diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,7 +2,7 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 1.0,
-  "global_step": 892,
+  "global_step": 446,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,5375 +10,2690 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.9906,
+      "loss": 9.058,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 8.7258,
+      "loss": 8.7613,
       "step": 2
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 2e-06,
-      "loss": 8.5849,
+      "loss": 8.9401,
       "step": 3
     },
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 2e-06,
-      "loss": 8.8189,
+      "loss": 8.9321,
       "step": 4
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2e-06,
-      "loss": 8.8606,
+      "learning_rate": 4e-06,
+      "loss": 8.8155,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4e-06,
-      "loss": 8.5207,
+      "learning_rate": 6e-06,
+      "loss": 8.4795,
       "step": 6
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 6e-06,
-      "loss": 9.075,
+      "epoch": 0.02,
+      "learning_rate": 8e-06,
+      "loss": 8.1861,
       "step": 7
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 8e-06,
-      "loss": 8.2021,
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 7.9638,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1e-05,
-      "loss": 8.0402,
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 7.7096,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.2e-05,
-      "loss": 8.3275,
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 7.8138,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.4e-05,
-      "loss": 7.8052,
+      "epoch": 0.02,
+      "learning_rate": 1.6e-05,
+      "loss": 7.6533,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.6e-05,
-      "loss": 7.7712,
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 7.4974,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-05,
-      "loss": 7.577,
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 7.4995,
       "step": 13
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2e-05,
-      "loss": 7.6142,
+      "epoch": 0.03,
+      "learning_rate": 2.2e-05,
+      "loss": 7.4919,
       "step": 14
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.2e-05,
-      "loss": 7.4003,
+      "epoch": 0.03,
+      "learning_rate": 2.4e-05,
+      "loss": 7.2911,
       "step": 15
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.4e-05,
-      "loss": 7.4515,
+      "epoch": 0.04,
+      "learning_rate": 2.6e-05,
+      "loss": 7.3818,
       "step": 16
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.6e-05,
-      "loss": 7.4263,
+      "epoch": 0.04,
+      "learning_rate": 2.8e-05,
+      "loss": 7.289,
       "step": 17
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 2.8e-05,
-      "loss": 7.3721,
+      "epoch": 0.04,
+      "learning_rate": 3e-05,
+      "loss": 7.2732,
       "step": 18
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3e-05,
-      "loss": 7.2789,
+      "epoch": 0.04,
+      "learning_rate": 3.2e-05,
+      "loss": 7.1699,
       "step": 19
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.2e-05,
-      "loss": 7.2996,
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 7.2997,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 7.2718,
+      "epoch": 0.05,
+      "learning_rate": 3.6e-05,
+      "loss": 7.2654,
       "step": 21
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 3.6e-05,
-      "loss": 7.2653,
+      "epoch": 0.05,
+      "learning_rate": 3.8e-05,
+      "loss": 7.2562,
       "step": 22
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 3.8e-05,
-      "loss": 7.2823,
+      "epoch": 0.05,
+      "learning_rate": 4e-05,
+      "loss": 7.2341,
       "step": 23
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4e-05,
-      "loss": 7.252,
+      "epoch": 0.05,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 7.1936,
       "step": 24
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.2000000000000004e-05,
-      "loss": 7.1921,
+      "epoch": 0.06,
+      "learning_rate": 4.4e-05,
+      "loss": 7.087,
       "step": 25
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.4e-05,
-      "loss": 7.2632,
+      "epoch": 0.06,
+      "learning_rate": 4.6e-05,
+      "loss": 7.2481,
       "step": 26
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.6e-05,
-      "loss": 7.2424,
+      "epoch": 0.06,
+      "learning_rate": 4.8e-05,
+      "loss": 7.1482,
       "step": 27
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.8e-05,
-      "loss": 7.0426,
+      "epoch": 0.06,
+      "learning_rate": 5e-05,
+      "loss": 7.0234,
       "step": 28
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5e-05,
-      "loss": 7.1775,
+      "epoch": 0.07,
+      "learning_rate": 5.2e-05,
+      "loss": 7.1107,
       "step": 29
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.2e-05,
-      "loss": 7.178,
+      "epoch": 0.07,
+      "learning_rate": 5.4e-05,
+      "loss": 7.1312,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 5.4e-05,
-      "loss": 7.244,
+      "epoch": 0.07,
+      "learning_rate": 5.6e-05,
+      "loss": 7.1319,
       "step": 31
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 5.6e-05,
-      "loss": 7.1643,
+      "epoch": 0.07,
+      "learning_rate": 5.800000000000001e-05,
+      "loss": 7.0867,
       "step": 32
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 5.800000000000001e-05,
-      "loss": 7.1528,
+      "epoch": 0.07,
+      "learning_rate": 6e-05,
+      "loss": 7.1029,
       "step": 33
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6e-05,
-      "loss": 7.0163,
+      "epoch": 0.08,
+      "learning_rate": 6.2e-05,
+      "loss": 6.9942,
       "step": 34
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.2e-05,
-      "loss": 7.1393,
+      "epoch": 0.08,
+      "learning_rate": 6.4e-05,
+      "loss": 7.0135,
       "step": 35
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.4e-05,
-      "loss": 7.0952,
+      "epoch": 0.08,
+      "learning_rate": 6.6e-05,
+      "loss": 7.0468,
       "step": 36
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.6e-05,
-      "loss": 7.1681,
+      "epoch": 0.08,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 7.1351,
       "step": 37
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 7.0933,
+      "epoch": 0.09,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 7.0239,
       "step": 38
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.000000000000001e-05,
-      "loss": 7.21,
+      "epoch": 0.09,
+      "learning_rate": 7.2e-05,
+      "loss": 7.1055,
       "step": 39
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 7.2e-05,
-      "loss": 7.0326,
+      "epoch": 0.09,
+      "learning_rate": 7.4e-05,
+      "loss": 7.0529,
       "step": 40
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.4e-05,
-      "loss": 6.997,
+      "epoch": 0.09,
+      "learning_rate": 7.6e-05,
+      "loss": 7.0928,
       "step": 41
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.6e-05,
-      "loss": 7.1275,
+      "epoch": 0.09,
+      "learning_rate": 7.8e-05,
+      "loss": 7.0995,
       "step": 42
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 7.8e-05,
-      "loss": 6.907,
+      "epoch": 0.1,
+      "learning_rate": 8e-05,
+      "loss": 6.868,
       "step": 43
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8e-05,
-      "loss": 7.0983,
+      "epoch": 0.1,
+      "learning_rate": 8.2e-05,
+      "loss": 7.0389,
       "step": 44
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.2e-05,
-      "loss": 7.1178,
+      "epoch": 0.1,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 7.0978,
       "step": 45
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.400000000000001e-05,
-      "loss": 7.0844,
+      "epoch": 0.1,
+      "learning_rate": 8.599999999999999e-05,
+      "loss": 7.0015,
       "step": 46
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.599999999999999e-05,
-      "loss": 6.8595,
+      "epoch": 0.11,
+      "learning_rate": 8.8e-05,
+      "loss": 6.9493,
       "step": 47
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.8e-05,
-      "loss": 6.9158,
+      "epoch": 0.11,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 6.9514,
       "step": 48
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 6.7471,
+      "epoch": 0.11,
+      "learning_rate": 9.2e-05,
+      "loss": 6.7024,
       "step": 49
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.2e-05,
-      "loss": 6.6454,
+      "epoch": 0.11,
+      "learning_rate": 9.400000000000001e-05,
+      "loss": 6.4352,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.400000000000001e-05,
-      "loss": 7.2296,
+      "epoch": 0.11,
+      "learning_rate": 9.6e-05,
+      "loss": 7.2917,
       "step": 51
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.6e-05,
-      "loss": 7.1616,
+      "epoch": 0.12,
+      "learning_rate": 9.800000000000001e-05,
+      "loss": 7.1611,
       "step": 52
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.800000000000001e-05,
-      "loss": 7.0183,
+      "epoch": 0.12,
+      "learning_rate": 0.0001,
+      "loss": 7.1632,
       "step": 53
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 0.0001,
-      "loss": 8.2149,
+      "epoch": 0.12,
+      "learning_rate": 0.000102,
+      "loss": 7.0577,
       "step": 54
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 0.000102,
-      "loss": 7.0033,
+      "epoch": 0.12,
+      "learning_rate": 0.000104,
+      "loss": 6.9254,
       "step": 55
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 0.000104,
-      "loss": 7.1229,
+      "epoch": 0.13,
+      "learning_rate": 0.000106,
+      "loss": 7.0441,
       "step": 56
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 0.000106,
-      "loss": 7.1342,
+      "epoch": 0.13,
+      "learning_rate": 0.000108,
+      "loss": 7.104,
       "step": 57
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.000108,
-      "loss": 7.1919,
+      "epoch": 0.13,
+      "learning_rate": 0.00011,
+      "loss": 7.1653,
       "step": 58
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.00011,
-      "loss": 6.9628,
+      "epoch": 0.13,
+      "learning_rate": 0.000112,
+      "loss": 7.0212,
       "step": 59
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.000112,
-      "loss": 7.1497,
+      "epoch": 0.13,
+      "learning_rate": 0.000114,
+      "loss": 7.041,
       "step": 60
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.000114,
-      "loss": 7.1235,
+      "epoch": 0.14,
+      "learning_rate": 0.00011600000000000001,
+      "loss": 7.0323,
       "step": 61
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.00011600000000000001,
-      "loss": 7.0763,
+      "epoch": 0.14,
+      "learning_rate": 0.000118,
+      "loss": 7.03,
       "step": 62
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.000118,
-      "loss": 7.0182,
+      "epoch": 0.14,
+      "learning_rate": 0.00012,
+      "loss": 6.9586,
       "step": 63
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.00012,
-      "loss": 7.0785,
+      "epoch": 0.14,
+      "learning_rate": 0.000122,
+      "loss": 7.0752,
       "step": 64
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.000122,
-      "loss": 6.9526,
+      "epoch": 0.15,
+      "learning_rate": 0.000124,
+      "loss": 6.9839,
       "step": 65
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.000124,
-      "loss": 7.1886,
+      "epoch": 0.15,
+      "learning_rate": 0.000126,
+      "loss": 6.9496,
       "step": 66
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.000126,
-      "loss": 7.0152,
+      "epoch": 0.15,
+      "learning_rate": 0.000128,
+      "loss": 6.9913,
       "step": 67
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.000128,
-      "loss": 6.9474,
+      "epoch": 0.15,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 7.0036,
       "step": 68
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.00013000000000000002,
-      "loss": 7.0152,
+      "epoch": 0.15,
+      "learning_rate": 0.000132,
+      "loss": 7.0529,
       "step": 69
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.000132,
-      "loss": 6.9903,
+      "epoch": 0.16,
+      "learning_rate": 0.000134,
+      "loss": 6.9548,
       "step": 70
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.000134,
-      "loss": 7.1365,
+      "epoch": 0.16,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 7.0742,
       "step": 71
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.00013600000000000003,
-      "loss": 7.1923,
+      "epoch": 0.16,
+      "learning_rate": 0.00013800000000000002,
+      "loss": 7.0652,
       "step": 72
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.00013800000000000002,
-      "loss": 6.9543,
+      "epoch": 0.16,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 7.1607,
       "step": 73
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.00014000000000000001,
-      "loss": 7.1106,
+      "epoch": 0.17,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 7.0535,
       "step": 74
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.00014199999999999998,
-      "loss": 6.9641,
+      "epoch": 0.17,
+      "learning_rate": 0.000144,
+      "loss": 7.0252,
       "step": 75
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.000144,
-      "loss": 7.0534,
+      "epoch": 0.17,
+      "learning_rate": 0.000146,
+      "loss": 7.0717,
       "step": 76
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.000146,
-      "loss": 6.9923,
+      "epoch": 0.17,
+      "learning_rate": 0.000148,
+      "loss": 7.0656,
       "step": 77
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.000148,
-      "loss": 6.9758,
+      "epoch": 0.17,
+      "learning_rate": 0.00015,
+      "loss": 7.0663,
       "step": 78
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.00015,
-      "loss": 6.9439,
+      "epoch": 0.18,
+      "learning_rate": 0.000152,
+      "loss": 6.9699,
       "step": 79
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.000152,
-      "loss": 6.9966,
+      "epoch": 0.18,
+      "learning_rate": 0.000154,
+      "loss": 7.0587,
       "step": 80
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.000154,
-      "loss": 7.0292,
+      "epoch": 0.18,
+      "learning_rate": 0.000156,
+      "loss": 7.0515,
       "step": 81
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.000156,
-      "loss": 6.9745,
+      "epoch": 0.18,
+      "learning_rate": 0.000158,
+      "loss": 7.0416,
       "step": 82
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.000158,
-      "loss": 7.1386,
+      "epoch": 0.19,
+      "learning_rate": 0.00016,
+      "loss": 7.0491,
       "step": 83
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.00016,
-      "loss": 6.9023,
+      "epoch": 0.19,
+      "learning_rate": 0.000162,
+      "loss": 7.0139,
       "step": 84
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000162,
-      "loss": 6.9209,
+      "epoch": 0.19,
+      "learning_rate": 0.000164,
+      "loss": 7.0157,
       "step": 85
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000164,
-      "loss": 7.0169,
+      "epoch": 0.19,
+      "learning_rate": 0.00016600000000000002,
+      "loss": 7.0541,
       "step": 86
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00016600000000000002,
-      "loss": 6.9453,
+      "epoch": 0.2,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 7.0067,
       "step": 87
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00016800000000000002,
-      "loss": 6.9099,
+      "epoch": 0.2,
+      "learning_rate": 0.00017,
+      "loss": 6.9629,
       "step": 88
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00017,
-      "loss": 6.9947,
+      "epoch": 0.2,
+      "learning_rate": 0.00017199999999999998,
+      "loss": 6.9218,
       "step": 89
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.00017199999999999998,
-      "loss": 7.0393,
+      "epoch": 0.2,
+      "learning_rate": 0.000174,
+      "loss": 7.0658,
       "step": 90
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000174,
-      "loss": 7.1708,
+      "epoch": 0.2,
+      "learning_rate": 0.000176,
+      "loss": 6.9446,
       "step": 91
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000176,
-      "loss": 7.0478,
+      "epoch": 0.21,
+      "learning_rate": 0.000178,
+      "loss": 7.1844,
       "step": 92
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000178,
-      "loss": 6.8305,
+      "epoch": 0.21,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 7.0148,
       "step": 93
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00017999999999999998,
-      "loss": 6.9167,
+      "epoch": 0.21,
+      "learning_rate": 0.000182,
+      "loss": 6.8312,
       "step": 94
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000182,
-      "loss": 7.0507,
+      "epoch": 0.21,
+      "learning_rate": 0.000184,
+      "loss": 6.7648,
       "step": 95
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000184,
-      "loss": 7.0312,
+      "epoch": 0.22,
+      "learning_rate": 0.000186,
+      "loss": 7.0495,
       "step": 96
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000186,
-      "loss": 6.8684,
+      "epoch": 0.22,
+      "learning_rate": 0.00018800000000000002,
+      "loss": 6.8445,
       "step": 97
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00018800000000000002,
-      "loss": 6.8327,
+      "epoch": 0.22,
+      "learning_rate": 0.00019,
+      "loss": 6.586,
       "step": 98
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00019,
-      "loss": 6.6334,
+      "epoch": 0.22,
+      "learning_rate": 0.000192,
+      "loss": 6.8264,
       "step": 99
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000192,
-      "loss": 6.2158,
+      "epoch": 0.22,
+      "learning_rate": 0.000194,
+      "loss": 6.2679,
       "step": 100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.000194,
-      "loss": 7.2374,
+      "epoch": 0.23,
+      "learning_rate": 0.00019600000000000002,
+      "loss": 7.232,
       "step": 101
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00019600000000000002,
-      "loss": 7.2875,
+      "epoch": 0.23,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 7.1135,
       "step": 102
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00019800000000000002,
-      "loss": 7.129,
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 7.3081,
       "step": 103
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.0002,
-      "loss": 7.0288,
+      "epoch": 0.23,
+      "learning_rate": 0.000202,
+      "loss": 7.1059,
       "step": 104
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000202,
-      "loss": 6.9156,
+      "epoch": 0.24,
+      "learning_rate": 0.000204,
+      "loss": 7.0181,
       "step": 105
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000204,
-      "loss": 7.0054,
+      "epoch": 0.24,
+      "learning_rate": 0.000206,
+      "loss": 7.042,
       "step": 106
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000206,
-      "loss": 7.1088,
+      "epoch": 0.24,
+      "learning_rate": 0.000208,
+      "loss": 6.9934,
       "step": 107
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000208,
-      "loss": 7.1409,
+      "epoch": 0.24,
+      "learning_rate": 0.00021,
+      "loss": 7.0081,
       "step": 108
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00021,
-      "loss": 6.8233,
+      "epoch": 0.24,
+      "learning_rate": 0.000212,
+      "loss": 6.9232,
       "step": 109
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000212,
-      "loss": 6.9543,
+      "epoch": 0.25,
+      "learning_rate": 0.000214,
+      "loss": 6.9753,
       "step": 110
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.000214,
-      "loss": 7.01,
+      "epoch": 0.25,
+      "learning_rate": 0.000216,
+      "loss": 6.8964,
       "step": 111
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000216,
-      "loss": 7.1038,
+      "epoch": 0.25,
+      "learning_rate": 0.000218,
+      "loss": 6.8813,
       "step": 112
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000218,
-      "loss": 6.9761,
+      "epoch": 0.25,
+      "learning_rate": 0.00022,
+      "loss": 6.9205,
       "step": 113
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00022,
-      "loss": 6.9252,
+      "epoch": 0.26,
+      "learning_rate": 0.000222,
+      "loss": 6.972,
       "step": 114
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000222,
-      "loss": 6.9818,
+      "epoch": 0.26,
+      "learning_rate": 0.000224,
+      "loss": 6.9011,
       "step": 115
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000224,
-      "loss": 6.8016,
+      "epoch": 0.26,
+      "learning_rate": 0.00022600000000000002,
+      "loss": 7.0466,
       "step": 116
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00022600000000000002,
-      "loss": 7.0191,
+      "epoch": 0.26,
+      "learning_rate": 0.000228,
+      "loss": 6.9956,
       "step": 117
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.000228,
-      "loss": 6.944,
+      "epoch": 0.26,
+      "learning_rate": 0.00023,
+      "loss": 6.9582,
       "step": 118
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00023,
-      "loss": 6.96,
+      "epoch": 0.27,
+      "learning_rate": 0.00023200000000000003,
+      "loss": 6.8957,
       "step": 119
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.00023200000000000003,
-      "loss": 7.0159,
+      "epoch": 0.27,
+      "learning_rate": 0.00023400000000000002,
+      "loss": 6.9889,
       "step": 120
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00023400000000000002,
-      "loss": 6.8737,
+      "epoch": 0.27,
+      "learning_rate": 0.000236,
+      "loss": 6.8994,
       "step": 121
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000236,
-      "loss": 7.0647,
+      "epoch": 0.27,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 6.8518,
       "step": 122
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00023799999999999998,
-      "loss": 6.8038,
+      "epoch": 0.28,
+      "learning_rate": 0.00024,
+      "loss": 6.9078,
       "step": 123
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00024,
-      "loss": 6.9569,
+      "epoch": 0.28,
+      "learning_rate": 0.000242,
+      "loss": 6.8499,
       "step": 124
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000242,
-      "loss": 7.0178,
+      "epoch": 0.28,
+      "learning_rate": 0.000244,
+      "loss": 6.9592,
       "step": 125
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000244,
-      "loss": 7.1276,
+      "epoch": 0.28,
+      "learning_rate": 0.000246,
+      "loss": 6.9507,
       "step": 126
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000246,
-      "loss": 7.1172,
+      "epoch": 0.28,
+      "learning_rate": 0.000248,
+      "loss": 6.9285,
       "step": 127
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.000248,
-      "loss": 7.0735,
+      "epoch": 0.29,
+      "learning_rate": 0.00025,
+      "loss": 6.9252,
       "step": 128
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00025,
-      "loss": 6.8208,
+      "epoch": 0.29,
+      "learning_rate": 0.000252,
+      "loss": 6.975,
       "step": 129
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000252,
-      "loss": 7.0163,
+      "epoch": 0.29,
+      "learning_rate": 0.000254,
+      "loss": 7.0424,
       "step": 130
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000254,
-      "loss": 6.9724,
+      "epoch": 0.29,
+      "learning_rate": 0.000256,
+      "loss": 6.8453,
       "step": 131
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000256,
-      "loss": 7.0759,
+      "epoch": 0.3,
+      "learning_rate": 0.00025800000000000004,
+      "loss": 7.0332,
       "step": 132
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00025800000000000004,
-      "loss": 6.9727,
+      "epoch": 0.3,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 6.8761,
       "step": 133
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00026000000000000003,
-      "loss": 7.0687,
+      "epoch": 0.3,
+      "learning_rate": 0.000262,
+      "loss": 6.9125,
       "step": 134
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000262,
-      "loss": 6.8244,
+      "epoch": 0.3,
+      "learning_rate": 0.000264,
+      "loss": 7.0235,
       "step": 135
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000264,
-      "loss": 6.946,
+      "epoch": 0.3,
+      "learning_rate": 0.000266,
+      "loss": 6.8194,
       "step": 136
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000266,
-      "loss": 7.1185,
+      "epoch": 0.31,
+      "learning_rate": 0.000268,
+      "loss": 6.9921,
       "step": 137
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000268,
-      "loss": 6.9646,
+      "epoch": 0.31,
+      "learning_rate": 0.00027,
+      "loss": 6.91,
       "step": 138
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00027,
-      "loss": 6.907,
+      "epoch": 0.31,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 6.8729,
       "step": 139
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00027200000000000005,
-      "loss": 6.8742,
+      "epoch": 0.31,
+      "learning_rate": 0.00027400000000000005,
+      "loss": 6.9956,
       "step": 140
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00027400000000000005,
-      "loss": 7.1073,
+      "epoch": 0.32,
+      "learning_rate": 0.00027600000000000004,
+      "loss": 7.0116,
       "step": 141
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00027600000000000004,
-      "loss": 6.989,
+      "epoch": 0.32,
+      "learning_rate": 0.00027800000000000004,
+      "loss": 6.9834,
       "step": 142
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00027800000000000004,
-      "loss": 6.949,
+      "epoch": 0.32,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 6.8219,
       "step": 143
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00028000000000000003,
-      "loss": 6.7263,
+      "epoch": 0.32,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 6.81,
       "step": 144
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 6.8278,
+      "epoch": 0.33,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 6.769,
       "step": 145
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00028399999999999996,
-      "loss": 6.7062,
+      "epoch": 0.33,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 6.7932,
       "step": 146
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00028599999999999996,
-      "loss": 7.0031,
+      "epoch": 0.33,
+      "learning_rate": 0.000288,
+      "loss": 6.8492,
       "step": 147
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000288,
-      "loss": 6.5011,
+      "epoch": 0.33,
+      "learning_rate": 0.00029,
+      "loss": 6.8182,
       "step": 148
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00029,
-      "loss": 6.7402,
+      "epoch": 0.33,
+      "learning_rate": 0.000292,
+      "loss": 6.6264,
       "step": 149
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000292,
-      "loss": 6.2222,
+      "epoch": 0.34,
+      "learning_rate": 0.000294,
+      "loss": 6.5046,
       "step": 150
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000294,
-      "loss": 7.0887,
+      "epoch": 0.34,
+      "learning_rate": 0.000296,
+      "loss": 7.0365,
       "step": 151
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000296,
-      "loss": 7.0349,
+      "epoch": 0.34,
+      "learning_rate": 0.000298,
+      "loss": 7.1078,
       "step": 152
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000298,
-      "loss": 7.0538,
+      "epoch": 0.34,
+      "learning_rate": 0.0003,
+      "loss": 7.0393,
       "step": 153
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.0003,
-      "loss": 6.8402,
+      "epoch": 0.35,
+      "learning_rate": 0.000302,
+      "loss": 6.9125,
       "step": 154
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000302,
-      "loss": 6.8826,
+      "epoch": 0.35,
+      "learning_rate": 0.000304,
+      "loss": 6.901,
       "step": 155
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000304,
-      "loss": 6.9756,
+      "epoch": 0.35,
+      "learning_rate": 0.000306,
+      "loss": 6.9173,
       "step": 156
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000306,
-      "loss": 7.1372,
+      "epoch": 0.35,
+      "learning_rate": 0.000308,
+      "loss": 7.0307,
       "step": 157
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000308,
-      "loss": 7.0853,
+      "epoch": 0.35,
+      "learning_rate": 0.00031,
+      "loss": 6.9548,
       "step": 158
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00031,
-      "loss": 7.086,
+      "epoch": 0.36,
+      "learning_rate": 0.000312,
+      "loss": 6.9309,
       "step": 159
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000312,
-      "loss": 7.0292,
+      "epoch": 0.36,
+      "learning_rate": 0.000314,
+      "loss": 6.988,
       "step": 160
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000314,
-      "loss": 6.8816,
+      "epoch": 0.36,
+      "learning_rate": 0.000316,
+      "loss": 6.9284,
       "step": 161
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000316,
-      "loss": 6.9211,
+      "epoch": 0.36,
+      "learning_rate": 0.00031800000000000003,
+      "loss": 6.9905,
       "step": 162
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00031800000000000003,
-      "loss": 6.969,
+      "epoch": 0.37,
+      "learning_rate": 0.00032,
+      "loss": 6.9599,
       "step": 163
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00032,
-      "loss": 6.9243,
+      "epoch": 0.37,
+      "learning_rate": 0.000322,
+      "loss": 6.9554,
       "step": 164
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.000322,
-      "loss": 6.9765,
+      "epoch": 0.37,
+      "learning_rate": 0.000324,
+      "loss": 6.9063,
       "step": 165
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000324,
-      "loss": 6.936,
+      "epoch": 0.37,
+      "learning_rate": 0.000326,
+      "loss": 6.9603,
       "step": 166
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000326,
-      "loss": 6.8521,
+      "epoch": 0.37,
+      "learning_rate": 0.000328,
+      "loss": 6.892,
       "step": 167
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.000328,
-      "loss": 6.9638,
+      "epoch": 0.38,
+      "learning_rate": 0.00033,
+      "loss": 6.82,
       "step": 168
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00033,
-      "loss": 7.0238,
+      "epoch": 0.38,
+      "learning_rate": 0.00033200000000000005,
+      "loss": 6.843,
       "step": 169
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00033200000000000005,
-      "loss": 6.9353,
+      "epoch": 0.38,
+      "learning_rate": 0.00033400000000000004,
+      "loss": 6.9878,
       "step": 170
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00033400000000000004,
-      "loss": 7.1132,
+      "epoch": 0.38,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 6.9122,
       "step": 171
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00033600000000000004,
-      "loss": 7.0827,
+      "epoch": 0.39,
+      "learning_rate": 0.00033800000000000003,
+      "loss": 6.9609,
       "step": 172
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00033800000000000003,
-      "loss": 7.0337,
+      "epoch": 0.39,
+      "learning_rate": 0.00034,
+      "loss": 6.9128,
       "step": 173
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00034,
-      "loss": 6.9524,
+      "epoch": 0.39,
+      "learning_rate": 0.000342,
+      "loss": 6.9686,
       "step": 174
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000342,
-      "loss": 6.7995,
+      "epoch": 0.39,
+      "learning_rate": 0.00034399999999999996,
+      "loss": 6.9677,
       "step": 175
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00034399999999999996,
-      "loss": 6.9197,
+      "epoch": 0.39,
+      "learning_rate": 0.000346,
+      "loss": 6.8522,
       "step": 176
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000346,
-      "loss": 6.9469,
+      "epoch": 0.4,
+      "learning_rate": 0.000348,
+      "loss": 6.8334,
       "step": 177
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000348,
-      "loss": 6.8256,
+      "epoch": 0.4,
+      "learning_rate": 0.00035,
+      "loss": 6.9131,
       "step": 178
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00035,
-      "loss": 7.0469,
+      "epoch": 0.4,
+      "learning_rate": 0.000352,
+      "loss": 6.9267,
       "step": 179
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000352,
-      "loss": 7.0116,
+      "epoch": 0.4,
+      "learning_rate": 0.000354,
+      "loss": 6.9799,
       "step": 180
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000354,
-      "loss": 7.0411,
+      "epoch": 0.41,
+      "learning_rate": 0.000356,
+      "loss": 6.9508,
       "step": 181
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.000356,
-      "loss": 6.8657,
+      "epoch": 0.41,
+      "learning_rate": 0.000358,
+      "loss": 6.8477,
       "step": 182
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000358,
-      "loss": 6.9022,
+      "epoch": 0.41,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 6.9096,
       "step": 183
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00035999999999999997,
-      "loss": 7.0531,
+      "epoch": 0.41,
+      "learning_rate": 0.000362,
+      "loss": 6.9855,
       "step": 184
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000362,
-      "loss": 6.9568,
+      "epoch": 0.41,
+      "learning_rate": 0.000364,
+      "loss": 6.9124,
       "step": 185
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000364,
-      "loss": 7.0912,
+      "epoch": 0.42,
+      "learning_rate": 0.000366,
+      "loss": 6.8208,
       "step": 186
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000366,
-      "loss": 7.0012,
+      "epoch": 0.42,
+      "learning_rate": 0.000368,
+      "loss": 6.8933,
       "step": 187
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000368,
-      "loss": 6.825,
-      "step": 188
+      "epoch": 0.42,
+      "learning_rate": 0.00037,
+      "loss": 6.8645,
+      "step": 188
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00037,
-      "loss": 6.8709,
+      "epoch": 0.42,
+      "learning_rate": 0.000372,
+      "loss": 6.9765,
       "step": 189
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000372,
-      "loss": 6.8816,
+      "epoch": 0.43,
+      "learning_rate": 0.000374,
+      "loss": 6.9949,
       "step": 190
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.000374,
-      "loss": 6.8838,
+      "epoch": 0.43,
+      "learning_rate": 0.00037600000000000003,
+      "loss": 6.9836,
       "step": 191
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00037600000000000003,
-      "loss": 7.027,
+      "epoch": 0.43,
+      "learning_rate": 0.000378,
+      "loss": 6.8604,
       "step": 192
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000378,
-      "loss": 7.1027,
+      "epoch": 0.43,
+      "learning_rate": 0.00038,
+      "loss": 6.8956,
       "step": 193
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00038,
-      "loss": 6.8466,
+      "epoch": 0.43,
+      "learning_rate": 0.000382,
+      "loss": 6.8332,
       "step": 194
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000382,
-      "loss": 6.6106,
+      "epoch": 0.44,
+      "learning_rate": 0.000384,
+      "loss": 6.8282,
       "step": 195
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000384,
-      "loss": 7.0549,
+      "epoch": 0.44,
+      "learning_rate": 0.000386,
+      "loss": 6.7299,
       "step": 196
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000386,
-      "loss": 6.6962,
+      "epoch": 0.44,
+      "learning_rate": 0.000388,
+      "loss": 6.6981,
       "step": 197
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.000388,
-      "loss": 6.6151,
+      "epoch": 0.44,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 6.7954,
       "step": 198
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00039000000000000005,
-      "loss": 6.6634,
+      "epoch": 0.45,
+      "learning_rate": 0.00039200000000000004,
+      "loss": 6.6457,
       "step": 199
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00039200000000000004,
-      "loss": 6.2976,
+      "epoch": 0.45,
+      "learning_rate": 0.00039400000000000004,
+      "loss": 6.5128,
       "step": 200
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.00039400000000000004,
-      "loss": 7.1151,
+      "epoch": 0.45,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 7.1254,
       "step": 201
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.00039600000000000003,
-      "loss": 7.1614,
+      "epoch": 0.45,
+      "learning_rate": 0.000398,
+      "loss": 7.0731,
       "step": 202
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000398,
-      "loss": 7.0492,
+      "epoch": 0.46,
+      "learning_rate": 0.0004,
+      "loss": 7.0289,
       "step": 203
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0004,
-      "loss": 7.0185,
+      "epoch": 0.46,
+      "learning_rate": 0.000402,
+      "loss": 6.9703,
       "step": 204
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000402,
-      "loss": 6.9719,
+      "epoch": 0.46,
+      "learning_rate": 0.000404,
+      "loss": 7.0268,
       "step": 205
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000404,
-      "loss": 6.9699,
+      "epoch": 0.46,
+      "learning_rate": 0.00040600000000000006,
+      "loss": 6.9987,
       "step": 206
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.00040600000000000006,
-      "loss": 6.9657,
+      "epoch": 0.46,
+      "learning_rate": 0.000408,
+      "loss": 6.9034,
       "step": 207
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.000408,
-      "loss": 6.954,
+      "epoch": 0.47,
+      "learning_rate": 0.00041,
+      "loss": 7.0335,
       "step": 208
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.00041,
-      "loss": 6.9257,
+      "epoch": 0.47,
+      "learning_rate": 0.000412,
+      "loss": 6.9846,
       "step": 209
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000412,
-      "loss": 6.9186,
+      "epoch": 0.47,
+      "learning_rate": 0.000414,
+      "loss": 7.0235,
       "step": 210
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000414,
-      "loss": 6.8191,
+      "epoch": 0.47,
+      "learning_rate": 0.000416,
+      "loss": 6.9098,
       "step": 211
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000416,
-      "loss": 6.8765,
+      "epoch": 0.48,
+      "learning_rate": 0.00041799999999999997,
+      "loss": 6.9513,
       "step": 212
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.00041799999999999997,
-      "loss": 6.8986,
+      "epoch": 0.48,
+      "learning_rate": 0.00042,
+      "loss": 6.8177,
       "step": 213
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.00042,
-      "loss": 6.9879,
+      "epoch": 0.48,
+      "learning_rate": 0.000422,
+      "loss": 6.9137,
       "step": 214
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000422,
-      "loss": 6.8871,
+      "epoch": 0.48,
+      "learning_rate": 0.000424,
+      "loss": 6.8912,
       "step": 215
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000424,
-      "loss": 7.0123,
+      "epoch": 0.48,
+      "learning_rate": 0.000426,
+      "loss": 6.9322,
       "step": 216
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000426,
-      "loss": 6.924,
+      "epoch": 0.49,
+      "learning_rate": 0.000428,
+      "loss": 6.8724,
       "step": 217
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.000428,
-      "loss": 6.9373,
+      "epoch": 0.49,
+      "learning_rate": 0.00043,
+      "loss": 6.9227,
       "step": 218
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00043,
-      "loss": 6.8466,
+      "epoch": 0.49,
+      "learning_rate": 0.000432,
+      "loss": 6.9208,
       "step": 219
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000432,
-      "loss": 7.0829,
+      "epoch": 0.49,
+      "learning_rate": 0.00043400000000000003,
+      "loss": 6.9525,
       "step": 220
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00043400000000000003,
-      "loss": 6.9005,
+      "epoch": 0.5,
+      "learning_rate": 0.000436,
+      "loss": 6.9423,
       "step": 221
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000436,
-      "loss": 6.7992,
+      "epoch": 0.5,
+      "learning_rate": 0.000438,
+      "loss": 6.9664,
       "step": 222
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000438,
-      "loss": 6.9272,
+      "epoch": 0.5,
+      "learning_rate": 0.00044,
+      "loss": 6.8986,
       "step": 223
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00044,
-      "loss": 6.8835,
+      "epoch": 0.5,
+      "learning_rate": 0.000442,
+      "loss": 6.8933,
       "step": 224
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000442,
-      "loss": 6.9156,
+      "epoch": 0.5,
+      "learning_rate": 0.000444,
+      "loss": 7.0104,
       "step": 225
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000444,
-      "loss": 7.0683,
+      "epoch": 0.51,
+      "learning_rate": 0.000446,
+      "loss": 6.893,
       "step": 226
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.000446,
-      "loss": 6.8844,
+      "epoch": 0.51,
+      "learning_rate": 0.000448,
+      "loss": 7.0228,
       "step": 227
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000448,
-      "loss": 6.9815,
+      "epoch": 0.51,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 6.9624,
       "step": 228
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00045000000000000004,
-      "loss": 7.0707,
+      "epoch": 0.51,
+      "learning_rate": 0.00045200000000000004,
+      "loss": 6.8561,
       "step": 229
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00045200000000000004,
-      "loss": 7.0897,
+      "epoch": 0.52,
+      "learning_rate": 0.00045400000000000003,
+      "loss": 6.9604,
       "step": 230
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00045400000000000003,
-      "loss": 6.9375,
+      "epoch": 0.52,
+      "learning_rate": 0.000456,
+      "loss": 6.9243,
       "step": 231
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000456,
-      "loss": 7.1025,
+      "epoch": 0.52,
+      "learning_rate": 0.000458,
+      "loss": 6.8045,
       "step": 232
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000458,
-      "loss": 6.954,
+      "epoch": 0.52,
+      "learning_rate": 0.00046,
+      "loss": 6.8398,
       "step": 233
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00046,
-      "loss": 6.9802,
+      "epoch": 0.52,
+      "learning_rate": 0.000462,
+      "loss": 6.8647,
       "step": 234
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.000462,
-      "loss": 7.0659,
+      "epoch": 0.53,
+      "learning_rate": 0.00046400000000000006,
+      "loss": 6.9927,
       "step": 235
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.00046400000000000006,
-      "loss": 6.8096,
+      "epoch": 0.53,
+      "learning_rate": 0.00046600000000000005,
+      "loss": 6.8641,
       "step": 236
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00046600000000000005,
-      "loss": 6.8746,
+      "epoch": 0.53,
+      "learning_rate": 0.00046800000000000005,
+      "loss": 6.9991,
       "step": 237
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00046800000000000005,
-      "loss": 7.1391,
+      "epoch": 0.53,
+      "learning_rate": 0.00047,
+      "loss": 6.8918,
       "step": 238
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00047,
-      "loss": 6.8512,
+      "epoch": 0.54,
+      "learning_rate": 0.000472,
+      "loss": 6.8122,
       "step": 239
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000472,
-      "loss": 6.9511,
+      "epoch": 0.54,
+      "learning_rate": 0.000474,
+      "loss": 7.0181,
       "step": 240
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000474,
-      "loss": 6.8613,
+      "epoch": 0.54,
+      "learning_rate": 0.00047599999999999997,
+      "loss": 6.8649,
       "step": 241
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00047599999999999997,
-      "loss": 7.0075,
+      "epoch": 0.54,
+      "learning_rate": 0.00047799999999999996,
+      "loss": 6.8825,
       "step": 242
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00047799999999999996,
-      "loss": 6.88,
+      "epoch": 0.54,
+      "learning_rate": 0.00048,
+      "loss": 6.9997,
       "step": 243
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00048,
-      "loss": 6.7439,
+      "epoch": 0.55,
+      "learning_rate": 0.000482,
+      "loss": 6.7921,
       "step": 244
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.000482,
-      "loss": 6.897,
+      "epoch": 0.55,
+      "learning_rate": 0.000484,
+      "loss": 6.8225,
       "step": 245
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000484,
-      "loss": 6.8805,
+      "epoch": 0.55,
+      "learning_rate": 0.000486,
+      "loss": 6.7595,
       "step": 246
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000486,
-      "loss": 6.9362,
+      "epoch": 0.55,
+      "learning_rate": 0.000488,
+      "loss": 6.5411,
       "step": 247
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000488,
-      "loss": 7.0005,
+      "epoch": 0.56,
+      "learning_rate": 0.00049,
+      "loss": 6.821,
       "step": 248
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.00049,
-      "loss": 6.7469,
+      "epoch": 0.56,
+      "learning_rate": 0.000492,
+      "loss": 6.727,
       "step": 249
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000492,
-      "loss": 6.5411,
+      "epoch": 0.56,
+      "learning_rate": 0.000494,
+      "loss": 6.4409,
       "step": 250
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000494,
-      "loss": 7.2376,
+      "epoch": 0.56,
+      "learning_rate": 0.000496,
+      "loss": 7.1683,
       "step": 251
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000496,
-      "loss": 7.0888,
+      "epoch": 0.57,
+      "learning_rate": 0.000498,
+      "loss": 7.0668,
       "step": 252
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.000498,
-      "loss": 6.9918,
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 7.0276,
       "step": 253
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0005,
-      "loss": 7.0821,
+      "epoch": 0.57,
+      "learning_rate": 0.0005020000000000001,
+      "loss": 7.0282,
       "step": 254
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0005020000000000001,
-      "loss": 6.8746,
+      "epoch": 0.57,
+      "learning_rate": 0.000504,
+      "loss": 7.0416,
       "step": 255
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000504,
-      "loss": 7.0239,
+      "epoch": 0.57,
+      "learning_rate": 0.000506,
+      "loss": 7.0002,
       "step": 256
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000506,
-      "loss": 7.0011,
+      "epoch": 0.58,
+      "learning_rate": 0.000508,
+      "loss": 7.0033,
       "step": 257
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000508,
-      "loss": 6.9455,
+      "epoch": 0.58,
+      "learning_rate": 0.00051,
+      "loss": 7.0092,
       "step": 258
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.00051,
-      "loss": 6.903,
+      "epoch": 0.58,
+      "learning_rate": 0.000512,
+      "loss": 6.928,
       "step": 259
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000512,
-      "loss": 7.0352,
+      "epoch": 0.58,
+      "learning_rate": 0.000514,
+      "loss": 6.9736,
       "step": 260
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000514,
-      "loss": 6.9299,
+      "epoch": 0.59,
+      "learning_rate": 0.0005160000000000001,
+      "loss": 6.9636,
       "step": 261
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0005160000000000001,
-      "loss": 6.7932,
+      "epoch": 0.59,
+      "learning_rate": 0.000518,
+      "loss": 6.9479,
       "step": 262
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.000518,
-      "loss": 6.9518,
+      "epoch": 0.59,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 7.0043,
       "step": 263
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0005200000000000001,
-      "loss": 6.9396,
+      "epoch": 0.59,
+      "learning_rate": 0.000522,
+      "loss": 6.9903,
       "step": 264
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000522,
-      "loss": 6.8727,
+      "epoch": 0.59,
+      "learning_rate": 0.000524,
+      "loss": 7.0038,
       "step": 265
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000524,
-      "loss": 7.0255,
+      "epoch": 0.6,
+      "learning_rate": 0.000526,
+      "loss": 6.7783,
       "step": 266
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000526,
-      "loss": 7.0435,
+      "epoch": 0.6,
+      "learning_rate": 0.000528,
+      "loss": 7.0107,
       "step": 267
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000528,
-      "loss": 7.0155,
+      "epoch": 0.6,
+      "learning_rate": 0.0005300000000000001,
+      "loss": 6.8296,
       "step": 268
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0005300000000000001,
-      "loss": 6.957,
+      "epoch": 0.6,
+      "learning_rate": 0.000532,
+      "loss": 6.9727,
       "step": 269
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000532,
-      "loss": 7.0204,
+      "epoch": 0.61,
+      "learning_rate": 0.0005340000000000001,
+      "loss": 6.9827,
       "step": 270
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0005340000000000001,
-      "loss": 6.8599,
+      "epoch": 0.61,
+      "learning_rate": 0.000536,
+      "loss": 6.8958,
       "step": 271
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.000536,
-      "loss": 6.8282,
+      "epoch": 0.61,
+      "learning_rate": 0.0005380000000000001,
+      "loss": 6.8921,
       "step": 272
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.0005380000000000001,
-      "loss": 6.8577,
+      "epoch": 0.61,
+      "learning_rate": 0.00054,
+      "loss": 6.8098,
       "step": 273
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00054,
-      "loss": 6.7691,
+      "epoch": 0.61,
+      "learning_rate": 0.0005420000000000001,
+      "loss": 6.7358,
       "step": 274
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.0005420000000000001,
-      "loss": 7.0144,
+      "epoch": 0.62,
+      "learning_rate": 0.0005440000000000001,
+      "loss": 7.0087,
       "step": 275
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.0005440000000000001,
-      "loss": 6.881,
+      "epoch": 0.62,
+      "learning_rate": 0.000546,
+      "loss": 6.8763,
       "step": 276
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.000546,
-      "loss": 6.9431,
+      "epoch": 0.62,
+      "learning_rate": 0.0005480000000000001,
+      "loss": 6.8399,
       "step": 277
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.0005480000000000001,
-      "loss": 6.8636,
+      "epoch": 0.62,
+      "learning_rate": 0.00055,
+      "loss": 6.934,
       "step": 278
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00055,
-      "loss": 6.954,
+      "epoch": 0.63,
+      "learning_rate": 0.0005520000000000001,
+      "loss": 6.9767,
       "step": 279
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.0005520000000000001,
-      "loss": 7.0149,
+      "epoch": 0.63,
+      "learning_rate": 0.000554,
+      "loss": 7.0013,
       "step": 280
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000554,
-      "loss": 6.7991,
+      "epoch": 0.63,
+      "learning_rate": 0.0005560000000000001,
+      "loss": 6.9378,
       "step": 281
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0005560000000000001,
-      "loss": 6.9592,
+      "epoch": 0.63,
+      "learning_rate": 0.000558,
+      "loss": 6.8736,
       "step": 282
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000558,
-      "loss": 6.7344,
+      "epoch": 0.63,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 6.8258,
       "step": 283
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0005600000000000001,
-      "loss": 6.9892,
+      "epoch": 0.64,
+      "learning_rate": 0.0005620000000000001,
+      "loss": 6.881,
       "step": 284
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0005620000000000001,
-      "loss": 6.9585,
+      "epoch": 0.64,
+      "learning_rate": 0.0005639999999999999,
+      "loss": 6.877,
       "step": 285
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0005639999999999999,
-      "loss": 6.765,
+      "epoch": 0.64,
+      "learning_rate": 0.000566,
+      "loss": 6.964,
       "step": 286
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000566,
-      "loss": 6.9463,
+      "epoch": 0.64,
+      "learning_rate": 0.0005679999999999999,
+      "loss": 6.924,
       "step": 287
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0005679999999999999,
-      "loss": 6.8843,
+      "epoch": 0.65,
+      "learning_rate": 0.00057,
+      "loss": 6.9248,
       "step": 288
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00057,
-      "loss": 7.0347,
+      "epoch": 0.65,
+      "learning_rate": 0.0005719999999999999,
+      "loss": 6.8809,
       "step": 289
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0005719999999999999,
-      "loss": 7.115,
+      "epoch": 0.65,
+      "learning_rate": 0.000574,
+      "loss": 6.9341,
       "step": 290
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000574,
-      "loss": 6.9952,
+      "epoch": 0.65,
+      "learning_rate": 0.000576,
+      "loss": 6.9908,
       "step": 291
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000576,
-      "loss": 6.9102,
+      "epoch": 0.65,
+      "learning_rate": 0.000578,
+      "loss": 6.9272,
       "step": 292
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000578,
-      "loss": 6.8741,
+      "epoch": 0.66,
+      "learning_rate": 0.00058,
+      "loss": 6.8068,
       "step": 293
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.00058,
-      "loss": 6.7783,
+      "epoch": 0.66,
+      "learning_rate": 0.0005819999999999999,
+      "loss": 6.7484,
       "step": 294
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0005819999999999999,
-      "loss": 6.6935,
+      "epoch": 0.66,
+      "learning_rate": 0.000584,
+      "loss": 6.7442,
       "step": 295
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000584,
-      "loss": 6.7783,
+      "epoch": 0.66,
+      "learning_rate": 0.0005859999999999999,
+      "loss": 6.7326,
       "step": 296
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0005859999999999999,
-      "loss": 6.7403,
+      "epoch": 0.67,
+      "learning_rate": 0.000588,
+      "loss": 6.7709,
       "step": 297
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.000588,
-      "loss": 6.556,
+      "epoch": 0.67,
+      "learning_rate": 0.00059,
+      "loss": 6.7711,
       "step": 298
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.00059,
-      "loss": 6.5849,
+      "epoch": 0.67,
+      "learning_rate": 0.000592,
+      "loss": 6.4007,
       "step": 299
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000592,
-      "loss": 6.2956,
+      "epoch": 0.67,
+      "learning_rate": 0.000594,
+      "loss": 6.4146,
       "step": 300
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000594,
-      "loss": 7.1285,
+      "epoch": 0.67,
+      "learning_rate": 0.000596,
+      "loss": 7.1881,
       "step": 301
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000596,
-      "loss": 7.1061,
+      "epoch": 0.68,
+      "learning_rate": 0.000598,
+      "loss": 7.0946,
       "step": 302
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000598,
-      "loss": 6.9421,
+      "epoch": 0.68,
+      "learning_rate": 0.0006,
+      "loss": 7.0366,
       "step": 303
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0006,
-      "loss": 7.0187,
+      "epoch": 0.68,
+      "learning_rate": 0.000602,
+      "loss": 6.9271,
       "step": 304
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000602,
-      "loss": 6.7961,
+      "epoch": 0.68,
+      "learning_rate": 0.000604,
+      "loss": 7.0897,
       "step": 305
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000604,
-      "loss": 6.8924,
+      "epoch": 0.69,
+      "learning_rate": 0.000606,
+      "loss": 6.8866,
       "step": 306
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.000606,
-      "loss": 6.9587,
+      "epoch": 0.69,
+      "learning_rate": 0.000608,
+      "loss": 6.9124,
       "step": 307
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000608,
-      "loss": 6.987,
+      "epoch": 0.69,
+      "learning_rate": 0.00061,
+      "loss": 7.0546,
       "step": 308
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.00061,
-      "loss": 6.8916,
+      "epoch": 0.69,
+      "learning_rate": 0.000612,
+      "loss": 6.8526,
       "step": 309
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000612,
-      "loss": 7.0186,
+      "epoch": 0.7,
+      "learning_rate": 0.000614,
+      "loss": 6.9688,
       "step": 310
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000614,
-      "loss": 7.0748,
+      "epoch": 0.7,
+      "learning_rate": 0.000616,
+      "loss": 6.9559,
       "step": 311
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000616,
-      "loss": 6.9488,
+      "epoch": 0.7,
+      "learning_rate": 0.0006180000000000001,
+      "loss": 6.9221,
       "step": 312
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0006180000000000001,
-      "loss": 6.9676,
+      "epoch": 0.7,
+      "learning_rate": 0.00062,
+      "loss": 6.8472,
       "step": 313
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.00062,
-      "loss": 6.9319,
+      "epoch": 0.7,
+      "learning_rate": 0.000622,
+      "loss": 6.8412,
       "step": 314
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000622,
-      "loss": 6.7853,
+      "epoch": 0.71,
+      "learning_rate": 0.000624,
+      "loss": 6.997,
       "step": 315
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000624,
-      "loss": 7.0159,
+      "epoch": 0.71,
+      "learning_rate": 0.000626,
+      "loss": 6.8871,
       "step": 316
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000626,
-      "loss": 6.9082,
+      "epoch": 0.71,
+      "learning_rate": 0.000628,
+      "loss": 6.856,
       "step": 317
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000628,
-      "loss": 6.8242,
+      "epoch": 0.71,
+      "learning_rate": 0.00063,
+      "loss": 6.7757,
       "step": 318
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.00063,
-      "loss": 6.7317,
+      "epoch": 0.72,
+      "learning_rate": 0.000632,
+      "loss": 6.8653,
       "step": 319
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000632,
-      "loss": 6.9907,
+      "epoch": 0.72,
+      "learning_rate": 0.000634,
+      "loss": 6.8784,
       "step": 320
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000634,
-      "loss": 6.9169,
+      "epoch": 0.72,
+      "learning_rate": 0.0006360000000000001,
+      "loss": 6.8707,
       "step": 321
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.0006360000000000001,
-      "loss": 6.7951,
+      "epoch": 0.72,
+      "learning_rate": 0.000638,
+      "loss": 6.826,
       "step": 322
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000638,
-      "loss": 6.9173,
+      "epoch": 0.72,
+      "learning_rate": 0.00064,
+      "loss": 6.9139,
       "step": 323
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.00064,
-      "loss": 6.9046,
+      "epoch": 0.73,
+      "learning_rate": 0.000642,
+      "loss": 6.8378,
       "step": 324
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.000642,
-      "loss": 6.98,
+      "epoch": 0.73,
+      "learning_rate": 0.000644,
+      "loss": 6.8941,
       "step": 325
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000644,
-      "loss": 6.9191,
+      "epoch": 0.73,
+      "learning_rate": 0.000646,
+      "loss": 6.9395,
       "step": 326
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000646,
-      "loss": 6.7969,
+      "epoch": 0.73,
+      "learning_rate": 0.000648,
+      "loss": 6.8839,
       "step": 327
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000648,
-      "loss": 6.9119,
+      "epoch": 0.74,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 6.825,
       "step": 328
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.0006500000000000001,
-      "loss": 6.946,
+      "epoch": 0.74,
+      "learning_rate": 0.000652,
+      "loss": 6.9028,
       "step": 329
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000652,
-      "loss": 6.9577,
+      "epoch": 0.74,
+      "learning_rate": 0.0006540000000000001,
+      "loss": 6.9174,
       "step": 330
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.0006540000000000001,
-      "loss": 7.0092,
+      "epoch": 0.74,
+      "learning_rate": 0.000656,
+      "loss": 6.8735,
       "step": 331
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.000656,
-      "loss": 6.6848,
+      "epoch": 0.74,
+      "learning_rate": 0.0006580000000000001,
+      "loss": 6.7823,
       "step": 332
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.0006580000000000001,
-      "loss": 6.8792,
+      "epoch": 0.75,
+      "learning_rate": 0.00066,
+      "loss": 6.9539,
       "step": 333
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.00066,
-      "loss": 6.9351,
+      "epoch": 0.75,
+      "learning_rate": 0.000662,
+      "loss": 6.8746,
       "step": 334
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000662,
-      "loss": 6.9868,
+      "epoch": 0.75,
+      "learning_rate": 0.0006640000000000001,
+      "loss": 6.9545,
       "step": 335
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0006640000000000001,
-      "loss": 6.7772,
+      "epoch": 0.75,
+      "learning_rate": 0.000666,
+      "loss": 6.8598,
       "step": 336
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000666,
-      "loss": 6.8113,
+      "epoch": 0.76,
+      "learning_rate": 0.0006680000000000001,
+      "loss": 6.731,
       "step": 337
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0006680000000000001,
-      "loss": 6.932,
+      "epoch": 0.76,
+      "learning_rate": 0.00067,
+      "loss": 6.9194,
       "step": 338
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.00067,
-      "loss": 6.9257,
+      "epoch": 0.76,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 6.8805,
       "step": 339
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0006720000000000001,
-      "loss": 6.8515,
+      "epoch": 0.76,
+      "learning_rate": 0.000674,
+      "loss": 6.9178,
       "step": 340
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.000674,
-      "loss": 6.9214,
+      "epoch": 0.76,
+      "learning_rate": 0.0006760000000000001,
+      "loss": 6.7853,
       "step": 341
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0006760000000000001,
-      "loss": 6.8224,
+      "epoch": 0.77,
+      "learning_rate": 0.0006780000000000001,
+      "loss": 6.9158,
       "step": 342
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0006780000000000001,
-      "loss": 6.8009,
+      "epoch": 0.77,
+      "learning_rate": 0.00068,
+      "loss": 6.7655,
       "step": 343
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.00068,
-      "loss": 6.7492,
+      "epoch": 0.77,
+      "learning_rate": 0.0006820000000000001,
+      "loss": 6.7349,
       "step": 344
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0006820000000000001,
-      "loss": 6.6741,
+      "epoch": 0.77,
+      "learning_rate": 0.000684,
+      "loss": 6.7429,
       "step": 345
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000684,
-      "loss": 6.7377,
+      "epoch": 0.78,
+      "learning_rate": 0.0006860000000000001,
+      "loss": 6.821,
       "step": 346
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0006860000000000001,
-      "loss": 6.7077,
+      "epoch": 0.78,
+      "learning_rate": 0.0006879999999999999,
+      "loss": 6.6749,
       "step": 347
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0006879999999999999,
-      "loss": 6.7429,
+      "epoch": 0.78,
+      "learning_rate": 0.00069,
+      "loss": 6.5191,
       "step": 348
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.00069,
-      "loss": 6.6844,
+      "epoch": 0.78,
+      "learning_rate": 0.000692,
+      "loss": 6.613,
       "step": 349
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000692,
-      "loss": 6.5364,
+      "epoch": 0.78,
+      "learning_rate": 0.000694,
+      "loss": 6.3143,
       "step": 350
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000694,
-      "loss": 6.9826,
+      "epoch": 0.79,
+      "learning_rate": 0.000696,
+      "loss": 7.042,
       "step": 351
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.000696,
-      "loss": 6.9642,
+      "epoch": 0.79,
+      "learning_rate": 0.0006979999999999999,
+      "loss": 7.0495,
       "step": 352
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0006979999999999999,
-      "loss": 6.9125,
+      "epoch": 0.79,
+      "learning_rate": 0.0007,
+      "loss": 6.9122,
       "step": 353
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0007,
-      "loss": 6.8606,
+      "epoch": 0.79,
+      "learning_rate": 0.0007019999999999999,
+      "loss": 6.8473,
       "step": 354
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0007019999999999999,
-      "loss": 6.9799,
+      "epoch": 0.8,
+      "learning_rate": 0.000704,
+      "loss": 6.9169,
       "step": 355
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000704,
-      "loss": 6.8791,
+      "epoch": 0.8,
+      "learning_rate": 0.0007059999999999999,
+      "loss": 7.0432,
       "step": 356
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0007059999999999999,
-      "loss": 7.1025,
+      "epoch": 0.8,
+      "learning_rate": 0.000708,
+      "loss": 6.9105,
       "step": 357
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000708,
-      "loss": 6.8395,
+      "epoch": 0.8,
+      "learning_rate": 0.00071,
+      "loss": 6.9928,
       "step": 358
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.00071,
-      "loss": 6.954,
+      "epoch": 0.8,
+      "learning_rate": 0.000712,
+      "loss": 6.8983,
       "step": 359
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000712,
-      "loss": 6.7964,
+      "epoch": 0.81,
+      "learning_rate": 0.000714,
+      "loss": 6.8808,
       "step": 360
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.000714,
-      "loss": 6.8786,
+      "epoch": 0.81,
+      "learning_rate": 0.000716,
+      "loss": 6.8656,
       "step": 361
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000716,
-      "loss": 6.965,
+      "epoch": 0.81,
+      "learning_rate": 0.000718,
+      "loss": 6.7979,
       "step": 362
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000718,
-      "loss": 6.8654,
+      "epoch": 0.81,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 6.754,
       "step": 363
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.0007199999999999999,
-      "loss": 6.9316,
+      "epoch": 0.82,
+      "learning_rate": 0.000722,
+      "loss": 6.8061,
       "step": 364
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000722,
-      "loss": 6.8547,
+      "epoch": 0.82,
+      "learning_rate": 0.000724,
+      "loss": 6.9021,
       "step": 365
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000724,
-      "loss": 6.8077,
+      "epoch": 0.82,
+      "learning_rate": 0.000726,
+      "loss": 6.8486,
       "step": 366
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000726,
-      "loss": 6.8695,
+      "epoch": 0.82,
+      "learning_rate": 0.000728,
+      "loss": 6.9699,
       "step": 367
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000728,
-      "loss": 6.8131,
+      "epoch": 0.83,
+      "learning_rate": 0.00073,
+      "loss": 6.9067,
       "step": 368
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.00073,
-      "loss": 6.7857,
+      "epoch": 0.83,
+      "learning_rate": 0.000732,
+      "loss": 6.7865,
       "step": 369
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.000732,
-      "loss": 6.9622,
+      "epoch": 0.83,
+      "learning_rate": 0.000734,
+      "loss": 6.8306,
       "step": 370
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000734,
-      "loss": 6.8707,
+      "epoch": 0.83,
+      "learning_rate": 0.000736,
+      "loss": 6.8887,
       "step": 371
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000736,
-      "loss": 7.0577,
+      "epoch": 0.83,
+      "learning_rate": 0.000738,
+      "loss": 6.9094,
       "step": 372
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000738,
-      "loss": 6.884,
+      "epoch": 0.84,
+      "learning_rate": 0.00074,
+      "loss": 6.8319,
       "step": 373
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.00074,
-      "loss": 6.9621,
+      "epoch": 0.84,
+      "learning_rate": 0.000742,
+      "loss": 6.7177,
       "step": 374
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000742,
-      "loss": 6.7632,
+      "epoch": 0.84,
+      "learning_rate": 0.000744,
+      "loss": 6.8512,
       "step": 375
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000744,
-      "loss": 6.868,
+      "epoch": 0.84,
+      "learning_rate": 0.000746,
+      "loss": 6.7854,
       "step": 376
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000746,
-      "loss": 6.7298,
+      "epoch": 0.85,
+      "learning_rate": 0.000748,
+      "loss": 6.8627,
       "step": 377
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.000748,
-      "loss": 6.8263,
+      "epoch": 0.85,
+      "learning_rate": 0.00075,
+      "loss": 6.8255,
       "step": 378
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.00075,
-      "loss": 6.8072,
+      "epoch": 0.85,
+      "learning_rate": 0.0007520000000000001,
+      "loss": 6.8259,
       "step": 379
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0007520000000000001,
-      "loss": 6.9167,
+      "epoch": 0.85,
+      "learning_rate": 0.000754,
+      "loss": 6.8465,
       "step": 380
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000754,
-      "loss": 6.9311,
+      "epoch": 0.85,
+      "learning_rate": 0.000756,
+      "loss": 6.9686,
       "step": 381
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000756,
-      "loss": 6.7368,
+      "epoch": 0.86,
+      "learning_rate": 0.000758,
+      "loss": 6.8864,
       "step": 382
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000758,
-      "loss": 6.9308,
+      "epoch": 0.86,
+      "learning_rate": 0.00076,
+      "loss": 6.8678,
       "step": 383
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.00076,
-      "loss": 6.9564,
+      "epoch": 0.86,
+      "learning_rate": 0.000762,
+      "loss": 6.8329,
       "step": 384
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000762,
-      "loss": 6.7991,
+      "epoch": 0.86,
+      "learning_rate": 0.000764,
+      "loss": 6.9062,
       "step": 385
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000764,
-      "loss": 6.9136,
+      "epoch": 0.87,
+      "learning_rate": 0.0007660000000000001,
+      "loss": 6.8693,
       "step": 386
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0007660000000000001,
-      "loss": 6.8608,
+      "epoch": 0.87,
+      "learning_rate": 0.000768,
+      "loss": 6.8085,
       "step": 387
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.000768,
-      "loss": 6.819,
+      "epoch": 0.87,
+      "learning_rate": 0.0007700000000000001,
+      "loss": 6.8915,
       "step": 388
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0007700000000000001,
-      "loss": 6.8589,
+      "epoch": 0.87,
+      "learning_rate": 0.000772,
+      "loss": 6.8316,
       "step": 389
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.000772,
-      "loss": 7.0872,
+      "epoch": 0.87,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 6.7548,
       "step": 390
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0007740000000000001,
-      "loss": 6.8762,
+      "epoch": 0.88,
+      "learning_rate": 0.000776,
+      "loss": 6.869,
       "step": 391
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.000776,
-      "loss": 6.8362,
+      "epoch": 0.88,
+      "learning_rate": 0.000778,
+      "loss": 6.8538,
       "step": 392
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.000778,
-      "loss": 6.8752,
+      "epoch": 0.88,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 6.7857,
       "step": 393
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0007800000000000001,
-      "loss": 7.0297,
+      "epoch": 0.88,
+      "learning_rate": 0.000782,
+      "loss": 6.7821,
       "step": 394
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.000782,
-      "loss": 6.8177,
+      "epoch": 0.89,
+      "learning_rate": 0.0007840000000000001,
+      "loss": 6.9268,
       "step": 395
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0007840000000000001,
-      "loss": 6.6555,
+      "epoch": 0.89,
+      "learning_rate": 0.000786,
+      "loss": 6.8703,
       "step": 396
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.000786,
-      "loss": 6.635,
+      "epoch": 0.89,
+      "learning_rate": 0.0007880000000000001,
+      "loss": 6.8004,
       "step": 397
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0007880000000000001,
-      "loss": 6.6191,
+      "epoch": 0.89,
+      "learning_rate": 0.00079,
+      "loss": 6.6531,
       "step": 398
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.00079,
-      "loss": 6.6694,
+      "epoch": 0.89,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 6.6996,
       "step": 399
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0007920000000000001,
-      "loss": 6.4359,
+      "epoch": 0.9,
+      "learning_rate": 0.0007940000000000001,
+      "loss": 6.371,
       "step": 400
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0007940000000000001,
-      "loss": 7.131,
+      "epoch": 0.9,
+      "learning_rate": 0.000796,
+      "loss": 7.1791,
       "step": 401
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.000796,
-      "loss": 7.0349,
+      "epoch": 0.9,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 7.0062,
       "step": 402
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0007980000000000001,
-      "loss": 6.9991,
+      "epoch": 0.9,
+      "learning_rate": 0.0008,
+      "loss": 7.0965,
       "step": 403
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0008,
-      "loss": 6.9237,
+      "epoch": 0.91,
+      "learning_rate": 0.0008020000000000001,
+      "loss": 6.9202,
       "step": 404
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0008020000000000001,
-      "loss": 7.0466,
+      "epoch": 0.91,
+      "learning_rate": 0.000804,
+      "loss": 6.868,
       "step": 405
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.000804,
-      "loss": 6.9054,
+      "epoch": 0.91,
+      "learning_rate": 0.0008060000000000001,
+      "loss": 6.9814,
       "step": 406
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0008060000000000001,
-      "loss": 6.9636,
+      "epoch": 0.91,
+      "learning_rate": 0.000808,
+      "loss": 7.017,
       "step": 407
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.000808,
-      "loss": 6.8282,
+      "epoch": 0.91,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 6.8681,
       "step": 408
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0008100000000000001,
-      "loss": 7.0725,
+      "epoch": 0.92,
+      "learning_rate": 0.0008120000000000001,
+      "loss": 6.8563,
       "step": 409
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0008120000000000001,
-      "loss": 7.052,
+      "epoch": 0.92,
+      "learning_rate": 0.0008139999999999999,
+      "loss": 6.8619,
       "step": 410
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0008139999999999999,
-      "loss": 6.8894,
+      "epoch": 0.92,
+      "learning_rate": 0.000816,
+      "loss": 6.951,
       "step": 411
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.000816,
-      "loss": 6.9373,
+      "epoch": 0.92,
+      "learning_rate": 0.0008179999999999999,
+      "loss": 7.847,
       "step": 412
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0008179999999999999,
-      "loss": 6.8408,
+      "epoch": 0.93,
+      "learning_rate": 0.00082,
+      "loss": 6.9371,
       "step": 413
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.00082,
-      "loss": 6.7604,
+      "epoch": 0.93,
+      "learning_rate": 0.0008219999999999999,
+      "loss": 6.9672,
       "step": 414
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.0008219999999999999,
-      "loss": 6.8211,
+      "epoch": 0.93,
+      "learning_rate": 0.000824,
+      "loss": 6.9922,
       "step": 415
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.000824,
-      "loss": 6.8882,
+      "epoch": 0.93,
+      "learning_rate": 0.000826,
+      "loss": 6.9618,
       "step": 416
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.000826,
-      "loss": 6.818,
+      "epoch": 0.93,
+      "learning_rate": 0.000828,
+      "loss": 6.9311,
       "step": 417
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.000828,
-      "loss": 6.9055,
+      "epoch": 0.94,
+      "learning_rate": 0.00083,
+      "loss": 6.855,
       "step": 418
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.00083,
-      "loss": 6.8693,
+      "epoch": 0.94,
+      "learning_rate": 0.000832,
+      "loss": 6.8993,
       "step": 419
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.000832,
-      "loss": 6.9117,
+      "epoch": 0.94,
+      "learning_rate": 0.000834,
+      "loss": 7.1197,
       "step": 420
     },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.000834,
-      "loss": 6.8999,
-      "step": 421
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.0008359999999999999,
-      "loss": 6.9682,
-      "step": 422
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 0.000838,
-      "loss": 6.9003,
-      "step": 423
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.00084,
-      "loss": 6.8791,
-      "step": 424
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.000842,
-      "loss": 6.9391,
-      "step": 425
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.000844,
-      "loss": 7.044,
-      "step": 426
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.000846,
-      "loss": 6.8145,
-      "step": 427
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.000848,
-      "loss": 6.9727,
-      "step": 428
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.00085,
-      "loss": 6.8726,
-      "step": 429
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.000852,
-      "loss": 6.8341,
-      "step": 430
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.000854,
-      "loss": 6.99,
-      "step": 431
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 0.000856,
-      "loss": 6.9364,
-      "step": 432
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000858,
-      "loss": 6.5882,
-      "step": 433
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00086,
-      "loss": 6.8557,
-      "step": 434
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000862,
-      "loss": 6.7636,
-      "step": 435
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000864,
-      "loss": 7.0034,
-      "step": 436
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000866,
-      "loss": 6.8664,
-      "step": 437
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.0008680000000000001,
-      "loss": 6.9297,
-      "step": 438
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.00087,
-      "loss": 6.7585,
-      "step": 439
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000872,
-      "loss": 6.8399,
-      "step": 440
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 0.000874,
-      "loss": 6.9393,
-      "step": 441
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000876,
-      "loss": 6.7923,
-      "step": 442
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000878,
-      "loss": 6.784,
-      "step": 443
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.00088,
-      "loss": 6.9066,
-      "step": 444
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000882,
-      "loss": 6.7017,
-      "step": 445
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000884,
-      "loss": 6.6741,
-      "step": 446
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008860000000000001,
-      "loss": 6.6476,
-      "step": 447
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000888,
-      "loss": 6.6637,
-      "step": 448
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.0008900000000000001,
-      "loss": 6.3006,
-      "step": 449
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 0.000892,
-      "loss": 6.2887,
-      "step": 450
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.000894,
-      "loss": 7.0309,
-      "step": 451
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.000896,
-      "loss": 7.1196,
-      "step": 452
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.000898,
-      "loss": 7.1074,
-      "step": 453
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0009000000000000001,
-      "loss": 6.9244,
-      "step": 454
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.000902,
-      "loss": 7.0235,
-      "step": 455
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0009040000000000001,
-      "loss": 7.0625,
-      "step": 456
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.000906,
-      "loss": 6.8702,
-      "step": 457
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.0009080000000000001,
-      "loss": 6.9522,
-      "step": 458
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 0.00091,
-      "loss": 6.9222,
-      "step": 459
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.000912,
-      "loss": 6.988,
-      "step": 460
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0009140000000000001,
-      "loss": 6.8307,
-      "step": 461
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.000916,
-      "loss": 6.935,
-      "step": 462
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0009180000000000001,
-      "loss": 6.8141,
-      "step": 463
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.00092,
-      "loss": 6.7427,
-      "step": 464
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0009220000000000001,
-      "loss": 6.9158,
-      "step": 465
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.000924,
-      "loss": 6.8655,
-      "step": 466
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0009260000000000001,
-      "loss": 6.8031,
-      "step": 467
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 0.0009280000000000001,
-      "loss": 6.8602,
-      "step": 468
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00093,
-      "loss": 6.8451,
-      "step": 469
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0009320000000000001,
-      "loss": 6.904,
-      "step": 470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.000934,
-      "loss": 6.7975,
-      "step": 471
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0009360000000000001,
-      "loss": 6.816,
-      "step": 472
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.0009379999999999999,
-      "loss": 6.8597,
-      "step": 473
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.00094,
-      "loss": 6.7457,
-      "step": 474
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.000942,
-      "loss": 6.9132,
-      "step": 475
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.000944,
-      "loss": 6.933,
-      "step": 476
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 0.000946,
-      "loss": 7.0363,
-      "step": 477
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000948,
-      "loss": 6.7903,
-      "step": 478
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00095,
-      "loss": 6.8356,
-      "step": 479
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0009519999999999999,
-      "loss": 6.8078,
-      "step": 480
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000954,
-      "loss": 6.8185,
-      "step": 481
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.0009559999999999999,
-      "loss": 6.8178,
-      "step": 482
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000958,
-      "loss": 6.8594,
-      "step": 483
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.00096,
-      "loss": 6.9006,
-      "step": 484
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000962,
-      "loss": 6.8851,
-      "step": 485
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 0.000964,
-      "loss": 6.8724,
-      "step": 486
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000966,
-      "loss": 6.9267,
-      "step": 487
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000968,
-      "loss": 6.7862,
-      "step": 488
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.0009699999999999999,
-      "loss": 7.0151,
-      "step": 489
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000972,
-      "loss": 7.0268,
-      "step": 490
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000974,
-      "loss": 6.8296,
-      "step": 491
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000976,
-      "loss": 6.8505,
-      "step": 492
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000978,
-      "loss": 6.8255,
-      "step": 493
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.00098,
-      "loss": 6.7825,
-      "step": 494
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 0.000982,
-      "loss": 6.9151,
-      "step": 495
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000984,
-      "loss": 6.5827,
-      "step": 496
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.0009860000000000001,
-      "loss": 6.2984,
-      "step": 497
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000988,
-      "loss": 6.9957,
-      "step": 498
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.00099,
-      "loss": 6.6962,
-      "step": 499
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000992,
-      "loss": 6.302,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "eval_loss": 6.801183223724365,
-      "eval_runtime": 462.9752,
-      "eval_samples_per_second": 5.707,
-      "eval_steps_per_second": 0.715,
-      "eval_wer": 1.0129769629144592,
-      "step": 500
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000994,
-      "loss": 7.2801,
-      "step": 501
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000996,
-      "loss": 6.8163,
-      "step": 502
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 0.000998,
-      "loss": 6.9428,
-      "step": 503
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.001,
-      "loss": 7.0299,
-      "step": 504
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0009974489795918369,
-      "loss": 6.9996,
-      "step": 505
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0009948979591836735,
-      "loss": 6.9882,
-      "step": 506
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.00099234693877551,
-      "loss": 6.9717,
-      "step": 507
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.000989795918367347,
-      "loss": 6.833,
-      "step": 508
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0009872448979591838,
-      "loss": 6.8901,
-      "step": 509
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0009846938775510204,
-      "loss": 6.9404,
-      "step": 510
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0009821428571428572,
-      "loss": 6.8448,
-      "step": 511
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 0.0009795918367346938,
-      "loss": 6.8849,
-      "step": 512
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009770408163265307,
-      "loss": 7.0128,
-      "step": 513
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009744897959183674,
-      "loss": 6.9506,
-      "step": 514
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009719387755102041,
-      "loss": 6.9002,
-      "step": 515
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009693877551020408,
-      "loss": 6.7096,
-      "step": 516
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009668367346938776,
-      "loss": 6.9798,
-      "step": 517
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009642857142857143,
-      "loss": 6.8173,
-      "step": 518
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009617346938775511,
-      "loss": 6.9689,
-      "step": 519
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009591836734693877,
-      "loss": 6.9611,
-      "step": 520
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 0.0009566326530612245,
-      "loss": 6.7893,
-      "step": 521
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009540816326530613,
-      "loss": 6.7837,
-      "step": 522
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.000951530612244898,
-      "loss": 6.6979,
-      "step": 523
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009489795918367348,
-      "loss": 6.7976,
-      "step": 524
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009464285714285714,
-      "loss": 6.9382,
-      "step": 525
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009438775510204082,
-      "loss": 6.767,
-      "step": 526
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009413265306122449,
-      "loss": 6.8538,
-      "step": 527
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009387755102040817,
-      "loss": 6.8889,
-      "step": 528
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009362244897959184,
-      "loss": 6.9027,
-      "step": 529
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 0.0009336734693877551,
-      "loss": 7.0115,
-      "step": 530
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009311224489795918,
-      "loss": 6.8262,
-      "step": 531
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009285714285714287,
-      "loss": 6.7705,
-      "step": 532
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009260204081632653,
-      "loss": 6.7862,
-      "step": 533
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.000923469387755102,
-      "loss": 6.7993,
-      "step": 534
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009209183673469387,
-      "loss": 6.7714,
-      "step": 535
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009183673469387756,
-      "loss": 6.9337,
-      "step": 536
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009158163265306123,
-      "loss": 6.8392,
-      "step": 537
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009132653061224489,
-      "loss": 6.9117,
-      "step": 538
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 0.0009107142857142857,
-      "loss": 6.9215,
-      "step": 539
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0009081632653061225,
-      "loss": 6.8691,
-      "step": 540
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0009056122448979592,
-      "loss": 6.9119,
-      "step": 541
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0009030612244897959,
-      "loss": 6.8511,
-      "step": 542
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0009005102040816327,
-      "loss": 6.7294,
-      "step": 543
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0008979591836734694,
-      "loss": 6.6228,
-      "step": 544
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0008954081632653062,
-      "loss": 6.7164,
-      "step": 545
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0008928571428571429,
-      "loss": 6.5826,
-      "step": 546
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0008903061224489796,
-      "loss": 6.8637,
-      "step": 547
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 0.0008877551020408163,
-      "loss": 6.6897,
-      "step": 548
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008852040816326531,
-      "loss": 6.2739,
-      "step": 549
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008826530612244898,
-      "loss": 6.5244,
-      "step": 550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008801020408163265,
-      "loss": 7.094,
-      "step": 551
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008775510204081633,
-      "loss": 7.1794,
-      "step": 552
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.000875,
-      "loss": 6.9146,
-      "step": 553
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008724489795918367,
-      "loss": 6.9065,
-      "step": 554
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008698979591836736,
-      "loss": 6.8946,
-      "step": 555
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008673469387755102,
-      "loss": 6.9136,
-      "step": 556
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 0.0008647959183673469,
-      "loss": 6.8388,
-      "step": 557
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008622448979591838,
-      "loss": 6.9811,
-      "step": 558
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008596938775510205,
-      "loss": 6.8711,
-      "step": 559
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008571428571428571,
-      "loss": 6.8007,
-      "step": 560
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008545918367346938,
-      "loss": 6.8772,
-      "step": 561
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008520408163265307,
-      "loss": 6.7457,
-      "step": 562
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008494897959183674,
-      "loss": 6.8083,
-      "step": 563
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008469387755102041,
-      "loss": 6.8157,
-      "step": 564
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008443877551020408,
-      "loss": 7.0156,
-      "step": 565
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 0.0008418367346938776,
-      "loss": 6.6783,
-      "step": 566
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008392857142857143,
-      "loss": 6.8623,
-      "step": 567
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008367346938775511,
-      "loss": 6.6903,
-      "step": 568
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008341836734693877,
-      "loss": 6.7772,
-      "step": 569
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008316326530612245,
-      "loss": 6.8977,
-      "step": 570
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008290816326530613,
-      "loss": 6.7691,
-      "step": 571
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.000826530612244898,
-      "loss": 6.8898,
-      "step": 572
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008239795918367348,
-      "loss": 6.7187,
-      "step": 573
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008214285714285714,
-      "loss": 6.6245,
-      "step": 574
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 0.0008188775510204082,
-      "loss": 6.8416,
-      "step": 575
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0008163265306122449,
-      "loss": 6.6741,
-      "step": 576
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0008137755102040817,
-      "loss": 6.8406,
-      "step": 577
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0008112244897959184,
-      "loss": 6.9118,
-      "step": 578
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0008086734693877551,
-      "loss": 6.7172,
-      "step": 579
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0008061224489795918,
-      "loss": 6.956,
-      "step": 580
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0008035714285714287,
-      "loss": 39.2708,
-      "step": 581
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0008010204081632653,
-      "loss": 6.8661,
-      "step": 582
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.000798469387755102,
-      "loss": 6.8078,
-      "step": 583
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 0.0007959183673469387,
-      "loss": 6.881,
-      "step": 584
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007933673469387756,
-      "loss": 6.913,
-      "step": 585
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007908163265306123,
-      "loss": 6.9422,
-      "step": 586
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007882653061224489,
-      "loss": 6.9401,
-      "step": 587
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007857142857142857,
-      "loss": 6.8329,
-      "step": 588
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007831632653061225,
-      "loss": 6.7724,
-      "step": 589
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007806122448979592,
-      "loss": 6.8107,
-      "step": 590
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007780612244897959,
-      "loss": 6.821,
-      "step": 591
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007755102040816326,
-      "loss": 6.8722,
-      "step": 592
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 0.0007729591836734694,
-      "loss": 6.9778,
-      "step": 593
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007704081632653062,
-      "loss": 6.6417,
-      "step": 594
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007678571428571429,
-      "loss": 6.7114,
-      "step": 595
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007653061224489796,
-      "loss": 6.5943,
-      "step": 596
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007627551020408163,
-      "loss": 6.6128,
-      "step": 597
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007602040816326531,
-      "loss": 6.4908,
-      "step": 598
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007576530612244898,
-      "loss": 6.3733,
-      "step": 599
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007551020408163265,
-      "loss": 5.994,
-      "step": 600
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.0007525510204081633,
-      "loss": 7.0877,
-      "step": 601
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 0.00075,
-      "loss": 6.8344,
-      "step": 602
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007474489795918367,
-      "loss": 6.9558,
-      "step": 603
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007448979591836736,
-      "loss": 6.9163,
-      "step": 604
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007423469387755102,
-      "loss": 7.0152,
-      "step": 605
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007397959183673469,
-      "loss": 6.7098,
-      "step": 606
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007372448979591838,
-      "loss": 6.8684,
-      "step": 607
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007346938775510205,
-      "loss": 6.924,
-      "step": 608
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007321428571428571,
-      "loss": 6.8781,
-      "step": 609
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007295918367346938,
-      "loss": 6.8367,
-      "step": 610
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 0.0007270408163265307,
-      "loss": 6.8337,
-      "step": 611
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007244897959183674,
-      "loss": 6.8307,
-      "step": 612
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007219387755102041,
-      "loss": 6.7412,
-      "step": 613
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007193877551020408,
-      "loss": 6.8499,
-      "step": 614
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007168367346938776,
-      "loss": 7.0124,
-      "step": 615
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007142857142857143,
-      "loss": 6.7071,
-      "step": 616
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007117346938775511,
-      "loss": 6.816,
-      "step": 617
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007091836734693877,
-      "loss": 6.6482,
-      "step": 618
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 0.0007066326530612245,
-      "loss": 6.7514,
-      "step": 619
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0007040816326530613,
-      "loss": 6.8163,
-      "step": 620
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.000701530612244898,
-      "loss": 6.8456,
-      "step": 621
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0006989795918367347,
-      "loss": 6.7182,
-      "step": 622
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0006964285714285714,
-      "loss": 6.9294,
-      "step": 623
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0006938775510204082,
-      "loss": 6.7307,
-      "step": 624
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0006913265306122449,
-      "loss": 6.9086,
-      "step": 625
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0006887755102040817,
-      "loss": 6.8001,
-      "step": 626
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0006862244897959184,
-      "loss": 6.8676,
-      "step": 627
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 0.0006836734693877551,
-      "loss": 6.8621,
-      "step": 628
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006811224489795918,
-      "loss": 6.8476,
-      "step": 629
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006785714285714287,
-      "loss": 6.9485,
-      "step": 630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006760204081632653,
-      "loss": 6.7802,
-      "step": 631
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.000673469387755102,
-      "loss": 6.6621,
-      "step": 632
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006709183673469387,
-      "loss": 6.8801,
-      "step": 633
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006683673469387756,
-      "loss": 6.7649,
-      "step": 634
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006658163265306123,
-      "loss": 6.8576,
-      "step": 635
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006632653061224489,
-      "loss": 6.8018,
-      "step": 636
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 0.0006607142857142857,
-      "loss": 6.8263,
-      "step": 637
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006581632653061225,
-      "loss": 6.6283,
-      "step": 638
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006556122448979592,
-      "loss": 6.9193,
-      "step": 639
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006530612244897959,
-      "loss": 6.6701,
-      "step": 640
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006505102040816326,
-      "loss": 6.7321,
-      "step": 641
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006479591836734694,
-      "loss": 6.9245,
-      "step": 642
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006454081632653062,
-      "loss": 6.831,
-      "step": 643
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006428571428571429,
-      "loss": 6.593,
-      "step": 644
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006403061224489795,
-      "loss": 6.5909,
-      "step": 645
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 0.0006377551020408163,
-      "loss": 6.7362,
-      "step": 646
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006352040816326531,
-      "loss": 6.5246,
-      "step": 647
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006326530612244898,
-      "loss": 6.4485,
-      "step": 648
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006301020408163265,
-      "loss": 6.6398,
-      "step": 649
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006275510204081633,
-      "loss": 6.3344,
-      "step": 650
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.000625,
-      "loss": 7.1452,
-      "step": 651
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006224489795918367,
-      "loss": 7.0537,
-      "step": 652
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006198979591836736,
-      "loss": 6.9121,
-      "step": 653
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006173469387755102,
-      "loss": 7.147,
-      "step": 654
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 0.0006147959183673469,
-      "loss": 6.9293,
-      "step": 655
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0006122448979591838,
-      "loss": 6.9389,
-      "step": 656
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0006096938775510205,
-      "loss": 6.7247,
-      "step": 657
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0006071428571428571,
-      "loss": 7.0184,
-      "step": 658
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0006045918367346938,
-      "loss": 6.7145,
-      "step": 659
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0006020408163265307,
-      "loss": 6.7504,
-      "step": 660
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0005994897959183674,
-      "loss": 6.7735,
-      "step": 661
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0005969387755102041,
-      "loss": 6.8037,
-      "step": 662
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0005943877551020408,
-      "loss": 6.9202,
-      "step": 663
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 0.0005918367346938776,
-      "loss": 6.6985,
-      "step": 664
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005892857142857143,
-      "loss": 6.756,
-      "step": 665
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005867346938775511,
-      "loss": 6.8799,
-      "step": 666
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005841836734693877,
-      "loss": 6.8059,
-      "step": 667
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005816326530612245,
-      "loss": 6.7008,
-      "step": 668
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005790816326530613,
-      "loss": 6.8647,
-      "step": 669
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.000576530612244898,
-      "loss": 6.7537,
-      "step": 670
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005739795918367347,
-      "loss": 6.6661,
-      "step": 671
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005714285714285714,
-      "loss": 6.8096,
-      "step": 672
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 0.0005688775510204082,
-      "loss": 6.7315,
-      "step": 673
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005663265306122449,
-      "loss": 6.7256,
-      "step": 674
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005637755102040816,
-      "loss": 6.8112,
-      "step": 675
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005612244897959184,
-      "loss": 6.8975,
-      "step": 676
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005586734693877551,
-      "loss": 6.7649,
-      "step": 677
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005561224489795918,
-      "loss": 6.573,
-      "step": 678
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005535714285714287,
-      "loss": 6.7579,
-      "step": 679
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005510204081632653,
-      "loss": 6.6988,
-      "step": 680
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.000548469387755102,
-      "loss": 6.8681,
-      "step": 681
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 0.0005459183673469387,
-      "loss": 6.759,
-      "step": 682
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005433673469387756,
-      "loss": 6.7182,
-      "step": 683
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005408163265306123,
-      "loss": 6.8221,
-      "step": 684
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005382653061224489,
-      "loss": 6.8208,
-      "step": 685
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005357142857142857,
-      "loss": 6.665,
-      "step": 686
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005331632653061225,
-      "loss": 6.6355,
-      "step": 687
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005306122448979592,
-      "loss": 6.8958,
-      "step": 688
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005280612244897959,
-      "loss": 6.771,
-      "step": 689
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005255102040816326,
-      "loss": 6.7969,
-      "step": 690
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 0.0005229591836734694,
-      "loss": 6.6106,
-      "step": 691
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005204081632653062,
-      "loss": 6.6995,
-      "step": 692
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005178571428571429,
-      "loss": 6.5481,
-      "step": 693
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005153061224489795,
-      "loss": 6.7331,
-      "step": 694
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005127551020408163,
-      "loss": 6.7038,
-      "step": 695
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005102040816326531,
-      "loss": 6.6983,
-      "step": 696
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005076530612244898,
-      "loss": 6.4672,
-      "step": 697
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005051020408163265,
-      "loss": 6.3126,
-      "step": 698
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005025510204081633,
-      "loss": 6.336,
-      "step": 699
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 0.0005,
-      "loss": 5.9357,
-      "step": 700
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0004974489795918367,
-      "loss": 6.9862,
-      "step": 701
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0004948979591836735,
-      "loss": 6.8882,
-      "step": 702
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0004923469387755102,
-      "loss": 6.7945,
-      "step": 703
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0004897959183673469,
-      "loss": 6.6457,
-      "step": 704
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0004872448979591837,
-      "loss": 6.7859,
-      "step": 705
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.0004846938775510204,
-      "loss": 6.9166,
-      "step": 706
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00048214285714285715,
-      "loss": 6.8526,
-      "step": 707
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00047959183673469387,
-      "loss": 6.9288,
-      "step": 708
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 0.00047704081632653065,
-      "loss": 6.8102,
-      "step": 709
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0004744897959183674,
-      "loss": 6.8213,
-      "step": 710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0004719387755102041,
-      "loss": 6.7929,
-      "step": 711
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00046938775510204083,
-      "loss": 6.6578,
-      "step": 712
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00046683673469387755,
-      "loss": 6.7111,
-      "step": 713
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00046428571428571433,
-      "loss": 6.7643,
-      "step": 714
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.000461734693877551,
-      "loss": 6.8162,
-      "step": 715
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.0004591836734693878,
-      "loss": 6.8418,
-      "step": 716
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00045663265306122446,
-      "loss": 6.9736,
-      "step": 717
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 0.00045408163265306124,
-      "loss": 6.8005,
-      "step": 718
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00045153061224489796,
-      "loss": 6.6271,
-      "step": 719
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0004489795918367347,
-      "loss": 6.734,
-      "step": 720
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00044642857142857147,
-      "loss": 6.782,
-      "step": 721
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00044387755102040814,
-      "loss": 6.702,
-      "step": 722
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0004413265306122449,
-      "loss": 6.6708,
-      "step": 723
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00043877551020408165,
-      "loss": 6.869,
-      "step": 724
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.00043622448979591837,
-      "loss": 6.7586,
-      "step": 725
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 0.0004336734693877551,
-      "loss": 6.8788,
-      "step": 726
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0004311224489795919,
-      "loss": 6.787,
-      "step": 727
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00042857142857142855,
-      "loss": 6.6816,
-      "step": 728
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00042602040816326533,
-      "loss": 6.7356,
-      "step": 729
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00042346938775510206,
-      "loss": 6.8602,
-      "step": 730
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0004209183673469388,
-      "loss": 6.9794,
-      "step": 731
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00041836734693877556,
-      "loss": 6.8256,
-      "step": 732
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.00041581632653061223,
-      "loss": 6.7333,
-      "step": 733
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.000413265306122449,
-      "loss": 6.9562,
-      "step": 734
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 0.0004107142857142857,
-      "loss": 6.6694,
-      "step": 735
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00040816326530612246,
-      "loss": 6.7187,
-      "step": 736
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0004056122448979592,
-      "loss": 6.7483,
-      "step": 737
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0004030612244897959,
-      "loss": 6.8004,
-      "step": 738
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00040051020408163264,
-      "loss": 6.692,
-      "step": 739
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00039795918367346937,
-      "loss": 6.7981,
-      "step": 740
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.00039540816326530615,
-      "loss": 6.746,
-      "step": 741
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003928571428571429,
-      "loss": 6.7633,
-      "step": 742
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003903061224489796,
-      "loss": 6.5778,
-      "step": 743
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 0.0003877551020408163,
-      "loss": 6.7006,
-      "step": 744
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0003852040816326531,
-      "loss": 6.6274,
-      "step": 745
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0003826530612244898,
-      "loss": 6.799,
-      "step": 746
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00038010204081632656,
-      "loss": 6.4994,
-      "step": 747
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00037755102040816323,
-      "loss": 6.4169,
-      "step": 748
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.000375,
-      "loss": 6.3974,
-      "step": 749
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0003724489795918368,
-      "loss": 6.244,
-      "step": 750
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00036989795918367346,
-      "loss": 6.8295,
-      "step": 751
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.00036734693877551024,
-      "loss": 6.8337,
-      "step": 752
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 0.0003647959183673469,
-      "loss": 6.8462,
-      "step": 753
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0003622448979591837,
-      "loss": 6.7029,
-      "step": 754
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0003596938775510204,
-      "loss": 6.7513,
-      "step": 755
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00035714285714285714,
-      "loss": 6.8398,
-      "step": 756
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00035459183673469387,
-      "loss": 6.7451,
-      "step": 757
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00035204081632653065,
-      "loss": 6.6772,
-      "step": 758
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0003494897959183674,
-      "loss": 6.7679,
-      "step": 759
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0003469387755102041,
-      "loss": 6.7179,
-      "step": 760
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.0003443877551020408,
-      "loss": 6.7113,
-      "step": 761
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 0.00034183673469387755,
-      "loss": 6.7243,
-      "step": 762
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00033928571428571433,
-      "loss": 6.5729,
-      "step": 763
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.000336734693877551,
-      "loss": 6.5423,
-      "step": 764
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0003341836734693878,
-      "loss": 6.7336,
-      "step": 765
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00033163265306122445,
-      "loss": 6.6695,
-      "step": 766
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00032908163265306123,
-      "loss": 6.6613,
-      "step": 767
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00032653061224489796,
-      "loss": 6.7331,
-      "step": 768
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.0003239795918367347,
-      "loss": 6.8202,
-      "step": 769
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00032142857142857147,
-      "loss": 6.6126,
-      "step": 770
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 0.00031887755102040814,
-      "loss": 6.7847,
-      "step": 771
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0003163265306122449,
-      "loss": 6.8486,
-      "step": 772
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00031377551020408164,
-      "loss": 6.7756,
-      "step": 773
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00031122448979591837,
-      "loss": 6.5979,
-      "step": 774
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0003086734693877551,
-      "loss": 6.5082,
-      "step": 775
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0003061224489795919,
-      "loss": 6.6384,
-      "step": 776
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00030357142857142855,
-      "loss": 6.5427,
-      "step": 777
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0003010204081632653,
-      "loss": 6.7552,
-      "step": 778
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.00029846938775510205,
-      "loss": 6.7084,
-      "step": 779
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 0.0002959183673469388,
-      "loss": 6.6791,
-      "step": 780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00029336734693877556,
-      "loss": 6.6975,
-      "step": 781
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00029081632653061223,
-      "loss": 6.6267,
-      "step": 782
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.000288265306122449,
-      "loss": 6.7477,
-      "step": 783
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.0002857142857142857,
-      "loss": 6.6193,
-      "step": 784
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00028316326530612246,
-      "loss": 6.7194,
-      "step": 785
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.0002806122448979592,
-      "loss": 6.7151,
-      "step": 786
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.0002780612244897959,
-      "loss": 6.6539,
-      "step": 787
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00027551020408163264,
-      "loss": 6.7117,
-      "step": 788
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 0.00027295918367346936,
-      "loss": 6.6578,
-      "step": 789
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00027040816326530614,
-      "loss": 6.6434,
-      "step": 790
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00026785714285714287,
-      "loss": 6.5353,
-      "step": 791
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.0002653061224489796,
-      "loss": 6.4925,
-      "step": 792
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.0002627551020408163,
-      "loss": 6.6255,
-      "step": 793
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.0002602040816326531,
-      "loss": 6.5412,
-      "step": 794
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.0002576530612244898,
-      "loss": 6.6882,
-      "step": 795
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00025510204081632655,
-      "loss": 6.5216,
-      "step": 796
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.0002525510204081632,
-      "loss": 6.5592,
-      "step": 797
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 0.00025,
-      "loss": 6.5048,
-      "step": 798
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00024744897959183673,
-      "loss": 6.5659,
-      "step": 799
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00024489795918367346,
-      "loss": 6.0286,
-      "step": 800
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.0002423469387755102,
-      "loss": 6.8181,
-      "step": 801
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00023979591836734694,
-      "loss": 6.7498,
-      "step": 802
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.0002372448979591837,
-      "loss": 6.8077,
-      "step": 803
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00023469387755102041,
-      "loss": 6.7592,
-      "step": 804
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00023214285714285717,
-      "loss": 6.5583,
-      "step": 805
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.0002295918367346939,
-      "loss": 6.7822,
-      "step": 806
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 0.00022704081632653062,
-      "loss": 6.7743,
-      "step": 807
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00022448979591836734,
-      "loss": 6.5667,
-      "step": 808
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00022193877551020407,
-      "loss": 6.7492,
-      "step": 809
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00021938775510204082,
-      "loss": 6.6462,
-      "step": 810
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00021683673469387755,
-      "loss": 6.78,
-      "step": 811
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00021428571428571427,
-      "loss": 6.7903,
-      "step": 812
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00021173469387755103,
-      "loss": 6.6255,
-      "step": 813
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00020918367346938778,
-      "loss": 6.7149,
-      "step": 814
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.0002066326530612245,
-      "loss": 6.7514,
-      "step": 815
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 0.00020408163265306123,
-      "loss": 6.6297,
-      "step": 816
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00020153061224489796,
-      "loss": 6.7634,
-      "step": 817
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00019897959183673468,
-      "loss": 6.6703,
-      "step": 818
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00019642857142857144,
-      "loss": 6.5988,
-      "step": 819
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00019387755102040816,
-      "loss": 6.7712,
-      "step": 820
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.0001913265306122449,
-      "loss": 6.7524,
-      "step": 821
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00018877551020408161,
-      "loss": 6.6462,
-      "step": 822
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.0001862244897959184,
-      "loss": 6.7647,
-      "step": 823
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00018367346938775512,
-      "loss": 6.6581,
-      "step": 824
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 0.00018112244897959185,
-      "loss": 6.8945,
-      "step": 825
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00017857142857142857,
-      "loss": 6.7968,
-      "step": 826
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00017602040816326532,
-      "loss": 6.663,
-      "step": 827
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00017346938775510205,
-      "loss": 6.5837,
-      "step": 828
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00017091836734693878,
-      "loss": 6.6076,
-      "step": 829
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.0001683673469387755,
-      "loss": 6.642,
-      "step": 830
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00016581632653061223,
-      "loss": 6.6105,
-      "step": 831
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00016326530612244898,
-      "loss": 6.643,
-      "step": 832
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00016071428571428573,
-      "loss": 6.6844,
-      "step": 833
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 0.00015816326530612246,
-      "loss": 6.7065,
-      "step": 834
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00015561224489795918,
-      "loss": 6.6498,
-      "step": 835
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00015306122448979594,
-      "loss": 6.7948,
-      "step": 836
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00015051020408163266,
-      "loss": 6.67,
-      "step": 837
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.0001479591836734694,
-      "loss": 6.7704,
-      "step": 838
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00014540816326530611,
-      "loss": 6.6423,
-      "step": 839
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.00014285714285714284,
-      "loss": 6.568,
-      "step": 840
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 0.0001403061224489796,
-      "loss": 6.7596,
-      "step": 841
-    },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00013775510204081632,
-      "loss": 6.5554,
-      "step": 842
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.00013520408163265307,
-      "loss": 6.7847,
-      "step": 843
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.0001326530612244898,
-      "loss": 6.5701,
-      "step": 844
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.00013010204081632655,
-      "loss": 6.5643,
-      "step": 845
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.00012755102040816328,
-      "loss": 6.4616,
-      "step": 846
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 0.000125,
-      "loss": 6.2713,
-      "step": 847
+      "learning_rate": 0.0008359999999999999,
+      "loss": 7.0555,
+      "step": 421
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00012244897959183673,
-      "loss": 6.5107,
-      "step": 848
+      "learning_rate": 0.000838,
+      "loss": 7.0712,
+      "step": 422
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00011989795918367347,
-      "loss": 6.3484,
-      "step": 849
+      "learning_rate": 0.00084,
+      "loss": 7.0318,
+      "step": 423
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00011734693877551021,
-      "loss": 6.0158,
-      "step": 850
+      "learning_rate": 0.000842,
+      "loss": 7.1324,
+      "step": 424
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00011479591836734695,
-      "loss": 6.7668,
-      "step": 851
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 0.00011224489795918367,
-      "loss": 6.6319,
-      "step": 852
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 0.00010969387755102041,
-      "loss": 6.9075,
-      "step": 853
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 0.00010714285714285714,
-      "loss": 6.8001,
-      "step": 854
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 0.00010459183673469389,
-      "loss": 6.7318,
-      "step": 855
+      "learning_rate": 0.000844,
+      "loss": 6.9384,
+      "step": 425
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00010204081632653062,
-      "loss": 6.8121,
-      "step": 856
+      "learning_rate": 0.000846,
+      "loss": 7.0145,
+      "step": 426
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.948979591836734e-05,
-      "loss": 6.8324,
-      "step": 857
+      "learning_rate": 0.000848,
+      "loss": 6.9932,
+      "step": 427
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.693877551020408e-05,
-      "loss": 6.7603,
-      "step": 858
+      "learning_rate": 0.00085,
+      "loss": 6.8553,
+      "step": 428
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.438775510204081e-05,
-      "loss": 6.5621,
-      "step": 859
+      "learning_rate": 0.000852,
+      "loss": 6.8817,
+      "step": 429
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.183673469387756e-05,
-      "loss": 6.7401,
-      "step": 860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.928571428571429e-05,
-      "loss": 6.8457,
-      "step": 861
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.673469387755102e-05,
-      "loss": 6.7298,
-      "step": 862
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.418367346938775e-05,
-      "loss": 6.6993,
-      "step": 863
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 8.163265306122449e-05,
-      "loss": 6.8711,
-      "step": 864
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 7.908163265306123e-05,
-      "loss": 6.7687,
-      "step": 865
+      "learning_rate": 0.000854,
+      "loss": 6.9164,
+      "step": 430
     },
     {
       "epoch": 0.97,
-      "learning_rate": 7.653061224489797e-05,
-      "loss": 6.6089,
-      "step": 866
+      "learning_rate": 0.000856,
+      "loss": 6.9114,
+      "step": 431
     },
     {
       "epoch": 0.97,
-      "learning_rate": 7.39795918367347e-05,
-      "loss": 6.8424,
-      "step": 867
+      "learning_rate": 0.000858,
+      "loss": 7.0238,
+      "step": 432
     },
     {
       "epoch": 0.97,
-      "learning_rate": 7.142857142857142e-05,
-      "loss": 6.6108,
-      "step": 868
+      "learning_rate": 0.00086,
+      "loss": 6.9004,
+      "step": 433
     },
     {
       "epoch": 0.97,
-      "learning_rate": 6.887755102040816e-05,
-      "loss": 6.6635,
-      "step": 869
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.63265306122449e-05,
-      "loss": 6.6289,
-      "step": 870
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.377551020408164e-05,
-      "loss": 6.6012,
-      "step": 871
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 6.122448979591836e-05,
-      "loss": 6.631,
-      "step": 872
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.8673469387755104e-05,
-      "loss": 6.6756,
-      "step": 873
+      "learning_rate": 0.000862,
+      "loss": 6.862,
+      "step": 434
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.6122448979591836e-05,
-      "loss": 6.7161,
-      "step": 874
+      "learning_rate": 0.000864,
+      "loss": 7.0077,
+      "step": 435
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.357142857142857e-05,
-      "loss": 6.81,
-      "step": 875
+      "learning_rate": 0.000866,
+      "loss": 6.8605,
+      "step": 436
     },
     {
       "epoch": 0.98,
-      "learning_rate": 5.102040816326531e-05,
-      "loss": 6.5769,
-      "step": 876
+      "learning_rate": 0.0008680000000000001,
+      "loss": 6.8455,
+      "step": 437
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.846938775510204e-05,
-      "loss": 6.5125,
-      "step": 877
+      "learning_rate": 0.00087,
+      "loss": 6.9177,
+      "step": 438
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.591836734693878e-05,
-      "loss": 6.628,
-      "step": 878
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.336734693877551e-05,
-      "loss": 6.6966,
-      "step": 879
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.0816326530612245e-05,
-      "loss": 6.7886,
-      "step": 880
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.8265306122448984e-05,
-      "loss": 6.474,
-      "step": 881
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.571428571428571e-05,
-      "loss": 6.7015,
-      "step": 882
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 3.316326530612245e-05,
-      "loss": 6.5668,
-      "step": 883
+      "learning_rate": 0.000872,
+      "loss": 6.8778,
+      "step": 439
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.061224489795918e-05,
-      "loss": 6.616,
-      "step": 884
+      "learning_rate": 0.000874,
+      "loss": 6.8986,
+      "step": 440
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.8061224489795918e-05,
-      "loss": 6.6092,
-      "step": 885
+      "learning_rate": 0.000876,
+      "loss": 6.7385,
+      "step": 441
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.5510204081632654e-05,
-      "loss": 6.6438,
-      "step": 886
+      "learning_rate": 0.000878,
+      "loss": 6.9416,
+      "step": 442
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.295918367346939e-05,
-      "loss": 6.4031,
-      "step": 887
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.0408163265306123e-05,
-      "loss": 6.5722,
-      "step": 888
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.7857142857142855e-05,
-      "loss": 6.3931,
-      "step": 889
+      "learning_rate": 0.00088,
+      "loss": 6.6524,
+      "step": 443
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.530612244897959e-05,
-      "loss": 6.2909,
-      "step": 890
+      "learning_rate": 0.000882,
+      "loss": 6.8695,
+      "step": 444
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.2755102040816327e-05,
-      "loss": 6.1501,
-      "step": 891
+      "learning_rate": 0.000884,
+      "loss": 6.5954,
+      "step": 445
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.0204081632653061e-05,
-      "loss": 5.9213,
-      "step": 892
+      "learning_rate": 0.0008860000000000001,
+      "loss": 6.4923,
+      "step": 446
     },
     {
       "epoch": 1.0,
-      "step": 892,
+      "step": 446,
       "total_flos": 0.0,
-      "train_loss": 6.905522221407014,
-      "train_runtime": 5524.3174,
-      "train_samples_per_second": 5.166,
-      "train_steps_per_second": 0.161
+      "train_loss": 6.968126589941871,
+      "train_runtime": 4906.9209,
+      "train_samples_per_second": 5.816,
+      "train_steps_per_second": 0.091
     }
   ],
-  "max_steps": 892,
+  "max_steps": 446,
   "num_train_epochs": 1,
   "total_flos": 0.0,
   "trial_name": null,