Commit
•
76283cb
1
Parent(s):
fee0113
cheers again
Browse files- README.md +1 -1
- trainer_state.json +720 -0
README.md
CHANGED
|
@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# whisper4
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [openai/whisper-tiny.en](https://huggingface.co/openai/whisper-tiny.en) on
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.5409
|
| 21 |
- Wer: 28.2719
|
|
|
|
| 15 |
|
| 16 |
# whisper4
|
| 17 |
|
| 18 |
+
This model is a fine-tuned version of [openai/whisper-tiny.en](https://huggingface.co/openai/whisper-tiny.en) on the tiny dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.5409
|
| 21 |
- Wer: 28.2719
|
trainer_state.json
ADDED
|
@@ -0,0 +1,720 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 23.09402795425667,
|
| 3 |
+
"best_model_checkpoint": "whisper4/checkpoint-130",
|
| 4 |
+
"epoch": 8.333333333333334,
|
| 5 |
+
"eval_steps": 10,
|
| 6 |
+
"global_step": 300,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.1388888888888889,
|
| 13 |
+
"grad_norm": 45.38771057128906,
|
| 14 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 15 |
+
"loss": 3.9755,
|
| 16 |
+
"step": 5
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.2777777777777778,
|
| 20 |
+
"grad_norm": 42.970829010009766,
|
| 21 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 22 |
+
"loss": 3.8231,
|
| 23 |
+
"step": 10
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.2777777777777778,
|
| 27 |
+
"eval_loss": 3.7088065147399902,
|
| 28 |
+
"eval_runtime": 261.4715,
|
| 29 |
+
"eval_samples_per_second": 1.912,
|
| 30 |
+
"eval_steps_per_second": 0.241,
|
| 31 |
+
"eval_wer": 76.93773824650572,
|
| 32 |
+
"step": 10
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.4166666666666667,
|
| 36 |
+
"grad_norm": 42.34159469604492,
|
| 37 |
+
"learning_rate": 3e-06,
|
| 38 |
+
"loss": 3.5982,
|
| 39 |
+
"step": 15
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.5555555555555556,
|
| 43 |
+
"grad_norm": 41.4112548828125,
|
| 44 |
+
"learning_rate": 4.000000000000001e-06,
|
| 45 |
+
"loss": 3.1925,
|
| 46 |
+
"step": 20
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.5555555555555556,
|
| 50 |
+
"eval_loss": 2.9438674449920654,
|
| 51 |
+
"eval_runtime": 254.9124,
|
| 52 |
+
"eval_samples_per_second": 1.961,
|
| 53 |
+
"eval_steps_per_second": 0.247,
|
| 54 |
+
"eval_wer": 65.56543837357052,
|
| 55 |
+
"step": 20
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.6944444444444444,
|
| 59 |
+
"grad_norm": 39.323768615722656,
|
| 60 |
+
"learning_rate": 5e-06,
|
| 61 |
+
"loss": 2.7453,
|
| 62 |
+
"step": 25
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.8333333333333334,
|
| 66 |
+
"grad_norm": 30.337932586669922,
|
| 67 |
+
"learning_rate": 6e-06,
|
| 68 |
+
"loss": 2.1383,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.8333333333333334,
|
| 73 |
+
"eval_loss": 1.722086787223816,
|
| 74 |
+
"eval_runtime": 256.1025,
|
| 75 |
+
"eval_samples_per_second": 1.952,
|
| 76 |
+
"eval_steps_per_second": 0.246,
|
| 77 |
+
"eval_wer": 61.53113087674714,
|
| 78 |
+
"step": 30
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 0.9722222222222222,
|
| 82 |
+
"grad_norm": 18.886234283447266,
|
| 83 |
+
"learning_rate": 7.000000000000001e-06,
|
| 84 |
+
"loss": 1.4936,
|
| 85 |
+
"step": 35
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 1.1111111111111112,
|
| 89 |
+
"grad_norm": 10.311893463134766,
|
| 90 |
+
"learning_rate": 8.000000000000001e-06,
|
| 91 |
+
"loss": 1.0671,
|
| 92 |
+
"step": 40
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.1111111111111112,
|
| 96 |
+
"eval_loss": 0.8320145606994629,
|
| 97 |
+
"eval_runtime": 254.169,
|
| 98 |
+
"eval_samples_per_second": 1.967,
|
| 99 |
+
"eval_steps_per_second": 0.248,
|
| 100 |
+
"eval_wer": 50.69885641677255,
|
| 101 |
+
"step": 40
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.25,
|
| 105 |
+
"grad_norm": 6.484711647033691,
|
| 106 |
+
"learning_rate": 9e-06,
|
| 107 |
+
"loss": 0.7616,
|
| 108 |
+
"step": 45
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 1.3888888888888888,
|
| 112 |
+
"grad_norm": 5.488741874694824,
|
| 113 |
+
"learning_rate": 1e-05,
|
| 114 |
+
"loss": 0.6947,
|
| 115 |
+
"step": 50
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 1.3888888888888888,
|
| 119 |
+
"eval_loss": 0.6587409377098083,
|
| 120 |
+
"eval_runtime": 256.6295,
|
| 121 |
+
"eval_samples_per_second": 1.948,
|
| 122 |
+
"eval_steps_per_second": 0.245,
|
| 123 |
+
"eval_wer": 41.01016518424396,
|
| 124 |
+
"step": 50
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 1.5277777777777777,
|
| 128 |
+
"grad_norm": 5.178290843963623,
|
| 129 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 130 |
+
"loss": 0.6099,
|
| 131 |
+
"step": 55
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.6666666666666665,
|
| 135 |
+
"grad_norm": 5.149004936218262,
|
| 136 |
+
"learning_rate": 1.2e-05,
|
| 137 |
+
"loss": 0.6263,
|
| 138 |
+
"step": 60
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 1.6666666666666665,
|
| 142 |
+
"eval_loss": 0.5873834490776062,
|
| 143 |
+
"eval_runtime": 254.3491,
|
| 144 |
+
"eval_samples_per_second": 1.966,
|
| 145 |
+
"eval_steps_per_second": 0.248,
|
| 146 |
+
"eval_wer": 29.796696315120712,
|
| 147 |
+
"step": 60
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 1.8055555555555556,
|
| 151 |
+
"grad_norm": 5.192800521850586,
|
| 152 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 153 |
+
"loss": 0.5654,
|
| 154 |
+
"step": 65
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 1.9444444444444444,
|
| 158 |
+
"grad_norm": 4.883419036865234,
|
| 159 |
+
"learning_rate": 1.4000000000000001e-05,
|
| 160 |
+
"loss": 0.5827,
|
| 161 |
+
"step": 70
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.9444444444444444,
|
| 165 |
+
"eval_loss": 0.5402054190635681,
|
| 166 |
+
"eval_runtime": 256.4853,
|
| 167 |
+
"eval_samples_per_second": 1.949,
|
| 168 |
+
"eval_steps_per_second": 0.246,
|
| 169 |
+
"eval_wer": 27.38246505717916,
|
| 170 |
+
"step": 70
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.0833333333333335,
|
| 174 |
+
"grad_norm": 4.878260612487793,
|
| 175 |
+
"learning_rate": 1.5e-05,
|
| 176 |
+
"loss": 0.5057,
|
| 177 |
+
"step": 75
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.2222222222222223,
|
| 181 |
+
"grad_norm": 4.442017555236816,
|
| 182 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 183 |
+
"loss": 0.4222,
|
| 184 |
+
"step": 80
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 2.2222222222222223,
|
| 188 |
+
"eval_loss": 0.5154020190238953,
|
| 189 |
+
"eval_runtime": 256.4153,
|
| 190 |
+
"eval_samples_per_second": 1.95,
|
| 191 |
+
"eval_steps_per_second": 0.246,
|
| 192 |
+
"eval_wer": 32.05209656925032,
|
| 193 |
+
"step": 80
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 2.361111111111111,
|
| 197 |
+
"grad_norm": 4.793032646179199,
|
| 198 |
+
"learning_rate": 1.7000000000000003e-05,
|
| 199 |
+
"loss": 0.3807,
|
| 200 |
+
"step": 85
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.5,
|
| 204 |
+
"grad_norm": 4.618255138397217,
|
| 205 |
+
"learning_rate": 1.8e-05,
|
| 206 |
+
"loss": 0.4065,
|
| 207 |
+
"step": 90
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 2.5,
|
| 211 |
+
"eval_loss": 0.49971044063568115,
|
| 212 |
+
"eval_runtime": 254.4642,
|
| 213 |
+
"eval_samples_per_second": 1.965,
|
| 214 |
+
"eval_steps_per_second": 0.248,
|
| 215 |
+
"eval_wer": 25.698856416772554,
|
| 216 |
+
"step": 90
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.638888888888889,
|
| 220 |
+
"grad_norm": 4.196300983428955,
|
| 221 |
+
"learning_rate": 1.9e-05,
|
| 222 |
+
"loss": 0.3807,
|
| 223 |
+
"step": 95
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 2.7777777777777777,
|
| 227 |
+
"grad_norm": 3.9986796379089355,
|
| 228 |
+
"learning_rate": 2e-05,
|
| 229 |
+
"loss": 0.3959,
|
| 230 |
+
"step": 100
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.7777777777777777,
|
| 234 |
+
"eval_loss": 0.48037058115005493,
|
| 235 |
+
"eval_runtime": 254.5231,
|
| 236 |
+
"eval_samples_per_second": 1.964,
|
| 237 |
+
"eval_steps_per_second": 0.248,
|
| 238 |
+
"eval_wer": 23.824650571791615,
|
| 239 |
+
"step": 100
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 2.9166666666666665,
|
| 243 |
+
"grad_norm": 4.1782402992248535,
|
| 244 |
+
"learning_rate": 2.1e-05,
|
| 245 |
+
"loss": 0.3847,
|
| 246 |
+
"step": 105
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 3.0555555555555554,
|
| 250 |
+
"grad_norm": 3.125694751739502,
|
| 251 |
+
"learning_rate": 2.2000000000000003e-05,
|
| 252 |
+
"loss": 0.3081,
|
| 253 |
+
"step": 110
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 3.0555555555555554,
|
| 257 |
+
"eval_loss": 0.46701571345329285,
|
| 258 |
+
"eval_runtime": 254.757,
|
| 259 |
+
"eval_samples_per_second": 1.963,
|
| 260 |
+
"eval_steps_per_second": 0.247,
|
| 261 |
+
"eval_wer": 24.841168996188056,
|
| 262 |
+
"step": 110
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.1944444444444446,
|
| 266 |
+
"grad_norm": 3.2956347465515137,
|
| 267 |
+
"learning_rate": 2.3000000000000003e-05,
|
| 268 |
+
"loss": 0.2701,
|
| 269 |
+
"step": 115
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.3333333333333335,
|
| 273 |
+
"grad_norm": 3.4233288764953613,
|
| 274 |
+
"learning_rate": 2.4e-05,
|
| 275 |
+
"loss": 0.2497,
|
| 276 |
+
"step": 120
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.3333333333333335,
|
| 280 |
+
"eval_loss": 0.4687294661998749,
|
| 281 |
+
"eval_runtime": 255.0619,
|
| 282 |
+
"eval_samples_per_second": 1.96,
|
| 283 |
+
"eval_steps_per_second": 0.247,
|
| 284 |
+
"eval_wer": 23.284625158831005,
|
| 285 |
+
"step": 120
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 3.4722222222222223,
|
| 289 |
+
"grad_norm": 3.383148193359375,
|
| 290 |
+
"learning_rate": 2.5e-05,
|
| 291 |
+
"loss": 0.268,
|
| 292 |
+
"step": 125
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 3.611111111111111,
|
| 296 |
+
"grad_norm": 3.4522266387939453,
|
| 297 |
+
"learning_rate": 2.6000000000000002e-05,
|
| 298 |
+
"loss": 0.2535,
|
| 299 |
+
"step": 130
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 3.611111111111111,
|
| 303 |
+
"eval_loss": 0.4594000279903412,
|
| 304 |
+
"eval_runtime": 254.9771,
|
| 305 |
+
"eval_samples_per_second": 1.961,
|
| 306 |
+
"eval_steps_per_second": 0.247,
|
| 307 |
+
"eval_wer": 23.09402795425667,
|
| 308 |
+
"step": 130
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 3.75,
|
| 312 |
+
"grad_norm": 3.3171420097351074,
|
| 313 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 314 |
+
"loss": 0.2174,
|
| 315 |
+
"step": 135
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 3.888888888888889,
|
| 319 |
+
"grad_norm": 3.3734307289123535,
|
| 320 |
+
"learning_rate": 2.8000000000000003e-05,
|
| 321 |
+
"loss": 0.2428,
|
| 322 |
+
"step": 140
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 3.888888888888889,
|
| 326 |
+
"eval_loss": 0.45448967814445496,
|
| 327 |
+
"eval_runtime": 252.6636,
|
| 328 |
+
"eval_samples_per_second": 1.979,
|
| 329 |
+
"eval_steps_per_second": 0.249,
|
| 330 |
+
"eval_wer": 23.506988564167724,
|
| 331 |
+
"step": 140
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 4.027777777777778,
|
| 335 |
+
"grad_norm": 3.152697801589966,
|
| 336 |
+
"learning_rate": 2.9e-05,
|
| 337 |
+
"loss": 0.2136,
|
| 338 |
+
"step": 145
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 4.166666666666667,
|
| 342 |
+
"grad_norm": 3.5505619049072266,
|
| 343 |
+
"learning_rate": 3e-05,
|
| 344 |
+
"loss": 0.1627,
|
| 345 |
+
"step": 150
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 4.166666666666667,
|
| 349 |
+
"eval_loss": 0.46508893370628357,
|
| 350 |
+
"eval_runtime": 253.2857,
|
| 351 |
+
"eval_samples_per_second": 1.974,
|
| 352 |
+
"eval_steps_per_second": 0.249,
|
| 353 |
+
"eval_wer": 24.49174078780178,
|
| 354 |
+
"step": 150
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 4.305555555555555,
|
| 358 |
+
"grad_norm": 2.473661184310913,
|
| 359 |
+
"learning_rate": 3.1e-05,
|
| 360 |
+
"loss": 0.1302,
|
| 361 |
+
"step": 155
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 4.444444444444445,
|
| 365 |
+
"grad_norm": 2.767871379852295,
|
| 366 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 367 |
+
"loss": 0.1224,
|
| 368 |
+
"step": 160
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 4.444444444444445,
|
| 372 |
+
"eval_loss": 0.46860620379447937,
|
| 373 |
+
"eval_runtime": 252.3676,
|
| 374 |
+
"eval_samples_per_second": 1.981,
|
| 375 |
+
"eval_steps_per_second": 0.25,
|
| 376 |
+
"eval_wer": 23.69758576874206,
|
| 377 |
+
"step": 160
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 4.583333333333333,
|
| 381 |
+
"grad_norm": 2.4746224880218506,
|
| 382 |
+
"learning_rate": 3.3e-05,
|
| 383 |
+
"loss": 0.1386,
|
| 384 |
+
"step": 165
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 4.722222222222222,
|
| 388 |
+
"grad_norm": 2.802751302719116,
|
| 389 |
+
"learning_rate": 3.4000000000000007e-05,
|
| 390 |
+
"loss": 0.1326,
|
| 391 |
+
"step": 170
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 4.722222222222222,
|
| 395 |
+
"eval_loss": 0.46528080105781555,
|
| 396 |
+
"eval_runtime": 252.901,
|
| 397 |
+
"eval_samples_per_second": 1.977,
|
| 398 |
+
"eval_steps_per_second": 0.249,
|
| 399 |
+
"eval_wer": 23.69758576874206,
|
| 400 |
+
"step": 170
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 4.861111111111111,
|
| 404 |
+
"grad_norm": 3.2861883640289307,
|
| 405 |
+
"learning_rate": 3.5e-05,
|
| 406 |
+
"loss": 0.1416,
|
| 407 |
+
"step": 175
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 5.0,
|
| 411 |
+
"grad_norm": 7.430758953094482,
|
| 412 |
+
"learning_rate": 3.6e-05,
|
| 413 |
+
"loss": 0.1334,
|
| 414 |
+
"step": 180
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 5.0,
|
| 418 |
+
"eval_loss": 0.474071204662323,
|
| 419 |
+
"eval_runtime": 253.2666,
|
| 420 |
+
"eval_samples_per_second": 1.974,
|
| 421 |
+
"eval_steps_per_second": 0.249,
|
| 422 |
+
"eval_wer": 24.74587039390089,
|
| 423 |
+
"step": 180
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 5.138888888888889,
|
| 427 |
+
"grad_norm": 2.6109585762023926,
|
| 428 |
+
"learning_rate": 3.7e-05,
|
| 429 |
+
"loss": 0.0807,
|
| 430 |
+
"step": 185
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 5.277777777777778,
|
| 434 |
+
"grad_norm": 1.921268343925476,
|
| 435 |
+
"learning_rate": 3.8e-05,
|
| 436 |
+
"loss": 0.0659,
|
| 437 |
+
"step": 190
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 5.277777777777778,
|
| 441 |
+
"eval_loss": 0.4791569113731384,
|
| 442 |
+
"eval_runtime": 252.5034,
|
| 443 |
+
"eval_samples_per_second": 1.98,
|
| 444 |
+
"eval_steps_per_second": 0.25,
|
| 445 |
+
"eval_wer": 24.68233799237611,
|
| 446 |
+
"step": 190
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 5.416666666666667,
|
| 450 |
+
"grad_norm": 1.5123528242111206,
|
| 451 |
+
"learning_rate": 3.9000000000000006e-05,
|
| 452 |
+
"loss": 0.0548,
|
| 453 |
+
"step": 195
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 5.555555555555555,
|
| 457 |
+
"grad_norm": 1.7394624948501587,
|
| 458 |
+
"learning_rate": 4e-05,
|
| 459 |
+
"loss": 0.0639,
|
| 460 |
+
"step": 200
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 5.555555555555555,
|
| 464 |
+
"eval_loss": 0.4760441780090332,
|
| 465 |
+
"eval_runtime": 254.7055,
|
| 466 |
+
"eval_samples_per_second": 1.963,
|
| 467 |
+
"eval_steps_per_second": 0.247,
|
| 468 |
+
"eval_wer": 33.38627700127065,
|
| 469 |
+
"step": 200
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 5.694444444444445,
|
| 473 |
+
"grad_norm": 2.006833553314209,
|
| 474 |
+
"learning_rate": 4.1e-05,
|
| 475 |
+
"loss": 0.0718,
|
| 476 |
+
"step": 205
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 5.833333333333333,
|
| 480 |
+
"grad_norm": 1.9101831912994385,
|
| 481 |
+
"learning_rate": 4.2e-05,
|
| 482 |
+
"loss": 0.0667,
|
| 483 |
+
"step": 210
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 5.833333333333333,
|
| 487 |
+
"eval_loss": 0.48197290301322937,
|
| 488 |
+
"eval_runtime": 252.8934,
|
| 489 |
+
"eval_samples_per_second": 1.977,
|
| 490 |
+
"eval_steps_per_second": 0.249,
|
| 491 |
+
"eval_wer": 25.47649301143583,
|
| 492 |
+
"step": 210
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 5.972222222222222,
|
| 496 |
+
"grad_norm": 2.15120267868042,
|
| 497 |
+
"learning_rate": 4.3e-05,
|
| 498 |
+
"loss": 0.0833,
|
| 499 |
+
"step": 215
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 6.111111111111111,
|
| 503 |
+
"grad_norm": 1.230398178100586,
|
| 504 |
+
"learning_rate": 4.4000000000000006e-05,
|
| 505 |
+
"loss": 0.042,
|
| 506 |
+
"step": 220
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 6.111111111111111,
|
| 510 |
+
"eval_loss": 0.4932812750339508,
|
| 511 |
+
"eval_runtime": 254.0189,
|
| 512 |
+
"eval_samples_per_second": 1.968,
|
| 513 |
+
"eval_steps_per_second": 0.248,
|
| 514 |
+
"eval_wer": 29.415501905972047,
|
| 515 |
+
"step": 220
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 6.25,
|
| 519 |
+
"grad_norm": 1.1129037141799927,
|
| 520 |
+
"learning_rate": 4.5e-05,
|
| 521 |
+
"loss": 0.0285,
|
| 522 |
+
"step": 225
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 6.388888888888889,
|
| 526 |
+
"grad_norm": 1.4872666597366333,
|
| 527 |
+
"learning_rate": 4.600000000000001e-05,
|
| 528 |
+
"loss": 0.0325,
|
| 529 |
+
"step": 230
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 6.388888888888889,
|
| 533 |
+
"eval_loss": 0.5065749883651733,
|
| 534 |
+
"eval_runtime": 253.3018,
|
| 535 |
+
"eval_samples_per_second": 1.974,
|
| 536 |
+
"eval_steps_per_second": 0.249,
|
| 537 |
+
"eval_wer": 29.987293519695047,
|
| 538 |
+
"step": 230
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 6.527777777777778,
|
| 542 |
+
"grad_norm": 1.527269959449768,
|
| 543 |
+
"learning_rate": 4.7e-05,
|
| 544 |
+
"loss": 0.0322,
|
| 545 |
+
"step": 235
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 6.666666666666667,
|
| 549 |
+
"grad_norm": 1.6628929376602173,
|
| 550 |
+
"learning_rate": 4.8e-05,
|
| 551 |
+
"loss": 0.0333,
|
| 552 |
+
"step": 240
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 6.666666666666667,
|
| 556 |
+
"eval_loss": 0.5125746130943298,
|
| 557 |
+
"eval_runtime": 252.7719,
|
| 558 |
+
"eval_samples_per_second": 1.978,
|
| 559 |
+
"eval_steps_per_second": 0.249,
|
| 560 |
+
"eval_wer": 26.08005082592122,
|
| 561 |
+
"step": 240
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 6.805555555555555,
|
| 565 |
+
"grad_norm": 2.2532899379730225,
|
| 566 |
+
"learning_rate": 4.9e-05,
|
| 567 |
+
"loss": 0.0314,
|
| 568 |
+
"step": 245
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 6.944444444444445,
|
| 572 |
+
"grad_norm": 3.415356397628784,
|
| 573 |
+
"learning_rate": 5e-05,
|
| 574 |
+
"loss": 0.0333,
|
| 575 |
+
"step": 250
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 6.944444444444445,
|
| 579 |
+
"eval_loss": 0.5072929859161377,
|
| 580 |
+
"eval_runtime": 252.6089,
|
| 581 |
+
"eval_samples_per_second": 1.979,
|
| 582 |
+
"eval_steps_per_second": 0.249,
|
| 583 |
+
"eval_wer": 24.618805590851334,
|
| 584 |
+
"step": 250
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 7.083333333333333,
|
| 588 |
+
"grad_norm": 0.6166062951087952,
|
| 589 |
+
"learning_rate": 5.1000000000000006e-05,
|
| 590 |
+
"loss": 0.0251,
|
| 591 |
+
"step": 255
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 7.222222222222222,
|
| 595 |
+
"grad_norm": 1.0601465702056885,
|
| 596 |
+
"learning_rate": 5.2000000000000004e-05,
|
| 597 |
+
"loss": 0.0187,
|
| 598 |
+
"step": 260
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 7.222222222222222,
|
| 602 |
+
"eval_loss": 0.5128570795059204,
|
| 603 |
+
"eval_runtime": 253.8562,
|
| 604 |
+
"eval_samples_per_second": 1.97,
|
| 605 |
+
"eval_steps_per_second": 0.248,
|
| 606 |
+
"eval_wer": 27.350698856416773,
|
| 607 |
+
"step": 260
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 7.361111111111111,
|
| 611 |
+
"grad_norm": 1.31247878074646,
|
| 612 |
+
"learning_rate": 5.300000000000001e-05,
|
| 613 |
+
"loss": 0.0219,
|
| 614 |
+
"step": 265
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 7.5,
|
| 618 |
+
"grad_norm": 0.7731389999389648,
|
| 619 |
+
"learning_rate": 5.4000000000000005e-05,
|
| 620 |
+
"loss": 0.0214,
|
| 621 |
+
"step": 270
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 7.5,
|
| 625 |
+
"eval_loss": 0.5208793878555298,
|
| 626 |
+
"eval_runtime": 254.7138,
|
| 627 |
+
"eval_samples_per_second": 1.963,
|
| 628 |
+
"eval_steps_per_second": 0.247,
|
| 629 |
+
"eval_wer": 28.208386277001267,
|
| 630 |
+
"step": 270
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 7.638888888888889,
|
| 634 |
+
"grad_norm": 2.0568604469299316,
|
| 635 |
+
"learning_rate": 5.500000000000001e-05,
|
| 636 |
+
"loss": 0.0248,
|
| 637 |
+
"step": 275
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 7.777777777777778,
|
| 641 |
+
"grad_norm": 1.7497611045837402,
|
| 642 |
+
"learning_rate": 5.6000000000000006e-05,
|
| 643 |
+
"loss": 0.0187,
|
| 644 |
+
"step": 280
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 7.777777777777778,
|
| 648 |
+
"eval_loss": 0.5212948322296143,
|
| 649 |
+
"eval_runtime": 254.6913,
|
| 650 |
+
"eval_samples_per_second": 1.963,
|
| 651 |
+
"eval_steps_per_second": 0.247,
|
| 652 |
+
"eval_wer": 29.320203303684877,
|
| 653 |
+
"step": 280
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 7.916666666666667,
|
| 657 |
+
"grad_norm": 2.408604383468628,
|
| 658 |
+
"learning_rate": 5.6999999999999996e-05,
|
| 659 |
+
"loss": 0.0303,
|
| 660 |
+
"step": 285
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 8.055555555555555,
|
| 664 |
+
"grad_norm": 1.3855714797973633,
|
| 665 |
+
"learning_rate": 5.8e-05,
|
| 666 |
+
"loss": 0.0312,
|
| 667 |
+
"step": 290
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 8.055555555555555,
|
| 671 |
+
"eval_loss": 0.5274094939231873,
|
| 672 |
+
"eval_runtime": 253.5749,
|
| 673 |
+
"eval_samples_per_second": 1.972,
|
| 674 |
+
"eval_steps_per_second": 0.248,
|
| 675 |
+
"eval_wer": 34.6569250317662,
|
| 676 |
+
"step": 290
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 8.194444444444445,
|
| 680 |
+
"grad_norm": 1.87900972366333,
|
| 681 |
+
"learning_rate": 5.9e-05,
|
| 682 |
+
"loss": 0.0153,
|
| 683 |
+
"step": 295
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 8.333333333333334,
|
| 687 |
+
"grad_norm": 1.318336009979248,
|
| 688 |
+
"learning_rate": 6e-05,
|
| 689 |
+
"loss": 0.0172,
|
| 690 |
+
"step": 300
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 8.333333333333334,
|
| 694 |
+
"eval_loss": 0.5408744812011719,
|
| 695 |
+
"eval_runtime": 252.2148,
|
| 696 |
+
"eval_samples_per_second": 1.982,
|
| 697 |
+
"eval_steps_per_second": 0.25,
|
| 698 |
+
"eval_wer": 28.27191867852605,
|
| 699 |
+
"step": 300
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 8.333333333333334,
|
| 703 |
+
"step": 300,
|
| 704 |
+
"total_flos": 9.2409447186432e+17,
|
| 705 |
+
"train_loss": 0.5446730978041887,
|
| 706 |
+
"train_runtime": 8175.9362,
|
| 707 |
+
"train_samples_per_second": 4.697,
|
| 708 |
+
"train_steps_per_second": 0.037
|
| 709 |
+
}
|
| 710 |
+
],
|
| 711 |
+
"logging_steps": 5,
|
| 712 |
+
"max_steps": 300,
|
| 713 |
+
"num_input_tokens_seen": 0,
|
| 714 |
+
"num_train_epochs": 9,
|
| 715 |
+
"save_steps": 10,
|
| 716 |
+
"total_flos": 9.2409447186432e+17,
|
| 717 |
+
"train_batch_size": 128,
|
| 718 |
+
"trial_name": null,
|
| 719 |
+
"trial_params": null
|
| 720 |
+
}
|