Commit
·
6312117
1
Parent(s):
d047de0
End of training
Browse files- all_results.json +16 -0
- eval_results.json +10 -0
- train_results.json +9 -0
- trainer_state.json +549 -0
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_cer": 0.03029062627916496,
|
| 4 |
+
"eval_loss": 0.17724749445915222,
|
| 5 |
+
"eval_runtime": 1.7983,
|
| 6 |
+
"eval_samples": 29,
|
| 7 |
+
"eval_samples_per_second": 16.126,
|
| 8 |
+
"eval_steps_per_second": 2.224,
|
| 9 |
+
"eval_wer": 0.11138613861386139,
|
| 10 |
+
"total_flos": 9.822732861444276e+18,
|
| 11 |
+
"train_loss": 1.1235487569173177,
|
| 12 |
+
"train_runtime": 3285.5807,
|
| 13 |
+
"train_samples": 226,
|
| 14 |
+
"train_samples_per_second": 6.879,
|
| 15 |
+
"train_steps_per_second": 0.213
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_cer": 0.03029062627916496,
|
| 4 |
+
"eval_loss": 0.17724749445915222,
|
| 5 |
+
"eval_runtime": 1.7983,
|
| 6 |
+
"eval_samples": 29,
|
| 7 |
+
"eval_samples_per_second": 16.126,
|
| 8 |
+
"eval_steps_per_second": 2.224,
|
| 9 |
+
"eval_wer": 0.11138613861386139
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"total_flos": 9.822732861444276e+18,
|
| 4 |
+
"train_loss": 1.1235487569173177,
|
| 5 |
+
"train_runtime": 3285.5807,
|
| 6 |
+
"train_samples": 226,
|
| 7 |
+
"train_samples_per_second": 6.879,
|
| 8 |
+
"train_steps_per_second": 0.213
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,549 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.17724749445915222,
|
| 3 |
+
"best_model_checkpoint": "wav2vec2-xlsr-1b-mecita-portuguese-all-text-protecao_aos_pandas/checkpoint-225",
|
| 4 |
+
"epoch": 50.0,
|
| 5 |
+
"global_step": 375,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 2.9957142857142858e-05,
|
| 13 |
+
"loss": 13.7229,
|
| 14 |
+
"step": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.93,
|
| 18 |
+
"eval_cer": 0.9995906672124437,
|
| 19 |
+
"eval_loss": 4.859188556671143,
|
| 20 |
+
"eval_runtime": 1.8002,
|
| 21 |
+
"eval_samples_per_second": 16.109,
|
| 22 |
+
"eval_steps_per_second": 2.222,
|
| 23 |
+
"eval_wer": 1.0,
|
| 24 |
+
"step": 7
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 2.0,
|
| 28 |
+
"eval_cer": 1.0,
|
| 29 |
+
"eval_loss": 3.002265691757202,
|
| 30 |
+
"eval_runtime": 1.7603,
|
| 31 |
+
"eval_samples_per_second": 16.475,
|
| 32 |
+
"eval_steps_per_second": 2.272,
|
| 33 |
+
"eval_wer": 1.0,
|
| 34 |
+
"step": 15
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 2.93,
|
| 38 |
+
"eval_cer": 1.0,
|
| 39 |
+
"eval_loss": 2.929029941558838,
|
| 40 |
+
"eval_runtime": 1.7878,
|
| 41 |
+
"eval_samples_per_second": 16.221,
|
| 42 |
+
"eval_steps_per_second": 2.237,
|
| 43 |
+
"eval_wer": 1.0,
|
| 44 |
+
"step": 22
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_cer": 1.0,
|
| 49 |
+
"eval_loss": 2.98417067527771,
|
| 50 |
+
"eval_runtime": 1.7731,
|
| 51 |
+
"eval_samples_per_second": 16.355,
|
| 52 |
+
"eval_steps_per_second": 2.256,
|
| 53 |
+
"eval_wer": 1.0,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 4.93,
|
| 58 |
+
"eval_cer": 1.0,
|
| 59 |
+
"eval_loss": 2.845335006713867,
|
| 60 |
+
"eval_runtime": 1.7685,
|
| 61 |
+
"eval_samples_per_second": 16.398,
|
| 62 |
+
"eval_steps_per_second": 2.262,
|
| 63 |
+
"eval_wer": 1.0,
|
| 64 |
+
"step": 37
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 6.0,
|
| 68 |
+
"eval_cer": 1.0,
|
| 69 |
+
"eval_loss": 2.8119678497314453,
|
| 70 |
+
"eval_runtime": 1.8211,
|
| 71 |
+
"eval_samples_per_second": 15.924,
|
| 72 |
+
"eval_steps_per_second": 2.196,
|
| 73 |
+
"eval_wer": 1.0,
|
| 74 |
+
"step": 45
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 6.93,
|
| 78 |
+
"eval_cer": 1.0,
|
| 79 |
+
"eval_loss": 2.816225528717041,
|
| 80 |
+
"eval_runtime": 1.805,
|
| 81 |
+
"eval_samples_per_second": 16.067,
|
| 82 |
+
"eval_steps_per_second": 2.216,
|
| 83 |
+
"eval_wer": 1.0,
|
| 84 |
+
"step": 52
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 8.0,
|
| 88 |
+
"eval_cer": 1.0,
|
| 89 |
+
"eval_loss": 2.7843384742736816,
|
| 90 |
+
"eval_runtime": 1.7945,
|
| 91 |
+
"eval_samples_per_second": 16.161,
|
| 92 |
+
"eval_steps_per_second": 2.229,
|
| 93 |
+
"eval_wer": 1.0,
|
| 94 |
+
"step": 60
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 8.93,
|
| 98 |
+
"eval_cer": 1.0,
|
| 99 |
+
"eval_loss": 2.782341957092285,
|
| 100 |
+
"eval_runtime": 1.7839,
|
| 101 |
+
"eval_samples_per_second": 16.256,
|
| 102 |
+
"eval_steps_per_second": 2.242,
|
| 103 |
+
"eval_wer": 1.0,
|
| 104 |
+
"step": 67
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 10.0,
|
| 108 |
+
"eval_cer": 1.0,
|
| 109 |
+
"eval_loss": 2.743428945541382,
|
| 110 |
+
"eval_runtime": 1.77,
|
| 111 |
+
"eval_samples_per_second": 16.384,
|
| 112 |
+
"eval_steps_per_second": 2.26,
|
| 113 |
+
"eval_wer": 1.0,
|
| 114 |
+
"step": 75
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 10.93,
|
| 118 |
+
"eval_cer": 1.0,
|
| 119 |
+
"eval_loss": 2.636403799057007,
|
| 120 |
+
"eval_runtime": 1.7839,
|
| 121 |
+
"eval_samples_per_second": 16.256,
|
| 122 |
+
"eval_steps_per_second": 2.242,
|
| 123 |
+
"eval_wer": 1.0,
|
| 124 |
+
"step": 82
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 12.0,
|
| 128 |
+
"eval_cer": 0.9860826852230864,
|
| 129 |
+
"eval_loss": 2.3797059059143066,
|
| 130 |
+
"eval_runtime": 1.7797,
|
| 131 |
+
"eval_samples_per_second": 16.295,
|
| 132 |
+
"eval_steps_per_second": 2.248,
|
| 133 |
+
"eval_wer": 0.9876237623762376,
|
| 134 |
+
"step": 90
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 12.93,
|
| 138 |
+
"eval_cer": 0.9770773638968482,
|
| 139 |
+
"eval_loss": 1.9516112804412842,
|
| 140 |
+
"eval_runtime": 1.7769,
|
| 141 |
+
"eval_samples_per_second": 16.32,
|
| 142 |
+
"eval_steps_per_second": 2.251,
|
| 143 |
+
"eval_wer": 0.995049504950495,
|
| 144 |
+
"step": 97
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 13.33,
|
| 148 |
+
"learning_rate": 2.58e-05,
|
| 149 |
+
"loss": 3.3197,
|
| 150 |
+
"step": 100
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 14.0,
|
| 154 |
+
"eval_cer": 0.7474416700777732,
|
| 155 |
+
"eval_loss": 1.5396052598953247,
|
| 156 |
+
"eval_runtime": 1.8171,
|
| 157 |
+
"eval_samples_per_second": 15.96,
|
| 158 |
+
"eval_steps_per_second": 2.201,
|
| 159 |
+
"eval_wer": 1.0,
|
| 160 |
+
"step": 105
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 14.93,
|
| 164 |
+
"eval_cer": 0.4273434302087597,
|
| 165 |
+
"eval_loss": 1.103837251663208,
|
| 166 |
+
"eval_runtime": 1.7792,
|
| 167 |
+
"eval_samples_per_second": 16.3,
|
| 168 |
+
"eval_steps_per_second": 2.248,
|
| 169 |
+
"eval_wer": 0.995049504950495,
|
| 170 |
+
"step": 112
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 16.0,
|
| 174 |
+
"eval_cer": 0.16905444126074498,
|
| 175 |
+
"eval_loss": 0.6535802483558655,
|
| 176 |
+
"eval_runtime": 1.7778,
|
| 177 |
+
"eval_samples_per_second": 16.313,
|
| 178 |
+
"eval_steps_per_second": 2.25,
|
| 179 |
+
"eval_wer": 0.6732673267326733,
|
| 180 |
+
"step": 120
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 16.93,
|
| 184 |
+
"eval_cer": 0.07286123618501841,
|
| 185 |
+
"eval_loss": 0.40873047709465027,
|
| 186 |
+
"eval_runtime": 1.7934,
|
| 187 |
+
"eval_samples_per_second": 16.171,
|
| 188 |
+
"eval_steps_per_second": 2.23,
|
| 189 |
+
"eval_wer": 0.3217821782178218,
|
| 190 |
+
"step": 127
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 18.0,
|
| 194 |
+
"eval_cer": 0.056078591895210805,
|
| 195 |
+
"eval_loss": 0.31188902258872986,
|
| 196 |
+
"eval_runtime": 1.7728,
|
| 197 |
+
"eval_samples_per_second": 16.358,
|
| 198 |
+
"eval_steps_per_second": 2.256,
|
| 199 |
+
"eval_wer": 0.22524752475247525,
|
| 200 |
+
"step": 135
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 18.93,
|
| 204 |
+
"eval_cer": 0.04789193614408514,
|
| 205 |
+
"eval_loss": 0.2720329761505127,
|
| 206 |
+
"eval_runtime": 1.785,
|
| 207 |
+
"eval_samples_per_second": 16.247,
|
| 208 |
+
"eval_steps_per_second": 2.241,
|
| 209 |
+
"eval_wer": 0.17574257425742573,
|
| 210 |
+
"step": 142
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 20.0,
|
| 214 |
+
"eval_cer": 0.04134261154318461,
|
| 215 |
+
"eval_loss": 0.24048186838626862,
|
| 216 |
+
"eval_runtime": 1.7775,
|
| 217 |
+
"eval_samples_per_second": 16.315,
|
| 218 |
+
"eval_steps_per_second": 2.25,
|
| 219 |
+
"eval_wer": 0.15841584158415842,
|
| 220 |
+
"step": 150
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 20.93,
|
| 224 |
+
"eval_cer": 0.04093327875562833,
|
| 225 |
+
"eval_loss": 0.23653259873390198,
|
| 226 |
+
"eval_runtime": 1.8549,
|
| 227 |
+
"eval_samples_per_second": 15.634,
|
| 228 |
+
"eval_steps_per_second": 2.156,
|
| 229 |
+
"eval_wer": 0.15841584158415842,
|
| 230 |
+
"step": 157
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 22.0,
|
| 234 |
+
"eval_cer": 0.03970528039295947,
|
| 235 |
+
"eval_loss": 0.22812487185001373,
|
| 236 |
+
"eval_runtime": 1.7767,
|
| 237 |
+
"eval_samples_per_second": 16.323,
|
| 238 |
+
"eval_steps_per_second": 2.251,
|
| 239 |
+
"eval_wer": 0.15099009900990099,
|
| 240 |
+
"step": 165
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 22.93,
|
| 244 |
+
"eval_cer": 0.036021285304952924,
|
| 245 |
+
"eval_loss": 0.19893357157707214,
|
| 246 |
+
"eval_runtime": 1.7827,
|
| 247 |
+
"eval_samples_per_second": 16.267,
|
| 248 |
+
"eval_steps_per_second": 2.244,
|
| 249 |
+
"eval_wer": 0.13613861386138615,
|
| 250 |
+
"step": 172
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 24.0,
|
| 254 |
+
"eval_cer": 0.036021285304952924,
|
| 255 |
+
"eval_loss": 0.2050793319940567,
|
| 256 |
+
"eval_runtime": 1.7818,
|
| 257 |
+
"eval_samples_per_second": 16.275,
|
| 258 |
+
"eval_steps_per_second": 2.245,
|
| 259 |
+
"eval_wer": 0.12871287128712872,
|
| 260 |
+
"step": 180
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 24.93,
|
| 264 |
+
"eval_cer": 0.03561195251739664,
|
| 265 |
+
"eval_loss": 0.22654221951961517,
|
| 266 |
+
"eval_runtime": 1.7854,
|
| 267 |
+
"eval_samples_per_second": 16.243,
|
| 268 |
+
"eval_steps_per_second": 2.24,
|
| 269 |
+
"eval_wer": 0.12871287128712872,
|
| 270 |
+
"step": 187
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 26.0,
|
| 274 |
+
"eval_cer": 0.03765861645517806,
|
| 275 |
+
"eval_loss": 0.22025753557682037,
|
| 276 |
+
"eval_runtime": 1.7733,
|
| 277 |
+
"eval_samples_per_second": 16.354,
|
| 278 |
+
"eval_steps_per_second": 2.256,
|
| 279 |
+
"eval_wer": 0.12871287128712872,
|
| 280 |
+
"step": 195
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 26.67,
|
| 284 |
+
"learning_rate": 2.1514285714285717e-05,
|
| 285 |
+
"loss": 0.5589,
|
| 286 |
+
"step": 200
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 26.93,
|
| 290 |
+
"eval_cer": 0.03397462136717151,
|
| 291 |
+
"eval_loss": 0.2180820107460022,
|
| 292 |
+
"eval_runtime": 1.787,
|
| 293 |
+
"eval_samples_per_second": 16.228,
|
| 294 |
+
"eval_steps_per_second": 2.238,
|
| 295 |
+
"eval_wer": 0.12128712871287128,
|
| 296 |
+
"step": 202
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 28.0,
|
| 300 |
+
"eval_cer": 0.03356528857961523,
|
| 301 |
+
"eval_loss": 0.20063219964504242,
|
| 302 |
+
"eval_runtime": 1.7961,
|
| 303 |
+
"eval_samples_per_second": 16.146,
|
| 304 |
+
"eval_steps_per_second": 2.227,
|
| 305 |
+
"eval_wer": 0.12376237623762376,
|
| 306 |
+
"step": 210
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 28.93,
|
| 310 |
+
"eval_cer": 0.03315595579205895,
|
| 311 |
+
"eval_loss": 0.18598036468029022,
|
| 312 |
+
"eval_runtime": 1.7925,
|
| 313 |
+
"eval_samples_per_second": 16.178,
|
| 314 |
+
"eval_steps_per_second": 2.231,
|
| 315 |
+
"eval_wer": 0.12128712871287128,
|
| 316 |
+
"step": 217
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 30.0,
|
| 320 |
+
"eval_cer": 0.03029062627916496,
|
| 321 |
+
"eval_loss": 0.17724749445915222,
|
| 322 |
+
"eval_runtime": 1.7802,
|
| 323 |
+
"eval_samples_per_second": 16.29,
|
| 324 |
+
"eval_steps_per_second": 2.247,
|
| 325 |
+
"eval_wer": 0.11138613861386139,
|
| 326 |
+
"step": 225
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 30.93,
|
| 330 |
+
"eval_cer": 0.032337290216946375,
|
| 331 |
+
"eval_loss": 0.19143883883953094,
|
| 332 |
+
"eval_runtime": 1.7768,
|
| 333 |
+
"eval_samples_per_second": 16.322,
|
| 334 |
+
"eval_steps_per_second": 2.251,
|
| 335 |
+
"eval_wer": 0.12376237623762376,
|
| 336 |
+
"step": 232
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 32.0,
|
| 340 |
+
"eval_cer": 0.032337290216946375,
|
| 341 |
+
"eval_loss": 0.19965380430221558,
|
| 342 |
+
"eval_runtime": 1.8097,
|
| 343 |
+
"eval_samples_per_second": 16.025,
|
| 344 |
+
"eval_steps_per_second": 2.21,
|
| 345 |
+
"eval_wer": 0.12376237623762376,
|
| 346 |
+
"step": 240
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 32.93,
|
| 350 |
+
"eval_cer": 0.03397462136717151,
|
| 351 |
+
"eval_loss": 0.19473285973072052,
|
| 352 |
+
"eval_runtime": 1.8384,
|
| 353 |
+
"eval_samples_per_second": 15.775,
|
| 354 |
+
"eval_steps_per_second": 2.176,
|
| 355 |
+
"eval_wer": 0.12623762376237624,
|
| 356 |
+
"step": 247
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 34.0,
|
| 360 |
+
"eval_cer": 0.03274662300450266,
|
| 361 |
+
"eval_loss": 0.20558013021945953,
|
| 362 |
+
"eval_runtime": 1.7938,
|
| 363 |
+
"eval_samples_per_second": 16.167,
|
| 364 |
+
"eval_steps_per_second": 2.23,
|
| 365 |
+
"eval_wer": 0.12128712871287128,
|
| 366 |
+
"step": 255
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 34.93,
|
| 370 |
+
"eval_cer": 0.03315595579205895,
|
| 371 |
+
"eval_loss": 0.1984556019306183,
|
| 372 |
+
"eval_runtime": 1.7744,
|
| 373 |
+
"eval_samples_per_second": 16.343,
|
| 374 |
+
"eval_steps_per_second": 2.254,
|
| 375 |
+
"eval_wer": 0.12128712871287128,
|
| 376 |
+
"step": 262
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 36.0,
|
| 380 |
+
"eval_cer": 0.03274662300450266,
|
| 381 |
+
"eval_loss": 0.20163008570671082,
|
| 382 |
+
"eval_runtime": 1.8047,
|
| 383 |
+
"eval_samples_per_second": 16.069,
|
| 384 |
+
"eval_steps_per_second": 2.216,
|
| 385 |
+
"eval_wer": 0.12128712871287128,
|
| 386 |
+
"step": 270
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 36.93,
|
| 390 |
+
"eval_cer": 0.031109291854277528,
|
| 391 |
+
"eval_loss": 0.19409456849098206,
|
| 392 |
+
"eval_runtime": 1.7987,
|
| 393 |
+
"eval_samples_per_second": 16.123,
|
| 394 |
+
"eval_steps_per_second": 2.224,
|
| 395 |
+
"eval_wer": 0.11386138613861387,
|
| 396 |
+
"step": 277
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 38.0,
|
| 400 |
+
"eval_cer": 0.03192795742939009,
|
| 401 |
+
"eval_loss": 0.18242661654949188,
|
| 402 |
+
"eval_runtime": 1.806,
|
| 403 |
+
"eval_samples_per_second": 16.058,
|
| 404 |
+
"eval_steps_per_second": 2.215,
|
| 405 |
+
"eval_wer": 0.12376237623762376,
|
| 406 |
+
"step": 285
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 38.93,
|
| 410 |
+
"eval_cer": 0.029471960704052395,
|
| 411 |
+
"eval_loss": 0.18223299086093903,
|
| 412 |
+
"eval_runtime": 1.8045,
|
| 413 |
+
"eval_samples_per_second": 16.071,
|
| 414 |
+
"eval_steps_per_second": 2.217,
|
| 415 |
+
"eval_wer": 0.10891089108910891,
|
| 416 |
+
"step": 292
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 40.0,
|
| 420 |
+
"learning_rate": 1.7271428571428574e-05,
|
| 421 |
+
"loss": 0.1503,
|
| 422 |
+
"step": 300
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 40.0,
|
| 426 |
+
"eval_cer": 0.031109291854277528,
|
| 427 |
+
"eval_loss": 0.19688135385513306,
|
| 428 |
+
"eval_runtime": 1.8319,
|
| 429 |
+
"eval_samples_per_second": 15.83,
|
| 430 |
+
"eval_steps_per_second": 2.184,
|
| 431 |
+
"eval_wer": 0.11633663366336634,
|
| 432 |
+
"step": 300
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 40.93,
|
| 436 |
+
"eval_cer": 0.029471960704052395,
|
| 437 |
+
"eval_loss": 0.19957467913627625,
|
| 438 |
+
"eval_runtime": 1.8232,
|
| 439 |
+
"eval_samples_per_second": 15.906,
|
| 440 |
+
"eval_steps_per_second": 2.194,
|
| 441 |
+
"eval_wer": 0.11633663366336634,
|
| 442 |
+
"step": 307
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 42.0,
|
| 446 |
+
"eval_cer": 0.029471960704052395,
|
| 447 |
+
"eval_loss": 0.18802578747272491,
|
| 448 |
+
"eval_runtime": 1.7999,
|
| 449 |
+
"eval_samples_per_second": 16.112,
|
| 450 |
+
"eval_steps_per_second": 2.222,
|
| 451 |
+
"eval_wer": 0.10891089108910891,
|
| 452 |
+
"step": 315
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 42.93,
|
| 456 |
+
"eval_cer": 0.034383954154727794,
|
| 457 |
+
"eval_loss": 0.20173507928848267,
|
| 458 |
+
"eval_runtime": 1.7941,
|
| 459 |
+
"eval_samples_per_second": 16.164,
|
| 460 |
+
"eval_steps_per_second": 2.23,
|
| 461 |
+
"eval_wer": 0.1311881188118812,
|
| 462 |
+
"step": 322
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 44.0,
|
| 466 |
+
"eval_cer": 0.03274662300450266,
|
| 467 |
+
"eval_loss": 0.19139862060546875,
|
| 468 |
+
"eval_runtime": 1.8064,
|
| 469 |
+
"eval_samples_per_second": 16.054,
|
| 470 |
+
"eval_steps_per_second": 2.214,
|
| 471 |
+
"eval_wer": 0.11633663366336634,
|
| 472 |
+
"step": 330
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 44.93,
|
| 476 |
+
"eval_cer": 0.03315595579205895,
|
| 477 |
+
"eval_loss": 0.19346125423908234,
|
| 478 |
+
"eval_runtime": 1.8407,
|
| 479 |
+
"eval_samples_per_second": 15.755,
|
| 480 |
+
"eval_steps_per_second": 2.173,
|
| 481 |
+
"eval_wer": 0.11633663366336634,
|
| 482 |
+
"step": 337
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 46.0,
|
| 486 |
+
"eval_cer": 0.03192795742939009,
|
| 487 |
+
"eval_loss": 0.19668900966644287,
|
| 488 |
+
"eval_runtime": 1.8343,
|
| 489 |
+
"eval_samples_per_second": 15.81,
|
| 490 |
+
"eval_steps_per_second": 2.181,
|
| 491 |
+
"eval_wer": 0.11386138613861387,
|
| 492 |
+
"step": 345
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 46.93,
|
| 496 |
+
"eval_cer": 0.029881293491608677,
|
| 497 |
+
"eval_loss": 0.19127391278743744,
|
| 498 |
+
"eval_runtime": 1.7943,
|
| 499 |
+
"eval_samples_per_second": 16.162,
|
| 500 |
+
"eval_steps_per_second": 2.229,
|
| 501 |
+
"eval_wer": 0.10643564356435643,
|
| 502 |
+
"step": 352
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 48.0,
|
| 506 |
+
"eval_cer": 0.03029062627916496,
|
| 507 |
+
"eval_loss": 0.1993999034166336,
|
| 508 |
+
"eval_runtime": 1.8021,
|
| 509 |
+
"eval_samples_per_second": 16.092,
|
| 510 |
+
"eval_steps_per_second": 2.22,
|
| 511 |
+
"eval_wer": 0.11138613861386139,
|
| 512 |
+
"step": 360
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 48.93,
|
| 516 |
+
"eval_cer": 0.029062627916496112,
|
| 517 |
+
"eval_loss": 0.18826055526733398,
|
| 518 |
+
"eval_runtime": 1.81,
|
| 519 |
+
"eval_samples_per_second": 16.022,
|
| 520 |
+
"eval_steps_per_second": 2.21,
|
| 521 |
+
"eval_wer": 0.10891089108910891,
|
| 522 |
+
"step": 367
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 50.0,
|
| 526 |
+
"eval_cer": 0.03029062627916496,
|
| 527 |
+
"eval_loss": 0.18807141482830048,
|
| 528 |
+
"eval_runtime": 1.8146,
|
| 529 |
+
"eval_samples_per_second": 15.982,
|
| 530 |
+
"eval_steps_per_second": 2.204,
|
| 531 |
+
"eval_wer": 0.11386138613861387,
|
| 532 |
+
"step": 375
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 50.0,
|
| 536 |
+
"step": 375,
|
| 537 |
+
"total_flos": 9.822732861444276e+18,
|
| 538 |
+
"train_loss": 1.1235487569173177,
|
| 539 |
+
"train_runtime": 3285.5807,
|
| 540 |
+
"train_samples_per_second": 6.879,
|
| 541 |
+
"train_steps_per_second": 0.213
|
| 542 |
+
}
|
| 543 |
+
],
|
| 544 |
+
"max_steps": 700,
|
| 545 |
+
"num_train_epochs": 100,
|
| 546 |
+
"total_flos": 9.822732861444276e+18,
|
| 547 |
+
"trial_name": null,
|
| 548 |
+
"trial_params": null
|
| 549 |
+
}
|