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 |
# whisper3
|
| 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: 0.5509
|
| 21 |
- Wer: 26.9488
|
|
|
|
| 15 |
|
| 16 |
# whisper3
|
| 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: 0.5509
|
| 21 |
- Wer: 26.9488
|
trainer_state.json
ADDED
|
@@ -0,0 +1,720 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 23.35348393254852,
|
| 3 |
+
"best_model_checkpoint": "whisper3/checkpoint-240",
|
| 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": 46.06148147583008,
|
| 14 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 15 |
+
"loss": 3.9402,
|
| 16 |
+
"step": 5
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.2777777777777778,
|
| 20 |
+
"grad_norm": 43.4765625,
|
| 21 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 22 |
+
"loss": 3.8281,
|
| 23 |
+
"step": 10
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.2777777777777778,
|
| 27 |
+
"eval_loss": 3.7929115295410156,
|
| 28 |
+
"eval_runtime": 253.0403,
|
| 29 |
+
"eval_samples_per_second": 1.976,
|
| 30 |
+
"eval_steps_per_second": 0.249,
|
| 31 |
+
"eval_wer": 80.40089086859689,
|
| 32 |
+
"step": 10
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.4166666666666667,
|
| 36 |
+
"grad_norm": 40.57815933227539,
|
| 37 |
+
"learning_rate": 3e-06,
|
| 38 |
+
"loss": 3.5929,
|
| 39 |
+
"step": 15
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.5555555555555556,
|
| 43 |
+
"grad_norm": 39.72583770751953,
|
| 44 |
+
"learning_rate": 4.000000000000001e-06,
|
| 45 |
+
"loss": 3.209,
|
| 46 |
+
"step": 20
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.5555555555555556,
|
| 50 |
+
"eval_loss": 3.0014312267303467,
|
| 51 |
+
"eval_runtime": 246.2101,
|
| 52 |
+
"eval_samples_per_second": 2.031,
|
| 53 |
+
"eval_steps_per_second": 0.256,
|
| 54 |
+
"eval_wer": 68.37416481069042,
|
| 55 |
+
"step": 20
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.6944444444444444,
|
| 59 |
+
"grad_norm": 39.53627395629883,
|
| 60 |
+
"learning_rate": 5e-06,
|
| 61 |
+
"loss": 2.7486,
|
| 62 |
+
"step": 25
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.8333333333333334,
|
| 66 |
+
"grad_norm": 30.079750061035156,
|
| 67 |
+
"learning_rate": 6e-06,
|
| 68 |
+
"loss": 2.1066,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.8333333333333334,
|
| 73 |
+
"eval_loss": 1.761271595954895,
|
| 74 |
+
"eval_runtime": 245.5315,
|
| 75 |
+
"eval_samples_per_second": 2.036,
|
| 76 |
+
"eval_steps_per_second": 0.257,
|
| 77 |
+
"eval_wer": 63.91982182628062,
|
| 78 |
+
"step": 30
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 0.9722222222222222,
|
| 82 |
+
"grad_norm": 19.831071853637695,
|
| 83 |
+
"learning_rate": 7.000000000000001e-06,
|
| 84 |
+
"loss": 1.5134,
|
| 85 |
+
"step": 35
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 1.1111111111111112,
|
| 89 |
+
"grad_norm": 9.755999565124512,
|
| 90 |
+
"learning_rate": 8.000000000000001e-06,
|
| 91 |
+
"loss": 0.9963,
|
| 92 |
+
"step": 40
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.1111111111111112,
|
| 96 |
+
"eval_loss": 0.8740884065628052,
|
| 97 |
+
"eval_runtime": 246.6146,
|
| 98 |
+
"eval_samples_per_second": 2.027,
|
| 99 |
+
"eval_steps_per_second": 0.255,
|
| 100 |
+
"eval_wer": 52.43398027362392,
|
| 101 |
+
"step": 40
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.25,
|
| 105 |
+
"grad_norm": 6.842897891998291,
|
| 106 |
+
"learning_rate": 9e-06,
|
| 107 |
+
"loss": 0.786,
|
| 108 |
+
"step": 45
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 1.3888888888888888,
|
| 112 |
+
"grad_norm": 5.720729351043701,
|
| 113 |
+
"learning_rate": 1e-05,
|
| 114 |
+
"loss": 0.6922,
|
| 115 |
+
"step": 50
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 1.3888888888888888,
|
| 119 |
+
"eval_loss": 0.7008740901947021,
|
| 120 |
+
"eval_runtime": 245.5713,
|
| 121 |
+
"eval_samples_per_second": 2.036,
|
| 122 |
+
"eval_steps_per_second": 0.257,
|
| 123 |
+
"eval_wer": 35.82564428889596,
|
| 124 |
+
"step": 50
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 1.5277777777777777,
|
| 128 |
+
"grad_norm": 4.806775093078613,
|
| 129 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 130 |
+
"loss": 0.6427,
|
| 131 |
+
"step": 55
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.6666666666666665,
|
| 135 |
+
"grad_norm": 5.128376483917236,
|
| 136 |
+
"learning_rate": 1.2e-05,
|
| 137 |
+
"loss": 0.5816,
|
| 138 |
+
"step": 60
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 1.6666666666666665,
|
| 142 |
+
"eval_loss": 0.6238442659378052,
|
| 143 |
+
"eval_runtime": 245.679,
|
| 144 |
+
"eval_samples_per_second": 2.035,
|
| 145 |
+
"eval_steps_per_second": 0.256,
|
| 146 |
+
"eval_wer": 31.148584155265667,
|
| 147 |
+
"step": 60
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 1.8055555555555556,
|
| 151 |
+
"grad_norm": 4.993675231933594,
|
| 152 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 153 |
+
"loss": 0.5805,
|
| 154 |
+
"step": 65
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 1.9444444444444444,
|
| 158 |
+
"grad_norm": 4.856825351715088,
|
| 159 |
+
"learning_rate": 1.4000000000000001e-05,
|
| 160 |
+
"loss": 0.5684,
|
| 161 |
+
"step": 70
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.9444444444444444,
|
| 165 |
+
"eval_loss": 0.5697694420814514,
|
| 166 |
+
"eval_runtime": 245.5413,
|
| 167 |
+
"eval_samples_per_second": 2.036,
|
| 168 |
+
"eval_steps_per_second": 0.257,
|
| 169 |
+
"eval_wer": 35.47566019726376,
|
| 170 |
+
"step": 70
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.0833333333333335,
|
| 174 |
+
"grad_norm": 4.464582443237305,
|
| 175 |
+
"learning_rate": 1.5e-05,
|
| 176 |
+
"loss": 0.4534,
|
| 177 |
+
"step": 75
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.2222222222222223,
|
| 181 |
+
"grad_norm": 4.251033306121826,
|
| 182 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 183 |
+
"loss": 0.427,
|
| 184 |
+
"step": 80
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 2.2222222222222223,
|
| 188 |
+
"eval_loss": 0.5380394458770752,
|
| 189 |
+
"eval_runtime": 244.4819,
|
| 190 |
+
"eval_samples_per_second": 2.045,
|
| 191 |
+
"eval_steps_per_second": 0.258,
|
| 192 |
+
"eval_wer": 27.266942411708563,
|
| 193 |
+
"step": 80
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 2.361111111111111,
|
| 197 |
+
"grad_norm": 4.489510536193848,
|
| 198 |
+
"learning_rate": 1.7000000000000003e-05,
|
| 199 |
+
"loss": 0.3929,
|
| 200 |
+
"step": 85
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.5,
|
| 204 |
+
"grad_norm": 4.552371025085449,
|
| 205 |
+
"learning_rate": 1.8e-05,
|
| 206 |
+
"loss": 0.4395,
|
| 207 |
+
"step": 90
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 2.5,
|
| 211 |
+
"eval_loss": 0.5162410140037537,
|
| 212 |
+
"eval_runtime": 245.2373,
|
| 213 |
+
"eval_samples_per_second": 2.039,
|
| 214 |
+
"eval_steps_per_second": 0.257,
|
| 215 |
+
"eval_wer": 32.73942093541203,
|
| 216 |
+
"step": 90
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.638888888888889,
|
| 220 |
+
"grad_norm": 4.691618919372559,
|
| 221 |
+
"learning_rate": 1.9e-05,
|
| 222 |
+
"loss": 0.3825,
|
| 223 |
+
"step": 95
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 2.7777777777777777,
|
| 227 |
+
"grad_norm": 4.219367027282715,
|
| 228 |
+
"learning_rate": 2e-05,
|
| 229 |
+
"loss": 0.3861,
|
| 230 |
+
"step": 100
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.7777777777777777,
|
| 234 |
+
"eval_loss": 0.495292991399765,
|
| 235 |
+
"eval_runtime": 243.4193,
|
| 236 |
+
"eval_samples_per_second": 2.054,
|
| 237 |
+
"eval_steps_per_second": 0.259,
|
| 238 |
+
"eval_wer": 24.530703149856826,
|
| 239 |
+
"step": 100
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 2.9166666666666665,
|
| 243 |
+
"grad_norm": 4.323045253753662,
|
| 244 |
+
"learning_rate": 2.1e-05,
|
| 245 |
+
"loss": 0.3669,
|
| 246 |
+
"step": 105
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 3.0555555555555554,
|
| 250 |
+
"grad_norm": 3.2159509658813477,
|
| 251 |
+
"learning_rate": 2.2000000000000003e-05,
|
| 252 |
+
"loss": 0.3745,
|
| 253 |
+
"step": 110
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 3.0555555555555554,
|
| 257 |
+
"eval_loss": 0.4837464392185211,
|
| 258 |
+
"eval_runtime": 244.5759,
|
| 259 |
+
"eval_samples_per_second": 2.044,
|
| 260 |
+
"eval_steps_per_second": 0.258,
|
| 261 |
+
"eval_wer": 24.626153356665608,
|
| 262 |
+
"step": 110
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.1944444444444446,
|
| 266 |
+
"grad_norm": 3.675457000732422,
|
| 267 |
+
"learning_rate": 2.3000000000000003e-05,
|
| 268 |
+
"loss": 0.257,
|
| 269 |
+
"step": 115
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.3333333333333335,
|
| 273 |
+
"grad_norm": 2.8939876556396484,
|
| 274 |
+
"learning_rate": 2.4e-05,
|
| 275 |
+
"loss": 0.2487,
|
| 276 |
+
"step": 120
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.3333333333333335,
|
| 280 |
+
"eval_loss": 0.4732927978038788,
|
| 281 |
+
"eval_runtime": 244.6891,
|
| 282 |
+
"eval_samples_per_second": 2.043,
|
| 283 |
+
"eval_steps_per_second": 0.257,
|
| 284 |
+
"eval_wer": 23.57620108176901,
|
| 285 |
+
"step": 120
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 3.4722222222222223,
|
| 289 |
+
"grad_norm": 3.4589827060699463,
|
| 290 |
+
"learning_rate": 2.5e-05,
|
| 291 |
+
"loss": 0.253,
|
| 292 |
+
"step": 125
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 3.611111111111111,
|
| 296 |
+
"grad_norm": 3.1798577308654785,
|
| 297 |
+
"learning_rate": 2.6000000000000002e-05,
|
| 298 |
+
"loss": 0.2343,
|
| 299 |
+
"step": 130
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 3.611111111111111,
|
| 303 |
+
"eval_loss": 0.46519017219543457,
|
| 304 |
+
"eval_runtime": 244.3925,
|
| 305 |
+
"eval_samples_per_second": 2.046,
|
| 306 |
+
"eval_steps_per_second": 0.258,
|
| 307 |
+
"eval_wer": 24.94432071269488,
|
| 308 |
+
"step": 130
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 3.75,
|
| 312 |
+
"grad_norm": 4.061887741088867,
|
| 313 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 314 |
+
"loss": 0.2354,
|
| 315 |
+
"step": 135
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 3.888888888888889,
|
| 319 |
+
"grad_norm": 4.474591255187988,
|
| 320 |
+
"learning_rate": 2.8000000000000003e-05,
|
| 321 |
+
"loss": 0.2429,
|
| 322 |
+
"step": 140
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 3.888888888888889,
|
| 326 |
+
"eval_loss": 0.4581267833709717,
|
| 327 |
+
"eval_runtime": 244.836,
|
| 328 |
+
"eval_samples_per_second": 2.042,
|
| 329 |
+
"eval_steps_per_second": 0.257,
|
| 330 |
+
"eval_wer": 24.085268851415844,
|
| 331 |
+
"step": 140
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 4.027777777777778,
|
| 335 |
+
"grad_norm": 2.3235318660736084,
|
| 336 |
+
"learning_rate": 2.9e-05,
|
| 337 |
+
"loss": 0.2728,
|
| 338 |
+
"step": 145
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 4.166666666666667,
|
| 342 |
+
"grad_norm": 2.3824808597564697,
|
| 343 |
+
"learning_rate": 3e-05,
|
| 344 |
+
"loss": 0.1286,
|
| 345 |
+
"step": 150
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 4.166666666666667,
|
| 349 |
+
"eval_loss": 0.46725359559059143,
|
| 350 |
+
"eval_runtime": 245.6982,
|
| 351 |
+
"eval_samples_per_second": 2.035,
|
| 352 |
+
"eval_steps_per_second": 0.256,
|
| 353 |
+
"eval_wer": 24.276169265033406,
|
| 354 |
+
"step": 150
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 4.305555555555555,
|
| 358 |
+
"grad_norm": 2.5686404705047607,
|
| 359 |
+
"learning_rate": 3.1e-05,
|
| 360 |
+
"loss": 0.1301,
|
| 361 |
+
"step": 155
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 4.444444444444445,
|
| 365 |
+
"grad_norm": 2.7436068058013916,
|
| 366 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 367 |
+
"loss": 0.1304,
|
| 368 |
+
"step": 160
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 4.444444444444445,
|
| 372 |
+
"eval_loss": 0.46984970569610596,
|
| 373 |
+
"eval_runtime": 245.0991,
|
| 374 |
+
"eval_samples_per_second": 2.04,
|
| 375 |
+
"eval_steps_per_second": 0.257,
|
| 376 |
+
"eval_wer": 31.72128539611836,
|
| 377 |
+
"step": 160
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 4.583333333333333,
|
| 381 |
+
"grad_norm": 2.83823823928833,
|
| 382 |
+
"learning_rate": 3.3e-05,
|
| 383 |
+
"loss": 0.1408,
|
| 384 |
+
"step": 165
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 4.722222222222222,
|
| 388 |
+
"grad_norm": 2.7204811573028564,
|
| 389 |
+
"learning_rate": 3.4000000000000007e-05,
|
| 390 |
+
"loss": 0.1361,
|
| 391 |
+
"step": 170
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 4.722222222222222,
|
| 395 |
+
"eval_loss": 0.4690161943435669,
|
| 396 |
+
"eval_runtime": 246.5232,
|
| 397 |
+
"eval_samples_per_second": 2.028,
|
| 398 |
+
"eval_steps_per_second": 0.256,
|
| 399 |
+
"eval_wer": 33.08940502704423,
|
| 400 |
+
"step": 170
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 4.861111111111111,
|
| 404 |
+
"grad_norm": 3.671097993850708,
|
| 405 |
+
"learning_rate": 3.5e-05,
|
| 406 |
+
"loss": 0.1511,
|
| 407 |
+
"step": 175
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 5.0,
|
| 411 |
+
"grad_norm": 6.484060764312744,
|
| 412 |
+
"learning_rate": 3.6e-05,
|
| 413 |
+
"loss": 0.1447,
|
| 414 |
+
"step": 180
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 5.0,
|
| 418 |
+
"eval_loss": 0.4811546802520752,
|
| 419 |
+
"eval_runtime": 244.9356,
|
| 420 |
+
"eval_samples_per_second": 2.041,
|
| 421 |
+
"eval_steps_per_second": 0.257,
|
| 422 |
+
"eval_wer": 24.657970092268535,
|
| 423 |
+
"step": 180
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 5.138888888888889,
|
| 427 |
+
"grad_norm": 1.9667352437973022,
|
| 428 |
+
"learning_rate": 3.7e-05,
|
| 429 |
+
"loss": 0.063,
|
| 430 |
+
"step": 185
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 5.277777777777778,
|
| 434 |
+
"grad_norm": 2.1828482151031494,
|
| 435 |
+
"learning_rate": 3.8e-05,
|
| 436 |
+
"loss": 0.0617,
|
| 437 |
+
"step": 190
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 5.277777777777778,
|
| 441 |
+
"eval_loss": 0.48713362216949463,
|
| 442 |
+
"eval_runtime": 244.9851,
|
| 443 |
+
"eval_samples_per_second": 2.041,
|
| 444 |
+
"eval_steps_per_second": 0.257,
|
| 445 |
+
"eval_wer": 29.939548202354437,
|
| 446 |
+
"step": 190
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 5.416666666666667,
|
| 450 |
+
"grad_norm": 1.8774911165237427,
|
| 451 |
+
"learning_rate": 3.9000000000000006e-05,
|
| 452 |
+
"loss": 0.0606,
|
| 453 |
+
"step": 195
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 5.555555555555555,
|
| 457 |
+
"grad_norm": 1.8562583923339844,
|
| 458 |
+
"learning_rate": 4e-05,
|
| 459 |
+
"loss": 0.0617,
|
| 460 |
+
"step": 200
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 5.555555555555555,
|
| 464 |
+
"eval_loss": 0.488438218832016,
|
| 465 |
+
"eval_runtime": 244.9014,
|
| 466 |
+
"eval_samples_per_second": 2.042,
|
| 467 |
+
"eval_steps_per_second": 0.257,
|
| 468 |
+
"eval_wer": 24.848870505886094,
|
| 469 |
+
"step": 200
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 5.694444444444445,
|
| 473 |
+
"grad_norm": 1.9106348752975464,
|
| 474 |
+
"learning_rate": 4.1e-05,
|
| 475 |
+
"loss": 0.0617,
|
| 476 |
+
"step": 205
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 5.833333333333333,
|
| 480 |
+
"grad_norm": 1.8114972114562988,
|
| 481 |
+
"learning_rate": 4.2e-05,
|
| 482 |
+
"loss": 0.0577,
|
| 483 |
+
"step": 210
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 5.833333333333333,
|
| 487 |
+
"eval_loss": 0.4998014569282532,
|
| 488 |
+
"eval_runtime": 244.1029,
|
| 489 |
+
"eval_samples_per_second": 2.048,
|
| 490 |
+
"eval_steps_per_second": 0.258,
|
| 491 |
+
"eval_wer": 26.853324848870507,
|
| 492 |
+
"step": 210
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 5.972222222222222,
|
| 496 |
+
"grad_norm": 5.00437593460083,
|
| 497 |
+
"learning_rate": 4.3e-05,
|
| 498 |
+
"loss": 0.078,
|
| 499 |
+
"step": 215
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 6.111111111111111,
|
| 503 |
+
"grad_norm": 1.4013047218322754,
|
| 504 |
+
"learning_rate": 4.4000000000000006e-05,
|
| 505 |
+
"loss": 0.038,
|
| 506 |
+
"step": 220
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 6.111111111111111,
|
| 510 |
+
"eval_loss": 0.500673770904541,
|
| 511 |
+
"eval_runtime": 247.5538,
|
| 512 |
+
"eval_samples_per_second": 2.02,
|
| 513 |
+
"eval_steps_per_second": 0.254,
|
| 514 |
+
"eval_wer": 24.848870505886094,
|
| 515 |
+
"step": 220
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 6.25,
|
| 519 |
+
"grad_norm": 1.4778488874435425,
|
| 520 |
+
"learning_rate": 4.5e-05,
|
| 521 |
+
"loss": 0.0243,
|
| 522 |
+
"step": 225
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 6.388888888888889,
|
| 526 |
+
"grad_norm": 1.3681198358535767,
|
| 527 |
+
"learning_rate": 4.600000000000001e-05,
|
| 528 |
+
"loss": 0.0269,
|
| 529 |
+
"step": 230
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 6.388888888888889,
|
| 533 |
+
"eval_loss": 0.5122880935668945,
|
| 534 |
+
"eval_runtime": 243.6648,
|
| 535 |
+
"eval_samples_per_second": 2.052,
|
| 536 |
+
"eval_steps_per_second": 0.259,
|
| 537 |
+
"eval_wer": 27.139675469296847,
|
| 538 |
+
"step": 230
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 6.527777777777778,
|
| 542 |
+
"grad_norm": 1.450726866722107,
|
| 543 |
+
"learning_rate": 4.7e-05,
|
| 544 |
+
"loss": 0.0297,
|
| 545 |
+
"step": 235
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 6.666666666666667,
|
| 549 |
+
"grad_norm": 1.4052125215530396,
|
| 550 |
+
"learning_rate": 4.8e-05,
|
| 551 |
+
"loss": 0.0321,
|
| 552 |
+
"step": 240
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 6.666666666666667,
|
| 556 |
+
"eval_loss": 0.500522792339325,
|
| 557 |
+
"eval_runtime": 247.602,
|
| 558 |
+
"eval_samples_per_second": 2.019,
|
| 559 |
+
"eval_steps_per_second": 0.254,
|
| 560 |
+
"eval_wer": 23.35348393254852,
|
| 561 |
+
"step": 240
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 6.805555555555555,
|
| 565 |
+
"grad_norm": 1.2223644256591797,
|
| 566 |
+
"learning_rate": 4.9e-05,
|
| 567 |
+
"loss": 0.0291,
|
| 568 |
+
"step": 245
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 6.944444444444445,
|
| 572 |
+
"grad_norm": 1.463398814201355,
|
| 573 |
+
"learning_rate": 5e-05,
|
| 574 |
+
"loss": 0.0296,
|
| 575 |
+
"step": 250
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 6.944444444444445,
|
| 579 |
+
"eval_loss": 0.5332342386245728,
|
| 580 |
+
"eval_runtime": 246.3422,
|
| 581 |
+
"eval_samples_per_second": 2.03,
|
| 582 |
+
"eval_steps_per_second": 0.256,
|
| 583 |
+
"eval_wer": 31.880369074132993,
|
| 584 |
+
"step": 250
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 7.083333333333333,
|
| 588 |
+
"grad_norm": 4.257472991943359,
|
| 589 |
+
"learning_rate": 5.1000000000000006e-05,
|
| 590 |
+
"loss": 0.027,
|
| 591 |
+
"step": 255
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 7.222222222222222,
|
| 595 |
+
"grad_norm": 2.294562339782715,
|
| 596 |
+
"learning_rate": 5.2000000000000004e-05,
|
| 597 |
+
"loss": 0.0207,
|
| 598 |
+
"step": 260
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 7.222222222222222,
|
| 602 |
+
"eval_loss": 0.5236981511116028,
|
| 603 |
+
"eval_runtime": 244.1894,
|
| 604 |
+
"eval_samples_per_second": 2.048,
|
| 605 |
+
"eval_steps_per_second": 0.258,
|
| 606 |
+
"eval_wer": 30.066815144766146,
|
| 607 |
+
"step": 260
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 7.361111111111111,
|
| 611 |
+
"grad_norm": 1.2468712329864502,
|
| 612 |
+
"learning_rate": 5.300000000000001e-05,
|
| 613 |
+
"loss": 0.0228,
|
| 614 |
+
"step": 265
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 7.5,
|
| 618 |
+
"grad_norm": 1.8487240076065063,
|
| 619 |
+
"learning_rate": 5.4000000000000005e-05,
|
| 620 |
+
"loss": 0.0215,
|
| 621 |
+
"step": 270
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 7.5,
|
| 625 |
+
"eval_loss": 0.5222529768943787,
|
| 626 |
+
"eval_runtime": 243.6778,
|
| 627 |
+
"eval_samples_per_second": 2.052,
|
| 628 |
+
"eval_steps_per_second": 0.259,
|
| 629 |
+
"eval_wer": 25.548838689150493,
|
| 630 |
+
"step": 270
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 7.638888888888889,
|
| 634 |
+
"grad_norm": 1.1909741163253784,
|
| 635 |
+
"learning_rate": 5.500000000000001e-05,
|
| 636 |
+
"loss": 0.0201,
|
| 637 |
+
"step": 275
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 7.777777777777778,
|
| 641 |
+
"grad_norm": 1.6141778230667114,
|
| 642 |
+
"learning_rate": 5.6000000000000006e-05,
|
| 643 |
+
"loss": 0.0198,
|
| 644 |
+
"step": 280
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 7.777777777777778,
|
| 648 |
+
"eval_loss": 0.5157026648521423,
|
| 649 |
+
"eval_runtime": 244.0734,
|
| 650 |
+
"eval_samples_per_second": 2.049,
|
| 651 |
+
"eval_steps_per_second": 0.258,
|
| 652 |
+
"eval_wer": 30.194082087177854,
|
| 653 |
+
"step": 280
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 7.916666666666667,
|
| 657 |
+
"grad_norm": 1.1372332572937012,
|
| 658 |
+
"learning_rate": 5.6999999999999996e-05,
|
| 659 |
+
"loss": 0.0193,
|
| 660 |
+
"step": 285
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 8.055555555555555,
|
| 664 |
+
"grad_norm": 2.210016965866089,
|
| 665 |
+
"learning_rate": 5.8e-05,
|
| 666 |
+
"loss": 0.0273,
|
| 667 |
+
"step": 290
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 8.055555555555555,
|
| 671 |
+
"eval_loss": 0.5289562940597534,
|
| 672 |
+
"eval_runtime": 243.9152,
|
| 673 |
+
"eval_samples_per_second": 2.05,
|
| 674 |
+
"eval_steps_per_second": 0.258,
|
| 675 |
+
"eval_wer": 27.553293032134903,
|
| 676 |
+
"step": 290
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 8.194444444444445,
|
| 680 |
+
"grad_norm": 1.942575454711914,
|
| 681 |
+
"learning_rate": 5.9e-05,
|
| 682 |
+
"loss": 0.0201,
|
| 683 |
+
"step": 295
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 8.333333333333334,
|
| 687 |
+
"grad_norm": 1.3640440702438354,
|
| 688 |
+
"learning_rate": 6e-05,
|
| 689 |
+
"loss": 0.0197,
|
| 690 |
+
"step": 300
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 8.333333333333334,
|
| 694 |
+
"eval_loss": 0.5509196519851685,
|
| 695 |
+
"eval_runtime": 243.9508,
|
| 696 |
+
"eval_samples_per_second": 2.05,
|
| 697 |
+
"eval_steps_per_second": 0.258,
|
| 698 |
+
"eval_wer": 26.948775055679285,
|
| 699 |
+
"step": 300
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 8.333333333333334,
|
| 703 |
+
"step": 300,
|
| 704 |
+
"total_flos": 9.2409447186432e+17,
|
| 705 |
+
"train_loss": 0.5431244759509961,
|
| 706 |
+
"train_runtime": 10016.0212,
|
| 707 |
+
"train_samples_per_second": 3.834,
|
| 708 |
+
"train_steps_per_second": 0.03
|
| 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 |
+
}
|