End of training
Browse files- all_results.json +15 -0
- eval_results.json +10 -0
- train_results.json +8 -0
- trainer_state.json +569 -0
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_cer": 0.04751177554781896,
|
| 4 |
+
"eval_loss": 0.1577913612127304,
|
| 5 |
+
"eval_runtime": 106.154,
|
| 6 |
+
"eval_samples": 3843,
|
| 7 |
+
"eval_samples_per_second": 36.202,
|
| 8 |
+
"eval_steps_per_second": 4.531,
|
| 9 |
+
"eval_wer": 0.6039552432995056,
|
| 10 |
+
"train_loss": 1.2140741569408473,
|
| 11 |
+
"train_runtime": 17169.0647,
|
| 12 |
+
"train_samples": 8841,
|
| 13 |
+
"train_samples_per_second": 25.747,
|
| 14 |
+
"train_steps_per_second": 0.402
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_cer": 0.04751177554781896,
|
| 4 |
+
"eval_loss": 0.1577913612127304,
|
| 5 |
+
"eval_runtime": 106.154,
|
| 6 |
+
"eval_samples": 3843,
|
| 7 |
+
"eval_samples_per_second": 36.202,
|
| 8 |
+
"eval_steps_per_second": 4.531,
|
| 9 |
+
"eval_wer": 0.6039552432995056
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"train_loss": 1.2140741569408473,
|
| 4 |
+
"train_runtime": 17169.0647,
|
| 5 |
+
"train_samples": 8841,
|
| 6 |
+
"train_samples_per_second": 25.747,
|
| 7 |
+
"train_steps_per_second": 0.402
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,569 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 49.99819168173599,
|
| 5 |
+
"global_step": 6900,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.72,
|
| 12 |
+
"learning_rate": 1.485e-05,
|
| 13 |
+
"loss": 10.7673,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.45,
|
| 18 |
+
"learning_rate": 2.985e-05,
|
| 19 |
+
"loss": 4.4331,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.17,
|
| 24 |
+
"learning_rate": 4.484999999999999e-05,
|
| 25 |
+
"loss": 3.4707,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.9,
|
| 30 |
+
"learning_rate": 5.985e-05,
|
| 31 |
+
"loss": 3.0807,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 3.62,
|
| 36 |
+
"learning_rate": 7.484999999999999e-05,
|
| 37 |
+
"loss": 2.9736,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 3.62,
|
| 42 |
+
"eval_cer": 1.0,
|
| 43 |
+
"eval_loss": 2.9508235454559326,
|
| 44 |
+
"eval_runtime": 114.1579,
|
| 45 |
+
"eval_samples_per_second": 33.664,
|
| 46 |
+
"eval_steps_per_second": 4.213,
|
| 47 |
+
"eval_wer": 1.0,
|
| 48 |
+
"step": 500
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 4.35,
|
| 52 |
+
"learning_rate": 8.984999999999999e-05,
|
| 53 |
+
"loss": 2.9229,
|
| 54 |
+
"step": 600
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 5.07,
|
| 58 |
+
"learning_rate": 0.00010484999999999999,
|
| 59 |
+
"loss": 2.8971,
|
| 60 |
+
"step": 700
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 5.8,
|
| 64 |
+
"learning_rate": 0.00011985,
|
| 65 |
+
"loss": 2.7156,
|
| 66 |
+
"step": 800
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 6.52,
|
| 70 |
+
"learning_rate": 0.00013485,
|
| 71 |
+
"loss": 1.8057,
|
| 72 |
+
"step": 900
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 7.25,
|
| 76 |
+
"learning_rate": 0.00014984999999999998,
|
| 77 |
+
"loss": 1.3293,
|
| 78 |
+
"step": 1000
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 7.25,
|
| 82 |
+
"eval_cer": 0.08615393306828052,
|
| 83 |
+
"eval_loss": 0.33302000164985657,
|
| 84 |
+
"eval_runtime": 114.4981,
|
| 85 |
+
"eval_samples_per_second": 33.564,
|
| 86 |
+
"eval_steps_per_second": 4.201,
|
| 87 |
+
"eval_wer": 0.8407494145199064,
|
| 88 |
+
"step": 1000
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 7.97,
|
| 92 |
+
"learning_rate": 0.00016485,
|
| 93 |
+
"loss": 1.1212,
|
| 94 |
+
"step": 1100
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 8.69,
|
| 98 |
+
"learning_rate": 0.00017984999999999998,
|
| 99 |
+
"loss": 1.0403,
|
| 100 |
+
"step": 1200
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 9.42,
|
| 104 |
+
"learning_rate": 0.00019484999999999997,
|
| 105 |
+
"loss": 0.9903,
|
| 106 |
+
"step": 1300
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 10.14,
|
| 110 |
+
"learning_rate": 0.00020984999999999998,
|
| 111 |
+
"loss": 0.9766,
|
| 112 |
+
"step": 1400
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 10.87,
|
| 116 |
+
"learning_rate": 0.00022485,
|
| 117 |
+
"loss": 0.956,
|
| 118 |
+
"step": 1500
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 10.87,
|
| 122 |
+
"eval_cer": 0.06022301157428658,
|
| 123 |
+
"eval_loss": 0.2042141854763031,
|
| 124 |
+
"eval_runtime": 114.0353,
|
| 125 |
+
"eval_samples_per_second": 33.7,
|
| 126 |
+
"eval_steps_per_second": 4.218,
|
| 127 |
+
"eval_wer": 0.6872235232890971,
|
| 128 |
+
"step": 1500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 11.59,
|
| 132 |
+
"learning_rate": 0.00023984999999999998,
|
| 133 |
+
"loss": 0.9616,
|
| 134 |
+
"step": 1600
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 12.32,
|
| 138 |
+
"learning_rate": 0.00025485,
|
| 139 |
+
"loss": 0.9653,
|
| 140 |
+
"step": 1700
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 13.04,
|
| 144 |
+
"learning_rate": 0.00026984999999999997,
|
| 145 |
+
"loss": 0.9653,
|
| 146 |
+
"step": 1800
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 13.77,
|
| 150 |
+
"learning_rate": 0.00028484999999999996,
|
| 151 |
+
"loss": 0.9486,
|
| 152 |
+
"step": 1900
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 14.49,
|
| 156 |
+
"learning_rate": 0.00029985,
|
| 157 |
+
"loss": 0.9509,
|
| 158 |
+
"step": 2000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 14.49,
|
| 162 |
+
"eval_cer": 0.06515214643343903,
|
| 163 |
+
"eval_loss": 0.21843121945858002,
|
| 164 |
+
"eval_runtime": 115.7242,
|
| 165 |
+
"eval_samples_per_second": 33.208,
|
| 166 |
+
"eval_steps_per_second": 4.156,
|
| 167 |
+
"eval_wer": 0.7088212334113974,
|
| 168 |
+
"step": 2000
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 15.22,
|
| 172 |
+
"learning_rate": 0.00029393877551020406,
|
| 173 |
+
"loss": 0.9402,
|
| 174 |
+
"step": 2100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 15.94,
|
| 178 |
+
"learning_rate": 0.0002878163265306122,
|
| 179 |
+
"loss": 0.9357,
|
| 180 |
+
"step": 2200
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 16.67,
|
| 184 |
+
"learning_rate": 0.0002816938775510204,
|
| 185 |
+
"loss": 0.9301,
|
| 186 |
+
"step": 2300
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 17.39,
|
| 190 |
+
"learning_rate": 0.00027563265306122445,
|
| 191 |
+
"loss": 0.9268,
|
| 192 |
+
"step": 2400
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 18.12,
|
| 196 |
+
"learning_rate": 0.00026951020408163266,
|
| 197 |
+
"loss": 0.9272,
|
| 198 |
+
"step": 2500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 18.12,
|
| 202 |
+
"eval_cer": 0.07027901163078096,
|
| 203 |
+
"eval_loss": 0.2312462031841278,
|
| 204 |
+
"eval_runtime": 107.6714,
|
| 205 |
+
"eval_samples_per_second": 35.692,
|
| 206 |
+
"eval_steps_per_second": 4.467,
|
| 207 |
+
"eval_wer": 0.7210512620348686,
|
| 208 |
+
"step": 2500
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 18.84,
|
| 212 |
+
"learning_rate": 0.0002633877551020408,
|
| 213 |
+
"loss": 0.9191,
|
| 214 |
+
"step": 2600
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 19.56,
|
| 218 |
+
"learning_rate": 0.00025726530612244896,
|
| 219 |
+
"loss": 0.9016,
|
| 220 |
+
"step": 2700
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 20.29,
|
| 224 |
+
"learning_rate": 0.0002511428571428571,
|
| 225 |
+
"loss": 0.9105,
|
| 226 |
+
"step": 2800
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 21.01,
|
| 230 |
+
"learning_rate": 0.00024502040816326527,
|
| 231 |
+
"loss": 0.8923,
|
| 232 |
+
"step": 2900
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 21.74,
|
| 236 |
+
"learning_rate": 0.00023889795918367345,
|
| 237 |
+
"loss": 0.8561,
|
| 238 |
+
"step": 3000
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 21.74,
|
| 242 |
+
"eval_cer": 0.06309716327582676,
|
| 243 |
+
"eval_loss": 0.21578675508499146,
|
| 244 |
+
"eval_runtime": 108.0601,
|
| 245 |
+
"eval_samples_per_second": 35.564,
|
| 246 |
+
"eval_steps_per_second": 4.451,
|
| 247 |
+
"eval_wer": 0.6838407494145199,
|
| 248 |
+
"step": 3000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 22.46,
|
| 252 |
+
"learning_rate": 0.0002327755102040816,
|
| 253 |
+
"loss": 0.8562,
|
| 254 |
+
"step": 3100
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 23.19,
|
| 258 |
+
"learning_rate": 0.00022665306122448975,
|
| 259 |
+
"loss": 0.855,
|
| 260 |
+
"step": 3200
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 23.91,
|
| 264 |
+
"learning_rate": 0.00022053061224489796,
|
| 265 |
+
"loss": 0.8398,
|
| 266 |
+
"step": 3300
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 24.64,
|
| 270 |
+
"learning_rate": 0.0002144081632653061,
|
| 271 |
+
"loss": 0.827,
|
| 272 |
+
"step": 3400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 25.36,
|
| 276 |
+
"learning_rate": 0.00020828571428571426,
|
| 277 |
+
"loss": 0.8258,
|
| 278 |
+
"step": 3500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 25.36,
|
| 282 |
+
"eval_cer": 0.060060590225059496,
|
| 283 |
+
"eval_loss": 0.19697847962379456,
|
| 284 |
+
"eval_runtime": 107.078,
|
| 285 |
+
"eval_samples_per_second": 35.89,
|
| 286 |
+
"eval_steps_per_second": 4.492,
|
| 287 |
+
"eval_wer": 0.6843611761644548,
|
| 288 |
+
"step": 3500
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 26.09,
|
| 292 |
+
"learning_rate": 0.00020216326530612242,
|
| 293 |
+
"loss": 0.8283,
|
| 294 |
+
"step": 3600
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 26.81,
|
| 298 |
+
"learning_rate": 0.00019604081632653057,
|
| 299 |
+
"loss": 0.8212,
|
| 300 |
+
"step": 3700
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 27.54,
|
| 304 |
+
"learning_rate": 0.00018991836734693878,
|
| 305 |
+
"loss": 0.8332,
|
| 306 |
+
"step": 3800
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 28.26,
|
| 310 |
+
"learning_rate": 0.00018379591836734693,
|
| 311 |
+
"loss": 0.8324,
|
| 312 |
+
"step": 3900
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 28.98,
|
| 316 |
+
"learning_rate": 0.00017767346938775508,
|
| 317 |
+
"loss": 0.7993,
|
| 318 |
+
"step": 4000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 28.98,
|
| 322 |
+
"eval_cer": 0.05770901156016298,
|
| 323 |
+
"eval_loss": 0.18949392437934875,
|
| 324 |
+
"eval_runtime": 106.6069,
|
| 325 |
+
"eval_samples_per_second": 36.048,
|
| 326 |
+
"eval_steps_per_second": 4.512,
|
| 327 |
+
"eval_wer": 0.6697892271662763,
|
| 328 |
+
"step": 4000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 29.71,
|
| 332 |
+
"learning_rate": 0.00017155102040816324,
|
| 333 |
+
"loss": 0.7777,
|
| 334 |
+
"step": 4100
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 30.43,
|
| 338 |
+
"learning_rate": 0.0001654285714285714,
|
| 339 |
+
"loss": 0.7764,
|
| 340 |
+
"step": 4200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 31.16,
|
| 344 |
+
"learning_rate": 0.0001593061224489796,
|
| 345 |
+
"loss": 0.7748,
|
| 346 |
+
"step": 4300
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 31.88,
|
| 350 |
+
"learning_rate": 0.00015318367346938775,
|
| 351 |
+
"loss": 0.7587,
|
| 352 |
+
"step": 4400
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 32.61,
|
| 356 |
+
"learning_rate": 0.00014712244897959183,
|
| 357 |
+
"loss": 0.7525,
|
| 358 |
+
"step": 4500
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 32.61,
|
| 362 |
+
"eval_cer": 0.05496197221888748,
|
| 363 |
+
"eval_loss": 0.18448850512504578,
|
| 364 |
+
"eval_runtime": 106.7665,
|
| 365 |
+
"eval_samples_per_second": 35.994,
|
| 366 |
+
"eval_steps_per_second": 4.505,
|
| 367 |
+
"eval_wer": 0.6453291699193339,
|
| 368 |
+
"step": 4500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 33.33,
|
| 372 |
+
"learning_rate": 0.00014099999999999998,
|
| 373 |
+
"loss": 0.7558,
|
| 374 |
+
"step": 4600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 34.06,
|
| 378 |
+
"learning_rate": 0.00013487755102040816,
|
| 379 |
+
"loss": 0.7444,
|
| 380 |
+
"step": 4700
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 34.78,
|
| 384 |
+
"learning_rate": 0.00012875510204081632,
|
| 385 |
+
"loss": 0.7302,
|
| 386 |
+
"step": 4800
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 35.51,
|
| 390 |
+
"learning_rate": 0.0001226326530612245,
|
| 391 |
+
"loss": 0.7391,
|
| 392 |
+
"step": 4900
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 36.23,
|
| 396 |
+
"learning_rate": 0.00011657142857142856,
|
| 397 |
+
"loss": 0.7211,
|
| 398 |
+
"step": 5000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 36.23,
|
| 402 |
+
"eval_cer": 0.05309765760167223,
|
| 403 |
+
"eval_loss": 0.17814987897872925,
|
| 404 |
+
"eval_runtime": 106.0316,
|
| 405 |
+
"eval_samples_per_second": 36.244,
|
| 406 |
+
"eval_steps_per_second": 4.536,
|
| 407 |
+
"eval_wer": 0.6273744470465782,
|
| 408 |
+
"step": 5000
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 36.95,
|
| 412 |
+
"learning_rate": 0.00011044897959183672,
|
| 413 |
+
"loss": 0.7074,
|
| 414 |
+
"step": 5100
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 37.68,
|
| 418 |
+
"learning_rate": 0.00010432653061224489,
|
| 419 |
+
"loss": 0.6976,
|
| 420 |
+
"step": 5200
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 38.41,
|
| 424 |
+
"learning_rate": 9.820408163265305e-05,
|
| 425 |
+
"loss": 0.6933,
|
| 426 |
+
"step": 5300
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 39.13,
|
| 430 |
+
"learning_rate": 9.214285714285714e-05,
|
| 431 |
+
"loss": 0.6845,
|
| 432 |
+
"step": 5400
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 39.85,
|
| 436 |
+
"learning_rate": 8.602040816326529e-05,
|
| 437 |
+
"loss": 0.677,
|
| 438 |
+
"step": 5500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 39.85,
|
| 442 |
+
"eval_cer": 0.05139576433368407,
|
| 443 |
+
"eval_loss": 0.17315863072872162,
|
| 444 |
+
"eval_runtime": 106.2239,
|
| 445 |
+
"eval_samples_per_second": 36.178,
|
| 446 |
+
"eval_steps_per_second": 4.528,
|
| 447 |
+
"eval_wer": 0.6187874056726516,
|
| 448 |
+
"step": 5500
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 40.58,
|
| 452 |
+
"learning_rate": 7.989795918367346e-05,
|
| 453 |
+
"loss": 0.6951,
|
| 454 |
+
"step": 5600
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 41.3,
|
| 458 |
+
"learning_rate": 7.377551020408162e-05,
|
| 459 |
+
"loss": 0.6814,
|
| 460 |
+
"step": 5700
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 42.03,
|
| 464 |
+
"learning_rate": 6.765306122448979e-05,
|
| 465 |
+
"loss": 0.6822,
|
| 466 |
+
"step": 5800
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 42.75,
|
| 470 |
+
"learning_rate": 6.153061224489796e-05,
|
| 471 |
+
"loss": 0.6658,
|
| 472 |
+
"step": 5900
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 43.48,
|
| 476 |
+
"learning_rate": 5.5408163265306116e-05,
|
| 477 |
+
"loss": 0.6517,
|
| 478 |
+
"step": 6000
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 43.48,
|
| 482 |
+
"eval_cer": 0.05030824747364184,
|
| 483 |
+
"eval_loss": 0.1690707802772522,
|
| 484 |
+
"eval_runtime": 105.9054,
|
| 485 |
+
"eval_samples_per_second": 36.287,
|
| 486 |
+
"eval_steps_per_second": 4.542,
|
| 487 |
+
"eval_wer": 0.6177465521727816,
|
| 488 |
+
"step": 6000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 44.2,
|
| 492 |
+
"learning_rate": 4.928571428571428e-05,
|
| 493 |
+
"loss": 0.6327,
|
| 494 |
+
"step": 6100
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 44.93,
|
| 498 |
+
"learning_rate": 4.316326530612245e-05,
|
| 499 |
+
"loss": 0.6355,
|
| 500 |
+
"step": 6200
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 45.65,
|
| 504 |
+
"learning_rate": 3.704081632653061e-05,
|
| 505 |
+
"loss": 0.6328,
|
| 506 |
+
"step": 6300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 46.38,
|
| 510 |
+
"learning_rate": 3.0918367346938774e-05,
|
| 511 |
+
"loss": 0.6388,
|
| 512 |
+
"step": 6400
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 47.1,
|
| 516 |
+
"learning_rate": 2.4795918367346937e-05,
|
| 517 |
+
"loss": 0.6326,
|
| 518 |
+
"step": 6500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 47.1,
|
| 522 |
+
"eval_cer": 0.0478507418418581,
|
| 523 |
+
"eval_loss": 0.1618689000606537,
|
| 524 |
+
"eval_runtime": 108.5769,
|
| 525 |
+
"eval_samples_per_second": 35.394,
|
| 526 |
+
"eval_steps_per_second": 4.43,
|
| 527 |
+
"eval_wer": 0.6044756700494406,
|
| 528 |
+
"step": 6500
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 47.82,
|
| 532 |
+
"learning_rate": 1.86734693877551e-05,
|
| 533 |
+
"loss": 0.6231,
|
| 534 |
+
"step": 6600
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 48.55,
|
| 538 |
+
"learning_rate": 1.2551020408163265e-05,
|
| 539 |
+
"loss": 0.6309,
|
| 540 |
+
"step": 6700
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 49.27,
|
| 544 |
+
"learning_rate": 6.428571428571428e-06,
|
| 545 |
+
"loss": 0.6341,
|
| 546 |
+
"step": 6800
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 50.0,
|
| 550 |
+
"learning_rate": 3.0612244897959183e-07,
|
| 551 |
+
"loss": 0.6141,
|
| 552 |
+
"step": 6900
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 50.0,
|
| 556 |
+
"step": 6900,
|
| 557 |
+
"total_flos": 4.523197640315366e+19,
|
| 558 |
+
"train_loss": 1.2140741569408473,
|
| 559 |
+
"train_runtime": 17169.0647,
|
| 560 |
+
"train_samples_per_second": 25.747,
|
| 561 |
+
"train_steps_per_second": 0.402
|
| 562 |
+
}
|
| 563 |
+
],
|
| 564 |
+
"max_steps": 6900,
|
| 565 |
+
"num_train_epochs": 50,
|
| 566 |
+
"total_flos": 4.523197640315366e+19,
|
| 567 |
+
"trial_name": null,
|
| 568 |
+
"trial_params": null
|
| 569 |
+
}
|