,
+      "loss": 1.1925,
+      "step": 14190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997550320244119e-05,
+      "loss": 1.1091,
+      "step": 14191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997549972594347e-05,
+      "loss": 1.0063,
+      "step": 14192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975496249199206e-05,
+      "loss": 1.0792,
+      "step": 14193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997549277220839e-05,
+      "loss": 0.9771,
+      "step": 14194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975489294971024e-05,
+      "loss": 1.3387,
+      "step": 14195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997548581748712e-05,
+      "loss": 1.0673,
+      "step": 14196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975482339756665e-05,
+      "loss": 0.997,
+      "step": 14197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997547886177966e-05,
+      "loss": 1.1778,
+      "step": 14198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99754753835561e-05,
+      "loss": 1.0958,
+      "step": 14199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975471905086e-05,
+      "loss": 1.2378,
+      "step": 14200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997546842636936e-05,
+      "loss": 1.1805,
+      "step": 14201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997546494740617e-05,
+      "loss": 1.2093,
+      "step": 14202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997546146819643e-05,
+      "loss": 1.1227,
+      "step": 14203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975457988740135e-05,
+      "loss": 1.2625,
+      "step": 14204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99754545090373e-05,
+      "loss": 1.0928,
+      "step": 14205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997545102908791e-05,
+      "loss": 1.148,
+      "step": 14206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997544754889199e-05,
+      "loss": 0.9633,
+      "step": 14207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975444068449515e-05,
+      "loss": 1.0229,
+      "step": 14208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997544058776049e-05,
+      "loss": 1.1975,
+      "step": 14209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997543710682492e-05,
+      "loss": 0.9083,
+      "step": 14210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99754336256428e-05,
+      "loss": 1.2467,
+      "step": 14211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975430144214134e-05,
+      "loss": 1.0118,
+      "step": 14212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975426662538926e-05,
+      "loss": 0.7819,
+      "step": 14213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975423180617164e-05,
+      "loss": 1.1299,
+      "step": 14214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997541969844886e-05,
+      "loss": 1.0817,
+      "step": 14215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997541621603401e-05,
+      "loss": 1.1529,
+      "step": 14216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997541273337261e-05,
+      "loss": 1.2344,
+      "step": 14217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997540925046467e-05,
+      "loss": 0.7974,
+      "step": 14218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997540576731017e-05,
+      "loss": 1.2864,
+      "step": 14219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975402283909135e-05,
+      "loss": 0.9274,
+      "step": 14220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997539880026155e-05,
+      "loss": 1.1259,
+      "step": 14221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997539531636741e-05,
+      "loss": 0.8732,
+      "step": 14222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975391832226735e-05,
+      "loss": 1.1752,
+      "step": 14223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997538834783951e-05,
+      "loss": 0.9616,
+      "step": 14224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997538486320574e-05,
+      "loss": 0.9051,
+      "step": 14225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997538137832542e-05,
+      "loss": 1.0937,
+      "step": 14226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997537789319855e-05,
+      "loss": 1.0993,
+      "step": 14227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975374407825146e-05,
+      "loss": 1.1841,
+      "step": 14228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975370922205184e-05,
+      "loss": 0.7506,
+      "step": 14229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997536743633868e-05,
+      "loss": 1.4471,
+      "step": 14230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997536395022563e-05,
+      "loss": 0.5568,
+      "step": 14231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997536046386603e-05,
+      "loss": 0.832,
+      "step": 14232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997535697725989e-05,
+      "loss": 1.2028,
+      "step": 14233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99753534904072e-05,
+      "loss": 1.0788,
+      "step": 14234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997535000330796e-05,
+      "loss": 0.7651,
+      "step": 14235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975346515962186e-05,
+      "loss": 0.9412,
+      "step": 14236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997534302836986e-05,
+      "loss": 1.28,
+      "step": 14237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997533954053099e-05,
+      "loss": 1.1412,
+      "step": 14238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975336052445564e-05,
+      "loss": 0.7331,
+      "step": 14239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99753325641136e-05,
+      "loss": 0.9254,
+      "step": 14240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975329075535086e-05,
+      "loss": 0.9398,
+      "step": 14241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997532558671003e-05,
+      "loss": 1.412,
+      "step": 14242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975322097638426e-05,
+      "loss": 1.1725,
+      "step": 14243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997531860832028e-05,
+      "loss": 1.2248,
+      "step": 14244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997531511875558e-05,
+      "loss": 1.0279,
+      "step": 14245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975311628944344e-05,
+      "loss": 1.0378,
+      "step": 14246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997530813888655e-05,
+      "loss": 1.2247,
+      "step": 14247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997530464858222e-05,
+      "loss": 0.4484,
+      "step": 14248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997530115803135e-05,
+      "loss": 0.9846,
+      "step": 14249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997529766723392e-05,
+      "loss": 0.8921,
+      "step": 14250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975294176189954e-05,
+      "loss": 1.0155,
+      "step": 14251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997529068489943e-05,
+      "loss": 0.9831,
+      "step": 14252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975287193362375e-05,
+      "loss": 1.0879,
+      "step": 14253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975283701578765e-05,
+      "loss": 1.1171,
+      "step": 14254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997528020954862e-05,
+      "loss": 1.0931,
+      "step": 14255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997527671727192e-05,
+      "loss": 1.1142,
+      "step": 14256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997527322474868e-05,
+      "loss": 1.0804,
+      "step": 14257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975269731978894e-05,
+      "loss": 1.1162,
+      "step": 14258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997526623896256e-05,
+      "loss": 1.1853,
+      "step": 14259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997526274569968e-05,
+      "loss": 1.082,
+      "step": 14260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997525925219026e-05,
+      "loss": 0.9478,
+      "step": 14261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997525575843429e-05,
+      "loss": 1.033,
+      "step": 14262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997525226443178e-05,
+      "loss": 1.1039,
+      "step": 14263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997524877018271e-05,
+      "loss": 1.0438,
+      "step": 14264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997524527568711e-05,
+      "loss": 1.1779,
+      "step": 14265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975241780944964e-05,
+      "loss": 1.1588,
+      "step": 14266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997523828595627e-05,
+      "loss": 1.1946,
+      "step": 14267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997523479072103e-05,
+      "loss": 1.1071,
+      "step": 14268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997523129523924e-05,
+      "loss": 1.0114,
+      "step": 14269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997522779951091e-05,
+      "loss": 0.9748,
+      "step": 14270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997522430353604e-05,
+      "loss": 1.3441,
+      "step": 14271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997522080731463e-05,
+      "loss": 1.0202,
+      "step": 14272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975217310846653e-05,
+      "loss": 1.1202,
+      "step": 14273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997521381413215e-05,
+      "loss": 0.757,
+      "step": 14274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99752103171711e-05,
+      "loss": 0.4076,
+      "step": 14275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99752068199635e-05,
+      "loss": 0.6337,
+      "step": 14276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997520332250936e-05,
+      "loss": 0.3864,
+      "step": 14277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997519982480867e-05,
+      "loss": 0.4376,
+      "step": 14278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997519632686144e-05,
+      "loss": 0.6838,
+      "step": 14279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997519282866766e-05,
+      "loss": 0.4501,
+      "step": 14280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997518933022733e-05,
+      "loss": 0.2631,
+      "step": 14281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997518583154047e-05,
+      "loss": 0.3098,
+      "step": 14282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997518233260706e-05,
+      "loss": 0.1209,
+      "step": 14283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975178833427106e-05,
+      "loss": 0.4564,
+      "step": 14284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997517533400061e-05,
+      "loss": 0.7584,
+      "step": 14285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975171834327564e-05,
+      "loss": 0.8234,
+      "step": 14286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997516833440797e-05,
+      "loss": 1.2015,
+      "step": 14287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997516483424184e-05,
+      "loss": 1.2761,
+      "step": 14288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975161333829165e-05,
+      "loss": 1.1576,
+      "step": 14289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975157833169936e-05,
+      "loss": 0.9933,
+      "step": 14290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975154332264174e-05,
+      "loss": 1.3237,
+      "step": 14291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975150831111864e-05,
+      "loss": 0.9063,
+      "step": 14292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997514732971301e-05,
+      "loss": 1.0895,
+      "step": 14293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975143828067615e-05,
+      "loss": 1.091,
+      "step": 14294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975140326175676e-05,
+      "loss": 1.0007,
+      "step": 14295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997513682403718e-05,
+      "loss": 1.4644,
+      "step": 14296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975133321652155e-05,
+      "loss": 0.9079,
+      "step": 14297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997512981902058e-05,
+      "loss": 1.1363,
+      "step": 14298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975126316142464e-05,
+      "loss": 1.3871,
+      "step": 14299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997512281301781e-05,
+      "loss": 1.1419,
+      "step": 14300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99751193096466e-05,
+      "loss": 1.207,
+      "step": 14301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997511580602885e-05,
+      "loss": 0.5183,
+      "step": 14302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997511230216455e-05,
+      "loss": 1.3572,
+      "step": 14303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997510879805372e-05,
+      "loss": 0.7195,
+      "step": 14304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997510529369633e-05,
+      "loss": 0.7084,
+      "step": 14305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997510178909241e-05,
+      "loss": 0.7775,
+      "step": 14306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997509828424194e-05,
+      "loss": 1.1075,
+      "step": 14307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997509477914493e-05,
+      "loss": 1.1303,
+      "step": 14308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997509127380138e-05,
+      "loss": 1.0402,
+      "step": 14309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997508776821127e-05,
+      "loss": 1.3226,
+      "step": 14310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997508426237463e-05,
+      "loss": 1.282,
+      "step": 14311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997508075629145e-05,
+      "loss": 1.2724,
+      "step": 14312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997507724996172e-05,
+      "loss": 1.2906,
+      "step": 14313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975073743385444e-05,
+      "loss": 1.2708,
+      "step": 14314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997507023656263e-05,
+      "loss": 0.7824,
+      "step": 14315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975066729493264e-05,
+      "loss": 1.1854,
+      "step": 14316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975063222177366e-05,
+      "loss": 0.9226,
+      "step": 14317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975059714614914e-05,
+      "loss": 0.824,
+      "step": 14318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997505620680593e-05,
+      "loss": 1.1231,
+      "step": 14319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997505269875039e-05,
+      "loss": 1.0022,
+      "step": 14320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997504919044831e-05,
+      "loss": 0.5094,
+      "step": 14321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99750456818997e-05,
+      "loss": 1.136,
+      "step": 14322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975042173104535e-05,
+      "loss": 1.4898,
+      "step": 14323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975038664062825e-05,
+      "loss": 1.0795,
+      "step": 14324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975035154774574e-05,
+      "loss": 1.8042,
+      "step": 14325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997503164523979e-05,
+      "loss": 1.0174,
+      "step": 14326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997502813545845e-05,
+      "loss": 1.4137,
+      "step": 14327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975024625430576e-05,
+      "loss": 1.2082,
+      "step": 14328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997502111515615e-05,
+      "loss": 1.0201,
+      "step": 14329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975017604635186e-05,
+      "loss": 1.3109,
+      "step": 14330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997501409386768e-05,
+      "loss": 1.0131,
+      "step": 14331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997501058285363e-05,
+      "loss": 0.9856,
+      "step": 14332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997500707159304e-05,
+      "loss": 1.3251,
+      "step": 14333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997500356008591e-05,
+      "loss": 1.2006,
+      "step": 14334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997500004833223e-05,
+      "loss": 1.5733,
+      "step": 14335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997499653633201e-05,
+      "loss": 1.4137,
+      "step": 14336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997499302408525e-05,
+      "loss": 1.1408,
+      "step": 14337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997498951159194e-05,
+      "loss": 1.2687,
+      "step": 14338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997498599885209e-05,
+      "loss": 1.5553,
+      "step": 14339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99749824858657e-05,
+      "loss": 0.52,
+      "step": 14340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997497897263277e-05,
+      "loss": 0.8567,
+      "step": 14341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99749754591533e-05,
+      "loss": 1.132,
+      "step": 14342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997497194542727e-05,
+      "loss": 1.0302,
+      "step": 14343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997496843145472e-05,
+      "loss": 0.9373,
+      "step": 14344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997496491723561e-05,
+      "loss": 1.1295,
+      "step": 14345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997496140276997e-05,
+      "loss": 1.3824,
+      "step": 14346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974957888057786e-05,
+      "loss": 1.354,
+      "step": 14347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997495437309905e-05,
+      "loss": 1.3216,
+      "step": 14348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997495085789378e-05,
+      "loss": 1.06,
+      "step": 14349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974947342441966e-05,
+      "loss": 1.1986,
+      "step": 14350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997494382674361e-05,
+      "loss": 1.0807,
+      "step": 14351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974940310798714e-05,
+      "loss": 1.24,
+      "step": 14352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974936794607266e-05,
+      "loss": 1.1952,
+      "step": 14353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997493327816929e-05,
+      "loss": 1.5186,
+      "step": 14354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997492976148476e-05,
+      "loss": 1.1785,
+      "step": 14355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99749262445537e-05,
+      "loss": 0.9407,
+      "step": 14356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997492272737608e-05,
+      "loss": 1.2702,
+      "step": 14357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974919209951936e-05,
+      "loss": 1.2608,
+      "step": 14358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997491569228124e-05,
+      "loss": 1.2639,
+      "step": 14359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997491217436401e-05,
+      "loss": 0.932,
+      "step": 14360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974908656200236e-05,
+      "loss": 1.3442,
+      "step": 14361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997490513778992e-05,
+      "loss": 1.2339,
+      "step": 14362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997490161913306e-05,
+      "loss": 1.2823,
+      "step": 14363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974898100229655e-05,
+      "loss": 1.1675,
+      "step": 14364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997489458107971e-05,
+      "loss": 0.7062,
+      "step": 14365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974891061683226e-05,
+      "loss": 1.0909,
+      "step": 14366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997488754204021e-05,
+      "loss": 1.2318,
+      "step": 14367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974884022150634e-05,
+      "loss": 1.6796,
+      "step": 14368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997488050201453e-05,
+      "loss": 1.1237,
+      "step": 14369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997487698163188e-05,
+      "loss": 1.7231,
+      "step": 14370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974873461002683e-05,
+      "loss": 0.9219,
+      "step": 14371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974869940126954e-05,
+      "loss": 0.7085,
+      "step": 14372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974866419004677e-05,
+      "loss": 1.2581,
+      "step": 14373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997486289763586e-05,
+      "loss": 1.0116,
+      "step": 14374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99748593760205e-05,
+      "loss": 1.3663,
+      "step": 14375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99748558541586e-05,
+      "loss": 1.0475,
+      "step": 14376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974852332050167e-05,
+      "loss": 0.9837,
+      "step": 14377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974848809695185e-05,
+      "loss": 0.8915,
+      "step": 14378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997484528709366e-05,
+      "loss": 1.2295,
+      "step": 14379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974841764245594e-05,
+      "loss": 1.2447,
+      "step": 14380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997483824115099e-05,
+      "loss": 1.198,
+      "step": 14381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974834717809846e-05,
+      "loss": 1.3973,
+      "step": 14382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997483119422216e-05,
+      "loss": 1.1056,
+      "step": 14383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997482767038793e-05,
+      "loss": 1.1403,
+      "step": 14384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997482414630716e-05,
+      "loss": 1.6749,
+      "step": 14385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974820621979853e-05,
+      "loss": 1.3194,
+      "step": 14386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974817097406e-05,
+      "loss": 0.8325,
+      "step": 14387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997481357258561e-05,
+      "loss": 1.031,
+      "step": 14388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997481004751868e-05,
+      "loss": 1.1606,
+      "step": 14389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997480652220521e-05,
+      "loss": 0.8679,
+      "step": 14390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997480299664519e-05,
+      "loss": 0.9806,
+      "step": 14391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997479947083864e-05,
+      "loss": 1.0224,
+      "step": 14392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997479594478554e-05,
+      "loss": 1.1977,
+      "step": 14393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974792418485904e-05,
+      "loss": 1.0153,
+      "step": 14394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997478889193973e-05,
+      "loss": 1.1233,
+      "step": 14395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974785365147013e-05,
+      "loss": 0.9666,
+      "step": 14396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997478183810776e-05,
+      "loss": 1.2151,
+      "step": 14397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997477831082196e-05,
+      "loss": 1.1125,
+      "step": 14398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974774783289615e-05,
+      "loss": 1.289,
+      "step": 14399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997477125551073e-05,
+      "loss": 1.0788,
+      "step": 14400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997476772748532e-05,
+      "loss": 0.9865,
+      "step": 14401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997476419921335e-05,
+      "loss": 0.8677,
+      "step": 14402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974760670694855e-05,
+      "loss": 1.1269,
+      "step": 14403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997475714192982e-05,
+      "loss": 1.0795,
+      "step": 14404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997475361291824e-05,
+      "loss": 1.1291,
+      "step": 14405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974750083660114e-05,
+      "loss": 1.2704,
+      "step": 14406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997474655415545e-05,
+      "loss": 1.1482,
+      "step": 14407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974743024404244e-05,
+      "loss": 1.0979,
+      "step": 14408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974739494406505e-05,
+      "loss": 0.8416,
+      "step": 14409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974735964162225e-05,
+      "loss": 0.921,
+      "step": 14410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99747324336714e-05,
+      "loss": 1.1402,
+      "step": 14411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997472890293404e-05,
+      "loss": 1.2226,
+      "step": 14412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997472537195014e-05,
+      "loss": 1.346,
+      "step": 14413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99747218407197e-05,
+      "loss": 0.9469,
+      "step": 14414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997471830924272e-05,
+      "loss": 1.1293,
+      "step": 14415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974714777519196e-05,
+      "loss": 0.9046,
+      "step": 14416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997471124554913e-05,
+      "loss": 0.8587,
+      "step": 14417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997470771333253e-05,
+      "loss": 1.1661,
+      "step": 14418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997470418086939e-05,
+      "loss": 0.9759,
+      "step": 14419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997470064815971e-05,
+      "loss": 1.3513,
+      "step": 14420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997469711520349e-05,
+      "loss": 0.7998,
+      "step": 14421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997469358200073e-05,
+      "loss": 0.4826,
+      "step": 14422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974690048551434e-05,
+      "loss": 0.681,
+      "step": 14423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997468651485558e-05,
+      "loss": 0.6375,
+      "step": 14424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997468298091321e-05,
+      "loss": 0.532,
+      "step": 14425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997467944672429e-05,
+      "loss": 0.3822,
+      "step": 14426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997467591228883e-05,
+      "loss": 0.2592,
+      "step": 14427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997467237760683e-05,
+      "loss": 0.264,
+      "step": 14428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99746688426783e-05,
+      "loss": 0.1805,
+      "step": 14429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997466530750322e-05,
+      "loss": 1.1733,
+      "step": 14430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974661772081605e-05,
+      "loss": 1.3639,
+      "step": 14431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974658236413455e-05,
+      "loss": 1.1364,
+      "step": 14432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974654700498757e-05,
+      "loss": 0.8945,
+      "step": 14433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997465116433752e-05,
+      "loss": 0.6625,
+      "step": 14434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997464762792975e-05,
+      "loss": 0.7756,
+      "step": 14435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997464409127544e-05,
+      "loss": 1.1861,
+      "step": 14436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974640554374584e-05,
+      "loss": 1.0582,
+      "step": 14437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997463701722719e-05,
+      "loss": 1.1725,
+      "step": 14438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974633479833266e-05,
+      "loss": 1.078,
+      "step": 14439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974629942192796e-05,
+      "loss": 0.9228,
+      "step": 14440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974626404305785e-05,
+      "loss": 0.5655,
+      "step": 14441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997462286617224e-05,
+      "loss": 0.4891,
+      "step": 14442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997461932779215e-05,
+      "loss": 0.484,
+      "step": 14443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997461578916552e-05,
+      "loss": 0.9818,
+      "step": 14444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997461225029236e-05,
+      "loss": 1.0619,
+      "step": 14445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997460871117267e-05,
+      "loss": 1.129,
+      "step": 14446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997460517180642e-05,
+      "loss": 1.2111,
+      "step": 14447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997460163219364e-05,
+      "loss": 0.4887,
+      "step": 14448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974598092334325e-05,
+      "loss": 1.174,
+      "step": 14449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997459455222847e-05,
+      "loss": 1.7233,
+      "step": 14450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997459101187607e-05,
+      "loss": 1.5597,
+      "step": 14451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974587471277137e-05,
+      "loss": 0.8462,
+      "step": 14452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997458393043166e-05,
+      "loss": 1.3633,
+      "step": 14453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997458038933965e-05,
+      "loss": 0.6299,
+      "step": 14454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99745768480011e-05,
+      "loss": 0.1442,
+      "step": 14455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974573306416005e-05,
+      "loss": 0.9584,
+      "step": 14456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974569764584376e-05,
+      "loss": 1.3043,
+      "step": 14457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997456622250621e-05,
+      "loss": 0.9104,
+      "step": 14458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974562680181514e-05,
+      "loss": 1.1106,
+      "step": 14459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997455913761027e-05,
+      "loss": 1.1653,
+      "step": 14460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997455559479249e-05,
+      "loss": 0.9697,
+      "step": 14461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997455205172817e-05,
+      "loss": 1.2126,
+      "step": 14462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997454850841731e-05,
+      "loss": 0.9386,
+      "step": 14463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974544964859914e-05,
+      "loss": 1.1982,
+      "step": 14464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974541421055985e-05,
+      "loss": 1.0994,
+      "step": 14465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974537877005515e-05,
+      "loss": 1.2372,
+      "step": 14466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99745343327085e-05,
+      "loss": 1.0711,
+      "step": 14467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997453078816495e-05,
+      "loss": 1.279,
+      "step": 14468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997452724337486e-05,
+      "loss": 1.1346,
+      "step": 14469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997452369833824e-05,
+      "loss": 1.0648,
+      "step": 14470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997452015305508e-05,
+      "loss": 1.367,
+      "step": 14471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997451660752538e-05,
+      "loss": 1.1264,
+      "step": 14472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997451306174914e-05,
+      "loss": 1.1409,
+      "step": 14473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997450951572636e-05,
+      "loss": 0.9711,
+      "step": 14474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997450596945705e-05,
+      "loss": 0.9547,
+      "step": 14475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99745024229412e-05,
+      "loss": 1.4691,
+      "step": 14476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997449887617881e-05,
+      "loss": 1.1123,
+      "step": 14477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997449532916988e-05,
+      "loss": 1.3364,
+      "step": 14478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997449178191442e-05,
+      "loss": 1.0358,
+      "step": 14479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997448823441241e-05,
+      "loss": 1.5155,
+      "step": 14480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997448468666388e-05,
+      "loss": 1.4649,
+      "step": 14481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974481138668796e-05,
+      "loss": 1.2042,
+      "step": 14482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997447759042718e-05,
+      "loss": 0.8581,
+      "step": 14483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997447404193903e-05,
+      "loss": 1.2848,
+      "step": 14484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997447049320434e-05,
+      "loss": 0.6986,
+      "step": 14485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997446694422311e-05,
+      "loss": 0.9755,
+      "step": 14486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997446339499535e-05,
+      "loss": 1.2767,
+      "step": 14487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997445984552104e-05,
+      "loss": 1.1844,
+      "step": 14488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99744562958002e-05,
+      "loss": 1.7483,
+      "step": 14489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997445274583282e-05,
+      "loss": 1.4285,
+      "step": 14490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997444919561891e-05,
+      "loss": 1.0301,
+      "step": 14491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974445645158464e-05,
+      "loss": 1.2763,
+      "step": 14492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997444209445147e-05,
+      "loss": 1.1808,
+      "step": 14493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997443854349795e-05,
+      "loss": 1.2185,
+      "step": 14494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997443499229788e-05,
+      "loss": 1.1837,
+      "step": 14495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997443144085128e-05,
+      "loss": 0.8031,
+      "step": 14496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974427889158146e-05,
+      "loss": 1.2143,
+      "step": 14497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997442433721847e-05,
+      "loss": 1.1006,
+      "step": 14498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997442078503226e-05,
+      "loss": 1.2395,
+      "step": 14499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997441723259951e-05,
+      "loss": 1.1118,
+      "step": 14500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.032680630683899,
+      "eval_runtime": 83.9404,
+      "eval_samples_per_second": 16.5,
+      "eval_steps_per_second": 4.134,
+      "step": 14500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997441367992023e-05,
+      "loss": 1.2032,
+      "step": 14501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974410126994404e-05,
+      "loss": 0.9744,
+      "step": 14502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997440657382204e-05,
+      "loss": 1.0883,
+      "step": 14503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997440302040315e-05,
+      "loss": 1.1888,
+      "step": 14504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997439946673772e-05,
+      "loss": 1.1701,
+      "step": 14505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974395912825744e-05,
+      "loss": 1.5214,
+      "step": 14506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974392358667243e-05,
+      "loss": 1.174,
+      "step": 14507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974388804262196e-05,
+      "loss": 1.1367,
+      "step": 14508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997438524961062e-05,
+      "loss": 0.7751,
+      "step": 14509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974381694712505e-05,
+      "loss": 1.1393,
+      "step": 14510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997437813956785e-05,
+      "loss": 1.1476,
+      "step": 14511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974374584176665e-05,
+      "loss": 1.2118,
+      "step": 14512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997437102853893e-05,
+      "loss": 1.3143,
+      "step": 14513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997436747265467e-05,
+      "loss": 1.0653,
+      "step": 14514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974363916523875e-05,
+      "loss": 1.2056,
+      "step": 14515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997436036014654e-05,
+      "loss": 0.8385,
+      "step": 14516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997435680352267e-05,
+      "loss": 1.083,
+      "step": 14517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997435324665226e-05,
+      "loss": 1.1865,
+      "step": 14518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997434968953532e-05,
+      "loss": 1.0839,
+      "step": 14519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974346132171834e-05,
+      "loss": 1.1288,
+      "step": 14520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997434257456182e-05,
+      "loss": 1.1401,
+      "step": 14521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974339016705265e-05,
+      "loss": 0.758,
+      "step": 14522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997433545860219e-05,
+      "loss": 1.0797,
+      "step": 14523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997433190025256e-05,
+      "loss": 1.3489,
+      "step": 14524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974328341656394e-05,
+      "loss": 1.1971,
+      "step": 14525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99743247828137e-05,
+      "loss": 0.9595,
+      "step": 14526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974321223724465e-05,
+      "loss": 0.982,
+      "step": 14527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997431766438869e-05,
+      "loss": 1.2661,
+      "step": 14528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974314104806394e-05,
+      "loss": 0.7192,
+      "step": 14529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997431054497756e-05,
+      "loss": 1.1649,
+      "step": 14530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997430698490217e-05,
+      "loss": 1.2051,
+      "step": 14531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997430342458026e-05,
+      "loss": 1.1533,
+      "step": 14532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974299864011815e-05,
+      "loss": 0.7428,
+      "step": 14533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974296303196835e-05,
+      "loss": 0.5662,
+      "step": 14534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997429274213531e-05,
+      "loss": 1.1641,
+      "step": 14535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974289180827253e-05,
+      "loss": 1.0921,
+      "step": 14536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974285619272665e-05,
+      "loss": 0.9288,
+      "step": 14537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974282057471535e-05,
+      "loss": 1.2262,
+      "step": 14538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997427849542388e-05,
+      "loss": 0.985,
+      "step": 14539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974274933129675e-05,
+      "loss": 1.1805,
+      "step": 14540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997427137058894e-05,
+      "loss": 1.1029,
+      "step": 14541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997426780780168e-05,
+      "loss": 0.5841,
+      "step": 14542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997426424476787e-05,
+      "loss": 0.9382,
+      "step": 14543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997426068148753e-05,
+      "loss": 0.9929,
+      "step": 14544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997425711796066e-05,
+      "loss": 1.0577,
+      "step": 14545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997425355418724e-05,
+      "loss": 1.0238,
+      "step": 14546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99742499901673e-05,
+      "loss": 1.0698,
+      "step": 14547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974246425900816e-05,
+      "loss": 1.0617,
+      "step": 14548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99742428613878e-05,
+      "loss": 0.9446,
+      "step": 14549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997423929662825e-05,
+      "loss": 0.715,
+      "step": 14550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997423573162216e-05,
+      "loss": 0.6863,
+      "step": 14551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997423216636954e-05,
+      "loss": 1.0478,
+      "step": 14552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997422860087039e-05,
+      "loss": 1.3427,
+      "step": 14553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974225035124696e-05,
+      "loss": 1.3418,
+      "step": 14554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997422146913247e-05,
+      "loss": 1.3523,
+      "step": 14555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974217902893706e-05,
+      "loss": 1.4194,
+      "step": 14556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997421433640841e-05,
+      "loss": 1.1343,
+      "step": 14557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997421076967658e-05,
+      "loss": 1.0051,
+      "step": 14558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997420720269821e-05,
+      "loss": 1.3802,
+      "step": 14559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974203635473304e-05,
+      "loss": 1.1661,
+      "step": 14560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997420006800187e-05,
+      "loss": 1.1622,
+      "step": 14561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99741965002839e-05,
+      "loss": 1.3033,
+      "step": 14562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99741929323194e-05,
+      "loss": 0.9048,
+      "step": 14563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997418936410835e-05,
+      "loss": 2.2987,
+      "step": 14564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974185795650776e-05,
+      "loss": 1.6775,
+      "step": 14565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997418222694667e-05,
+      "loss": 0.9796,
+      "step": 14566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974178657996024e-05,
+      "loss": 1.2108,
+      "step": 14567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974175088798846e-05,
+      "loss": 0.3747,
+      "step": 14568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997417151935513e-05,
+      "loss": 1.1472,
+      "step": 14569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997416794966488e-05,
+      "loss": 1.1262,
+      "step": 14570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99741643797281e-05,
+      "loss": 0.9259,
+      "step": 14571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974160809544784e-05,
+      "loss": 0.2318,
+      "step": 14572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997415723911493e-05,
+      "loss": 0.7396,
+      "step": 14573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997415366843855e-05,
+      "loss": 1.0218,
+      "step": 14574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997415009751563e-05,
+      "loss": 1.2151,
+      "step": 14575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997414652634617e-05,
+      "loss": 1.0314,
+      "step": 14576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997414295493018e-05,
+      "loss": 1.1444,
+      "step": 14577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974139383267657e-05,
+      "loss": 1.1257,
+      "step": 14578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974135811358605e-05,
+      "loss": 1.1311,
+      "step": 14579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974132239203006e-05,
+      "loss": 1.2159,
+      "step": 14580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997412866680089e-05,
+      "loss": 1.1511,
+      "step": 14581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997412509415223e-05,
+      "loss": 1.0625,
+      "step": 14582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997412152125703e-05,
+      "loss": 1.8244,
+      "step": 14583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974117948115304e-05,
+      "loss": 1.1516,
+      "step": 14584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997411437472704e-05,
+      "loss": 0.9846,
+      "step": 14585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997411080109225e-05,
+      "loss": 0.9495,
+      "step": 14586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997410722721092e-05,
+      "loss": 1.0236,
+      "step": 14587
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997410365308306e-05,
+      "loss": 1.0757,
+      "step": 14588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974100078708666e-05,
+      "loss": 1.1619,
+      "step": 14589
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974096504087734e-05,
+      "loss": 0.9697,
+      "step": 14590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997409292922027e-05,
+      "loss": 1.085,
+      "step": 14591
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997408935410627e-05,
+      "loss": 1.4168,
+      "step": 14592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997408577874574e-05,
+      "loss": 1.0358,
+      "step": 14593
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997408220313867e-05,
+      "loss": 0.9766,
+      "step": 14594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997407862728507e-05,
+      "loss": 1.0095,
+      "step": 14595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974075051184945e-05,
+      "loss": 1.1314,
+      "step": 14596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974071474838275e-05,
+      "loss": 1.4483,
+      "step": 14597
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997406789824508e-05,
+      "loss": 0.9681,
+      "step": 14598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997406432140534e-05,
+      "loss": 1.0439,
+      "step": 14599
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997406074431907e-05,
+      "loss": 1.0032,
+      "step": 14600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997405716698628e-05,
+      "loss": 1.1184,
+      "step": 14601
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974053589406946e-05,
+      "loss": 1.1359,
+      "step": 14602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997405001158108e-05,
+      "loss": 1.0905,
+      "step": 14603
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997404643350867e-05,
+      "loss": 1.1548,
+      "step": 14604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997404285518974e-05,
+      "loss": 0.956,
+      "step": 14605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997403927662428e-05,
+      "loss": 1.1926,
+      "step": 14606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997403569781227e-05,
+      "loss": 1.0824,
+      "step": 14607
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997403211875374e-05,
+      "loss": 1.0924,
+      "step": 14608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997402853944868e-05,
+      "loss": 1.0906,
+      "step": 14609
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997402495989708e-05,
+      "loss": 1.2025,
+      "step": 14610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974021380098944e-05,
+      "loss": 1.3244,
+      "step": 14611
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997401780005428e-05,
+      "loss": 1.0882,
+      "step": 14612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997401421976308e-05,
+      "loss": 1.0168,
+      "step": 14613
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997401063922535e-05,
+      "loss": 0.8935,
+      "step": 14614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974007058441084e-05,
+      "loss": 1.1162,
+      "step": 14615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9974003477410295e-05,
+      "loss": 0.8198,
+      "step": 14616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973999896132964e-05,
+      "loss": 0.7165,
+      "step": 14617
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997399631460909e-05,
+      "loss": 1.4271,
+      "step": 14618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99739927328387e-05,
+      "loss": 1.2403,
+      "step": 14619
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973989150821775e-05,
+      "loss": 1.1047,
+      "step": 14620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973985568558315e-05,
+      "loss": 0.9785,
+      "step": 14621
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997398198604832e-05,
+      "loss": 0.9175,
+      "step": 14622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997397840329179e-05,
+      "loss": 0.9679,
+      "step": 14623
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997397482028874e-05,
+      "loss": 1.4641,
+      "step": 14624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997397123703914e-05,
+      "loss": 1.4326,
+      "step": 14625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997396765354302e-05,
+      "loss": 0.6138,
+      "step": 14626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997396406980036e-05,
+      "loss": 0.5195,
+      "step": 14627
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973960485811176e-05,
+      "loss": 0.5505,
+      "step": 14628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997395690157546e-05,
+      "loss": 1.2871,
+      "step": 14629
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973953317093205e-05,
+      "loss": 1.3319,
+      "step": 14630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997394973236442e-05,
+      "loss": 1.171,
+      "step": 14631
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997394614738909e-05,
+      "loss": 1.0855,
+      "step": 14632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997394256216725e-05,
+      "loss": 0.9017,
+      "step": 14633
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997393897669886e-05,
+      "loss": 0.8217,
+      "step": 14634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997393539098395e-05,
+      "loss": 0.5106,
+      "step": 14635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99739318050225e-05,
+      "loss": 0.6605,
+      "step": 14636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973928218814525e-05,
+      "loss": 1.1345,
+      "step": 14637
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973924632360014e-05,
+      "loss": 1.3363,
+      "step": 14638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997392104565897e-05,
+      "loss": 1.2334,
+      "step": 14639
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973917458711397e-05,
+      "loss": 1.2053,
+      "step": 14640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997391387151728e-05,
+      "loss": 1.3028,
+      "step": 14641
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997391028407665e-05,
+      "loss": 1.1134,
+      "step": 14642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973906696389475e-05,
+      "loss": 1.096,
+      "step": 14643
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997390310845578e-05,
+      "loss": 1.1165,
+      "step": 14644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973899520275544e-05,
+      "loss": 1.0725,
+      "step": 14645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973895931848774e-05,
+      "loss": 0.9818,
+      "step": 14646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997389234317548e-05,
+      "loss": 0.951,
+      "step": 14647
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973888754255646e-05,
+      "loss": 1.2996,
+      "step": 14648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997388516508929e-05,
+      "loss": 1.0835,
+      "step": 14649
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997388157567639e-05,
+      "loss": 1.3664,
+      "step": 14650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997387798601696e-05,
+      "loss": 1.3241,
+      "step": 14651
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997387439611101e-05,
+      "loss": 0.752,
+      "step": 14652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997387080595852e-05,
+      "loss": 0.9959,
+      "step": 14653
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99738672155595e-05,
+      "loss": 1.0986,
+      "step": 14654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973863624913944e-05,
+      "loss": 1.1846,
+      "step": 14655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997386003402187e-05,
+      "loss": 1.1293,
+      "step": 14656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997385644288325e-05,
+      "loss": 1.0859,
+      "step": 14657
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973852851498107e-05,
+      "loss": 1.0269,
+      "step": 14658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997384925986643e-05,
+      "loss": 1.2403,
+      "step": 14659
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973845667988216e-05,
+      "loss": 0.9942,
+      "step": 14660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973842075863483e-05,
+      "loss": 1.0663,
+      "step": 14661
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997383848349221e-05,
+      "loss": 1.0268,
+      "step": 14662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99738348908744e-05,
+      "loss": 1.1241,
+      "step": 14663
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973831298010074e-05,
+      "loss": 1.0425,
+      "step": 14664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997382770489921e-05,
+      "loss": 0.7101,
+      "step": 14665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997382411154181e-05,
+      "loss": 0.8925,
+      "step": 14666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973820517937886e-05,
+      "loss": 1.0695,
+      "step": 14667
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997381692408743e-05,
+      "loss": 1.1682,
+      "step": 14668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973813329990444e-05,
+      "loss": 1.0653,
+      "step": 14669
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973809735646926e-05,
+      "loss": 1.0438,
+      "step": 14670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997380614105687e-05,
+      "loss": 0.848,
+      "step": 14671
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973802546220293e-05,
+      "loss": 0.9973,
+      "step": 14672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997379895113718e-05,
+      "loss": 1.1369,
+      "step": 14673
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997379535580754e-05,
+      "loss": 1.0308,
+      "step": 14674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997379176023136e-05,
+      "loss": 1.2479,
+      "step": 14675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997378816440866e-05,
+      "loss": 0.8342,
+      "step": 14676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997378456833942e-05,
+      "loss": 1.2351,
+      "step": 14677
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973780972023654e-05,
+      "loss": 1.0975,
+      "step": 14678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997377737546136e-05,
+      "loss": 1.1363,
+      "step": 14679
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997377377865253e-05,
+      "loss": 0.975,
+      "step": 14680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997377018159718e-05,
+      "loss": 0.724,
+      "step": 14681
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973766584295287e-05,
+      "loss": 0.3843,
+      "step": 14682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973762986746866e-05,
+      "loss": 0.3052,
+      "step": 14683
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997375938895192e-05,
+      "loss": 0.3796,
+      "step": 14684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997375579091044e-05,
+      "loss": 0.3063,
+      "step": 14685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997375219262243e-05,
+      "loss": 0.8041,
+      "step": 14686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973748594087886e-05,
+      "loss": 1.13,
+      "step": 14687
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973744995306816e-05,
+      "loss": 1.1825,
+      "step": 14688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997374139627922e-05,
+      "loss": 0.9396,
+      "step": 14689
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997373779700508e-05,
+      "loss": 1.1783,
+      "step": 14690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997373419748442e-05,
+      "loss": 1.1474,
+      "step": 14691
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997373059771723e-05,
+      "loss": 1.1786,
+      "step": 14692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997372699770351e-05,
+      "loss": 1.0563,
+      "step": 14693
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997372339744326e-05,
+      "loss": 0.9318,
+      "step": 14694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997371979693648e-05,
+      "loss": 0.858,
+      "step": 14695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997371619618316e-05,
+      "loss": 0.8403,
+      "step": 14696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997371259518332e-05,
+      "loss": 0.8338,
+      "step": 14697
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997370899393695e-05,
+      "loss": 1.2301,
+      "step": 14698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973705392444045e-05,
+      "loss": 1.0208,
+      "step": 14699
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997370179070462e-05,
+      "loss": 1.033,
+      "step": 14700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973698188718656e-05,
+      "loss": 0.8073,
+      "step": 14701
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973694586486164e-05,
+      "loss": 1.0833,
+      "step": 14702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973690984007145e-05,
+      "loss": 1.1689,
+      "step": 14703
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997368738128159e-05,
+      "loss": 1.225,
+      "step": 14704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997368377830951e-05,
+      "loss": 1.1056,
+      "step": 14705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973680175090896e-05,
+      "loss": 0.8393,
+      "step": 14706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997367657162576e-05,
+      "loss": 0.91,
+      "step": 14707
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973672967914085e-05,
+      "loss": 0.9734,
+      "step": 14708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997366936395589e-05,
+      "loss": 1.1687,
+      "step": 14709
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973665759751166e-05,
+      "loss": 0.9984,
+      "step": 14710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997366215529991e-05,
+      "loss": 1.1054,
+      "step": 14711
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997365855060212e-05,
+      "loss": 1.0267,
+      "step": 14712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99736549456578e-05,
+      "loss": 1.1135,
+      "step": 14713
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997365134046695e-05,
+      "loss": 0.9772,
+      "step": 14714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997364773502958e-05,
+      "loss": 1.2222,
+      "step": 14715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997364412934567e-05,
+      "loss": 0.9664,
+      "step": 14716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973640523415236e-05,
+      "loss": 0.8003,
+      "step": 14717
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973636917238274e-05,
+      "loss": 0.6631,
+      "step": 14718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973633310814785e-05,
+      "loss": 0.7529,
+      "step": 14719
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997362970414476e-05,
+      "loss": 0.9658,
+      "step": 14720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997362609722821e-05,
+      "loss": 1.2275,
+      "step": 14721
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997362249006513e-05,
+      "loss": 1.1193,
+      "step": 14722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973618882655515e-05,
+      "loss": 1.2217,
+      "step": 14723
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973615274999377e-05,
+      "loss": 1.0647,
+      "step": 14724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997361166709671e-05,
+      "loss": 0.6829,
+      "step": 14725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997360805894751e-05,
+      "loss": 1.0552,
+      "step": 14726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997360445055179e-05,
+      "loss": 1.0175,
+      "step": 14727
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997360084190953e-05,
+      "loss": 1.0626,
+      "step": 14728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997359723302075e-05,
+      "loss": 0.8943,
+      "step": 14729
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997359362388544e-05,
+      "loss": 1.1485,
+      "step": 14730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973590014503594e-05,
+      "loss": 1.4467,
+      "step": 14731
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997358640487523e-05,
+      "loss": 0.9085,
+      "step": 14732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973582795000326e-05,
+      "loss": 1.0796,
+      "step": 14733
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997357918487891e-05,
+      "loss": 1.1598,
+      "step": 14734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997357557451094e-05,
+      "loss": 1.1493,
+      "step": 14735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973571963896456e-05,
+      "loss": 1.0093,
+      "step": 14736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997356835303545e-05,
+      "loss": 1.0047,
+      "step": 14737
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99735647419279e-05,
+      "loss": 1.1759,
+      "step": 14738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973561130573834e-05,
+      "loss": 1.0909,
+      "step": 14739
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997355751897323e-05,
+      "loss": 0.8766,
+      "step": 14740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973553907126117e-05,
+      "loss": 1.4397,
+      "step": 14741
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973550295032453e-05,
+      "loss": 0.9025,
+      "step": 14742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973546682692277e-05,
+      "loss": 1.2771,
+      "step": 14743
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997354307010556e-05,
+      "loss": 1.2829,
+      "step": 14744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997353945727232e-05,
+      "loss": 0.755,
+      "step": 14745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973535844192555e-05,
+      "loss": 1.2094,
+      "step": 14746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973532230866256e-05,
+      "loss": 1.1977,
+      "step": 14747
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997352861729343e-05,
+      "loss": 1.0674,
+      "step": 14748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973525003474075e-05,
+      "loss": 1.1,
+      "step": 14749
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973521389408194e-05,
+      "loss": 1.5613,
+      "step": 14750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973517775095786e-05,
+      "loss": 1.5573,
+      "step": 14751
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997351416053685e-05,
+      "loss": 1.4173,
+      "step": 14752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997351054573138e-05,
+      "loss": 1.2348,
+      "step": 14753
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997350693067939e-05,
+      "loss": 1.2222,
+      "step": 14754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973503315380866e-05,
+      "loss": 1.1506,
+      "step": 14755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997349969983582e-05,
+      "loss": 0.746,
+      "step": 14756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997349608404424e-05,
+      "loss": 0.935,
+      "step": 14757
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997349246800613e-05,
+      "loss": 1.1727,
+      "step": 14758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973488851721504e-05,
+      "loss": 1.2878,
+      "step": 14759
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973485235190344e-05,
+      "loss": 1.2053,
+      "step": 14760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973481618412656e-05,
+      "loss": 1.1233,
+      "step": 14761
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997347800138844e-05,
+      "loss": 1.3196,
+      "step": 14762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99734743841177e-05,
+      "loss": 1.0109,
+      "step": 14763
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997347076660043e-05,
+      "loss": 1.2367,
+      "step": 14764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973467148836626e-05,
+      "loss": 1.101,
+      "step": 14765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99734635308263e-05,
+      "loss": 1.1594,
+      "step": 14766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997345991256945e-05,
+      "loss": 1.2238,
+      "step": 14767
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973456294066066e-05,
+      "loss": 1.1648,
+      "step": 14768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973452675316154e-05,
+      "loss": 1.1207,
+      "step": 14769
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997344905631972e-05,
+      "loss": 0.9738,
+      "step": 14770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997344543707676e-05,
+      "loss": 1.0473,
+      "step": 14771
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997344181758726e-05,
+      "loss": 0.9454,
+      "step": 14772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997343819785125e-05,
+      "loss": 0.8451,
+      "step": 14773
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973434577868704e-05,
+      "loss": 1.2283,
+      "step": 14774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973430957639635e-05,
+      "loss": 0.9389,
+      "step": 14775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997342733716403e-05,
+      "loss": 1.1241,
+      "step": 14776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99734237164419e-05,
+      "loss": 0.9569,
+      "step": 14777
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997342009547325e-05,
+      "loss": 1.0957,
+      "step": 14778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997341647425807e-05,
+      "loss": 0.6861,
+      "step": 14779
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997341285279637e-05,
+      "loss": 1.0632,
+      "step": 14780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997340923108813e-05,
+      "loss": 1.0148,
+      "step": 14781
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997340560913336e-05,
+      "loss": 1.6072,
+      "step": 14782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997340198693208e-05,
+      "loss": 1.3777,
+      "step": 14783
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997339836448426e-05,
+      "loss": 1.1734,
+      "step": 14784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997339474178991e-05,
+      "loss": 0.4705,
+      "step": 14785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997339111884904e-05,
+      "loss": 1.7411,
+      "step": 14786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997338749566165e-05,
+      "loss": 2.3054,
+      "step": 14787
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997338387222773e-05,
+      "loss": 2.3027,
+      "step": 14788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973380248547273e-05,
+      "loss": 2.2947,
+      "step": 14789
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99733766246203e-05,
+      "loss": 2.2729,
+      "step": 14790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99733730004468e-05,
+      "loss": 2.2439,
+      "step": 14791
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973369376026765e-05,
+      "loss": 2.2121,
+      "step": 14792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997336575136021e-05,
+      "loss": 1.6621,
+      "step": 14793
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997336212644713e-05,
+      "loss": 0.1036,
+      "step": 14794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997335850128751e-05,
+      "loss": 0.0036,
+      "step": 14795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973354875881384e-05,
+      "loss": 0.4035,
+      "step": 14796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973351250228716e-05,
+      "loss": 1.617,
+      "step": 14797
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973347624329534e-05,
+      "loss": 1.0411,
+      "step": 14798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997334399818382e-05,
+      "loss": 1.1408,
+      "step": 14799
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973340371791574e-05,
+      "loss": 1.0667,
+      "step": 14800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997333674515281e-05,
+      "loss": 1.0492,
+      "step": 14801
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997333311826751e-05,
+      "loss": 1.1013,
+      "step": 14802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997332949113569e-05,
+      "loss": 1.6289,
+      "step": 14803
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997332586375735e-05,
+      "loss": 1.1567,
+      "step": 14804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973322236132475e-05,
+      "loss": 0.9798,
+      "step": 14805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973318608261075e-05,
+      "loss": 1.0451,
+      "step": 14806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997331498014315e-05,
+      "loss": 1.3002,
+      "step": 14807
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973311351778693e-05,
+      "loss": 0.9813,
+      "step": 14808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973307723167725e-05,
+      "loss": 0.9681,
+      "step": 14809
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973304094310216e-05,
+      "loss": 0.894,
+      "step": 14810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997330046520619e-05,
+      "loss": 0.7351,
+      "step": 14811
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973296835855636e-05,
+      "loss": 0.6698,
+      "step": 14812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997329320625855e-05,
+      "loss": 0.6868,
+      "step": 14813
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997328957641495e-05,
+      "loss": 1.1907,
+      "step": 14814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973285946324816e-05,
+      "loss": 1.1407,
+      "step": 14815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997328231598816e-05,
+      "loss": 1.2337,
+      "step": 14816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997327868540498e-05,
+      "loss": 1.1565,
+      "step": 14817
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973275054575264e-05,
+      "loss": 1.3925,
+      "step": 14818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997327142349904e-05,
+      "loss": 1.2359,
+      "step": 14819
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997326779217627e-05,
+      "loss": 1.3999,
+      "step": 14820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997326416060699e-05,
+      "loss": 0.9791,
+      "step": 14821
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997326052879118e-05,
+      "loss": 0.7765,
+      "step": 14822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973256896728836e-05,
+      "loss": 1.2074,
+      "step": 14823
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997325326441997e-05,
+      "loss": 1.1995,
+      "step": 14824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997324963186458e-05,
+      "loss": 0.9219,
+      "step": 14825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997324599906267e-05,
+      "loss": 1.2767,
+      "step": 14826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997324236601423e-05,
+      "loss": 1.108,
+      "step": 14827
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973238732719266e-05,
+      "loss": 0.6057,
+      "step": 14828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997323509917778e-05,
+      "loss": 0.6592,
+      "step": 14829
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997323146538977e-05,
+      "loss": 1.0524,
+      "step": 14830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973227831355226e-05,
+      "loss": 1.1696,
+      "step": 14831
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997322419707416e-05,
+      "loss": 0.9979,
+      "step": 14832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997322056254657e-05,
+      "loss": 1.1966,
+      "step": 14833
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973216927772455e-05,
+      "loss": 1.2316,
+      "step": 14834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997321329275181e-05,
+      "loss": 1.3542,
+      "step": 14835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997320965748465e-05,
+      "loss": 1.1776,
+      "step": 14836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997320602197096e-05,
+      "loss": 1.0939,
+      "step": 14837
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997320238621074e-05,
+      "loss": 1.1933,
+      "step": 14838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973198750204e-05,
+      "loss": 0.9844,
+      "step": 14839
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997319511395074e-05,
+      "loss": 1.0345,
+      "step": 14840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997319147745095e-05,
+      "loss": 0.9952,
+      "step": 14841
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997318784070463e-05,
+      "loss": 1.0174,
+      "step": 14842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997318420371179e-05,
+      "loss": 0.8177,
+      "step": 14843
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997318056647243e-05,
+      "loss": 1.0,
+      "step": 14844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997317692898654e-05,
+      "loss": 1.2882,
+      "step": 14845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973173291254126e-05,
+      "loss": 1.1994,
+      "step": 14846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997316965327519e-05,
+      "loss": 0.9474,
+      "step": 14847
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997316601504972e-05,
+      "loss": 1.3273,
+      "step": 14848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997316237657774e-05,
+      "loss": 0.9667,
+      "step": 14849
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973158737859226e-05,
+      "loss": 0.8481,
+      "step": 14850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997315509889419e-05,
+      "loss": 1.2891,
+      "step": 14851
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997315145968263e-05,
+      "loss": 1.218,
+      "step": 14852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973147820224544e-05,
+      "loss": 1.0303,
+      "step": 14853
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973144180519924e-05,
+      "loss": 0.9202,
+      "step": 14854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99731405405688e-05,
+      "loss": 1.045,
+      "step": 14855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973136900371137e-05,
+      "loss": 1.4301,
+      "step": 14856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997313325992695e-05,
+      "loss": 0.9022,
+      "step": 14857
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997312961923625e-05,
+      "loss": 1.0168,
+      "step": 14858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997312597829902e-05,
+      "loss": 1.0202,
+      "step": 14859
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997312233711526e-05,
+      "loss": 1.1546,
+      "step": 14860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973118695684985e-05,
+      "loss": 1.2422,
+      "step": 14861
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997311505400818e-05,
+      "loss": 0.9089,
+      "step": 14862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997311141208485e-05,
+      "loss": 1.1787,
+      "step": 14863
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997310776991499e-05,
+      "loss": 1.0265,
+      "step": 14864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997310412749862e-05,
+      "loss": 1.1003,
+      "step": 14865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997310048483572e-05,
+      "loss": 1.1564,
+      "step": 14866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973096841926294e-05,
+      "loss": 1.2079,
+      "step": 14867
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997309319877035e-05,
+      "loss": 1.1985,
+      "step": 14868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997308955536788e-05,
+      "loss": 0.9917,
+      "step": 14869
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997308591171888e-05,
+      "loss": 1.0653,
+      "step": 14870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997308226782336e-05,
+      "loss": 1.0926,
+      "step": 14871
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997307862368132e-05,
+      "loss": 0.8831,
+      "step": 14872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997307497929276e-05,
+      "loss": 1.342,
+      "step": 14873
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997307133465766e-05,
+      "loss": 0.8392,
+      "step": 14874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997306768977605e-05,
+      "loss": 1.1513,
+      "step": 14875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997306404464791e-05,
+      "loss": 1.1801,
+      "step": 14876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973060399273254e-05,
+      "loss": 1.3827,
+      "step": 14877
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997305675365206e-05,
+      "loss": 1.0725,
+      "step": 14878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973053107784355e-05,
+      "loss": 0.8855,
+      "step": 14879
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997304946167013e-05,
+      "loss": 1.2834,
+      "step": 14880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997304581530937e-05,
+      "loss": 1.0089,
+      "step": 14881
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997304216870209e-05,
+      "loss": 0.8269,
+      "step": 14882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997303852184829e-05,
+      "loss": 1.0531,
+      "step": 14883
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973034874747964e-05,
+      "loss": 1.0446,
+      "step": 14884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973031227401115e-05,
+      "loss": 2.3159,
+      "step": 14885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973027579807745e-05,
+      "loss": 1.4127,
+      "step": 14886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973023931967855e-05,
+      "loss": 1.1488,
+      "step": 14887
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997302028388143e-05,
+      "loss": 1.1259,
+      "step": 14888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997301663554849e-05,
+      "loss": 1.1292,
+      "step": 14889
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997301298696903e-05,
+      "loss": 1.0684,
+      "step": 14890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997300933814304e-05,
+      "loss": 0.9875,
+      "step": 14891
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997300568907053e-05,
+      "loss": 1.0519,
+      "step": 14892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9973002039751496e-05,
+      "loss": 1.092,
+      "step": 14893
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997299839018594e-05,
+      "loss": 1.2802,
+      "step": 14894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997299474037386e-05,
+      "loss": 1.1129,
+      "step": 14895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997299109031526e-05,
+      "loss": 1.0056,
+      "step": 14896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997298744001013e-05,
+      "loss": 1.1671,
+      "step": 14897
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997298378945849e-05,
+      "loss": 0.9901,
+      "step": 14898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997298013866032e-05,
+      "loss": 0.94,
+      "step": 14899
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997297648761562e-05,
+      "loss": 0.9399,
+      "step": 14900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972972836324405e-05,
+      "loss": 0.7191,
+      "step": 14901
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997296918478667e-05,
+      "loss": 0.909,
+      "step": 14902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997296553300241e-05,
+      "loss": 1.1014,
+      "step": 14903
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972961880971624e-05,
+      "loss": 1.0286,
+      "step": 14904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972958228694324e-05,
+      "loss": 1.1349,
+      "step": 14905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997295457617049e-05,
+      "loss": 0.6595,
+      "step": 14906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972950923400144e-05,
+      "loss": 0.7242,
+      "step": 14907
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997294727038326e-05,
+      "loss": 0.7094,
+      "step": 14908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997294361711987e-05,
+      "loss": 0.6031,
+      "step": 14909
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997293996360995e-05,
+      "loss": 0.452,
+      "step": 14910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972936309853504e-05,
+      "loss": 0.3931,
+      "step": 14911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997293265585055e-05,
+      "loss": 0.7613,
+      "step": 14912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997292900160106e-05,
+      "loss": 0.9411,
+      "step": 14913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997292534710505e-05,
+      "loss": 0.8584,
+      "step": 14914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997292169236252e-05,
+      "loss": 0.9696,
+      "step": 14915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997291803737347e-05,
+      "loss": 0.8084,
+      "step": 14916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99729143821379e-05,
+      "loss": 0.9684,
+      "step": 14917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972910726655794e-05,
+      "loss": 1.0021,
+      "step": 14918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997290707092718e-05,
+      "loss": 1.1548,
+      "step": 14919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997290341495204e-05,
+      "loss": 1.0268,
+      "step": 14920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997289975873038e-05,
+      "loss": 0.952,
+      "step": 14921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997289610226219e-05,
+      "loss": 0.8096,
+      "step": 14922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997289244554749e-05,
+      "loss": 0.567,
+      "step": 14923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972888788586255e-05,
+      "loss": 1.1943,
+      "step": 14924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997288513137851e-05,
+      "loss": 1.4243,
+      "step": 14925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972881473924236e-05,
+      "loss": 0.9679,
+      "step": 14926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972877816223446e-05,
+      "loss": 1.1361,
+      "step": 14927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997287415827613e-05,
+      "loss": 1.0682,
+      "step": 14928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997287050008229e-05,
+      "loss": 1.1659,
+      "step": 14929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997286684164193e-05,
+      "loss": 1.0461,
+      "step": 14930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972863182955054e-05,
+      "loss": 1.2201,
+      "step": 14931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972859524021654e-05,
+      "loss": 1.1834,
+      "step": 14932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997285586484173e-05,
+      "loss": 1.1125,
+      "step": 14933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997285220541529e-05,
+      "loss": 1.0901,
+      "step": 14934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997284854574233e-05,
+      "loss": 1.0592,
+      "step": 14935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997284488582283e-05,
+      "loss": 1.0942,
+      "step": 14936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997284122565682e-05,
+      "loss": 0.9409,
+      "step": 14937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997283756524429e-05,
+      "loss": 1.2872,
+      "step": 14938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997283390458524e-05,
+      "loss": 1.0326,
+      "step": 14939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997283024367967e-05,
+      "loss": 1.243,
+      "step": 14940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997282658252758e-05,
+      "loss": 1.1183,
+      "step": 14941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997282292112896e-05,
+      "loss": 1.112,
+      "step": 14942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997281925948382e-05,
+      "loss": 1.2697,
+      "step": 14943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997281559759216e-05,
+      "loss": 1.11,
+      "step": 14944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972811935453985e-05,
+      "loss": 0.5532,
+      "step": 14945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972808273069286e-05,
+      "loss": 1.1706,
+      "step": 14946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997280461043806e-05,
+      "loss": 1.176,
+      "step": 14947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997280094756032e-05,
+      "loss": 1.0269,
+      "step": 14948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997279728443606e-05,
+      "loss": 1.1924,
+      "step": 14949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997279362106527e-05,
+      "loss": 1.3168,
+      "step": 14950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997278995744797e-05,
+      "loss": 1.3441,
+      "step": 14951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997278629358414e-05,
+      "loss": 1.2574,
+      "step": 14952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972782629473796e-05,
+      "loss": 1.0412,
+      "step": 14953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972778965116926e-05,
+      "loss": 1.0197,
+      "step": 14954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997277530051354e-05,
+      "loss": 1.0773,
+      "step": 14955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997277163566363e-05,
+      "loss": 1.0283,
+      "step": 14956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99727679705672e-05,
+      "loss": 1.0716,
+      "step": 14957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997276430522425e-05,
+      "loss": 1.0224,
+      "step": 14958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997276063963478e-05,
+      "loss": 0.9932,
+      "step": 14959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997275697379879e-05,
+      "loss": 0.7764,
+      "step": 14960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997275330771628e-05,
+      "loss": 1.1036,
+      "step": 14961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997274964138724e-05,
+      "loss": 0.9545,
+      "step": 14962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972745974811696e-05,
+      "loss": 1.0358,
+      "step": 14963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972742307989614e-05,
+      "loss": 1.1154,
+      "step": 14964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972738640921026e-05,
+      "loss": 1.2059,
+      "step": 14965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997273497360591e-05,
+      "loss": 1.0128,
+      "step": 14966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997273130604428e-05,
+      "loss": 0.8856,
+      "step": 14967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972727638236125e-05,
+      "loss": 0.4957,
+      "step": 14968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997272397018145e-05,
+      "loss": 0.2882,
+      "step": 14969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997272030188026e-05,
+      "loss": 0.2383,
+      "step": 14970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997271663333254e-05,
+      "loss": 0.5858,
+      "step": 14971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997271296453831e-05,
+      "loss": 1.2508,
+      "step": 14972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997270929549754e-05,
+      "loss": 1.0206,
+      "step": 14973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997270562621027e-05,
+      "loss": 1.0349,
+      "step": 14974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972701956676476e-05,
+      "loss": 0.2786,
+      "step": 14975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972698286896156e-05,
+      "loss": 0.2352,
+      "step": 14976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997269461686933e-05,
+      "loss": 0.1715,
+      "step": 14977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997269094659597e-05,
+      "loss": 0.1294,
+      "step": 14978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997268727607609e-05,
+      "loss": 0.1131,
+      "step": 14979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99726836053097e-05,
+      "loss": 0.1052,
+      "step": 14980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997267993429678e-05,
+      "loss": 0.0949,
+      "step": 14981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997267626303735e-05,
+      "loss": 0.5213,
+      "step": 14982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99726725915314e-05,
+      "loss": 1.1957,
+      "step": 14983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997266891977892e-05,
+      "loss": 0.9684,
+      "step": 14984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997266524777993e-05,
+      "loss": 1.0212,
+      "step": 14985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972661575534415e-05,
+      "loss": 1.2718,
+      "step": 14986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972657903042383e-05,
+      "loss": 1.1513,
+      "step": 14987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997265423030383e-05,
+      "loss": 1.2228,
+      "step": 14988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997265055731876e-05,
+      "loss": 1.1608,
+      "step": 14989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972646884087165e-05,
+      "loss": 0.9096,
+      "step": 14990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997264321060906e-05,
+      "loss": 1.2469,
+      "step": 14991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972639536884424e-05,
+      "loss": 1.3127,
+      "step": 14992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972635862913276e-05,
+      "loss": 1.7783,
+      "step": 14993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997263218869561e-05,
+      "loss": 0.7732,
+      "step": 14994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997262851423142e-05,
+      "loss": 1.0013,
+      "step": 14995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997262483952071e-05,
+      "loss": 2.2452,
+      "step": 14996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997262116456348e-05,
+      "loss": 6.0102,
+      "step": 14997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972617489359736e-05,
+      "loss": 3.1733,
+      "step": 14998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997261381390947e-05,
+      "loss": 1.0003,
+      "step": 14999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997261013821268e-05,
+      "loss": 1.2808,
+      "step": 15000
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0408920049667358,
+      "eval_runtime": 84.1304,
+      "eval_samples_per_second": 16.463,
+      "eval_steps_per_second": 4.125,
+      "step": 15000
+    }
+  ],
+  "max_steps": 1000000,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 7.5185365450752e+17,
+  "trial_name": null,
+  "trial_params": null
+}