diff --git "a/lora-checkpoints-3/trainer_state.json" "b/lora-checkpoints-3/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/lora-checkpoints-3/trainer_state.json"
@@ -0,0 +1,11647 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 1937,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6949152542372882e-06,
+      "loss": 1.2899,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3898305084745763e-06,
+      "loss": 1.0182,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.084745762711865e-06,
+      "loss": 1.1648,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.779661016949153e-06,
+      "loss": 1.0653,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.47457627118644e-06,
+      "loss": 1.0308,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.016949152542373e-05,
+      "loss": 1.4027,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1864406779661018e-05,
+      "loss": 1.2993,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3559322033898305e-05,
+      "loss": 1.032,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5254237288135596e-05,
+      "loss": 0.9718,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.694915254237288e-05,
+      "loss": 1.0297,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.864406779661017e-05,
+      "loss": 1.5744,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.033898305084746e-05,
+      "loss": 1.2003,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2033898305084748e-05,
+      "loss": 1.0179,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3728813559322036e-05,
+      "loss": 0.9415,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5423728813559322e-05,
+      "loss": 1.0417,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.711864406779661e-05,
+      "loss": 0.9565,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88135593220339e-05,
+      "loss": 1.0186,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.050847457627119e-05,
+      "loss": 1.021,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2203389830508473e-05,
+      "loss": 1.094,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.389830508474576e-05,
+      "loss": 0.9752,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.559322033898305e-05,
+      "loss": 0.8284,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.728813559322034e-05,
+      "loss": 0.8938,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.898305084745763e-05,
+      "loss": 1.013,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.067796610169492e-05,
+      "loss": 1.0115,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2372881355932206e-05,
+      "loss": 1.4292,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4067796610169495e-05,
+      "loss": 0.896,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5762711864406784e-05,
+      "loss": 0.9729,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.745762711864407e-05,
+      "loss": 0.8836,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.915254237288136e-05,
+      "loss": 0.8197,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0847457627118643e-05,
+      "loss": 1.5967,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.254237288135594e-05,
+      "loss": 0.8357,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.423728813559322e-05,
+      "loss": 1.5921,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.593220338983051e-05,
+      "loss": 0.7769,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.76271186440678e-05,
+      "loss": 0.9933,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.932203389830509e-05,
+      "loss": 0.9779,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.101694915254238e-05,
+      "loss": 0.8397,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.271186440677966e-05,
+      "loss": 0.7402,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.440677966101695e-05,
+      "loss": 0.8627,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.610169491525424e-05,
+      "loss": 1.1349,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.779661016949152e-05,
+      "loss": 0.8746,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.949152542372882e-05,
+      "loss": 1.0791,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.11864406779661e-05,
+      "loss": 0.7917,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.288135593220338e-05,
+      "loss": 0.847,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.457627118644068e-05,
+      "loss": 0.7737,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.627118644067796e-05,
+      "loss": 0.723,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.796610169491526e-05,
+      "loss": 1.338,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.966101694915254e-05,
+      "loss": 0.8153,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.135593220338983e-05,
+      "loss": 0.7999,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.305084745762712e-05,
+      "loss": 1.0855,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.474576271186441e-05,
+      "loss": 0.7852,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.644067796610171e-05,
+      "loss": 0.9082,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.813559322033899e-05,
+      "loss": 0.8591,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.983050847457629e-05,
+      "loss": 0.7523,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.152542372881357e-05,
+      "loss": 0.7452,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.322033898305085e-05,
+      "loss": 0.7614,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.491525423728815e-05,
+      "loss": 0.8243,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.661016949152543e-05,
+      "loss": 0.8477,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.830508474576272e-05,
+      "loss": 1.5161,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 1.1987,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999993004021326e-05,
+      "loss": 0.9333,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999972016104884e-05,
+      "loss": 0.825,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999937036309401e-05,
+      "loss": 1.3294,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99988806473277e-05,
+      "loss": 0.7035,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99982510151203e-05,
+      "loss": 0.8338,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999748146823376e-05,
+      "loss": 0.8713,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999657200882158e-05,
+      "loss": 0.8963,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99955226394288e-05,
+      "loss": 0.8235,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999433336299195e-05,
+      "loss": 0.729,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999300418283908e-05,
+      "loss": 1.1474,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999153510268979e-05,
+      "loss": 0.9153,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998992612665511e-05,
+      "loss": 0.7829,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99881772592376e-05,
+      "loss": 0.8356,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998628850533127e-05,
+      "loss": 0.8217,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99842598702216e-05,
+      "loss": 0.833,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99820913595855e-05,
+      "loss": 0.9078,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997978297949129e-05,
+      "loss": 0.8192,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997733473639876e-05,
+      "loss": 0.8498,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997474663715904e-05,
+      "loss": 1.3945,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997201868901463e-05,
+      "loss": 0.8609,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996915089959942e-05,
+      "loss": 0.7291,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996614327693858e-05,
+      "loss": 0.7774,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996299582944864e-05,
+      "loss": 0.8991,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995970856593738e-05,
+      "loss": 0.9675,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995628149560385e-05,
+      "loss": 0.8515,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995271462803834e-05,
+      "loss": 0.786,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994900797322233e-05,
+      "loss": 0.856,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994516154152849e-05,
+      "loss": 0.9323,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994117534372066e-05,
+      "loss": 0.8206,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993704939095376e-05,
+      "loss": 0.9141,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993278369477386e-05,
+      "loss": 0.8119,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9928378267118e-05,
+      "loss": 0.7887,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99238331203143e-05,
+      "loss": 1.4362,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991914826708188e-05,
+      "loss": 0.7523,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99143237205308e-05,
+      "loss": 0.7705,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9909359494162e-05,
+      "loss": 0.906,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990425560186736e-05,
+      "loss": 0.7624,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989901205792952e-05,
+      "loss": 0.9882,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989362887702203e-05,
+      "loss": 0.989,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988810607420911e-05,
+      "loss": 1.0716,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98824436649457e-05,
+      "loss": 0.7495,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987664166507748e-05,
+      "loss": 0.7852,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987070009084069e-05,
+      "loss": 0.7205,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986461895886218e-05,
+      "loss": 0.7737,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985839828615936e-05,
+      "loss": 0.8129,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985203809014008e-05,
+      "loss": 0.7981,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984553838860269e-05,
+      "loss": 0.9896,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983889919973586e-05,
+      "loss": 0.9212,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983212054211867e-05,
+      "loss": 0.7133,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982520243472045e-05,
+      "loss": 0.8014,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981814489690076e-05,
+      "loss": 0.8013,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981094794840938e-05,
+      "loss": 0.8484,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980361160938616e-05,
+      "loss": 0.7524,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979613590036108e-05,
+      "loss": 0.8436,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978852084225408e-05,
+      "loss": 0.8621,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978076645637505e-05,
+      "loss": 0.9756,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977287276442385e-05,
+      "loss": 1.0361,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976483978849007e-05,
+      "loss": 0.7949,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975666755105317e-05,
+      "loss": 0.9241,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974835607498222e-05,
+      "loss": 0.762,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973990538353603e-05,
+      "loss": 0.9376,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973131550036292e-05,
+      "loss": 1.0385,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972258644950074e-05,
+      "loss": 0.7703,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971371825537679e-05,
+      "loss": 0.85,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970471094280778e-05,
+      "loss": 0.8349,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969556453699965e-05,
+      "loss": 0.8393,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968627906354765e-05,
+      "loss": 0.88,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967685454843618e-05,
+      "loss": 0.9723,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966729101803871e-05,
+      "loss": 0.757,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965758849911774e-05,
+      "loss": 0.7839,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96477470188247e-05,
+      "loss": 0.8269,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963776660469995e-05,
+      "loss": 0.9267,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962764728467256e-05,
+      "loss": 0.8362,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961738908706036e-05,
+      "loss": 0.8023,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960699204056978e-05,
+      "loss": 0.8623,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959645617429587e-05,
+      "loss": 0.8488,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958578151772207e-05,
+      "loss": 1.0176,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957496810072027e-05,
+      "loss": 0.758,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956401595355063e-05,
+      "loss": 0.9498,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955292510686156e-05,
+      "loss": 0.6545,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954169559168958e-05,
+      "loss": 0.9842,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953032743945927e-05,
+      "loss": 0.8481,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951882068198318e-05,
+      "loss": 0.8975,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95071753514617e-05,
+      "loss": 0.7408,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949539148048304e-05,
+      "loss": 0.7109,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948346910202309e-05,
+      "loss": 0.7935,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947140824944531e-05,
+      "loss": 0.6937,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945920895650071e-05,
+      "loss": 0.7701,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944687125732767e-05,
+      "loss": 0.8839,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943439518645192e-05,
+      "loss": 0.9858,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942178077878639e-05,
+      "loss": 0.7852,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94090280696311e-05,
+      "loss": 0.7666,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939613709467316e-05,
+      "loss": 0.7178,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938310788998654e-05,
+      "loss": 0.9647,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936994049203206e-05,
+      "loss": 1.3293,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935663493765726e-05,
+      "loss": 0.8796,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93431912640963e-05,
+      "loss": 0.7018,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932960950896981e-05,
+      "loss": 0.7409,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931588971028489e-05,
+      "loss": 0.709,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93020319064349e-05,
+      "loss": 0.8357,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928803613619937e-05,
+      "loss": 0.8627,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9273902438744e-05,
+      "loss": 0.8429,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925963085362034e-05,
+      "loss": 0.7428,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924522142076593e-05,
+      "loss": 0.7954,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923067418050398e-05,
+      "loss": 0.7703,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921598917354337e-05,
+      "loss": 0.9293,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920116644097849e-05,
+      "loss": 0.8455,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918620602428915e-05,
+      "loss": 0.8137,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917110796534046e-05,
+      "loss": 0.8333,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91558723063827e-05,
+      "loss": 0.8867,
+      "step": 169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914049909005119e-05,
+      "loss": 0.7276,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912498835936621e-05,
+      "loss": 0.8077,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910934015773288e-05,
+      "loss": 0.8248,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909355452894098e-05,
+      "loss": 1.2503,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907763151716487e-05,
+      "loss": 0.9288,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90615711669634e-05,
+      "loss": 0.8973,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904537352327967e-05,
+      "loss": 0.8246,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902903863144107e-05,
+      "loss": 0.807,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9012566537159e-05,
+      "loss": 0.7596,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899595728652884e-05,
+      "loss": 0.7234,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897921092602978e-05,
+      "loss": 0.8362,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896232750252466e-05,
+      "loss": 0.7451,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894530706325995e-05,
+      "loss": 0.9197,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892814965586546e-05,
+      "loss": 0.7198,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891085532835437e-05,
+      "loss": 0.8119,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889342412912295e-05,
+      "loss": 0.7306,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887585610695055e-05,
+      "loss": 0.7503,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885815131099934e-05,
+      "loss": 0.7387,
+      "step": 187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88403097908143e-05,
+      "loss": 0.7598,
+      "step": 188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882233159632297e-05,
+      "loss": 0.695,
+      "step": 189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880421677783537e-05,
+      "loss": 0.7013,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878596538604388e-05,
+      "loss": 0.7889,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876757747202301e-05,
+      "loss": 0.7214,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874905308722935e-05,
+      "loss": 0.9459,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87303922835014e-05,
+      "loss": 0.9456,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871159511305937e-05,
+      "loss": 0.8062,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869266162850511e-05,
+      "loss": 0.8158,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867359188282192e-05,
+      "loss": 0.6718,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865438592937443e-05,
+      "loss": 1.2274,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863504382190838e-05,
+      "loss": 0.8339,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86155656145506e-05,
+      "loss": 0.8034,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859595136180871e-05,
+      "loss": 0.7716,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857620111857107e-05,
+      "loss": 0.669,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855631494010661e-05,
+      "loss": 0.7323,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853629288206462e-05,
+      "loss": 0.8591,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851613500047467e-05,
+      "loss": 0.7584,
+      "step": 205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849584135174641e-05,
+      "loss": 0.7521,
+      "step": 206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847541199266941e-05,
+      "loss": 0.785,
+      "step": 207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845484698041301e-05,
+      "loss": 0.8438,
+      "step": 208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843414637252615e-05,
+      "loss": 0.9328,
+      "step": 209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841331022693725e-05,
+      "loss": 0.6478,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8392338601954e-05,
+      "loss": 0.7885,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837123155626322e-05,
+      "loss": 0.9145,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83499891489307e-05,
+      "loss": 0.7773,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832861143940098e-05,
+      "loss": 0.7819,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830709848749727e-05,
+      "loss": 0.6504,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828545035342125e-05,
+      "loss": 0.8006,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826366709775286e-05,
+      "loss": 0.8074,
+      "step": 217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824174878145016e-05,
+      "loss": 0.8039,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821969546584922e-05,
+      "loss": 0.853,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81975072126638e-05,
+      "loss": 0.9003,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817518408398536e-05,
+      "loss": 1.0273,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815272614228274e-05,
+      "loss": 0.6749,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.813013345040205e-05,
+      "loss": 0.8692,
+      "step": 223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810740607156648e-05,
+      "loss": 0.9198,
+      "step": 224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808454406937614e-05,
+      "loss": 0.7347,
+      "step": 225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806154750780787e-05,
+      "loss": 1.0314,
+      "step": 226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803841645121504e-05,
+      "loss": 0.8253,
+      "step": 227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801515096432741e-05,
+      "loss": 0.8665,
+      "step": 228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799175111225089e-05,
+      "loss": 0.7043,
+      "step": 229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796821696046748e-05,
+      "loss": 0.9182,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79445485748349e-05,
+      "loss": 0.7463,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792074602158661e-05,
+      "loss": 0.8029,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789680936733142e-05,
+      "loss": 0.8576,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787273867905347e-05,
+      "loss": 0.7052,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784853402411199e-05,
+      "loss": 0.7294,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782419547024107e-05,
+      "loss": 0.8211,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779972308554952e-05,
+      "loss": 0.9741,
+      "step": 237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777511693852063e-05,
+      "loss": 0.9359,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775037709801204e-05,
+      "loss": 0.8012,
+      "step": 239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772550363325552e-05,
+      "loss": 0.7964,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770049661385676e-05,
+      "loss": 0.7818,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767535610979519e-05,
+      "loss": 0.6649,
+      "step": 242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765008219142377e-05,
+      "loss": 0.9135,
+      "step": 243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762467492946881e-05,
+      "loss": 0.6994,
+      "step": 244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759913439502981e-05,
+      "loss": 0.6691,
+      "step": 245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757346065957916e-05,
+      "loss": 0.8557,
+      "step": 246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754765379496202e-05,
+      "loss": 0.6125,
+      "step": 247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752171387339611e-05,
+      "loss": 0.8432,
+      "step": 248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749564096747148e-05,
+      "loss": 0.8036,
+      "step": 249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746943515015033e-05,
+      "loss": 0.7109,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744309649476679e-05,
+      "loss": 0.9144,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741662507502674e-05,
+      "loss": 0.8545,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739002096500755e-05,
+      "loss": 0.839,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736328423915798e-05,
+      "loss": 0.7039,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733641497229781e-05,
+      "loss": 0.6235,
+      "step": 255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730941323961778e-05,
+      "loss": 0.7025,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728227911667934e-05,
+      "loss": 0.657,
+      "step": 257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725501267941432e-05,
+      "loss": 0.8421,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722761400412495e-05,
+      "loss": 0.7547,
+      "step": 259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720008316748343e-05,
+      "loss": 0.8173,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717242024653181e-05,
+      "loss": 0.8449,
+      "step": 261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.714462531868178e-05,
+      "loss": 0.7769,
+      "step": 262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711669846171442e-05,
+      "loss": 1.0137,
+      "step": 263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708863975378002e-05,
+      "loss": 0.6748,
+      "step": 264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706044927339782e-05,
+      "loss": 1.0747,
+      "step": 265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703212709945581e-05,
+      "loss": 0.6852,
+      "step": 266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700367331121054e-05,
+      "loss": 0.8057,
+      "step": 267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697508798828685e-05,
+      "loss": 0.8148,
+      "step": 268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694637121067765e-05,
+      "loss": 0.7064,
+      "step": 269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691752305874373e-05,
+      "loss": 1.2035,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688854361321351e-05,
+      "loss": 0.7395,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685943295518283e-05,
+      "loss": 0.9433,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683019116611471e-05,
+      "loss": 0.6698,
+      "step": 273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68008183278391e-05,
+      "loss": 0.846,
+      "step": 274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677131452255272e-05,
+      "loss": 0.6935,
+      "step": 275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674167983281877e-05,
+      "loss": 0.6262,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67119143415667e-05,
+      "loss": 0.7154,
+      "step": 277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668201813209201e-05,
+      "loss": 0.8448,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6651991288056e-05,
+      "loss": 0.8244,
+      "step": 279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662183389348555e-05,
+      "loss": 1.0565,
+      "step": 280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.659154603277283e-05,
+      "loss": 0.7413,
+      "step": 281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656112779067514e-05,
+      "loss": 0.8999,
+      "step": 282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653057925231462e-05,
+      "loss": 0.6289,
+      "step": 283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649990050317806e-05,
+      "loss": 0.8713,
+      "step": 284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646909162911661e-05,
+      "loss": 0.6732,
+      "step": 285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643815271634552e-05,
+      "loss": 0.6883,
+      "step": 286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640708385144403e-05,
+      "loss": 0.6388,
+      "step": 287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637588512135498e-05,
+      "loss": 0.8915,
+      "step": 288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63445566133846e-05,
+      "loss": 0.7575,
+      "step": 289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631309841520234e-05,
+      "loss": 0.7048,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628151061484054e-05,
+      "loss": 0.6772,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624979330069425e-05,
+      "loss": 0.9256,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621794656152091e-05,
+      "loss": 0.7899,
+      "step": 293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618597048644018e-05,
+      "loss": 0.7135,
+      "step": 294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615386516493363e-05,
+      "loss": 0.6944,
+      "step": 295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612163068684452e-05,
+      "loss": 0.7981,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608926714237754e-05,
+      "loss": 0.6931,
+      "step": 297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605677462209855e-05,
+      "loss": 0.6743,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602415321693434e-05,
+      "loss": 1.08,
+      "step": 299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599140301817239e-05,
+      "loss": 1.1371,
+      "step": 300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595852411746055e-05,
+      "loss": 0.7409,
+      "step": 301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592551660680688e-05,
+      "loss": 1.1411,
+      "step": 302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58923805785793e-05,
+      "loss": 0.7701,
+      "step": 303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585911612550539e-05,
+      "loss": 0.835,
+      "step": 304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582572334067212e-05,
+      "loss": 0.7429,
+      "step": 305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579220231752557e-05,
+      "loss": 0.751,
+      "step": 306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575855314987068e-05,
+      "loss": 0.6873,
+      "step": 307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572477593187101e-05,
+      "loss": 0.7636,
+      "step": 308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569087075804842e-05,
+      "loss": 0.7033,
+      "step": 309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565683772328287e-05,
+      "loss": 0.7045,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56226769228121e-05,
+      "loss": 0.6315,
+      "step": 311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558838845223142e-05,
+      "loss": 0.8513,
+      "step": 312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555397240749339e-05,
+      "loss": 0.6809,
+      "step": 313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551942888490757e-05,
+      "loss": 0.6088,
+      "step": 314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548475798114027e-05,
+      "loss": 0.6352,
+      "step": 315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544995979321425e-05,
+      "loss": 0.6517,
+      "step": 316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541503441850843e-05,
+      "loss": 0.8924,
+      "step": 317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537998195475772e-05,
+      "loss": 0.6756,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534480250005262e-05,
+      "loss": 0.6294,
+      "step": 319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.530949615283903e-05,
+      "loss": 0.5993,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52740630119179e-05,
+      "loss": 0.6593,
+      "step": 321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523850317644508e-05,
+      "loss": 0.7287,
+      "step": 322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520281674593083e-05,
+      "loss": 0.7517,
+      "step": 323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516700382023984e-05,
+      "loss": 1.0167,
+      "step": 324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513106449959064e-05,
+      "loss": 0.7037,
+      "step": 325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509499888455554e-05,
+      "loss": 0.9184,
+      "step": 326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505880707606024e-05,
+      "loss": 0.6021,
+      "step": 327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502248917538359e-05,
+      "loss": 0.6689,
+      "step": 328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498604528415731e-05,
+      "loss": 0.6521,
+      "step": 329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494947550436565e-05,
+      "loss": 0.6721,
+      "step": 330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491277993834518e-05,
+      "loss": 0.8637,
+      "step": 331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487595868878446e-05,
+      "loss": 0.7222,
+      "step": 332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483901185872377e-05,
+      "loss": 1.0388,
+      "step": 333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480193955155478e-05,
+      "loss": 0.6427,
+      "step": 334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476474187102033e-05,
+      "loss": 0.7128,
+      "step": 335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47274189212141e-05,
+      "loss": 0.689,
+      "step": 336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468997080658031e-05,
+      "loss": 0.6733,
+      "step": 337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465239763191345e-05,
+      "loss": 0.8392,
+      "step": 338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.461469950235795e-05,
+      "loss": 0.8487,
+      "step": 339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457687652340796e-05,
+      "loss": 0.9243,
+      "step": 340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453892880090696e-05,
+      "loss": 0.8291,
+      "step": 341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450085644104754e-05,
+      "loss": 0.7225,
+      "step": 342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446265955037108e-05,
+      "loss": 0.7428,
+      "step": 343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442433823576741e-05,
+      "loss": 0.8138,
+      "step": 344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438589260447461e-05,
+      "loss": 1.3266,
+      "step": 345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434732276407856e-05,
+      "loss": 0.7189,
+      "step": 346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430862882251278e-05,
+      "loss": 0.7981,
+      "step": 347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42698108880581e-05,
+      "loss": 0.6702,
+      "step": 348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423086906934227e-05,
+      "loss": 0.6601,
+      "step": 349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419180347533976e-05,
+      "loss": 0.8036,
+      "step": 350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415261421537136e-05,
+      "loss": 0.9533,
+      "step": 351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4113301399104e-05,
+      "loss": 0.7378,
+      "step": 352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40738651365503e-05,
+      "loss": 0.8535,
+      "step": 353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403430553806838e-05,
+      "loss": 1.0126,
+      "step": 354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399462271436148e-05,
+      "loss": 0.762,
+      "step": 355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395481677647766e-05,
+      "loss": 0.7806,
+      "step": 356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391488783580955e-05,
+      "loss": 0.7473,
+      "step": 357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387483600409393e-05,
+      "loss": 0.7455,
+      "step": 358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38346613934115e-05,
+      "loss": 0.8515,
+      "step": 359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379436411618656e-05,
+      "loss": 0.7325,
+      "step": 360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375394428518667e-05,
+      "loss": 0.7797,
+      "step": 361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371340201352233e-05,
+      "loss": 0.7342,
+      "step": 362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36727374146467e-05,
+      "loss": 0.948,
+      "step": 363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363195060235524e-05,
+      "loss": 0.8774,
+      "step": 364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35910416907854e-05,
+      "loss": 0.7479,
+      "step": 365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355001079441637e-05,
+      "loss": 0.7011,
+      "step": 366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350885802806863e-05,
+      "loss": 0.7051,
+      "step": 367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346758350690373e-05,
+      "loss": 1.1496,
+      "step": 368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342618734642395e-05,
+      "loss": 0.7525,
+      "step": 369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338466966247194e-05,
+      "loss": 0.7006,
+      "step": 370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334303057123043e-05,
+      "loss": 0.796,
+      "step": 371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330127018922194e-05,
+      "loss": 0.7928,
+      "step": 372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325938863330832e-05,
+      "loss": 0.8915,
+      "step": 373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321738602069056e-05,
+      "loss": 0.826,
+      "step": 374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317526246890844e-05,
+      "loss": 0.6599,
+      "step": 375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313301809584012e-05,
+      "loss": 0.7449,
+      "step": 376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309065301970193e-05,
+      "loss": 0.7667,
+      "step": 377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.304816735904789e-05,
+      "loss": 0.8491,
+      "step": 378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300556123276955e-05,
+      "loss": 0.7297,
+      "step": 379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29628347600955e-05,
+      "loss": 0.9033,
+      "step": 380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291998806059117e-05,
+      "loss": 0.884,
+      "step": 381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287702125415837e-05,
+      "loss": 0.6596,
+      "step": 382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283393446103505e-05,
+      "loss": 0.8277,
+      "step": 383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279072780179494e-05,
+      "loss": 1.0674,
+      "step": 384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274740139734716e-05,
+      "loss": 0.6318,
+      "step": 385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270395536893599e-05,
+      "loss": 0.7691,
+      "step": 386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266038983814039e-05,
+      "loss": 0.7684,
+      "step": 387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261670492687377e-05,
+      "loss": 0.6624,
+      "step": 388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257290075738365e-05,
+      "loss": 0.6667,
+      "step": 389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252897745225121e-05,
+      "loss": 1.1117,
+      "step": 390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248493513439104e-05,
+      "loss": 1.1533,
+      "step": 391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244077392705084e-05,
+      "loss": 0.6694,
+      "step": 392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23964939538109e-05,
+      "loss": 0.729,
+      "step": 393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235209533858395e-05,
+      "loss": 0.7589,
+      "step": 394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23075782056147e-05,
+      "loss": 0.6862,
+      "step": 395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22629426794795e-05,
+      "loss": 0.8407,
+      "step": 396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221818888508602e-05,
+      "loss": 0.6076,
+      "step": 397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21733169476729e-05,
+      "loss": 0.6425,
+      "step": 398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212832699280941e-05,
+      "loss": 0.7151,
+      "step": 399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208321914639503e-05,
+      "loss": 0.9149,
+      "step": 400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203799353465918e-05,
+      "loss": 0.9245,
+      "step": 401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199265028416084e-05,
+      "loss": 0.7992,
+      "step": 402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194718952178816e-05,
+      "loss": 0.7628,
+      "step": 403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190161137475815e-05,
+      "loss": 0.8478,
+      "step": 404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185591597061631e-05,
+      "loss": 0.865,
+      "step": 405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181010343723628e-05,
+      "loss": 0.741,
+      "step": 406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176417390281944e-05,
+      "loss": 0.9099,
+      "step": 407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171812749589461e-05,
+      "loss": 0.617,
+      "step": 408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167196434531768e-05,
+      "loss": 0.7038,
+      "step": 409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162568458027122e-05,
+      "loss": 0.7678,
+      "step": 410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157928833026411e-05,
+      "loss": 1.0452,
+      "step": 411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153277572513124e-05,
+      "loss": 0.8723,
+      "step": 412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148614689503307e-05,
+      "loss": 1.0691,
+      "step": 413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143940197045532e-05,
+      "loss": 0.8668,
+      "step": 414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139254108220859e-05,
+      "loss": 0.9241,
+      "step": 415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1345564361428e-05,
+      "loss": 1.2006,
+      "step": 416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.129847193957282e-05,
+      "loss": 0.8637,
+      "step": 417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125126394842603e-05,
+      "loss": 0.7258,
+      "step": 418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120394052009411e-05,
+      "loss": 0.786,
+      "step": 419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115650178700654e-05,
+      "loss": 0.8204,
+      "step": 420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110894788191544e-05,
+      "loss": 0.791,
+      "step": 421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106127893789528e-05,
+      "loss": 0.8476,
+      "step": 422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101349508834241e-05,
+      "loss": 0.9043,
+      "step": 423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096559646697475e-05,
+      "loss": 0.8241,
+      "step": 424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091758320783139e-05,
+      "loss": 0.778,
+      "step": 425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086945544527224e-05,
+      "loss": 0.7087,
+      "step": 426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08212133139776e-05,
+      "loss": 0.7837,
+      "step": 427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077285694894784e-05,
+      "loss": 0.7288,
+      "step": 428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072438648550303e-05,
+      "loss": 0.7253,
+      "step": 429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067580205928246e-05,
+      "loss": 0.6455,
+      "step": 430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06271038062444e-05,
+      "loss": 0.5991,
+      "step": 431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057829186266563e-05,
+      "loss": 0.7127,
+      "step": 432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052936636514104e-05,
+      "loss": 0.661,
+      "step": 433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048032745058335e-05,
+      "loss": 0.8448,
+      "step": 434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043117525622265e-05,
+      "loss": 0.714,
+      "step": 435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.038190991960601e-05,
+      "loss": 0.6194,
+      "step": 436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033253157859714e-05,
+      "loss": 0.7792,
+      "step": 437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028304037137594e-05,
+      "loss": 0.667,
+      "step": 438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02334364364382e-05,
+      "loss": 1.3762,
+      "step": 439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018371991259515e-05,
+      "loss": 0.9788,
+      "step": 440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01338909389731e-05,
+      "loss": 0.7807,
+      "step": 441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008394965501299e-05,
+      "loss": 0.6526,
+      "step": 442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003389620047011e-05,
+      "loss": 0.8908,
+      "step": 443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998373071541361e-05,
+      "loss": 0.6614,
+      "step": 444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993345334022617e-05,
+      "loss": 1.193,
+      "step": 445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988306421560353e-05,
+      "loss": 0.8207,
+      "step": 446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983256348255423e-05,
+      "loss": 0.8381,
+      "step": 447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978195128239906e-05,
+      "loss": 0.7207,
+      "step": 448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973122775677078e-05,
+      "loss": 0.7046,
+      "step": 449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968039304761368e-05,
+      "loss": 0.7751,
+      "step": 450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962944729718318e-05,
+      "loss": 0.8346,
+      "step": 451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957839064804541e-05,
+      "loss": 0.7912,
+      "step": 452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952722324307687e-05,
+      "loss": 0.5897,
+      "step": 453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947594522546401e-05,
+      "loss": 0.7067,
+      "step": 454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942455673870278e-05,
+      "loss": 0.6755,
+      "step": 455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937305792659826e-05,
+      "loss": 1.2562,
+      "step": 456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932144893326432e-05,
+      "loss": 0.77,
+      "step": 457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926972990312313e-05,
+      "loss": 0.6978,
+      "step": 458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921790098090476e-05,
+      "loss": 0.7979,
+      "step": 459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916596231164685e-05,
+      "loss": 0.8076,
+      "step": 460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911391404069408e-05,
+      "loss": 0.7289,
+      "step": 461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.906175631369796e-05,
+      "loss": 0.6616,
+      "step": 462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900948927661616e-05,
+      "loss": 0.6206,
+      "step": 463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895711307571234e-05,
+      "loss": 0.6951,
+      "step": 464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890462785755562e-05,
+      "loss": 0.6637,
+      "step": 465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885203376902017e-05,
+      "loss": 0.8318,
+      "step": 466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879933095728485e-05,
+      "loss": 0.7468,
+      "step": 467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874651956983278e-05,
+      "loss": 0.9372,
+      "step": 468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869359975445084e-05,
+      "loss": 0.7312,
+      "step": 469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864057165922944e-05,
+      "loss": 0.6979,
+      "step": 470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858743543256194e-05,
+      "loss": 0.7235,
+      "step": 471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853419122314428e-05,
+      "loss": 0.937,
+      "step": 472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848083917997463e-05,
+      "loss": 0.613,
+      "step": 473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842737945235287e-05,
+      "loss": 0.5299,
+      "step": 474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.837381218988025e-05,
+      "loss": 1.164,
+      "step": 475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832013754245896e-05,
+      "loss": 0.7655,
+      "step": 476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826635566029166e-05,
+      "loss": 0.6038,
+      "step": 477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82124666938811e-05,
+      "loss": 0.7363,
+      "step": 478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815847079402973e-05,
+      "loss": 0.7472,
+      "step": 479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81043681118392e-05,
+      "loss": 0.7547,
+      "step": 480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805015879870997e-05,
+      "loss": 0.7574,
+      "step": 481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799584300634096e-05,
+      "loss": 0.7524,
+      "step": 482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.794142088672902e-05,
+      "loss": 0.845,
+      "step": 483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788689259216852e-05,
+      "loss": 0.6912,
+      "step": 484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7832258275251e-05,
+      "loss": 0.5932,
+      "step": 485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777751808886461e-05,
+      "loss": 0.7495,
+      "step": 486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772267218619388e-05,
+      "loss": 0.9373,
+      "step": 487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766772072071912e-05,
+      "loss": 0.7662,
+      "step": 488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7612663846216e-05,
+      "loss": 0.6622,
+      "step": 489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755750171675523e-05,
+      "loss": 0.7893,
+      "step": 490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750223448670203e-05,
+      "loss": 1.0085,
+      "step": 491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744686231071577e-05,
+      "loss": 0.8498,
+      "step": 492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739138534374945e-05,
+      "loss": 0.777,
+      "step": 493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.733580374104936e-05,
+      "loss": 0.8197,
+      "step": 494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728011765815457e-05,
+      "loss": 0.7217,
+      "step": 495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722432725089654e-05,
+      "loss": 0.6047,
+      "step": 496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716843267539869e-05,
+      "loss": 0.7486,
+      "step": 497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711243408807588e-05,
+      "loss": 0.6957,
+      "step": 498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705633164563412e-05,
+      "loss": 0.8438,
+      "step": 499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700012550507e-05,
+      "loss": 0.6151,
+      "step": 500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694381582367028e-05,
+      "loss": 0.6998,
+      "step": 501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688740275901152e-05,
+      "loss": 0.8622,
+      "step": 502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683088646895957e-05,
+      "loss": 0.9324,
+      "step": 503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677426711166907e-05,
+      "loss": 0.7546,
+      "step": 504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67175448455832e-05,
+      "loss": 0.6858,
+      "step": 505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666071982943305e-05,
+      "loss": 0.6722,
+      "step": 506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660379222223727e-05,
+      "loss": 0.7772,
+      "step": 507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654676218330159e-05,
+      "loss": 0.9929,
+      "step": 508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648962987221837e-05,
+      "loss": 0.7864,
+      "step": 509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643239544886619e-05,
+      "loss": 0.7881,
+      "step": 510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63750590734094e-05,
+      "loss": 0.5956,
+      "step": 511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631762090629756e-05,
+      "loss": 0.6408,
+      "step": 512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62600811082652e-05,
+      "loss": 0.7026,
+      "step": 513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.620243984033118e-05,
+      "loss": 0.7407,
+      "step": 514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614469726379833e-05,
+      "loss": 0.7934,
+      "step": 515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608685354025299e-05,
+      "loss": 0.6647,
+      "step": 516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602890883156454e-05,
+      "loss": 0.9164,
+      "step": 517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597086329988498e-05,
+      "loss": 0.6783,
+      "step": 518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591271710764838e-05,
+      "loss": 0.6909,
+      "step": 519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585447041757061e-05,
+      "loss": 0.646,
+      "step": 520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579612339264867e-05,
+      "loss": 0.8818,
+      "step": 521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573767619616038e-05,
+      "loss": 0.6604,
+      "step": 522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56791289916639e-05,
+      "loss": 0.7729,
+      "step": 523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562048194299719e-05,
+      "loss": 0.7489,
+      "step": 524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556173521427769e-05,
+      "loss": 0.7805,
+      "step": 525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55028889699017e-05,
+      "loss": 0.6011,
+      "step": 526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544394337454409e-05,
+      "loss": 1.0065,
+      "step": 527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538489859315771e-05,
+      "loss": 0.6413,
+      "step": 528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532575479097293e-05,
+      "loss": 0.7189,
+      "step": 529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52665121334973e-05,
+      "loss": 0.6208,
+      "step": 530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520717078651496e-05,
+      "loss": 0.63,
+      "step": 531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514773091608623e-05,
+      "loss": 0.7174,
+      "step": 532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.508819268854713e-05,
+      "loss": 0.7031,
+      "step": 533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502855627050892e-05,
+      "loss": 0.8368,
+      "step": 534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496882182885766e-05,
+      "loss": 0.6855,
+      "step": 535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49089895307537e-05,
+      "loss": 0.7907,
+      "step": 536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484905954363123e-05,
+      "loss": 0.8165,
+      "step": 537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478903203519781e-05,
+      "loss": 0.7256,
+      "step": 538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472890717343391e-05,
+      "loss": 0.6989,
+      "step": 539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466868512659243e-05,
+      "loss": 0.7523,
+      "step": 540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460836606319823e-05,
+      "loss": 0.7545,
+      "step": 541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454795015204766e-05,
+      "loss": 0.7589,
+      "step": 542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448743756220811e-05,
+      "loss": 0.7928,
+      "step": 543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442682846301749e-05,
+      "loss": 0.7246,
+      "step": 544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436612302408376e-05,
+      "loss": 0.7479,
+      "step": 545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430532141528451e-05,
+      "loss": 1.1287,
+      "step": 546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424442380676647e-05,
+      "loss": 0.7267,
+      "step": 547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418343036894497e-05,
+      "loss": 0.7104,
+      "step": 548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412234127250352e-05,
+      "loss": 0.904,
+      "step": 549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406115668839333e-05,
+      "loss": 0.6523,
+      "step": 550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399987678783284e-05,
+      "loss": 0.5657,
+      "step": 551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393850174230717e-05,
+      "loss": 1.1838,
+      "step": 552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387703172356774e-05,
+      "loss": 0.9385,
+      "step": 553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381546690363173e-05,
+      "loss": 0.5962,
+      "step": 554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37538074547816e-05,
+      "loss": 0.6541,
+      "step": 555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369205354956461e-05,
+      "loss": 0.8049,
+      "step": 556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363020536079239e-05,
+      "loss": 1.0934,
+      "step": 557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356826306154038e-05,
+      "loss": 0.8095,
+      "step": 558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350622682514735e-05,
+      "loss": 0.7668,
+      "step": 559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3444096825215e-05,
+      "loss": 0.8709,
+      "step": 560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338187323560739e-05,
+      "loss": 0.9203,
+      "step": 561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331955623045049e-05,
+      "loss": 0.6805,
+      "step": 562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325714598413168e-05,
+      "loss": 1.3393,
+      "step": 563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319464267129921e-05,
+      "loss": 0.8702,
+      "step": 564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31320464668619e-05,
+      "loss": 0.8204,
+      "step": 565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306935754598837e-05,
+      "loss": 0.7047,
+      "step": 566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300657608410678e-05,
+      "loss": 0.6449,
+      "step": 567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294370225690423e-05,
+      "loss": 0.6832,
+      "step": 568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288073624032633e-05,
+      "loss": 0.8335,
+      "step": 569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281767821057661e-05,
+      "loss": 0.8539,
+      "step": 570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275452834411614e-05,
+      "loss": 0.7909,
+      "step": 571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.269128681766296e-05,
+      "loss": 0.88,
+      "step": 572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262795380819164e-05,
+      "loss": 0.7781,
+      "step": 573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25645294929327e-05,
+      "loss": 0.7222,
+      "step": 574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250101404937222e-05,
+      "loss": 0.7437,
+      "step": 575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243740765525128e-05,
+      "loss": 0.6647,
+      "step": 576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237371048856546e-05,
+      "loss": 1.0254,
+      "step": 577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230992272756439e-05,
+      "loss": 0.6396,
+      "step": 578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224604455075116e-05,
+      "loss": 0.8816,
+      "step": 579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218207613688194e-05,
+      "loss": 0.6079,
+      "step": 580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211801766496537e-05,
+      "loss": 0.7198,
+      "step": 581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205386931426217e-05,
+      "loss": 0.6571,
+      "step": 582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.198963126428452e-05,
+      "loss": 0.7092,
+      "step": 583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19253036947956e-05,
+      "loss": 0.6693,
+      "step": 584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18608867858092e-05,
+      "loss": 0.8582,
+      "step": 585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179638071758897e-05,
+      "loss": 1.4449,
+      "step": 586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17317856706482e-05,
+      "loss": 0.6011,
+      "step": 587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16671018257491e-05,
+      "loss": 0.6765,
+      "step": 588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160232936390239e-05,
+      "loss": 0.7202,
+      "step": 589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153746846636675e-05,
+      "loss": 0.6929,
+      "step": 590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.147251931464841e-05,
+      "loss": 0.7039,
+      "step": 591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.140748209050047e-05,
+      "loss": 0.8022,
+      "step": 592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13423569759226e-05,
+      "loss": 1.1322,
+      "step": 593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12771441531603e-05,
+      "loss": 0.7887,
+      "step": 594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121184380470464e-05,
+      "loss": 0.7872,
+      "step": 595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114645611329152e-05,
+      "loss": 0.7069,
+      "step": 596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108098126190129e-05,
+      "loss": 0.7596,
+      "step": 597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101541943375826e-05,
+      "loss": 0.6837,
+      "step": 598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094977081233006e-05,
+      "loss": 0.73,
+      "step": 599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088403558132723e-05,
+      "loss": 0.6149,
+      "step": 600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081821392470268e-05,
+      "loss": 0.7287,
+      "step": 601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075230602665118e-05,
+      "loss": 0.6549,
+      "step": 602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068631207160882e-05,
+      "loss": 1.415,
+      "step": 603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.062023224425255e-05,
+      "loss": 0.8031,
+      "step": 604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055406672949957e-05,
+      "loss": 0.8043,
+      "step": 605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048781571250688e-05,
+      "loss": 0.8259,
+      "step": 606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042147937867079e-05,
+      "loss": 0.6148,
+      "step": 607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03550579136263e-05,
+      "loss": 0.6482,
+      "step": 608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02885515032467e-05,
+      "loss": 0.9864,
+      "step": 609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022196033364296e-05,
+      "loss": 0.6039,
+      "step": 610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015528459116321e-05,
+      "loss": 0.769,
+      "step": 611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008852446239232e-05,
+      "loss": 0.6823,
+      "step": 612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002168013415123e-05,
+      "loss": 0.9339,
+      "step": 613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995475179349656e-05,
+      "loss": 0.8612,
+      "step": 614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988773962772001e-05,
+      "loss": 0.856,
+      "step": 615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.982064382434782e-05,
+      "loss": 0.6503,
+      "step": 616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975346457114034e-05,
+      "loss": 0.663,
+      "step": 617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968620205609144e-05,
+      "loss": 0.6827,
+      "step": 618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961885646742793e-05,
+      "loss": 0.8037,
+      "step": 619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955142799360914e-05,
+      "loss": 0.8055,
+      "step": 620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948391682332635e-05,
+      "loss": 0.7533,
+      "step": 621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941632314550222e-05,
+      "loss": 0.6966,
+      "step": 622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934864714929036e-05,
+      "loss": 0.69,
+      "step": 623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928088902407465e-05,
+      "loss": 0.6784,
+      "step": 624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921304895946889e-05,
+      "loss": 0.7253,
+      "step": 625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914512714531613e-05,
+      "loss": 0.7221,
+      "step": 626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907712377168817e-05,
+      "loss": 0.7556,
+      "step": 627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90090390288851e-05,
+      "loss": 0.7881,
+      "step": 628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894087310743467e-05,
+      "loss": 0.715,
+      "step": 629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.887262619809182e-05,
+      "loss": 0.8899,
+      "step": 630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880429849183813e-05,
+      "loss": 1.2894,
+      "step": 631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873589017988123e-05,
+      "loss": 0.9658,
+      "step": 632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86674014536544e-05,
+      "loss": 0.6487,
+      "step": 633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859883250481588e-05,
+      "loss": 0.639,
+      "step": 634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853018352524845e-05,
+      "loss": 0.7318,
+      "step": 635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846145470705881e-05,
+      "loss": 0.977,
+      "step": 636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839264624257712e-05,
+      "loss": 0.7883,
+      "step": 637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832375832435637e-05,
+      "loss": 0.7167,
+      "step": 638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825479114517196e-05,
+      "loss": 0.7096,
+      "step": 639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818574489802102e-05,
+      "loss": 0.6378,
+      "step": 640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811661977612201e-05,
+      "loss": 0.7142,
+      "step": 641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804741597291405e-05,
+      "loss": 1.0176,
+      "step": 642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79781336820565e-05,
+      "loss": 1.0909,
+      "step": 643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790877309742832e-05,
+      "loss": 1.1751,
+      "step": 644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78393344131276e-05,
+      "loss": 0.6235,
+      "step": 645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776981782347093e-05,
+      "loss": 0.9633,
+      "step": 646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770022352299293e-05,
+      "loss": 0.9332,
+      "step": 647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763055170644575e-05,
+      "loss": 0.6678,
+      "step": 648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.756080256879836e-05,
+      "loss": 0.6606,
+      "step": 649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749097630523619e-05,
+      "loss": 0.5863,
+      "step": 650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74210731111604e-05,
+      "loss": 0.73,
+      "step": 651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735109318218756e-05,
+      "loss": 0.7003,
+      "step": 652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.728103671414889e-05,
+      "loss": 0.6538,
+      "step": 653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721090390308978e-05,
+      "loss": 0.8853,
+      "step": 654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714069494526931e-05,
+      "loss": 0.6819,
+      "step": 655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707041003715963e-05,
+      "loss": 0.6169,
+      "step": 656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.700004937544542e-05,
+      "loss": 0.6688,
+      "step": 657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.692961315702337e-05,
+      "loss": 0.8008,
+      "step": 658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685910157900158e-05,
+      "loss": 0.6,
+      "step": 659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678851483869907e-05,
+      "loss": 0.6532,
+      "step": 660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671785313364514e-05,
+      "loss": 0.6474,
+      "step": 661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664711666157892e-05,
+      "loss": 0.7001,
+      "step": 662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657630562044875e-05,
+      "loss": 0.631,
+      "step": 663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650542020841163e-05,
+      "loss": 0.6157,
+      "step": 664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643446062383273e-05,
+      "loss": 0.819,
+      "step": 665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63634270652847e-05,
+      "loss": 0.7021,
+      "step": 666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629231973154725e-05,
+      "loss": 0.8842,
+      "step": 667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622113882160657e-05,
+      "loss": 0.6434,
+      "step": 668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614988453465467e-05,
+      "loss": 0.7251,
+      "step": 669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607855707008899e-05,
+      "loss": 0.6231,
+      "step": 670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600715662751165e-05,
+      "loss": 0.975,
+      "step": 671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593568340672907e-05,
+      "loss": 0.6886,
+      "step": 672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586413760775129e-05,
+      "loss": 0.7481,
+      "step": 673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579251943079145e-05,
+      "loss": 0.7798,
+      "step": 674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572082907626526e-05,
+      "loss": 0.6886,
+      "step": 675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564906674479039e-05,
+      "loss": 0.5953,
+      "step": 676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.557723263718596e-05,
+      "loss": 0.6638,
+      "step": 677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.550532695447188e-05,
+      "loss": 0.6821,
+      "step": 678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543334989786845e-05,
+      "loss": 0.6337,
+      "step": 679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53613016687956e-05,
+      "loss": 0.7084,
+      "step": 680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528918246887252e-05,
+      "loss": 0.6558,
+      "step": 681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521699249991693e-05,
+      "loss": 0.623,
+      "step": 682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514473196394467e-05,
+      "loss": 0.6077,
+      "step": 683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507240106316895e-05,
+      "loss": 0.6701,
+      "step": 684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.6089,
+      "step": 685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492752897704432e-05,
+      "loss": 0.6755,
+      "step": 686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485498819710417e-05,
+      "loss": 0.7427,
+      "step": 687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.478237786317707e-05,
+      "loss": 0.7109,
+      "step": 688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470969817845517e-05,
+      "loss": 0.6293,
+      "step": 689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46369493463247e-05,
+      "loss": 0.7214,
+      "step": 690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456413157036532e-05,
+      "loss": 0.6823,
+      "step": 691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.449124505434971e-05,
+      "loss": 0.7448,
+      "step": 692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441829000224285e-05,
+      "loss": 0.8388,
+      "step": 693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434526661820156e-05,
+      "loss": 0.6722,
+      "step": 694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.427217510657382e-05,
+      "loss": 0.649,
+      "step": 695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419901567189835e-05,
+      "loss": 0.784,
+      "step": 696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412578851890384e-05,
+      "loss": 0.7902,
+      "step": 697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405249385250853e-05,
+      "loss": 0.7655,
+      "step": 698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397913187781962e-05,
+      "loss": 0.6876,
+      "step": 699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39057028001326e-05,
+      "loss": 0.5443,
+      "step": 700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38322068249308e-05,
+      "loss": 0.9235,
+      "step": 701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375864415788473e-05,
+      "loss": 0.7844,
+      "step": 702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368501500485152e-05,
+      "loss": 0.5843,
+      "step": 703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361131957187434e-05,
+      "loss": 0.695,
+      "step": 704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353755806518193e-05,
+      "loss": 0.7965,
+      "step": 705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34637306911878e-05,
+      "loss": 1.2161,
+      "step": 706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338983765648985e-05,
+      "loss": 0.6911,
+      "step": 707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331587916786973e-05,
+      "loss": 0.6212,
+      "step": 708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324185543229225e-05,
+      "loss": 0.8265,
+      "step": 709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316776665690479e-05,
+      "loss": 0.6513,
+      "step": 710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309361304903675e-05,
+      "loss": 0.7858,
+      "step": 711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301939481619895e-05,
+      "loss": 0.6381,
+      "step": 712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294511216608308e-05,
+      "loss": 0.666,
+      "step": 713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.287076530656104e-05,
+      "loss": 0.8325,
+      "step": 714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279635444568447e-05,
+      "loss": 1.1954,
+      "step": 715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272187979168408e-05,
+      "loss": 1.1145,
+      "step": 716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264734155296912e-05,
+      "loss": 0.665,
+      "step": 717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257273993812674e-05,
+      "loss": 0.6932,
+      "step": 718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249807515592148e-05,
+      "loss": 0.8137,
+      "step": 719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242334741529463e-05,
+      "loss": 0.7237,
+      "step": 720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234855692536365e-05,
+      "loss": 0.8108,
+      "step": 721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227370389542161e-05,
+      "loss": 0.7394,
+      "step": 722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21987885349366e-05,
+      "loss": 0.7286,
+      "step": 723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.212381105355113e-05,
+      "loss": 0.7406,
+      "step": 724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.204877166108152e-05,
+      "loss": 0.8387,
+      "step": 725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197367056751738e-05,
+      "loss": 0.7604,
+      "step": 726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189850798302099e-05,
+      "loss": 0.7217,
+      "step": 727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.182328411792663e-05,
+      "loss": 0.717,
+      "step": 728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174799918274017e-05,
+      "loss": 1.2244,
+      "step": 729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167265338813834e-05,
+      "loss": 0.7056,
+      "step": 730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159724694496815e-05,
+      "loss": 0.7018,
+      "step": 731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.152178006424632e-05,
+      "loss": 0.6331,
+      "step": 732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144625295715877e-05,
+      "loss": 0.8269,
+      "step": 733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137066583505987e-05,
+      "loss": 0.6852,
+      "step": 734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129501890947201e-05,
+      "loss": 0.6414,
+      "step": 735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12193123920849e-05,
+      "loss": 0.6235,
+      "step": 736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.114354649475499e-05,
+      "loss": 0.7774,
+      "step": 737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106772142950492e-05,
+      "loss": 0.6197,
+      "step": 738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099183740852295e-05,
+      "loss": 0.5626,
+      "step": 739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091589464416225e-05,
+      "loss": 0.6417,
+      "step": 740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083989334894037e-05,
+      "loss": 0.8258,
+      "step": 741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.076383373553875e-05,
+      "loss": 0.644,
+      "step": 742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06877160168019e-05,
+      "loss": 0.6316,
+      "step": 743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061154040573705e-05,
+      "loss": 0.6501,
+      "step": 744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053530711551333e-05,
+      "loss": 0.6584,
+      "step": 745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.045901635946139e-05,
+      "loss": 0.7773,
+      "step": 746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038266835107257e-05,
+      "loss": 0.8391,
+      "step": 747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030626330399852e-05,
+      "loss": 0.9583,
+      "step": 748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022980143205045e-05,
+      "loss": 0.7905,
+      "step": 749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015328294919863e-05,
+      "loss": 0.6637,
+      "step": 750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007670806957174e-05,
+      "loss": 0.6704,
+      "step": 751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.000007700745622e-05,
+      "loss": 0.7979,
+      "step": 752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992338997729585e-05,
+      "loss": 0.6048,
+      "step": 753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.984664719369089e-05,
+      "loss": 0.5794,
+      "step": 754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976984887139775e-05,
+      "loss": 0.9097,
+      "step": 755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969299522532816e-05,
+      "loss": 0.7026,
+      "step": 756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961608647054873e-05,
+      "loss": 0.81,
+      "step": 757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953912282228025e-05,
+      "loss": 0.6444,
+      "step": 758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946210449589714e-05,
+      "loss": 0.7306,
+      "step": 759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938503170692684e-05,
+      "loss": 0.744,
+      "step": 760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.930790467104917e-05,
+      "loss": 0.7262,
+      "step": 761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923072360409578e-05,
+      "loss": 0.7082,
+      "step": 762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915348872204949e-05,
+      "loss": 0.7493,
+      "step": 763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907620024104376e-05,
+      "loss": 0.7305,
+      "step": 764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899885837736201e-05,
+      "loss": 0.7242,
+      "step": 765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892146334743704e-05,
+      "loss": 0.6619,
+      "step": 766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884401536785045e-05,
+      "loss": 0.6666,
+      "step": 767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8766514655332e-05,
+      "loss": 0.5772,
+      "step": 768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868896142675902e-05,
+      "loss": 0.6456,
+      "step": 769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861135589915583e-05,
+      "loss": 0.6527,
+      "step": 770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853369828969304e-05,
+      "loss": 0.7701,
+      "step": 771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845598881568705e-05,
+      "loss": 0.7946,
+      "step": 772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837822769459942e-05,
+      "loss": 0.6946,
+      "step": 773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830041514403618e-05,
+      "loss": 0.553,
+      "step": 774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82225513817473e-05,
+      "loss": 0.7436,
+      "step": 775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.81446366256261e-05,
+      "loss": 0.746,
+      "step": 776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806667109370853e-05,
+      "loss": 0.8388,
+      "step": 777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798865500417269e-05,
+      "loss": 0.6666,
+      "step": 778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791058857533814e-05,
+      "loss": 0.6999,
+      "step": 779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.783247202566532e-05,
+      "loss": 0.6496,
+      "step": 780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77543055737549e-05,
+      "loss": 0.6556,
+      "step": 781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76760894383472e-05,
+      "loss": 0.691,
+      "step": 782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759782383832161e-05,
+      "loss": 1.0277,
+      "step": 783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751950899269592e-05,
+      "loss": 1.0601,
+      "step": 784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.74411451206257e-05,
+      "loss": 0.7415,
+      "step": 785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736273244140376e-05,
+      "loss": 0.822,
+      "step": 786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728427117445948e-05,
+      "loss": 0.6683,
+      "step": 787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720576153935818e-05,
+      "loss": 0.714,
+      "step": 788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712720375580056e-05,
+      "loss": 0.6756,
+      "step": 789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704859804362206e-05,
+      "loss": 0.5719,
+      "step": 790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696994462279222e-05,
+      "loss": 0.6679,
+      "step": 791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689124371341411e-05,
+      "loss": 0.6411,
+      "step": 792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68124955357237e-05,
+      "loss": 0.6863,
+      "step": 793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673370031008919e-05,
+      "loss": 1.1815,
+      "step": 794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665485825701047e-05,
+      "loss": 0.7435,
+      "step": 795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.657596959711849e-05,
+      "loss": 0.6682,
+      "step": 796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649703455117458e-05,
+      "loss": 0.6426,
+      "step": 797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.641805334006992e-05,
+      "loss": 0.7946,
+      "step": 798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633902618482483e-05,
+      "loss": 0.6259,
+      "step": 799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.625995330658827e-05,
+      "loss": 0.6798,
+      "step": 800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618083492663707e-05,
+      "loss": 0.8379,
+      "step": 801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610167126637545e-05,
+      "loss": 0.7309,
+      "step": 802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602246254733431e-05,
+      "loss": 0.62,
+      "step": 803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.594320899117064e-05,
+      "loss": 0.7205,
+      "step": 804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.586391081966696e-05,
+      "loss": 0.6035,
+      "step": 805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578456825473055e-05,
+      "loss": 0.7457,
+      "step": 806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5705181518393e-05,
+      "loss": 0.6389,
+      "step": 807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562575083280944e-05,
+      "loss": 0.7861,
+      "step": 808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554627642025807e-05,
+      "loss": 0.7383,
+      "step": 809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.546675850313937e-05,
+      "loss": 0.6148,
+      "step": 810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.538719730397561e-05,
+      "loss": 0.8163,
+      "step": 811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.53075930454102e-05,
+      "loss": 0.6445,
+      "step": 812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522794595020696e-05,
+      "loss": 0.9001,
+      "step": 813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514825624124969e-05,
+      "loss": 0.5888,
+      "step": 814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506852414154138e-05,
+      "loss": 0.6862,
+      "step": 815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.498874987420366e-05,
+      "loss": 0.69,
+      "step": 816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490893366247612e-05,
+      "loss": 0.5893,
+      "step": 817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482907572971583e-05,
+      "loss": 0.698,
+      "step": 818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474917629939652e-05,
+      "loss": 0.6344,
+      "step": 819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466923559510805e-05,
+      "loss": 0.7056,
+      "step": 820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458925384055585e-05,
+      "loss": 0.8364,
+      "step": 821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450923125956014e-05,
+      "loss": 0.7158,
+      "step": 822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442916807605546e-05,
+      "loss": 0.9092,
+      "step": 823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43490645140899e-05,
+      "loss": 0.6564,
+      "step": 824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.426892079782464e-05,
+      "loss": 0.6227,
+      "step": 825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.418873715153312e-05,
+      "loss": 0.7018,
+      "step": 826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41085137996006e-05,
+      "loss": 0.8611,
+      "step": 827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.40282509665234e-05,
+      "loss": 0.7145,
+      "step": 828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394794887690838e-05,
+      "loss": 0.7405,
+      "step": 829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.38676077554722e-05,
+      "loss": 0.7406,
+      "step": 830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378722782704078e-05,
+      "loss": 0.7028,
+      "step": 831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.370680931654862e-05,
+      "loss": 0.5977,
+      "step": 832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362635244903818e-05,
+      "loss": 0.619,
+      "step": 833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.35458574496593e-05,
+      "loss": 0.8205,
+      "step": 834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346532454366849e-05,
+      "loss": 0.8077,
+      "step": 835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.338475395642834e-05,
+      "loss": 0.6353,
+      "step": 836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330414591340689e-05,
+      "loss": 0.6219,
+      "step": 837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322350064017701e-05,
+      "loss": 0.7186,
+      "step": 838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.314281836241573e-05,
+      "loss": 0.6162,
+      "step": 839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306209930590368e-05,
+      "loss": 0.9909,
+      "step": 840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298134369652435e-05,
+      "loss": 0.644,
+      "step": 841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.290055176026356e-05,
+      "loss": 0.8664,
+      "step": 842
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.281972372320877e-05,
+      "loss": 0.8076,
+      "step": 843
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.273885981154845e-05,
+      "loss": 0.767,
+      "step": 844
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.265796025157153e-05,
+      "loss": 0.7237,
+      "step": 845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.257702526966663e-05,
+      "loss": 0.6748,
+      "step": 846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249605509232149e-05,
+      "loss": 1.0827,
+      "step": 847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241504994612236e-05,
+      "loss": 0.6998,
+      "step": 848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.23340100577534e-05,
+      "loss": 0.9568,
+      "step": 849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.225293565399588e-05,
+      "loss": 0.6854,
+      "step": 850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.217182696172776e-05,
+      "loss": 0.7397,
+      "step": 851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20906842079229e-05,
+      "loss": 0.6521,
+      "step": 852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.200950761965049e-05,
+      "loss": 0.6951,
+      "step": 853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.192829742407442e-05,
+      "loss": 0.5945,
+      "step": 854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184705384845259e-05,
+      "loss": 0.7416,
+      "step": 855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.176577712013631e-05,
+      "loss": 0.8404,
+      "step": 856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168446746656973e-05,
+      "loss": 0.8041,
+      "step": 857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.160312511528905e-05,
+      "loss": 0.6246,
+      "step": 858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152175029392203e-05,
+      "loss": 0.592,
+      "step": 859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.144034323018728e-05,
+      "loss": 0.6281,
+      "step": 860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.135890415189363e-05,
+      "loss": 0.6565,
+      "step": 861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.127743328693949e-05,
+      "loss": 1.0469,
+      "step": 862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.119593086331224e-05,
+      "loss": 0.7029,
+      "step": 863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111439710908757e-05,
+      "loss": 0.656,
+      "step": 864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.103283225242885e-05,
+      "loss": 0.6102,
+      "step": 865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0951236521586475e-05,
+      "loss": 0.7728,
+      "step": 866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0869610144897215e-05,
+      "loss": 0.6496,
+      "step": 867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0787953350783665e-05,
+      "loss": 0.6788,
+      "step": 868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.070626636775348e-05,
+      "loss": 0.6299,
+      "step": 869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.062454942439883e-05,
+      "loss": 1.0324,
+      "step": 870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.054280274939569e-05,
+      "loss": 0.6478,
+      "step": 871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0461026571503274e-05,
+      "loss": 0.7462,
+      "step": 872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0379221119563346e-05,
+      "loss": 0.5819,
+      "step": 873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0297386622499575e-05,
+      "loss": 0.7511,
+      "step": 874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0215523309316926e-05,
+      "loss": 0.729,
+      "step": 875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013363140910098e-05,
+      "loss": 0.6679,
+      "step": 876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005171115101735e-05,
+      "loss": 0.6153,
+      "step": 877
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.996976276431096e-05,
+      "loss": 0.5648,
+      "step": 878
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.988778647830554e-05,
+      "loss": 0.6678,
+      "step": 879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.980578252240275e-05,
+      "loss": 0.6203,
+      "step": 880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9723751126081815e-05,
+      "loss": 0.6972,
+      "step": 881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.964169251889867e-05,
+      "loss": 1.0719,
+      "step": 882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.955960693048543e-05,
+      "loss": 0.5555,
+      "step": 883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947749459054972e-05,
+      "loss": 0.8525,
+      "step": 884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.939535572887398e-05,
+      "loss": 0.6711,
+      "step": 885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.931319057531493e-05,
+      "loss": 0.6708,
+      "step": 886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9230999359802784e-05,
+      "loss": 0.5776,
+      "step": 887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9148782312340813e-05,
+      "loss": 0.7152,
+      "step": 888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9066539663004436e-05,
+      "loss": 0.6858,
+      "step": 889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.898427164194084e-05,
+      "loss": 0.6312,
+      "step": 890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.890197847936809e-05,
+      "loss": 0.6705,
+      "step": 891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881966040557471e-05,
+      "loss": 0.6125,
+      "step": 892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.87373176509189e-05,
+      "loss": 0.8353,
+      "step": 893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8654950445827894e-05,
+      "loss": 0.6625,
+      "step": 894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.85725590207974e-05,
+      "loss": 0.7143,
+      "step": 895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.849014360639087e-05,
+      "loss": 0.6857,
+      "step": 896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84077044332389e-05,
+      "loss": 0.6109,
+      "step": 897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832524173203855e-05,
+      "loss": 0.5455,
+      "step": 898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.824275573355278e-05,
+      "loss": 0.6046,
+      "step": 899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.816024666860966e-05,
+      "loss": 0.6985,
+      "step": 900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.807771476810189e-05,
+      "loss": 0.5648,
+      "step": 901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7995160262986004e-05,
+      "loss": 0.8881,
+      "step": 902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.791258338428184e-05,
+      "loss": 0.8201,
+      "step": 903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.782998436307184e-05,
+      "loss": 0.6708,
+      "step": 904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774736343050039e-05,
+      "loss": 0.6896,
+      "step": 905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766472081777321e-05,
+      "loss": 0.6153,
+      "step": 906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7582056756156665e-05,
+      "loss": 0.7003,
+      "step": 907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749937147697719e-05,
+      "loss": 0.6181,
+      "step": 908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.741666521162055e-05,
+      "loss": 0.6456,
+      "step": 909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.733393819153123e-05,
+      "loss": 0.9188,
+      "step": 910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725119064821185e-05,
+      "loss": 0.7887,
+      "step": 911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7168422813222414e-05,
+      "loss": 0.72,
+      "step": 912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.708563491817972e-05,
+      "loss": 0.6714,
+      "step": 913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7002827194756717e-05,
+      "loss": 0.6003,
+      "step": 914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.691999987468183e-05,
+      "loss": 0.5491,
+      "step": 915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.683715318973834e-05,
+      "loss": 0.654,
+      "step": 916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.675428737176367e-05,
+      "loss": 0.7454,
+      "step": 917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.667140265264884e-05,
+      "loss": 0.5358,
+      "step": 918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658849926433774e-05,
+      "loss": 0.6561,
+      "step": 919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.650557743882649e-05,
+      "loss": 0.5904,
+      "step": 920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.642263740816285e-05,
+      "loss": 0.6298,
+      "step": 921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.633967940444547e-05,
+      "loss": 0.8532,
+      "step": 922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625670365982332e-05,
+      "loss": 0.7346,
+      "step": 923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.617371040649503e-05,
+      "loss": 0.6063,
+      "step": 924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.609069987670818e-05,
+      "loss": 0.8683,
+      "step": 925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600767230275879e-05,
+      "loss": 0.6333,
+      "step": 926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5924627916990446e-05,
+      "loss": 0.7952,
+      "step": 927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.584156695179389e-05,
+      "loss": 0.5769,
+      "step": 928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57584896396062e-05,
+      "loss": 0.6642,
+      "step": 929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567539621291024e-05,
+      "loss": 0.7095,
+      "step": 930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.559228690423392e-05,
+      "loss": 0.8606,
+      "step": 931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550916194614963e-05,
+      "loss": 0.8595,
+      "step": 932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5426021571273557e-05,
+      "loss": 0.7933,
+      "step": 933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5342866012265005e-05,
+      "loss": 0.6008,
+      "step": 934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525969550182577e-05,
+      "loss": 0.8082,
+      "step": 935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.517651027269952e-05,
+      "loss": 0.8274,
+      "step": 936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5093310557671074e-05,
+      "loss": 0.7621,
+      "step": 937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.501009658956583e-05,
+      "loss": 0.5651,
+      "step": 938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492686860124901e-05,
+      "loss": 1.0995,
+      "step": 939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.484362682562513e-05,
+      "loss": 0.8572,
+      "step": 940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4760371495637256e-05,
+      "loss": 0.6738,
+      "step": 941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.467710284426639e-05,
+      "loss": 0.7547,
+      "step": 942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.459382110453084e-05,
+      "loss": 0.6363,
+      "step": 943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.451052650948549e-05,
+      "loss": 0.5655,
+      "step": 944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4427219292221235e-05,
+      "loss": 0.5943,
+      "step": 945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.434389968586427e-05,
+      "loss": 0.7651,
+      "step": 946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.426056792357551e-05,
+      "loss": 0.5885,
+      "step": 947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4177224238549787e-05,
+      "loss": 0.7539,
+      "step": 948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.409386886401541e-05,
+      "loss": 0.6401,
+      "step": 949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40105020332333e-05,
+      "loss": 0.537,
+      "step": 950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392712397949652e-05,
+      "loss": 0.712,
+      "step": 951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.384373493612951e-05,
+      "loss": 0.6524,
+      "step": 952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.376033513648743e-05,
+      "loss": 0.6185,
+      "step": 953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.36769248139556e-05,
+      "loss": 0.6494,
+      "step": 954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3593504201948706e-05,
+      "loss": 0.6773,
+      "step": 955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3510073533910346e-05,
+      "loss": 0.7451,
+      "step": 956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.342663304331211e-05,
+      "loss": 0.6765,
+      "step": 957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3343182963653235e-05,
+      "loss": 0.7846,
+      "step": 958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.325972352845965e-05,
+      "loss": 0.7859,
+      "step": 959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.317625497128357e-05,
+      "loss": 0.7214,
+      "step": 960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.309277752570264e-05,
+      "loss": 0.6104,
+      "step": 961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.30092914253195e-05,
+      "loss": 0.7541,
+      "step": 962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.292579690376088e-05,
+      "loss": 0.6619,
+      "step": 963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2842294194677164e-05,
+      "loss": 0.6433,
+      "step": 964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.275878353174165e-05,
+      "loss": 0.6814,
+      "step": 965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.26752651486498e-05,
+      "loss": 0.6301,
+      "step": 966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25917392791188e-05,
+      "loss": 0.5868,
+      "step": 967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25082061568867e-05,
+      "loss": 0.6593,
+      "step": 968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.24246660157119e-05,
+      "loss": 0.7268,
+      "step": 969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.234111908937238e-05,
+      "loss": 0.5097,
+      "step": 970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.22575656116652e-05,
+      "loss": 0.6001,
+      "step": 971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.217400581640566e-05,
+      "loss": 1.2799,
+      "step": 972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2090439937426784e-05,
+      "loss": 0.7302,
+      "step": 973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.200686820857862e-05,
+      "loss": 0.6127,
+      "step": 974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.192329086372757e-05,
+      "loss": 0.6029,
+      "step": 975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183970813675581e-05,
+      "loss": 0.7055,
+      "step": 976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1756120261560446e-05,
+      "loss": 0.7145,
+      "step": 977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167252747205317e-05,
+      "loss": 0.5818,
+      "step": 978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1588930002159255e-05,
+      "loss": 0.6912,
+      "step": 979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150532808581718e-05,
+      "loss": 0.6136,
+      "step": 980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142172195697783e-05,
+      "loss": 0.629,
+      "step": 981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.133811184960389e-05,
+      "loss": 0.6808,
+      "step": 982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.125449799766916e-05,
+      "loss": 0.8666,
+      "step": 983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1170880635157935e-05,
+      "loss": 0.5483,
+      "step": 984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108725999606433e-05,
+      "loss": 0.6326,
+      "step": 985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1003636314391626e-05,
+      "loss": 0.7396,
+      "step": 986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.092000982415162e-05,
+      "loss": 0.6907,
+      "step": 987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083638075936398e-05,
+      "loss": 0.6827,
+      "step": 988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.075274935405554e-05,
+      "loss": 0.6645,
+      "step": 989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.066911584225974e-05,
+      "loss": 0.6932,
+      "step": 990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.058548045801588e-05,
+      "loss": 0.5932,
+      "step": 991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050184343536849e-05,
+      "loss": 0.5621,
+      "step": 992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.041820500836671e-05,
+      "loss": 0.606,
+      "step": 993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0334565411063614e-05,
+      "loss": 0.9303,
+      "step": 994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.025092487751552e-05,
+      "loss": 0.6832,
+      "step": 995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0167283641781384e-05,
+      "loss": 0.6864,
+      "step": 996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0083641937922145e-05,
+      "loss": 0.8748,
+      "step": 997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-05,
+      "loss": 0.6611,
+      "step": 998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.991635806207787e-05,
+      "loss": 0.6891,
+      "step": 999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9832716358218614e-05,
+      "loss": 0.6384,
+      "step": 1000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974907512248451e-05,
+      "loss": 0.6414,
+      "step": 1001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.96654345889364e-05,
+      "loss": 0.613,
+      "step": 1002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.958179499163329e-05,
+      "loss": 0.7273,
+      "step": 1003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.949815656463151e-05,
+      "loss": 0.8013,
+      "step": 1004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941451954198415e-05,
+      "loss": 0.7702,
+      "step": 1005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.933088415774027e-05,
+      "loss": 1.0239,
+      "step": 1006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.924725064594447e-05,
+      "loss": 0.7265,
+      "step": 1007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916361924063604e-05,
+      "loss": 1.068,
+      "step": 1008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.90799901758484e-05,
+      "loss": 0.6591,
+      "step": 1009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.89963636856084e-05,
+      "loss": 0.6818,
+      "step": 1010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.891274000393568e-05,
+      "loss": 0.6277,
+      "step": 1011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.882911936484208e-05,
+      "loss": 0.7054,
+      "step": 1012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.874550200233085e-05,
+      "loss": 0.7728,
+      "step": 1013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866188815039614e-05,
+      "loss": 1.0237,
+      "step": 1014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.857827804302218e-05,
+      "loss": 0.6794,
+      "step": 1015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8494671914182835e-05,
+      "loss": 0.6693,
+      "step": 1016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8411069997840756e-05,
+      "loss": 0.5508,
+      "step": 1017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.832747252794686e-05,
+      "loss": 0.6308,
+      "step": 1018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824387973843956e-05,
+      "loss": 0.6705,
+      "step": 1019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816029186324421e-05,
+      "loss": 0.801,
+      "step": 1020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.807670913627243e-05,
+      "loss": 0.5725,
+      "step": 1021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.799313179142138e-05,
+      "loss": 0.5967,
+      "step": 1022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.790956006257324e-05,
+      "loss": 0.7431,
+      "step": 1023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7825994183594353e-05,
+      "loss": 0.6773,
+      "step": 1024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774243438833481e-05,
+      "loss": 0.6474,
+      "step": 1025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7658880910627616e-05,
+      "loss": 0.9102,
+      "step": 1026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757533398428812e-05,
+      "loss": 0.7081,
+      "step": 1027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749179384311331e-05,
+      "loss": 0.569,
+      "step": 1028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74082607208812e-05,
+      "loss": 0.769,
+      "step": 1029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73247348513502e-05,
+      "loss": 0.8312,
+      "step": 1030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7241216468258375e-05,
+      "loss": 0.6373,
+      "step": 1031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715770580532284e-05,
+      "loss": 0.6813,
+      "step": 1032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7074203096239123e-05,
+      "loss": 0.7834,
+      "step": 1033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6990708574680514e-05,
+      "loss": 0.5317,
+      "step": 1034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6907222474297356e-05,
+      "loss": 0.6346,
+      "step": 1035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682374502871646e-05,
+      "loss": 0.6228,
+      "step": 1036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.674027647154037e-05,
+      "loss": 0.662,
+      "step": 1037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665681703634678e-05,
+      "loss": 0.7538,
+      "step": 1038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.657336695668788e-05,
+      "loss": 0.6148,
+      "step": 1039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.648992646608968e-05,
+      "loss": 0.6278,
+      "step": 1040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.64064957980513e-05,
+      "loss": 0.5806,
+      "step": 1041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.632307518604441e-05,
+      "loss": 0.5842,
+      "step": 1042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623966486351257e-05,
+      "loss": 0.6359,
+      "step": 1043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615626506387051e-05,
+      "loss": 0.5894,
+      "step": 1044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.607287602050349e-05,
+      "loss": 1.2229,
+      "step": 1045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5989497966766715e-05,
+      "loss": 0.6321,
+      "step": 1046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.590613113598461e-05,
+      "loss": 0.7916,
+      "step": 1047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.582277576145022e-05,
+      "loss": 0.7797,
+      "step": 1048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5739432076424515e-05,
+      "loss": 0.7666,
+      "step": 1049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565610031413573e-05,
+      "loss": 0.807,
+      "step": 1050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5572780707778776e-05,
+      "loss": 0.6925,
+      "step": 1051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5489473490514515e-05,
+      "loss": 0.8361,
+      "step": 1052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.540617889546917e-05,
+      "loss": 0.584,
+      "step": 1053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.532289715573361e-05,
+      "loss": 0.6331,
+      "step": 1054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5239628504362756e-05,
+      "loss": 0.5994,
+      "step": 1055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.515637317437488e-05,
+      "loss": 0.5717,
+      "step": 1056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.507313139875102e-05,
+      "loss": 0.5856,
+      "step": 1057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4989903410434195e-05,
+      "loss": 0.7519,
+      "step": 1058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490668944232894e-05,
+      "loss": 0.8109,
+      "step": 1059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4823489727300495e-05,
+      "loss": 0.5756,
+      "step": 1060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.474030449817423e-05,
+      "loss": 0.6553,
+      "step": 1061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.465713398773501e-05,
+      "loss": 0.5967,
+      "step": 1062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4573978428726455e-05,
+      "loss": 0.7621,
+      "step": 1063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.449083805385038e-05,
+      "loss": 0.8609,
+      "step": 1064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.440771309576608e-05,
+      "loss": 0.6263,
+      "step": 1065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432460378708979e-05,
+      "loss": 0.624,
+      "step": 1066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4241510360393804e-05,
+      "loss": 0.7993,
+      "step": 1067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.415843304820612e-05,
+      "loss": 0.5931,
+      "step": 1068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4075372083009566e-05,
+      "loss": 0.6184,
+      "step": 1069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399232769724122e-05,
+      "loss": 0.6129,
+      "step": 1070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.390930012329182e-05,
+      "loss": 0.8197,
+      "step": 1071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.382628959350499e-05,
+      "loss": 1.0878,
+      "step": 1072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.374329634017669e-05,
+      "loss": 0.6523,
+      "step": 1073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366032059555453e-05,
+      "loss": 1.143,
+      "step": 1074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357736259183717e-05,
+      "loss": 0.5628,
+      "step": 1075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.349442256117351e-05,
+      "loss": 0.8073,
+      "step": 1076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.341150073566227e-05,
+      "loss": 0.6239,
+      "step": 1077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332859734735117e-05,
+      "loss": 0.6679,
+      "step": 1078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.324571262823636e-05,
+      "loss": 0.5747,
+      "step": 1079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.316284681026168e-05,
+      "loss": 0.6223,
+      "step": 1080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.308000012531817e-05,
+      "loss": 0.6146,
+      "step": 1081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.299717280524329e-05,
+      "loss": 0.602,
+      "step": 1082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.291436508182029e-05,
+      "loss": 0.6334,
+      "step": 1083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283157718677762e-05,
+      "loss": 0.6853,
+      "step": 1084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.274880935178816e-05,
+      "loss": 0.636,
+      "step": 1085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2666061808468785e-05,
+      "loss": 0.7777,
+      "step": 1086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.258333478837947e-05,
+      "loss": 0.8741,
+      "step": 1087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.250062852302283e-05,
+      "loss": 0.817,
+      "step": 1088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.241794324384335e-05,
+      "loss": 0.661,
+      "step": 1089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2335279182226796e-05,
+      "loss": 0.6399,
+      "step": 1090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2252636569499615e-05,
+      "loss": 0.6608,
+      "step": 1091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.217001563692817e-05,
+      "loss": 0.7172,
+      "step": 1092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.208741661571817e-05,
+      "loss": 0.7308,
+      "step": 1093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.200483973701401e-05,
+      "loss": 1.1123,
+      "step": 1094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.192228523189813e-05,
+      "loss": 0.5995,
+      "step": 1095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.183975333139035e-05,
+      "loss": 0.6186,
+      "step": 1096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1757244266447245e-05,
+      "loss": 0.6408,
+      "step": 1097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.167475826796146e-05,
+      "loss": 0.688,
+      "step": 1098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.159229556676111e-05,
+      "loss": 0.5002,
+      "step": 1099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1509856393609136e-05,
+      "loss": 0.6628,
+      "step": 1100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142744097920261e-05,
+      "loss": 0.549,
+      "step": 1101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1345049554172124e-05,
+      "loss": 0.6119,
+      "step": 1102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.126268234908111e-05,
+      "loss": 0.6546,
+      "step": 1103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.118033959442529e-05,
+      "loss": 0.7044,
+      "step": 1104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1098021520631914e-05,
+      "loss": 0.9533,
+      "step": 1105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.101572835805919e-05,
+      "loss": 0.6624,
+      "step": 1106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.093346033699557e-05,
+      "loss": 0.5716,
+      "step": 1107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.085121768765919e-05,
+      "loss": 0.718,
+      "step": 1108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076900064019721e-05,
+      "loss": 0.6159,
+      "step": 1109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0686809424685104e-05,
+      "loss": 0.6532,
+      "step": 1110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060464427112604e-05,
+      "loss": 0.6534,
+      "step": 1111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052250540945028e-05,
+      "loss": 0.6069,
+      "step": 1112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0440393069514565e-05,
+      "loss": 0.6979,
+      "step": 1113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035830748110134e-05,
+      "loss": 0.6934,
+      "step": 1114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02762488739182e-05,
+      "loss": 0.5529,
+      "step": 1115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019421747759726e-05,
+      "loss": 0.7759,
+      "step": 1116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.011221352169447e-05,
+      "loss": 0.6441,
+      "step": 1117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.003023723568903e-05,
+      "loss": 0.7517,
+      "step": 1118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994828884898267e-05,
+      "loss": 0.5551,
+      "step": 1119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.986636859089904e-05,
+      "loss": 0.6469,
+      "step": 1120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9784476690683085e-05,
+      "loss": 0.6625,
+      "step": 1121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.970261337750043e-05,
+      "loss": 0.6498,
+      "step": 1122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9620778880436666e-05,
+      "loss": 0.6558,
+      "step": 1123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953897342849673e-05,
+      "loss": 0.6598,
+      "step": 1124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9457197250604324e-05,
+      "loss": 0.6784,
+      "step": 1125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.937545057560118e-05,
+      "loss": 0.6394,
+      "step": 1126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9293733632246544e-05,
+      "loss": 0.8244,
+      "step": 1127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921204664921635e-05,
+      "loss": 0.9815,
+      "step": 1128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91303898551028e-05,
+      "loss": 0.7077,
+      "step": 1129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9048763478413536e-05,
+      "loss": 0.6245,
+      "step": 1130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8967167747571145e-05,
+      "loss": 0.6405,
+      "step": 1131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.888560289091244e-05,
+      "loss": 0.6813,
+      "step": 1132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.880406913668777e-05,
+      "loss": 0.66,
+      "step": 1133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8722566713060525e-05,
+      "loss": 0.6545,
+      "step": 1134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.864109584810637e-05,
+      "loss": 0.6582,
+      "step": 1135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8559656769812744e-05,
+      "loss": 0.5931,
+      "step": 1136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.847824970607797e-05,
+      "loss": 0.7702,
+      "step": 1137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.839687488471096e-05,
+      "loss": 0.5634,
+      "step": 1138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8315532533430285e-05,
+      "loss": 0.7906,
+      "step": 1139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.823422287986371e-05,
+      "loss": 0.9326,
+      "step": 1140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815294615154743e-05,
+      "loss": 0.6626,
+      "step": 1141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8071702575925594e-05,
+      "loss": 0.6447,
+      "step": 1142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.799049238034951e-05,
+      "loss": 0.6307,
+      "step": 1143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.790931579207709e-05,
+      "loss": 0.5985,
+      "step": 1144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7828173038272266e-05,
+      "loss": 0.698,
+      "step": 1145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7747064346004134e-05,
+      "loss": 0.7164,
+      "step": 1146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7665989942246625e-05,
+      "loss": 0.6179,
+      "step": 1147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.758495005387764e-05,
+      "loss": 0.648,
+      "step": 1148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.750394490767854e-05,
+      "loss": 0.7596,
+      "step": 1149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.742297473033339e-05,
+      "loss": 0.6304,
+      "step": 1150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.734203974842847e-05,
+      "loss": 0.5912,
+      "step": 1151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.726114018845154e-05,
+      "loss": 0.6971,
+      "step": 1152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.718027627679124e-05,
+      "loss": 0.6798,
+      "step": 1153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7099448239736464e-05,
+      "loss": 0.6779,
+      "step": 1154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.701865630347566e-05,
+      "loss": 0.7109,
+      "step": 1155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.693790069409633e-05,
+      "loss": 0.6341,
+      "step": 1156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.685718163758427e-05,
+      "loss": 0.7161,
+      "step": 1157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.677649935982301e-05,
+      "loss": 0.5863,
+      "step": 1158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.669585408659312e-05,
+      "loss": 0.6647,
+      "step": 1159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6615246043571674e-05,
+      "loss": 0.7414,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6534675456331525e-05,
+      "loss": 0.6782,
+      "step": 1161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.645414255034071e-05,
+      "loss": 0.4977,
+      "step": 1162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637364755096183e-05,
+      "loss": 0.6401,
+      "step": 1163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6293190683451396e-05,
+      "loss": 0.5972,
+      "step": 1164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621277217295923e-05,
+      "loss": 0.5484,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6132392244527805e-05,
+      "loss": 0.6078,
+      "step": 1166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6052051123091634e-05,
+      "loss": 0.771,
+      "step": 1167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.59717490334766e-05,
+      "loss": 0.6865,
+      "step": 1168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.589148620039941e-05,
+      "loss": 0.6765,
+      "step": 1169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.581126284846689e-05,
+      "loss": 0.7025,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.573107920217538e-05,
+      "loss": 0.8099,
+      "step": 1171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.56509354859101e-05,
+      "loss": 0.8641,
+      "step": 1172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.557083192394455e-05,
+      "loss": 0.6743,
+      "step": 1173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5490768740439856e-05,
+      "loss": 0.7511,
+      "step": 1174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.541074615944416e-05,
+      "loss": 0.655,
+      "step": 1175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.533076440489195e-05,
+      "loss": 0.705,
+      "step": 1176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5250823700603496e-05,
+      "loss": 0.6132,
+      "step": 1177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5170924270284166e-05,
+      "loss": 0.5971,
+      "step": 1178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.509106633752387e-05,
+      "loss": 0.6179,
+      "step": 1179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.501125012579637e-05,
+      "loss": 0.5519,
+      "step": 1180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4931475858458637e-05,
+      "loss": 0.8887,
+      "step": 1181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.485174375875031e-05,
+      "loss": 0.7502,
+      "step": 1182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.477205404979304e-05,
+      "loss": 0.5989,
+      "step": 1183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.469240695458983e-05,
+      "loss": 0.7226,
+      "step": 1184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.46128026960244e-05,
+      "loss": 0.7141,
+      "step": 1185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4533241496860645e-05,
+      "loss": 0.7222,
+      "step": 1186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445372357974194e-05,
+      "loss": 0.7579,
+      "step": 1187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437424916719056e-05,
+      "loss": 0.7341,
+      "step": 1188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4294818481607015e-05,
+      "loss": 0.7555,
+      "step": 1189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.421543174526946e-05,
+      "loss": 0.6422,
+      "step": 1190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.413608918033305e-05,
+      "loss": 0.5561,
+      "step": 1191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4056791008829354e-05,
+      "loss": 0.6411,
+      "step": 1192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.397753745266571e-05,
+      "loss": 0.7213,
+      "step": 1193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3898328733624563e-05,
+      "loss": 0.6591,
+      "step": 1194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3819165073362934e-05,
+      "loss": 0.5854,
+      "step": 1195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.374004669341173e-05,
+      "loss": 0.6272,
+      "step": 1196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3660973815175165e-05,
+      "loss": 0.7072,
+      "step": 1197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.358194665993009e-05,
+      "loss": 0.8244,
+      "step": 1198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350296544882543e-05,
+      "loss": 0.7464,
+      "step": 1199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.342403040288151e-05,
+      "loss": 0.5511,
+      "step": 1200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3345141742989525e-05,
+      "loss": 0.6961,
+      "step": 1201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.326629968991083e-05,
+      "loss": 0.6826,
+      "step": 1202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.318750446427632e-05,
+      "loss": 0.6737,
+      "step": 1203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.31087562865859e-05,
+      "loss": 0.6608,
+      "step": 1204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303005537720778e-05,
+      "loss": 0.7103,
+      "step": 1205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2951401956377955e-05,
+      "loss": 0.8048,
+      "step": 1206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.287279624419945e-05,
+      "loss": 0.6836,
+      "step": 1207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2794238460641835e-05,
+      "loss": 0.6472,
+      "step": 1208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271572882554053e-05,
+      "loss": 0.6131,
+      "step": 1209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.263726755859625e-05,
+      "loss": 0.6809,
+      "step": 1210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2558854879374303e-05,
+      "loss": 1.1769,
+      "step": 1211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.248049100730409e-05,
+      "loss": 0.7557,
+      "step": 1212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.24021761616784e-05,
+      "loss": 0.6462,
+      "step": 1213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.23239105616528e-05,
+      "loss": 0.6005,
+      "step": 1214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.224569442624512e-05,
+      "loss": 0.7562,
+      "step": 1215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2167527974334695e-05,
+      "loss": 0.6903,
+      "step": 1216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.208941142466187e-05,
+      "loss": 0.7525,
+      "step": 1217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2011344995827305e-05,
+      "loss": 0.5442,
+      "step": 1218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.19333289062915e-05,
+      "loss": 0.6347,
+      "step": 1219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1855363374373925e-05,
+      "loss": 0.6587,
+      "step": 1220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177744861825271e-05,
+      "loss": 0.7804,
+      "step": 1221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.169958485596384e-05,
+      "loss": 0.6996,
+      "step": 1222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.16217723054006e-05,
+      "loss": 0.6585,
+      "step": 1223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.154401118431296e-05,
+      "loss": 0.7651,
+      "step": 1224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.146630171030697e-05,
+      "loss": 0.726,
+      "step": 1225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.138864410084419e-05,
+      "loss": 0.6889,
+      "step": 1226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1311038573240975e-05,
+      "loss": 0.6055,
+      "step": 1227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1233485344668024e-05,
+      "loss": 1.1057,
+      "step": 1228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1155984632149566e-05,
+      "loss": 0.658,
+      "step": 1229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1078536652562974e-05,
+      "loss": 0.6795,
+      "step": 1230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1001141622638005e-05,
+      "loss": 0.5996,
+      "step": 1231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.092379975895626e-05,
+      "loss": 0.5896,
+      "step": 1232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.084651127795053e-05,
+      "loss": 1.0985,
+      "step": 1233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0769276395904234e-05,
+      "loss": 0.7935,
+      "step": 1234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0692095328950846e-05,
+      "loss": 0.7361,
+      "step": 1235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061496829307318e-05,
+      "loss": 0.7563,
+      "step": 1236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0537895504102874e-05,
+      "loss": 0.6403,
+      "step": 1237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.046087717771976e-05,
+      "loss": 0.5886,
+      "step": 1238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0383913529451284e-05,
+      "loss": 0.877,
+      "step": 1239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0307004774671855e-05,
+      "loss": 0.5997,
+      "step": 1240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0230151128602278e-05,
+      "loss": 0.7012,
+      "step": 1241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0153352806309125e-05,
+      "loss": 0.5938,
+      "step": 1242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0076610022704166e-05,
+      "loss": 0.6572,
+      "step": 1243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.999992299254378e-05,
+      "loss": 0.6511,
+      "step": 1244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9923291930428288e-05,
+      "loss": 0.6108,
+      "step": 1245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9846717050801377e-05,
+      "loss": 0.6978,
+      "step": 1246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9770198567949546e-05,
+      "loss": 0.506,
+      "step": 1247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.969373669600149e-05,
+      "loss": 0.6184,
+      "step": 1248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9617331648927438e-05,
+      "loss": 0.7933,
+      "step": 1249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9540983640538635e-05,
+      "loss": 0.6084,
+      "step": 1250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9464692884486668e-05,
+      "loss": 0.692,
+      "step": 1251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9388459594262957e-05,
+      "loss": 0.9853,
+      "step": 1252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9312283983198095e-05,
+      "loss": 0.5455,
+      "step": 1253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9236166264461274e-05,
+      "loss": 0.6419,
+      "step": 1254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9160106651059625e-05,
+      "loss": 0.5982,
+      "step": 1255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908410535583777e-05,
+      "loss": 0.6656,
+      "step": 1256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.900816259147705e-05,
+      "loss": 0.6699,
+      "step": 1257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8932278570495075e-05,
+      "loss": 0.588,
+      "step": 1258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8856453505245018e-05,
+      "loss": 0.6484,
+      "step": 1259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8780687607915124e-05,
+      "loss": 0.6272,
+      "step": 1260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8704981090527993e-05,
+      "loss": 0.9103,
+      "step": 1261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8629334164940124e-05,
+      "loss": 0.6597,
+      "step": 1262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8553747042841266e-05,
+      "loss": 0.8548,
+      "step": 1263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8478219935753692e-05,
+      "loss": 0.7337,
+      "step": 1264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8402753055031863e-05,
+      "loss": 0.6234,
+      "step": 1265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.832734661186167e-05,
+      "loss": 0.5167,
+      "step": 1266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8252000817259837e-05,
+      "loss": 0.5103,
+      "step": 1267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8176715882073377e-05,
+      "loss": 0.6417,
+      "step": 1268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8101492016979026e-05,
+      "loss": 0.6057,
+      "step": 1269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8026329432482635e-05,
+      "loss": 0.7817,
+      "step": 1270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7951228338918507e-05,
+      "loss": 0.6306,
+      "step": 1271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.787618894644889e-05,
+      "loss": 0.5103,
+      "step": 1272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7801211465063403e-05,
+      "loss": 0.6558,
+      "step": 1273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.77262961045784e-05,
+      "loss": 0.546,
+      "step": 1274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.765144307463635e-05,
+      "loss": 0.7613,
+      "step": 1275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.757665258470538e-05,
+      "loss": 0.7232,
+      "step": 1276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7501924844078534e-05,
+      "loss": 0.9178,
+      "step": 1277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7427260061873268e-05,
+      "loss": 0.5952,
+      "step": 1278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7352658447030882e-05,
+      "loss": 0.6436,
+      "step": 1279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7278120208315926e-05,
+      "loss": 0.555,
+      "step": 1280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.720364555431555e-05,
+      "loss": 0.7267,
+      "step": 1281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7129234693438976e-05,
+      "loss": 0.694,
+      "step": 1282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7054887833916932e-05,
+      "loss": 0.5615,
+      "step": 1283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6980605183801055e-05,
+      "loss": 0.7224,
+      "step": 1284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.690638695096327e-05,
+      "loss": 0.6575,
+      "step": 1285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6832233343095226e-05,
+      "loss": 0.7557,
+      "step": 1286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6758144567707754e-05,
+      "loss": 0.6503,
+      "step": 1287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6684120832130282e-05,
+      "loss": 0.6837,
+      "step": 1288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.661016234351018e-05,
+      "loss": 0.558,
+      "step": 1289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.653626930881222e-05,
+      "loss": 0.8141,
+      "step": 1290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6462441934818078e-05,
+      "loss": 0.6632,
+      "step": 1291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6388680428125657e-05,
+      "loss": 0.6125,
+      "step": 1292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6314984995148507e-05,
+      "loss": 0.5203,
+      "step": 1293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6241355842115277e-05,
+      "loss": 0.6148,
+      "step": 1294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.616779317506921e-05,
+      "loss": 1.1579,
+      "step": 1295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6094297199867406e-05,
+      "loss": 0.644,
+      "step": 1296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6020868122180385e-05,
+      "loss": 0.6236,
+      "step": 1297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.594750614749148e-05,
+      "loss": 0.6619,
+      "step": 1298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5874211481096188e-05,
+      "loss": 0.664,
+      "step": 1299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5800984328101663e-05,
+      "loss": 0.5913,
+      "step": 1300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5727824893426165e-05,
+      "loss": 0.6251,
+      "step": 1301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5654733381798458e-05,
+      "loss": 0.6608,
+      "step": 1302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5581709997757165e-05,
+      "loss": 0.7783,
+      "step": 1303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5508754945650303e-05,
+      "loss": 0.6961,
+      "step": 1304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5435868429634678e-05,
+      "loss": 0.5615,
+      "step": 1305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.536305065367533e-05,
+      "loss": 0.5907,
+      "step": 1306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5290301821544825e-05,
+      "loss": 0.6223,
+      "step": 1307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5217622136822927e-05,
+      "loss": 0.6756,
+      "step": 1308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5145011802895833e-05,
+      "loss": 0.682,
+      "step": 1309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5072471022955702e-05,
+      "loss": 0.5578,
+      "step": 1310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.5109,
+      "step": 1311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4927598936831043e-05,
+      "loss": 0.6691,
+      "step": 1312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4855268036055345e-05,
+      "loss": 0.6285,
+      "step": 1313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4783007500083067e-05,
+      "loss": 0.5377,
+      "step": 1314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.47108175311275e-05,
+      "loss": 0.5422,
+      "step": 1315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.46386983312044e-05,
+      "loss": 0.6637,
+      "step": 1316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4566650102131573e-05,
+      "loss": 0.749,
+      "step": 1317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4494673045528123e-05,
+      "loss": 0.6091,
+      "step": 1318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4422767362814043e-05,
+      "loss": 0.6874,
+      "step": 1319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4350933255209612e-05,
+      "loss": 1.0919,
+      "step": 1320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4279170923734756e-05,
+      "loss": 0.6509,
+      "step": 1321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4207480569208556e-05,
+      "loss": 0.5495,
+      "step": 1322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4135862392248716e-05,
+      "loss": 1.2367,
+      "step": 1323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4064316593270953e-05,
+      "loss": 0.7585,
+      "step": 1324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3992843372488355e-05,
+      "loss": 0.5854,
+      "step": 1325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3921442929911007e-05,
+      "loss": 1.0369,
+      "step": 1326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3850115465345324e-05,
+      "loss": 0.7245,
+      "step": 1327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3778861178393453e-05,
+      "loss": 0.6134,
+      "step": 1328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.370768026845276e-05,
+      "loss": 0.6384,
+      "step": 1329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.363657293471531e-05,
+      "loss": 0.5563,
+      "step": 1330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3565539376167295e-05,
+      "loss": 0.845,
+      "step": 1331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3494579791588366e-05,
+      "loss": 0.7787,
+      "step": 1332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3423694379551263e-05,
+      "loss": 0.6048,
+      "step": 1333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3352883338421082e-05,
+      "loss": 0.6129,
+      "step": 1334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.328214686635487e-05,
+      "loss": 0.5578,
+      "step": 1335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3211485161300933e-05,
+      "loss": 0.613,
+      "step": 1336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3140898420998426e-05,
+      "loss": 0.7515,
+      "step": 1337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3070386842976625e-05,
+      "loss": 0.8714,
+      "step": 1338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.299995062455459e-05,
+      "loss": 0.6053,
+      "step": 1339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2929589962840374e-05,
+      "loss": 0.6345,
+      "step": 1340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2859305054730705e-05,
+      "loss": 0.5907,
+      "step": 1341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278909609691025e-05,
+      "loss": 0.5223,
+      "step": 1342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2718963285851135e-05,
+      "loss": 0.9458,
+      "step": 1343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2648906817812437e-05,
+      "loss": 0.6566,
+      "step": 1344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2578926888839603e-05,
+      "loss": 0.5944,
+      "step": 1345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2509023694763844e-05,
+      "loss": 0.6328,
+      "step": 1346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2439197431201646e-05,
+      "loss": 0.6273,
+      "step": 1347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2369448293554258e-05,
+      "loss": 0.6929,
+      "step": 1348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2299776477007072e-05,
+      "loss": 0.7056,
+      "step": 1349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2230182176529103e-05,
+      "loss": 0.5412,
+      "step": 1350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2160665586872416e-05,
+      "loss": 0.6275,
+      "step": 1351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2091226902571672e-05,
+      "loss": 1.0919,
+      "step": 1352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.20218663179435e-05,
+      "loss": 0.6073,
+      "step": 1353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1952584027085966e-05,
+      "loss": 0.6184,
+      "step": 1354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1883380223878004e-05,
+      "loss": 0.5697,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1814255101978998e-05,
+      "loss": 0.8387,
+      "step": 1356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1745208854828058e-05,
+      "loss": 0.6923,
+      "step": 1357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1676241675643626e-05,
+      "loss": 0.6782,
+      "step": 1358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.16073537574229e-05,
+      "loss": 0.5526,
+      "step": 1359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.153854529294121e-05,
+      "loss": 0.551,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1469816474751565e-05,
+      "loss": 0.681,
+      "step": 1361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1401167495184115e-05,
+      "loss": 0.6222,
+      "step": 1362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.133259854634561e-05,
+      "loss": 0.6625,
+      "step": 1363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1264109820118784e-05,
+      "loss": 0.5897,
+      "step": 1364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1195701508161886e-05,
+      "loss": 0.5496,
+      "step": 1365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1127373801908168e-05,
+      "loss": 0.6375,
+      "step": 1366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.105912689256533e-05,
+      "loss": 0.6762,
+      "step": 1367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0990960971114916e-05,
+      "loss": 0.7821,
+      "step": 1368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0922876228311832e-05,
+      "loss": 0.5734,
+      "step": 1369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0854872854683877e-05,
+      "loss": 0.6351,
+      "step": 1370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0786951040531115e-05,
+      "loss": 0.6822,
+      "step": 1371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.071911097592536e-05,
+      "loss": 0.4984,
+      "step": 1372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0651352850709655e-05,
+      "loss": 0.77,
+      "step": 1373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.058367685449779e-05,
+      "loss": 0.5808,
+      "step": 1374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0516083176673663e-05,
+      "loss": 0.672,
+      "step": 1375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0448572006390877e-05,
+      "loss": 0.551,
+      "step": 1376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0381143532572082e-05,
+      "loss": 0.5572,
+      "step": 1377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.031379794390858e-05,
+      "loss": 0.7255,
+      "step": 1378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0246535428859655e-05,
+      "loss": 0.5673,
+      "step": 1379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0179356175652175e-05,
+      "loss": 0.5795,
+      "step": 1380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0112260372280007e-05,
+      "loss": 0.8024,
+      "step": 1381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.004524820650345e-05,
+      "loss": 0.5827,
+      "step": 1382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.997831986584877e-05,
+      "loss": 0.4556,
+      "step": 1383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9911475537607677e-05,
+      "loss": 0.563,
+      "step": 1384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.984471540883679e-05,
+      "loss": 0.6131,
+      "step": 1385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9778039666357058e-05,
+      "loss": 0.6858,
+      "step": 1386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9711448496753297e-05,
+      "loss": 0.6787,
+      "step": 1387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.964494208637369e-05,
+      "loss": 0.6166,
+      "step": 1388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9578520621329238e-05,
+      "loss": 0.5733,
+      "step": 1389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9512184287493135e-05,
+      "loss": 0.6814,
+      "step": 1390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9445933270500443e-05,
+      "loss": 0.9088,
+      "step": 1391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9379767755747465e-05,
+      "loss": 0.7184,
+      "step": 1392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931368792839118e-05,
+      "loss": 0.5754,
+      "step": 1393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9247693973348836e-05,
+      "loss": 1.0854,
+      "step": 1394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9181786075297324e-05,
+      "loss": 0.4896,
+      "step": 1395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9115964418672793e-05,
+      "loss": 1.1189,
+      "step": 1396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.905022918766995e-05,
+      "loss": 0.7269,
+      "step": 1397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8984580566241755e-05,
+      "loss": 1.2124,
+      "step": 1398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8919018738098705e-05,
+      "loss": 0.8544,
+      "step": 1399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.88535438867085e-05,
+      "loss": 0.5966,
+      "step": 1400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.878815619529537e-05,
+      "loss": 0.5057,
+      "step": 1401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8722855846839705e-05,
+      "loss": 0.7093,
+      "step": 1402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865764302407743e-05,
+      "loss": 0.5618,
+      "step": 1403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8592517909499535e-05,
+      "loss": 1.1138,
+      "step": 1404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.85274806853516e-05,
+      "loss": 1.0355,
+      "step": 1405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8462531533633237e-05,
+      "loss": 0.8037,
+      "step": 1406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8397670636097636e-05,
+      "loss": 0.6349,
+      "step": 1407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8332898174250913e-05,
+      "loss": 0.5222,
+      "step": 1408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8268214329351794e-05,
+      "loss": 0.6375,
+      "step": 1409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.820361928241103e-05,
+      "loss": 0.754,
+      "step": 1410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8139113214190823e-05,
+      "loss": 0.6604,
+      "step": 1411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8074696305204396e-05,
+      "loss": 0.6871,
+      "step": 1412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.801036873571549e-05,
+      "loss": 0.6144,
+      "step": 1413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.794613068573784e-05,
+      "loss": 0.7549,
+      "step": 1414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7881982335034626e-05,
+      "loss": 0.6808,
+      "step": 1415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.781792386311808e-05,
+      "loss": 0.5928,
+      "step": 1416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.775395544924885e-05,
+      "loss": 0.7381,
+      "step": 1417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690077272435634e-05,
+      "loss": 0.5819,
+      "step": 1418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.762628951143454e-05,
+      "loss": 0.5979,
+      "step": 1419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.756259234474873e-05,
+      "loss": 0.6012,
+      "step": 1420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7498985950627793e-05,
+      "loss": 0.8363,
+      "step": 1421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7435470507067313e-05,
+      "loss": 0.732,
+      "step": 1422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7372046191808366e-05,
+      "loss": 0.7693,
+      "step": 1423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7308713182337043e-05,
+      "loss": 0.5792,
+      "step": 1424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7245471655883876e-05,
+      "loss": 0.66,
+      "step": 1425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71823217894234e-05,
+      "loss": 0.6041,
+      "step": 1426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7119263759673675e-05,
+      "loss": 0.6703,
+      "step": 1427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7056297743095777e-05,
+      "loss": 0.703,
+      "step": 1428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.699342391589324e-05,
+      "loss": 0.5963,
+      "step": 1429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6930642454011647e-05,
+      "loss": 0.5416,
+      "step": 1430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.686795353313811e-05,
+      "loss": 0.6381,
+      "step": 1431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6805357328700784e-05,
+      "loss": 0.6267,
+      "step": 1432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.674285401586835e-05,
+      "loss": 0.5709,
+      "step": 1433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6680443769549515e-05,
+      "loss": 0.6576,
+      "step": 1434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.66181267643926e-05,
+      "loss": 0.6846,
+      "step": 1435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6555903174785008e-05,
+      "loss": 0.7574,
+      "step": 1436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6493773174852673e-05,
+      "loss": 0.7729,
+      "step": 1437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.643173693845964e-05,
+      "loss": 0.7262,
+      "step": 1438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6369794639207624e-05,
+      "loss": 0.7694,
+      "step": 1439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6307946450435397e-05,
+      "loss": 0.5581,
+      "step": 1440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.624619254521841e-05,
+      "loss": 0.6087,
+      "step": 1441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6184533096368278e-05,
+      "loss": 0.6781,
+      "step": 1442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6122968276432272e-05,
+      "loss": 0.5679,
+      "step": 1443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.606149825769283e-05,
+      "loss": 0.6113,
+      "step": 1444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6000123212167156e-05,
+      "loss": 0.6733,
+      "step": 1445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5938843311606665e-05,
+      "loss": 0.7617,
+      "step": 1446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587765872749649e-05,
+      "loss": 0.8823,
+      "step": 1447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.581656963105504e-05,
+      "loss": 0.6149,
+      "step": 1448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5755576193233528e-05,
+      "loss": 0.5521,
+      "step": 1449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5694678584715487e-05,
+      "loss": 0.8606,
+      "step": 1450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563387697591626e-05,
+      "loss": 0.6672,
+      "step": 1451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.557317153698253e-05,
+      "loss": 1.128,
+      "step": 1452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.55125624377919e-05,
+      "loss": 0.6329,
+      "step": 1453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545204984795234e-05,
+      "loss": 0.6188,
+      "step": 1454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5391633936801787e-05,
+      "loss": 0.6445,
+      "step": 1455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5331314873407583e-05,
+      "loss": 0.8137,
+      "step": 1456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.527109282656611e-05,
+      "loss": 0.8617,
+      "step": 1457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5210967964802198e-05,
+      "loss": 0.5789,
+      "step": 1458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5150940456368784e-05,
+      "loss": 0.6206,
+      "step": 1459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5091010469246303e-05,
+      "loss": 0.615,
+      "step": 1460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5031178171142345e-05,
+      "loss": 0.6543,
+      "step": 1461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4971443729491081e-05,
+      "loss": 0.6044,
+      "step": 1462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4911807311452874e-05,
+      "loss": 0.6336,
+      "step": 1463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4852269083913777e-05,
+      "loss": 0.6587,
+      "step": 1464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4792829213485048e-05,
+      "loss": 0.7014,
+      "step": 1465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4733487866502699e-05,
+      "loss": 0.5498,
+      "step": 1466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4674245209027066e-05,
+      "loss": 0.8556,
+      "step": 1467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4615101406842325e-05,
+      "loss": 0.8976,
+      "step": 1468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4556056625455922e-05,
+      "loss": 0.6175,
+      "step": 1469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4497111030098303e-05,
+      "loss": 0.5897,
+      "step": 1470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4438264785722333e-05,
+      "loss": 0.6564,
+      "step": 1471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4379518057002833e-05,
+      "loss": 0.6867,
+      "step": 1472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4320871008336123e-05,
+      "loss": 0.6274,
+      "step": 1473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4262323803839622e-05,
+      "loss": 0.5565,
+      "step": 1474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4203876607351347e-05,
+      "loss": 0.7861,
+      "step": 1475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4145529582429395e-05,
+      "loss": 0.6716,
+      "step": 1476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4087282892351623e-05,
+      "loss": 0.7838,
+      "step": 1477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4029136700115031e-05,
+      "loss": 0.6627,
+      "step": 1478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3971091168435462e-05,
+      "loss": 0.7501,
+      "step": 1479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3913146459747012e-05,
+      "loss": 1.016,
+      "step": 1480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3855302736201686e-05,
+      "loss": 0.6677,
+      "step": 1481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3797560159668831e-05,
+      "loss": 0.8432,
+      "step": 1482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3739918891734821e-05,
+      "loss": 0.9453,
+      "step": 1483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3682379093702446e-05,
+      "loss": 0.601,
+      "step": 1484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3624940926590629e-05,
+      "loss": 0.6408,
+      "step": 1485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3567604551133823e-05,
+      "loss": 0.6858,
+      "step": 1486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3510370127781635e-05,
+      "loss": 1.0232,
+      "step": 1487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.345323781669841e-05,
+      "loss": 0.6936,
+      "step": 1488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3396207777762731e-05,
+      "loss": 0.6148,
+      "step": 1489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3339280170566958e-05,
+      "loss": 0.6237,
+      "step": 1490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3282455154416812e-05,
+      "loss": 0.7689,
+      "step": 1491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3225732888330938e-05,
+      "loss": 0.8155,
+      "step": 1492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3169113531040462e-05,
+      "loss": 0.6557,
+      "step": 1493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3112597240988489e-05,
+      "loss": 1.0433,
+      "step": 1494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3056184176329722e-05,
+      "loss": 0.679,
+      "step": 1495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2999874494930004e-05,
+      "loss": 0.6501,
+      "step": 1496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2943668354365878e-05,
+      "loss": 0.5296,
+      "step": 1497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2887565911924115e-05,
+      "loss": 0.6283,
+      "step": 1498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2831567324601324e-05,
+      "loss": 0.6803,
+      "step": 1499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2775672749103474e-05,
+      "loss": 0.6468,
+      "step": 1500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.271988234184544e-05,
+      "loss": 0.7384,
+      "step": 1501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2664196258950639e-05,
+      "loss": 0.6976,
+      "step": 1502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2608614656250556e-05,
+      "loss": 0.6334,
+      "step": 1503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.255313768928425e-05,
+      "loss": 0.5855,
+      "step": 1504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2497765513297977e-05,
+      "loss": 0.5426,
+      "step": 1505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.244249828324478e-05,
+      "loss": 0.6361,
+      "step": 1506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2387336153784018e-05,
+      "loss": 0.5923,
+      "step": 1507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2332279279280906e-05,
+      "loss": 0.8197,
+      "step": 1508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2277327813806122e-05,
+      "loss": 0.6641,
+      "step": 1509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2222481911135386e-05,
+      "loss": 0.6913,
+      "step": 1510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2167741724749026e-05,
+      "loss": 0.5779,
+      "step": 1511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2113107407831497e-05,
+      "loss": 0.5901,
+      "step": 1512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.205857911327099e-05,
+      "loss": 0.5751,
+      "step": 1513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2004156993659027e-05,
+      "loss": 0.5417,
+      "step": 1514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.194984120129003e-05,
+      "loss": 0.6537,
+      "step": 1515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1895631888160825e-05,
+      "loss": 0.7752,
+      "step": 1516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.184152920597028e-05,
+      "loss": 0.5437,
+      "step": 1517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1787533306118909e-05,
+      "loss": 0.6674,
+      "step": 1518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.173364433970835e-05,
+      "loss": 0.5658,
+      "step": 1519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1679862457541053e-05,
+      "loss": 0.5986,
+      "step": 1520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1626187810119748e-05,
+      "loss": 0.5852,
+      "step": 1521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1572620547647144e-05,
+      "loss": 0.7353,
+      "step": 1522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1519160820025382e-05,
+      "loss": 0.6822,
+      "step": 1523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1465808776855718e-05,
+      "loss": 0.5627,
+      "step": 1524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1412564567438072e-05,
+      "loss": 0.5628,
+      "step": 1525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1359428340770567e-05,
+      "loss": 0.5995,
+      "step": 1526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1306400245549158e-05,
+      "loss": 0.9902,
+      "step": 1527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1253480430167229e-05,
+      "loss": 0.9453,
+      "step": 1528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1200669042715161e-05,
+      "loss": 0.7629,
+      "step": 1529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.114796623097984e-05,
+      "loss": 0.5959,
+      "step": 1530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1095372142444388e-05,
+      "loss": 0.6253,
+      "step": 1531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.104288692428766e-05,
+      "loss": 0.5884,
+      "step": 1532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0990510723383852e-05,
+      "loss": 0.6529,
+      "step": 1533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0938243686302063e-05,
+      "loss": 0.5801,
+      "step": 1534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0886085959305914e-05,
+      "loss": 0.749,
+      "step": 1535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.083403768835317e-05,
+      "loss": 0.7223,
+      "step": 1536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0782099019095238e-05,
+      "loss": 0.5465,
+      "step": 1537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0730270096876877e-05,
+      "loss": 0.722,
+      "step": 1538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.067855106673567e-05,
+      "loss": 0.5644,
+      "step": 1539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.062694207340174e-05,
+      "loss": 0.6127,
+      "step": 1540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.057544326129723e-05,
+      "loss": 0.9066,
+      "step": 1541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0524054774536003e-05,
+      "loss": 0.5984,
+      "step": 1542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0472776756923124e-05,
+      "loss": 0.7822,
+      "step": 1543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0421609351954597e-05,
+      "loss": 0.6566,
+      "step": 1544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0370552702816822e-05,
+      "loss": 0.5993,
+      "step": 1545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0319606952386307e-05,
+      "loss": 0.5139,
+      "step": 1546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.026877224322923e-05,
+      "loss": 0.5946,
+      "step": 1547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0218048717600947e-05,
+      "loss": 0.5854,
+      "step": 1548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0167436517445777e-05,
+      "loss": 0.5503,
+      "step": 1549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.011693578439648e-05,
+      "loss": 0.6746,
+      "step": 1550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0066546659773856e-05,
+      "loss": 0.5272,
+      "step": 1551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0016269284586398e-05,
+      "loss": 0.7075,
+      "step": 1552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.96610379952989e-06,
+      "loss": 0.6588,
+      "step": 1553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.916050344987015e-06,
+      "loss": 0.6256,
+      "step": 1554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.866109061026923e-06,
+      "loss": 0.5927,
+      "step": 1555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.81628008740485e-06,
+      "loss": 0.6049,
+      "step": 1556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.766563563561799e-06,
+      "loss": 0.8245,
+      "step": 1557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.716959628624074e-06,
+      "loss": 0.6148,
+      "step": 1558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66746842140287e-06,
+      "loss": 0.805,
+      "step": 1559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.618090080393999e-06,
+      "loss": 0.9268,
+      "step": 1560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.568824743777355e-06,
+      "loss": 0.6675,
+      "step": 1561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.519672549416659e-06,
+      "loss": 0.7722,
+      "step": 1562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.470633634858972e-06,
+      "loss": 0.8001,
+      "step": 1563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.421708137334395e-06,
+      "loss": 0.6838,
+      "step": 1564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.372896193755621e-06,
+      "loss": 0.6169,
+      "step": 1565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.324197940717549e-06,
+      "loss": 0.6077,
+      "step": 1566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.275613514496977e-06,
+      "loss": 0.692,
+      "step": 1567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.227143051052163e-06,
+      "loss": 0.5948,
+      "step": 1568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.178786686022417e-06,
+      "loss": 0.5687,
+      "step": 1569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.130544554727771e-06,
+      "loss": 0.5831,
+      "step": 1570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.082416792168607e-06,
+      "loss": 0.61,
+      "step": 1571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.034403533025255e-06,
+      "loss": 0.5686,
+      "step": 1572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.9865049116576e-06,
+      "loss": 0.5782,
+      "step": 1573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.938721062104727e-06,
+      "loss": 0.6446,
+      "step": 1574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.891052118084552e-06,
+      "loss": 0.6103,
+      "step": 1575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.843498212993467e-06,
+      "loss": 0.7363,
+      "step": 1576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.7960594799059e-06,
+      "loss": 0.9043,
+      "step": 1577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.748736051573974e-06,
+      "loss": 0.5863,
+      "step": 1578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.701528060427194e-06,
+      "loss": 1.0775,
+      "step": 1579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.654435638572e-06,
+      "loss": 0.6323,
+      "step": 1580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.607458917791405e-06,
+      "loss": 0.6035,
+      "step": 1581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.560598029544692e-06,
+      "loss": 0.5429,
+      "step": 1582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.513853104966952e-06,
+      "loss": 0.6117,
+      "step": 1583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.467224274868774e-06,
+      "loss": 0.5925,
+      "step": 1584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.420711669735887e-06,
+      "loss": 0.5963,
+      "step": 1585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.374315419728784e-06,
+      "loss": 0.58,
+      "step": 1586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.328035654682325e-06,
+      "loss": 0.7642,
+      "step": 1587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.281872504105392e-06,
+      "loss": 0.6191,
+      "step": 1588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.235826097180566e-06,
+      "loss": 0.677,
+      "step": 1589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189896562763738e-06,
+      "loss": 0.7149,
+      "step": 1590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.144084029383703e-06,
+      "loss": 0.554,
+      "step": 1591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.098388625241853e-06,
+      "loss": 0.5801,
+      "step": 1592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.052810478211842e-06,
+      "loss": 0.553,
+      "step": 1593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.007349715839163e-06,
+      "loss": 0.6055,
+      "step": 1594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.962006465340822e-06,
+      "loss": 0.5489,
+      "step": 1595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.916780853604971e-06,
+      "loss": 0.6814,
+      "step": 1596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.871673007190599e-06,
+      "loss": 0.7129,
+      "step": 1597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.8266830523271e-06,
+      "loss": 0.6635,
+      "step": 1598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.781811114913994e-06,
+      "loss": 0.6279,
+      "step": 1599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.737057320520513e-06,
+      "loss": 0.6416,
+      "step": 1600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.692421794385313e-06,
+      "loss": 0.7995,
+      "step": 1601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.64790466141605e-06,
+      "loss": 0.5964,
+      "step": 1602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.603506046189107e-06,
+      "loss": 0.6646,
+      "step": 1603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.559226072949166e-06,
+      "loss": 0.5786,
+      "step": 1604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.515064865608956e-06,
+      "loss": 0.6034,
+      "step": 1605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4710225477488016e-06,
+      "loss": 0.56,
+      "step": 1606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.427099242616348e-06,
+      "loss": 0.9708,
+      "step": 1607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.383295073126223e-06,
+      "loss": 0.5327,
+      "step": 1608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.339610161859617e-06,
+      "loss": 0.5451,
+      "step": 1609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.296044631064014e-06,
+      "loss": 0.6898,
+      "step": 1610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.252598602652827e-06,
+      "loss": 0.9563,
+      "step": 1611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.209272198205075e-06,
+      "loss": 0.833,
+      "step": 1612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.166065538964955e-06,
+      "loss": 0.6937,
+      "step": 1613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.122978745841635e-06,
+      "loss": 0.7317,
+      "step": 1614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.080011939408849e-06,
+      "loss": 0.6463,
+      "step": 1615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.037165239904514e-06,
+      "loss": 0.5547,
+      "step": 1616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.994438767230466e-06,
+      "loss": 0.7994,
+      "step": 1617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.951832640952116e-06,
+      "loss": 0.6188,
+      "step": 1618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.909346980298093e-06,
+      "loss": 0.8244,
+      "step": 1619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.86698190415988e-06,
+      "loss": 0.6917,
+      "step": 1620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.824737531091574e-06,
+      "loss": 0.6755,
+      "step": 1621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.782613979309443e-06,
+      "loss": 0.693,
+      "step": 1622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.740611366691696e-06,
+      "loss": 0.7396,
+      "step": 1623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.698729810778065e-06,
+      "loss": 0.5254,
+      "step": 1624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.656969428769566e-06,
+      "loss": 0.6479,
+      "step": 1625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.615330337528075e-06,
+      "loss": 0.6702,
+      "step": 1626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.573812653576062e-06,
+      "loss": 0.6208,
+      "step": 1627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.532416493096272e-06,
+      "loss": 0.6338,
+      "step": 1628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.491141971931369e-06,
+      "loss": 0.9772,
+      "step": 1629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.449989205583645e-06,
+      "loss": 0.5441,
+      "step": 1630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.408958309214597e-06,
+      "loss": 0.5868,
+      "step": 1631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.368049397644765e-06,
+      "loss": 0.6133,
+      "step": 1632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.327262585353311e-06,
+      "loss": 0.9098,
+      "step": 1633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.286597986477683e-06,
+      "loss": 0.7711,
+      "step": 1634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2460557148133425e-06,
+      "loss": 0.73,
+      "step": 1635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.205635883813443e-06,
+      "loss": 0.7534,
+      "step": 1636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1653386065885165e-06,
+      "loss": 0.5437,
+      "step": 1637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.125163995906097e-06,
+      "loss": 0.5347,
+      "step": 1638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.085112164190465e-06,
+      "loss": 0.5038,
+      "step": 1639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.045183223522338e-06,
+      "loss": 0.6054,
+      "step": 1640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.005377285638536e-06,
+      "loss": 0.556,
+      "step": 1641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.965694461931626e-06,
+      "loss": 0.5683,
+      "step": 1642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.926134863449712e-06,
+      "loss": 0.55,
+      "step": 1643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8866986008960235e-06,
+      "loss": 0.7372,
+      "step": 1644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.84738578462865e-06,
+      "loss": 0.8199,
+      "step": 1645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.808196524660253e-06,
+      "loss": 0.603,
+      "step": 1646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.769130930657734e-06,
+      "loss": 0.5363,
+      "step": 1647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.730189111941908e-06,
+      "loss": 0.6102,
+      "step": 1648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.691371177487214e-06,
+      "loss": 0.5851,
+      "step": 1649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.652677235921444e-06,
+      "loss": 0.7504,
+      "step": 1650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6141073955254035e-06,
+      "loss": 0.4806,
+      "step": 1651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.575661764232592e-06,
+      "loss": 0.674,
+      "step": 1652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.537340449628925e-06,
+      "loss": 0.7068,
+      "step": 1653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4991435589524575e-06,
+      "loss": 0.5454,
+      "step": 1654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.461071199093049e-06,
+      "loss": 0.6069,
+      "step": 1655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.423123476592062e-06,
+      "loss": 0.609,
+      "step": 1656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.38530049764206e-06,
+      "loss": 0.6341,
+      "step": 1657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347602368086563e-06,
+      "loss": 0.5777,
+      "step": 1658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.310029193419697e-06,
+      "loss": 0.5778,
+      "step": 1659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.27258107878591e-06,
+      "loss": 0.6369,
+      "step": 1660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235258128979675e-06,
+      "loss": 0.6423,
+      "step": 1661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.198060448445235e-06,
+      "loss": 0.4892,
+      "step": 1662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1609881412762465e-06,
+      "loss": 0.7153,
+      "step": 1663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.124041311215544e-06,
+      "loss": 0.48,
+      "step": 1664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.087220061654829e-06,
+      "loss": 0.655,
+      "step": 1665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.050524495634368e-06,
+      "loss": 0.6603,
+      "step": 1666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0139547158427e-06,
+      "loss": 0.5681,
+      "step": 1667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.97751082461641e-06,
+      "loss": 0.7316,
+      "step": 1668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9411929239397696e-06,
+      "loss": 0.6291,
+      "step": 1669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.905001115444474e-06,
+      "loss": 0.6439,
+      "step": 1670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8689355004093665e-06,
+      "loss": 0.6448,
+      "step": 1671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.832996179760163e-06,
+      "loss": 0.8305,
+      "step": 1672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.797183254069176e-06,
+      "loss": 0.751,
+      "step": 1673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.761496823554945e-06,
+      "loss": 0.6019,
+      "step": 1674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.725936988082097e-06,
+      "loss": 0.618,
+      "step": 1675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.690503847160982e-06,
+      "loss": 0.5951,
+      "step": 1676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.655197499947378e-06,
+      "loss": 0.6326,
+      "step": 1677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.620018045242286e-06,
+      "loss": 0.5648,
+      "step": 1678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5849655814915685e-06,
+      "loss": 0.8371,
+      "step": 1679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.550040206785766e-06,
+      "loss": 0.7224,
+      "step": 1680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5152420188597335e-06,
+      "loss": 0.7014,
+      "step": 1681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.48057111509243e-06,
+      "loss": 0.6973,
+      "step": 1682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4460275925066074e-06,
+      "loss": 0.5611,
+      "step": 1683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.411611547768585e-06,
+      "loss": 0.5655,
+      "step": 1684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3773230771879e-06,
+      "loss": 0.5545,
+      "step": 1685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343162276717144e-06,
+      "loss": 0.5601,
+      "step": 1686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.309129241951587e-06,
+      "loss": 0.5062,
+      "step": 1687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.275224068129002e-06,
+      "loss": 0.6073,
+      "step": 1688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.241446850129322e-06,
+      "loss": 0.7621,
+      "step": 1689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.207797682474435e-06,
+      "loss": 0.7813,
+      "step": 1690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.174276659327897e-06,
+      "loss": 0.6778,
+      "step": 1691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1408838744946225e-06,
+      "loss": 0.5884,
+      "step": 1692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1076194214207106e-06,
+      "loss": 0.618,
+      "step": 1693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0744833931931344e-06,
+      "loss": 0.6338,
+      "step": 1694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.041475882539464e-06,
+      "loss": 0.7675,
+      "step": 1695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.008596981827622e-06,
+      "loss": 0.6902,
+      "step": 1696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.975846783065662e-06,
+      "loss": 0.5053,
+      "step": 1697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9432253779014606e-06,
+      "loss": 0.6072,
+      "step": 1698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.910732857622473e-06,
+      "loss": 0.6656,
+      "step": 1699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.878369313155483e-06,
+      "loss": 0.7221,
+      "step": 1700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.846134835066367e-06,
+      "loss": 0.6798,
+      "step": 1701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.814029513559825e-06,
+      "loss": 0.6313,
+      "step": 1702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.782053438479094e-06,
+      "loss": 0.5532,
+      "step": 1703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.750206699305764e-06,
+      "loss": 0.6359,
+      "step": 1704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.718489385159463e-06,
+      "loss": 0.7563,
+      "step": 1705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.686901584797675e-06,
+      "loss": 0.6924,
+      "step": 1706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6554433866154036e-06,
+      "loss": 0.6583,
+      "step": 1707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6241148786450297e-06,
+      "loss": 0.5119,
+      "step": 1708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.592916148555969e-06,
+      "loss": 0.629,
+      "step": 1709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5618472836544736e-06,
+      "loss": 0.6944,
+      "step": 1710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530908370883401e-06,
+      "loss": 0.5558,
+      "step": 1711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5000994968219405e-06,
+      "loss": 0.6659,
+      "step": 1712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4694207476853866e-06,
+      "loss": 1.093,
+      "step": 1713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.438872209324873e-06,
+      "loss": 0.721,
+      "step": 1714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4084539672271764e-06,
+      "loss": 0.7448,
+      "step": 1715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.378166106514463e-06,
+      "loss": 0.8866,
+      "step": 1716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3480087119440063e-06,
+      "loss": 0.7252,
+      "step": 1717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3179818679079933e-06,
+      "loss": 0.6269,
+      "step": 1718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.288085658433304e-06,
+      "loss": 0.6271,
+      "step": 1719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.258320167181239e-06,
+      "loss": 0.6799,
+      "step": 1720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.228685477447291e-06,
+      "loss": 0.5432,
+      "step": 1721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.199181672160906e-06,
+      "loss": 0.5444,
+      "step": 1722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1698088338853073e-06,
+      "loss": 0.5155,
+      "step": 1723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.140567044817172e-06,
+      "loss": 0.5837,
+      "step": 1724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1114563867864876e-06,
+      "loss": 0.6204,
+      "step": 1725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.082476941256274e-06,
+      "loss": 0.6825,
+      "step": 1726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0536287893223604e-06,
+      "loss": 0.5945,
+      "step": 1727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.024912011713155e-06,
+      "loss": 0.6262,
+      "step": 1728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9963266887894528e-06,
+      "loss": 0.6707,
+      "step": 1729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.967872900544194e-06,
+      "loss": 0.6858,
+      "step": 1730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.939550726602197e-06,
+      "loss": 0.645,
+      "step": 1731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9113602462199907e-06,
+      "loss": 0.6888,
+      "step": 1732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.883301538285582e-06,
+      "loss": 0.7345,
+      "step": 1733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.855374681318229e-06,
+      "loss": 0.8766,
+      "step": 1734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8275797534682012e-06,
+      "loss": 0.684,
+      "step": 1735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7999168325165746e-06,
+      "loss": 0.5868,
+      "step": 1736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7723859958750486e-06,
+      "loss": 0.6525,
+      "step": 1737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.744987320585679e-06,
+      "loss": 0.8771,
+      "step": 1738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7177208833206844e-06,
+      "loss": 0.805,
+      "step": 1739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6905867603822244e-06,
+      "loss": 0.8001,
+      "step": 1740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6635850277022032e-06,
+      "loss": 0.5787,
+      "step": 1741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6367157608420347e-06,
+      "loss": 0.7909,
+      "step": 1742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.609979034992449e-06,
+      "loss": 0.5636,
+      "step": 1743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.58337492497327e-06,
+      "loss": 0.7192,
+      "step": 1744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5569035052332156e-06,
+      "loss": 0.6797,
+      "step": 1745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530564849849676e-06,
+      "loss": 0.6406,
+      "step": 1746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5043590325285195e-06,
+      "loss": 0.7401,
+      "step": 1747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4782861266038904e-06,
+      "loss": 0.7069,
+      "step": 1748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4523462050379865e-06,
+      "loss": 0.6137,
+      "step": 1749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4265393404208415e-06,
+      "loss": 0.5926,
+      "step": 1750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4008656049701872e-06,
+      "loss": 0.7391,
+      "step": 1751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3753250705311926e-06,
+      "loss": 0.5465,
+      "step": 1752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3499178085762407e-06,
+      "loss": 0.5876,
+      "step": 1753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3246438902048197e-06,
+      "loss": 0.6111,
+      "step": 1754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.299503386143237e-06,
+      "loss": 0.6714,
+      "step": 1755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.274496366744483e-06,
+      "loss": 0.6197,
+      "step": 1756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.249622901987963e-06,
+      "loss": 0.9702,
+      "step": 1757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.224883061479377e-06,
+      "loss": 0.7501,
+      "step": 1758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2002769144504943e-06,
+      "loss": 0.6987,
+      "step": 1759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.175804529758929e-06,
+      "loss": 0.7725,
+      "step": 1760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.151465975888012e-06,
+      "loss": 0.9676,
+      "step": 1761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.127261320946533e-06,
+      "loss": 0.6337,
+      "step": 1762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.103190632668595e-06,
+      "loss": 0.6418,
+      "step": 1763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0792539784134036e-06,
+      "loss": 0.6244,
+      "step": 1764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0554514251650947e-06,
+      "loss": 0.7371,
+      "step": 1765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0317830395325255e-06,
+      "loss": 0.5383,
+      "step": 1766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0082488877491033e-06,
+      "loss": 0.6587,
+      "step": 1767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9848490356726036e-06,
+      "loss": 0.6314,
+      "step": 1768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9615835487849675e-06,
+      "loss": 0.6262,
+      "step": 1769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9384524921921386e-06,
+      "loss": 0.6394,
+      "step": 1770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9154559306238607e-06,
+      "loss": 0.6073,
+      "step": 1771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8925939284335226e-06,
+      "loss": 0.6004,
+      "step": 1772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8698665495979595e-06,
+      "loss": 0.6664,
+      "step": 1773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8472738577172688e-06,
+      "loss": 0.6618,
+      "step": 1774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.824815916014644e-06,
+      "loss": 0.6551,
+      "step": 1775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8024927873362019e-06,
+      "loss": 0.6048,
+      "step": 1776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7803045341507952e-06,
+      "loss": 0.5539,
+      "step": 1777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7582512185498445e-06,
+      "loss": 0.6995,
+      "step": 1778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7363329022471563e-06,
+      "loss": 0.5324,
+      "step": 1779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7145496465787557e-06,
+      "loss": 1.0313,
+      "step": 1780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6929015125027315e-06,
+      "loss": 0.5642,
+      "step": 1781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6713885605990298e-06,
+      "loss": 0.7519,
+      "step": 1782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6500108510693114e-06,
+      "loss": 0.5797,
+      "step": 1783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6287684437367723e-06,
+      "loss": 0.5622,
+      "step": 1784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.607661398046001e-06,
+      "loss": 0.6151,
+      "step": 1785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.586689773062755e-06,
+      "loss": 0.5507,
+      "step": 1786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5658536274738621e-06,
+      "loss": 0.7044,
+      "step": 1787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.545153019587009e-06,
+      "loss": 0.7851,
+      "step": 1788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5245880073305963e-06,
+      "loss": 0.6497,
+      "step": 1789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5041586482535842e-06,
+      "loss": 0.6659,
+      "step": 1790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.483864999525325e-06,
+      "loss": 0.5955,
+      "step": 1791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4637071179353812e-06,
+      "loss": 0.5542,
+      "step": 1792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4436850598933959e-06,
+      "loss": 0.6986,
+      "step": 1793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4237988814289338e-06,
+      "loss": 0.5518,
+      "step": 1794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.404048638191302e-06,
+      "loss": 0.5987,
+      "step": 1795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3844343854494123e-06,
+      "loss": 0.6415,
+      "step": 1796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3649561780916199e-06,
+      "loss": 0.5982,
+      "step": 1797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3456140706255782e-06,
+      "loss": 0.5766,
+      "step": 1798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3264081171780796e-06,
+      "loss": 0.6158,
+      "step": 1799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.307338371494893e-06,
+      "loss": 0.7097,
+      "step": 1800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2884048869406307e-06,
+      "loss": 0.6462,
+      "step": 1801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2696077164985997e-06,
+      "loss": 0.6446,
+      "step": 1802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2509469127706453e-06,
+      "loss": 0.8568,
+      "step": 1803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2324225279769953e-06,
+      "loss": 0.6328,
+      "step": 1804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2140346139561276e-06,
+      "loss": 0.5183,
+      "step": 1805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1957832221646314e-06,
+      "loss": 0.6317,
+      "step": 1806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1776684036770347e-06,
+      "loss": 0.6166,
+      "step": 1807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1596902091857042e-06,
+      "loss": 0.6068,
+      "step": 1808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1418486890006574e-06,
+      "loss": 0.6054,
+      "step": 1809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.124143893049462e-06,
+      "loss": 0.5411,
+      "step": 1810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1065758708770468e-06,
+      "loss": 0.6196,
+      "step": 1811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0891446716456367e-06,
+      "loss": 0.7014,
+      "step": 1812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0718503441345452e-06,
+      "loss": 0.8717,
+      "step": 1813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0546929367400705e-06,
+      "loss": 0.7218,
+      "step": 1814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0376724974753448e-06,
+      "loss": 0.5209,
+      "step": 1815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0207890739702298e-06,
+      "loss": 0.5226,
+      "step": 1816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.004042713471165e-06,
+      "loss": 0.6316,
+      "step": 1817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.87433462841003e-07,
+      "loss": 0.7564,
+      "step": 1818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.709613685589314e-07,
+      "loss": 0.9512,
+      "step": 1819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.546264767203327e-07,
+      "loss": 0.5372,
+      "step": 1820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.384288330366086e-07,
+      "loss": 0.551,
+      "step": 1821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.223684828351287e-07,
+      "loss": 0.5527,
+      "step": 1822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.064454710590253e-07,
+      "loss": 0.7486,
+      "step": 1823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.906598422671275e-07,
+      "loss": 0.6886,
+      "step": 1824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.750116406337939e-07,
+      "loss": 0.6347,
+      "step": 1825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.595009099488239e-07,
+      "loss": 0.6005,
+      "step": 1826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.441276936173193e-07,
+      "loss": 0.6199,
+      "step": 1827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.288920346595508e-07,
+      "loss": 0.5772,
+      "step": 1828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.137939757108526e-07,
+      "loss": 0.5548,
+      "step": 1829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.988335590215168e-07,
+      "loss": 0.5858,
+      "step": 1830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.840108264566326e-07,
+      "loss": 0.6369,
+      "step": 1831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.693258194960251e-07,
+      "loss": 0.6988,
+      "step": 1832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.547785792340722e-07,
+      "loss": 0.916,
+      "step": 1833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.403691463796547e-07,
+      "loss": 0.7369,
+      "step": 1834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.260975612560173e-07,
+      "loss": 0.6307,
+      "step": 1835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.119638638006298e-07,
+      "loss": 0.5573,
+      "step": 1836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.9796809356511e-07,
+      "loss": 0.6352,
+      "step": 1837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.841102897151119e-07,
+      "loss": 0.5487,
+      "step": 1838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.703904910301928e-07,
+      "loss": 0.724,
+      "step": 1839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.568087359037078e-07,
+      "loss": 0.6451,
+      "step": 1840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.433650623427379e-07,
+      "loss": 0.5983,
+      "step": 1841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.300595079679395e-07,
+      "loss": 0.6324,
+      "step": 1842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.168921100134617e-07,
+      "loss": 0.6338,
+      "step": 1843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.038629053268463e-07,
+      "loss": 0.5728,
+      "step": 1844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.909719303688999e-07,
+      "loss": 0.6297,
+      "step": 1845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.782192212136217e-07,
+      "loss": 0.5647,
+      "step": 1846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.656048135480763e-07,
+      "loss": 0.6882,
+      "step": 1847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.531287426723264e-07,
+      "loss": 0.5783,
+      "step": 1848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.407910434992947e-07,
+      "loss": 0.714,
+      "step": 1849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.285917505546967e-07,
+      "loss": 0.6167,
+      "step": 1850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.16530897976919e-07,
+      "loss": 0.6579,
+      "step": 1851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.046085195169692e-07,
+      "loss": 0.7213,
+      "step": 1852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.928246485383148e-07,
+      "loss": 0.6089,
+      "step": 1853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.811793180168389e-07,
+      "loss": 0.5782,
+      "step": 1854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.696725605407404e-07,
+      "loss": 0.5609,
+      "step": 1855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5830440831042823e-07,
+      "loss": 0.5983,
+      "step": 1856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.470748931384494e-07,
+      "loss": 0.6472,
+      "step": 1857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.359840464493725e-07,
+      "loss": 0.5815,
+      "step": 1858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.250318992797375e-07,
+      "loss": 0.5603,
+      "step": 1859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.142184822779338e-07,
+      "loss": 0.6832,
+      "step": 1860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0354382570413905e-07,
+      "loss": 0.602,
+      "step": 1861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.930079594302194e-07,
+      "loss": 0.7551,
+      "step": 1862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.826109129396516e-07,
+      "loss": 0.6593,
+      "step": 1863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7235271532745084e-07,
+      "loss": 0.9152,
+      "step": 1864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6223339530006006e-07,
+      "loss": 0.7662,
+      "step": 1865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5225298117529395e-07,
+      "loss": 0.6458,
+      "step": 1866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.424115008822726e-07,
+      "loss": 0.8273,
+      "step": 1867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.327089819612994e-07,
+      "loss": 0.6416,
+      "step": 1868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2314545156382213e-07,
+      "loss": 0.7048,
+      "step": 1869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.137209364523497e-07,
+      "loss": 0.671,
+      "step": 1870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0443546300035766e-07,
+      "loss": 0.8479,
+      "step": 1871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9528905719223287e-07,
+      "loss": 0.5738,
+      "step": 1872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.862817446232069e-07,
+      "loss": 0.6091,
+      "step": 1873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.77413550499267e-07,
+      "loss": 0.9964,
+      "step": 1874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6868449963708965e-07,
+      "loss": 0.6683,
+      "step": 1875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6009461646397395e-07,
+      "loss": 0.5777,
+      "step": 1876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5164392501777487e-07,
+      "loss": 0.6993,
+      "step": 1877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.433324489468369e-07,
+      "loss": 0.7755,
+      "step": 1878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3516021150992717e-07,
+      "loss": 0.7033,
+      "step": 1879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2712723557616332e-07,
+      "loss": 0.6144,
+      "step": 1880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1923354362495263e-07,
+      "loss": 1.0398,
+      "step": 1881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.114791577459363e-07,
+      "loss": 0.676,
+      "step": 1882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0386409963892849e-07,
+      "loss": 1.0523,
+      "step": 1883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9638839061383862e-07,
+      "loss": 0.6502,
+      "step": 1884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8905205159062688e-07,
+      "loss": 0.6378,
+      "step": 1885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.818551030992377e-07,
+      "loss": 0.6179,
+      "step": 1886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7479756527955527e-07,
+      "loss": 0.6192,
+      "step": 1887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.678794578813314e-07,
+      "loss": 0.6855,
+      "step": 1888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.611008002641412e-07,
+      "loss": 0.9148,
+      "step": 1889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5446161139732184e-07,
+      "loss": 0.6414,
+      "step": 1890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.479619098599172e-07,
+      "loss": 1.0145,
+      "step": 1891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4160171384064446e-07,
+      "loss": 0.647,
+      "step": 1892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3538104113782203e-07,
+      "loss": 0.6884,
+      "step": 1893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.292999091593139e-07,
+      "loss": 0.6512,
+      "step": 1894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2335833492252424e-07,
+      "loss": 0.6756,
+      "step": 1895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1755633505429741e-07,
+      "loss": 0.6279,
+      "step": 1896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1189392579090129e-07,
+      "loss": 0.6393,
+      "step": 1897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.063711229779718e-07,
+      "loss": 0.7214,
+      "step": 1898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0098794207047402e-07,
+      "loss": 0.7021,
+      "step": 1899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.574439813265223e-08,
+      "loss": 0.7683,
+      "step": 1900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.064050583800221e-08,
+      "loss": 0.72,
+      "step": 1901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.567627946920453e-08,
+      "loss": 0.8174,
+      "step": 1902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.085173291811354e-08,
+      "loss": 0.5643,
+      "step": 1903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.616687968570179e-08,
+      "loss": 0.5164,
+      "step": 1904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.162173288201568e-08,
+      "loss": 0.6345,
+      "step": 1905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.721630522615318e-08,
+      "loss": 0.7414,
+      "step": 1906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.295060904623617e-08,
+      "loss": 0.8533,
+      "step": 1907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8824656279343745e-08,
+      "loss": 0.6023,
+      "step": 1908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.483845847151226e-08,
+      "loss": 0.5837,
+      "step": 1909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.099202677767978e-08,
+      "loss": 0.6618,
+      "step": 1910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.728537196166949e-08,
+      "loss": 0.7821,
+      "step": 1911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.371850439615633e-08,
+      "loss": 0.6033,
+      "step": 1912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.029143406262259e-08,
+      "loss": 1.0673,
+      "step": 1913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.700417055136352e-08,
+      "loss": 0.5849,
+      "step": 1914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3856723061420627e-08,
+      "loss": 0.7062,
+      "step": 1915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0849100400587304e-08,
+      "loss": 0.8676,
+      "step": 1916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7981310985369935e-08,
+      "loss": 0.6108,
+      "step": 1917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5253362840965687e-08,
+      "loss": 0.6532,
+      "step": 1918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2665263601240327e-08,
+      "loss": 0.6745,
+      "step": 1919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.021702050871155e-08,
+      "loss": 0.5714,
+      "step": 1920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7908640414515676e-08,
+      "loss": 0.5756,
+      "step": 1921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5740129778413215e-08,
+      "loss": 0.5569,
+      "step": 1922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.371149466873889e-08,
+      "loss": 0.5758,
+      "step": 1923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1822740762407192e-08,
+      "loss": 0.5624,
+      "step": 1924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0073873344895735e-08,
+      "loss": 0.6566,
+      "step": 1925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.46489731021749e-09,
+      "loss": 0.6897,
+      "step": 1926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.995817160920792e-09,
+      "loss": 0.5478,
+      "step": 1927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.666637008061582e-09,
+      "loss": 0.9162,
+      "step": 1928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.477360571208955e-09,
+      "loss": 0.6098,
+      "step": 1929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4279911784174113e-09,
+      "loss": 0.783,
+      "step": 1930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.518531766249055e-09,
+      "loss": 0.5654,
+      "step": 1931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7489848797125342e-09,
+      "loss": 0.6701,
+      "step": 1932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1193526723074499e-09,
+      "loss": 0.7362,
+      "step": 1933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.296369059854979e-10,
+      "loss": 0.837,
+      "step": 1934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.798389511726729e-10,
+      "loss": 0.6,
+      "step": 1935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.995978673596248e-11,
+      "loss": 0.61,
+      "step": 1936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5308,
+      "step": 1937
+    },
+    {
+      "epoch": 1.0,
+      "step": 1937,
+      "total_flos": 99720888164352.0,
+      "train_loss": 0.7323010703861313,
+      "train_runtime": 21684.5356,
+      "train_samples_per_second": 1.429,
+      "train_steps_per_second": 0.089
+    }
+  ],
+  "max_steps": 1937,
+  "num_train_epochs": 1,
+  "total_flos": 99720888164352.0,
+  "trial_name": null,
+  "trial_params": null
+}