Commit
•
3644968
1
Parent(s):
df76731
clean training step
Browse files- checkpoint-360/config.json +0 -142
- checkpoint-360/optimizer.pt +0 -3
- checkpoint-360/preprocessor_config.json +0 -0
- checkpoint-360/pytorch_model.bin +0 -3
- checkpoint-360/rng_state.pth +0 -3
- checkpoint-360/scaler.pt +0 -3
- checkpoint-360/scheduler.pt +0 -3
- checkpoint-360/trainer_state.json +0 -637
- checkpoint-360/training_args.bin +0 -3
- checkpoint-400/config.json +0 -142
- checkpoint-400/optimizer.pt +0 -3
- checkpoint-400/preprocessor_config.json +0 -0
- checkpoint-400/pytorch_model.bin +0 -3
- checkpoint-400/rng_state.pth +0 -3
- checkpoint-400/scaler.pt +0 -3
- checkpoint-400/scheduler.pt +0 -3
- checkpoint-400/trainer_state.json +0 -706
- checkpoint-400/training_args.bin +0 -3
- huggingface_training.py +15 -14
checkpoint-360/config.json
DELETED
|
@@ -1,142 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "openai/whisper-tiny",
|
| 3 |
-
"activation_dropout": 0.0,
|
| 4 |
-
"activation_function": "gelu",
|
| 5 |
-
"architectures": [
|
| 6 |
-
"WhisperForConditionalGeneration"
|
| 7 |
-
],
|
| 8 |
-
"attention_dropout": 0.0,
|
| 9 |
-
"begin_suppress_tokens": [
|
| 10 |
-
220,
|
| 11 |
-
50257
|
| 12 |
-
],
|
| 13 |
-
"bos_token_id": 50257,
|
| 14 |
-
"d_model": 384,
|
| 15 |
-
"decoder_attention_heads": 6,
|
| 16 |
-
"decoder_ffn_dim": 1536,
|
| 17 |
-
"decoder_layerdrop": 0.0,
|
| 18 |
-
"decoder_layers": 4,
|
| 19 |
-
"decoder_start_token_id": 50258,
|
| 20 |
-
"dropout": 0.0,
|
| 21 |
-
"encoder_attention_heads": 6,
|
| 22 |
-
"encoder_ffn_dim": 1536,
|
| 23 |
-
"encoder_layerdrop": 0.0,
|
| 24 |
-
"encoder_layers": 4,
|
| 25 |
-
"eos_token_id": 50257,
|
| 26 |
-
"forced_decoder_ids": [
|
| 27 |
-
[
|
| 28 |
-
1,
|
| 29 |
-
50259
|
| 30 |
-
],
|
| 31 |
-
[
|
| 32 |
-
2,
|
| 33 |
-
50359
|
| 34 |
-
],
|
| 35 |
-
[
|
| 36 |
-
3,
|
| 37 |
-
50363
|
| 38 |
-
]
|
| 39 |
-
],
|
| 40 |
-
"init_std": 0.02,
|
| 41 |
-
"is_encoder_decoder": true,
|
| 42 |
-
"max_length": 448,
|
| 43 |
-
"max_source_positions": 1500,
|
| 44 |
-
"max_target_positions": 448,
|
| 45 |
-
"model_type": "whisper",
|
| 46 |
-
"num_hidden_layers": 4,
|
| 47 |
-
"num_mel_bins": 80,
|
| 48 |
-
"pad_token_id": 50257,
|
| 49 |
-
"scale_embedding": false,
|
| 50 |
-
"suppress_tokens": [
|
| 51 |
-
1,
|
| 52 |
-
2,
|
| 53 |
-
7,
|
| 54 |
-
8,
|
| 55 |
-
9,
|
| 56 |
-
10,
|
| 57 |
-
14,
|
| 58 |
-
25,
|
| 59 |
-
26,
|
| 60 |
-
27,
|
| 61 |
-
28,
|
| 62 |
-
29,
|
| 63 |
-
31,
|
| 64 |
-
58,
|
| 65 |
-
59,
|
| 66 |
-
60,
|
| 67 |
-
61,
|
| 68 |
-
62,
|
| 69 |
-
63,
|
| 70 |
-
90,
|
| 71 |
-
91,
|
| 72 |
-
92,
|
| 73 |
-
93,
|
| 74 |
-
359,
|
| 75 |
-
503,
|
| 76 |
-
522,
|
| 77 |
-
542,
|
| 78 |
-
873,
|
| 79 |
-
893,
|
| 80 |
-
902,
|
| 81 |
-
918,
|
| 82 |
-
922,
|
| 83 |
-
931,
|
| 84 |
-
1350,
|
| 85 |
-
1853,
|
| 86 |
-
1982,
|
| 87 |
-
2460,
|
| 88 |
-
2627,
|
| 89 |
-
3246,
|
| 90 |
-
3253,
|
| 91 |
-
3268,
|
| 92 |
-
3536,
|
| 93 |
-
3846,
|
| 94 |
-
3961,
|
| 95 |
-
4183,
|
| 96 |
-
4667,
|
| 97 |
-
6585,
|
| 98 |
-
6647,
|
| 99 |
-
7273,
|
| 100 |
-
9061,
|
| 101 |
-
9383,
|
| 102 |
-
10428,
|
| 103 |
-
10929,
|
| 104 |
-
11938,
|
| 105 |
-
12033,
|
| 106 |
-
12331,
|
| 107 |
-
12562,
|
| 108 |
-
13793,
|
| 109 |
-
14157,
|
| 110 |
-
14635,
|
| 111 |
-
15265,
|
| 112 |
-
15618,
|
| 113 |
-
16553,
|
| 114 |
-
16604,
|
| 115 |
-
18362,
|
| 116 |
-
18956,
|
| 117 |
-
20075,
|
| 118 |
-
21675,
|
| 119 |
-
22520,
|
| 120 |
-
26130,
|
| 121 |
-
26161,
|
| 122 |
-
26435,
|
| 123 |
-
28279,
|
| 124 |
-
29464,
|
| 125 |
-
31650,
|
| 126 |
-
32302,
|
| 127 |
-
32470,
|
| 128 |
-
36865,
|
| 129 |
-
42863,
|
| 130 |
-
47425,
|
| 131 |
-
49870,
|
| 132 |
-
50254,
|
| 133 |
-
50258,
|
| 134 |
-
50360,
|
| 135 |
-
50361,
|
| 136 |
-
50362
|
| 137 |
-
],
|
| 138 |
-
"torch_dtype": "float32",
|
| 139 |
-
"transformers_version": "4.26.0.dev0",
|
| 140 |
-
"use_cache": true,
|
| 141 |
-
"vocab_size": 51865
|
| 142 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-360/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:513fbc5d03bd07f32d77cf2f5dcc0d8298575b96fbda2ed1de30f1cb859889ae
|
| 3 |
-
size 302183173
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-360/preprocessor_config.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-360/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:48879b1ce776151b602f3a1bdf10683d776d3f0765214b322443dddb1d951006
|
| 3 |
-
size 151098921
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-360/rng_state.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:89f9781ff6e5ab617d91036a7029d39a2832fa624ae853afb0f238fb19535016
|
| 3 |
-
size 14575
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-360/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:12d681e2b2a56f2134611cbb1679a9f32470e4cf3a48f4a2243741f0852b30ae
|
| 3 |
-
size 557
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-360/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:922e864e56c484925ddcd495d1c992405fc4f95d13329256b422ef0f40cc0891
|
| 3 |
-
size 627
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-360/trainer_state.json
DELETED
|
@@ -1,637 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 8.07862407862408,
|
| 5 |
-
"global_step": 360,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.01,
|
| 12 |
-
"learning_rate": 0.0,
|
| 13 |
-
"loss": 1.8118,
|
| 14 |
-
"step": 4
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.02,
|
| 18 |
-
"learning_rate": 3.6585365853658536e-07,
|
| 19 |
-
"loss": 1.8122,
|
| 20 |
-
"step": 8
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.03,
|
| 24 |
-
"learning_rate": 8.53658536585366e-07,
|
| 25 |
-
"loss": 1.8174,
|
| 26 |
-
"step": 12
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.04,
|
| 30 |
-
"learning_rate": 1.3414634146341465e-06,
|
| 31 |
-
"loss": 1.7616,
|
| 32 |
-
"step": 16
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.05,
|
| 36 |
-
"learning_rate": 1.8292682926829268e-06,
|
| 37 |
-
"loss": 1.6875,
|
| 38 |
-
"step": 20
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.06,
|
| 42 |
-
"learning_rate": 2.317073170731708e-06,
|
| 43 |
-
"loss": 1.5201,
|
| 44 |
-
"step": 24
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.07,
|
| 48 |
-
"learning_rate": 2.8048780487804884e-06,
|
| 49 |
-
"loss": 1.3982,
|
| 50 |
-
"step": 28
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.08,
|
| 54 |
-
"learning_rate": 3.292682926829269e-06,
|
| 55 |
-
"loss": 1.3541,
|
| 56 |
-
"step": 32
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.09,
|
| 60 |
-
"learning_rate": 3.780487804878049e-06,
|
| 61 |
-
"loss": 1.2092,
|
| 62 |
-
"step": 36
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.1,
|
| 66 |
-
"learning_rate": 4.268292682926829e-06,
|
| 67 |
-
"loss": 1.1599,
|
| 68 |
-
"step": 40
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.1,
|
| 72 |
-
"eval_loss": 1.142654299736023,
|
| 73 |
-
"eval_runtime": 101.9854,
|
| 74 |
-
"eval_samples_per_second": 3.863,
|
| 75 |
-
"eval_steps_per_second": 0.127,
|
| 76 |
-
"eval_wer": 15.213946117274169,
|
| 77 |
-
"step": 40
|
| 78 |
-
},
|
| 79 |
-
{
|
| 80 |
-
"epoch": 1.01,
|
| 81 |
-
"learning_rate": 4.75609756097561e-06,
|
| 82 |
-
"loss": 1.0124,
|
| 83 |
-
"step": 44
|
| 84 |
-
},
|
| 85 |
-
{
|
| 86 |
-
"epoch": 1.02,
|
| 87 |
-
"learning_rate": 5.243902439024391e-06,
|
| 88 |
-
"loss": 0.9171,
|
| 89 |
-
"step": 48
|
| 90 |
-
},
|
| 91 |
-
{
|
| 92 |
-
"epoch": 1.03,
|
| 93 |
-
"learning_rate": 5.731707317073171e-06,
|
| 94 |
-
"loss": 0.8027,
|
| 95 |
-
"step": 52
|
| 96 |
-
},
|
| 97 |
-
{
|
| 98 |
-
"epoch": 1.04,
|
| 99 |
-
"learning_rate": 6.219512195121951e-06,
|
| 100 |
-
"loss": 0.7284,
|
| 101 |
-
"step": 56
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 1.05,
|
| 105 |
-
"learning_rate": 6.707317073170733e-06,
|
| 106 |
-
"loss": 0.6185,
|
| 107 |
-
"step": 60
|
| 108 |
-
},
|
| 109 |
-
{
|
| 110 |
-
"epoch": 1.06,
|
| 111 |
-
"learning_rate": 7.1951219512195125e-06,
|
| 112 |
-
"loss": 0.57,
|
| 113 |
-
"step": 64
|
| 114 |
-
},
|
| 115 |
-
{
|
| 116 |
-
"epoch": 1.07,
|
| 117 |
-
"learning_rate": 7.682926829268293e-06,
|
| 118 |
-
"loss": 0.4985,
|
| 119 |
-
"step": 68
|
| 120 |
-
},
|
| 121 |
-
{
|
| 122 |
-
"epoch": 1.08,
|
| 123 |
-
"learning_rate": 8.170731707317073e-06,
|
| 124 |
-
"loss": 0.488,
|
| 125 |
-
"step": 72
|
| 126 |
-
},
|
| 127 |
-
{
|
| 128 |
-
"epoch": 1.09,
|
| 129 |
-
"learning_rate": 8.658536585365854e-06,
|
| 130 |
-
"loss": 0.4569,
|
| 131 |
-
"step": 76
|
| 132 |
-
},
|
| 133 |
-
{
|
| 134 |
-
"epoch": 1.1,
|
| 135 |
-
"learning_rate": 9.146341463414635e-06,
|
| 136 |
-
"loss": 0.4655,
|
| 137 |
-
"step": 80
|
| 138 |
-
},
|
| 139 |
-
{
|
| 140 |
-
"epoch": 1.1,
|
| 141 |
-
"eval_loss": 0.5613037943840027,
|
| 142 |
-
"eval_runtime": 91.9697,
|
| 143 |
-
"eval_samples_per_second": 4.284,
|
| 144 |
-
"eval_steps_per_second": 0.141,
|
| 145 |
-
"eval_wer": 17.591125198098258,
|
| 146 |
-
"step": 80
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 2.0,
|
| 150 |
-
"learning_rate": 9.634146341463415e-06,
|
| 151 |
-
"loss": 0.425,
|
| 152 |
-
"step": 84
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 2.01,
|
| 156 |
-
"learning_rate": 9.96923076923077e-06,
|
| 157 |
-
"loss": 0.4162,
|
| 158 |
-
"step": 88
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 2.02,
|
| 162 |
-
"learning_rate": 9.846153846153848e-06,
|
| 163 |
-
"loss": 0.3809,
|
| 164 |
-
"step": 92
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 2.03,
|
| 168 |
-
"learning_rate": 9.723076923076924e-06,
|
| 169 |
-
"loss": 0.3533,
|
| 170 |
-
"step": 96
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 2.04,
|
| 174 |
-
"learning_rate": 9.600000000000001e-06,
|
| 175 |
-
"loss": 0.3511,
|
| 176 |
-
"step": 100
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 2.05,
|
| 180 |
-
"learning_rate": 9.476923076923079e-06,
|
| 181 |
-
"loss": 0.3475,
|
| 182 |
-
"step": 104
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 2.06,
|
| 186 |
-
"learning_rate": 9.353846153846155e-06,
|
| 187 |
-
"loss": 0.321,
|
| 188 |
-
"step": 108
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 2.07,
|
| 192 |
-
"learning_rate": 9.230769230769232e-06,
|
| 193 |
-
"loss": 0.2859,
|
| 194 |
-
"step": 112
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 2.08,
|
| 198 |
-
"learning_rate": 9.107692307692308e-06,
|
| 199 |
-
"loss": 0.3191,
|
| 200 |
-
"step": 116
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 2.09,
|
| 204 |
-
"learning_rate": 8.984615384615386e-06,
|
| 205 |
-
"loss": 0.2753,
|
| 206 |
-
"step": 120
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 2.09,
|
| 210 |
-
"eval_loss": 0.5241264700889587,
|
| 211 |
-
"eval_runtime": 88.0526,
|
| 212 |
-
"eval_samples_per_second": 4.475,
|
| 213 |
-
"eval_steps_per_second": 0.148,
|
| 214 |
-
"eval_wer": 17.21321467755699,
|
| 215 |
-
"step": 120
|
| 216 |
-
},
|
| 217 |
-
{
|
| 218 |
-
"epoch": 3.0,
|
| 219 |
-
"learning_rate": 8.861538461538463e-06,
|
| 220 |
-
"loss": 0.3104,
|
| 221 |
-
"step": 124
|
| 222 |
-
},
|
| 223 |
-
{
|
| 224 |
-
"epoch": 3.01,
|
| 225 |
-
"learning_rate": 8.73846153846154e-06,
|
| 226 |
-
"loss": 0.2734,
|
| 227 |
-
"step": 128
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 3.02,
|
| 231 |
-
"learning_rate": 8.615384615384617e-06,
|
| 232 |
-
"loss": 0.2608,
|
| 233 |
-
"step": 132
|
| 234 |
-
},
|
| 235 |
-
{
|
| 236 |
-
"epoch": 3.03,
|
| 237 |
-
"learning_rate": 8.492307692307693e-06,
|
| 238 |
-
"loss": 0.2509,
|
| 239 |
-
"step": 136
|
| 240 |
-
},
|
| 241 |
-
{
|
| 242 |
-
"epoch": 3.04,
|
| 243 |
-
"learning_rate": 8.36923076923077e-06,
|
| 244 |
-
"loss": 0.2548,
|
| 245 |
-
"step": 140
|
| 246 |
-
},
|
| 247 |
-
{
|
| 248 |
-
"epoch": 3.05,
|
| 249 |
-
"learning_rate": 8.246153846153848e-06,
|
| 250 |
-
"loss": 0.2469,
|
| 251 |
-
"step": 144
|
| 252 |
-
},
|
| 253 |
-
{
|
| 254 |
-
"epoch": 3.06,
|
| 255 |
-
"learning_rate": 8.123076923076924e-06,
|
| 256 |
-
"loss": 0.2231,
|
| 257 |
-
"step": 148
|
| 258 |
-
},
|
| 259 |
-
{
|
| 260 |
-
"epoch": 3.07,
|
| 261 |
-
"learning_rate": 8.000000000000001e-06,
|
| 262 |
-
"loss": 0.2138,
|
| 263 |
-
"step": 152
|
| 264 |
-
},
|
| 265 |
-
{
|
| 266 |
-
"epoch": 3.08,
|
| 267 |
-
"learning_rate": 7.876923076923077e-06,
|
| 268 |
-
"loss": 0.2349,
|
| 269 |
-
"step": 156
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 3.09,
|
| 273 |
-
"learning_rate": 7.753846153846155e-06,
|
| 274 |
-
"loss": 0.2077,
|
| 275 |
-
"step": 160
|
| 276 |
-
},
|
| 277 |
-
{
|
| 278 |
-
"epoch": 3.09,
|
| 279 |
-
"eval_loss": 0.5241798758506775,
|
| 280 |
-
"eval_runtime": 88.5317,
|
| 281 |
-
"eval_samples_per_second": 4.45,
|
| 282 |
-
"eval_steps_per_second": 0.147,
|
| 283 |
-
"eval_wer": 17.26197732536877,
|
| 284 |
-
"step": 160
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 3.1,
|
| 288 |
-
"learning_rate": 7.630769230769232e-06,
|
| 289 |
-
"loss": 0.2322,
|
| 290 |
-
"step": 164
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 4.01,
|
| 294 |
-
"learning_rate": 7.507692307692308e-06,
|
| 295 |
-
"loss": 0.2036,
|
| 296 |
-
"step": 168
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 4.02,
|
| 300 |
-
"learning_rate": 7.384615384615386e-06,
|
| 301 |
-
"loss": 0.2058,
|
| 302 |
-
"step": 172
|
| 303 |
-
},
|
| 304 |
-
{
|
| 305 |
-
"epoch": 4.03,
|
| 306 |
-
"learning_rate": 7.261538461538462e-06,
|
| 307 |
-
"loss": 0.1797,
|
| 308 |
-
"step": 176
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 4.04,
|
| 312 |
-
"learning_rate": 7.1384615384615385e-06,
|
| 313 |
-
"loss": 0.186,
|
| 314 |
-
"step": 180
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 4.05,
|
| 318 |
-
"learning_rate": 7.015384615384616e-06,
|
| 319 |
-
"loss": 0.2035,
|
| 320 |
-
"step": 184
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 4.06,
|
| 324 |
-
"learning_rate": 6.892307692307693e-06,
|
| 325 |
-
"loss": 0.1794,
|
| 326 |
-
"step": 188
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 4.07,
|
| 330 |
-
"learning_rate": 6.76923076923077e-06,
|
| 331 |
-
"loss": 0.1589,
|
| 332 |
-
"step": 192
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 4.08,
|
| 336 |
-
"learning_rate": 6.646153846153846e-06,
|
| 337 |
-
"loss": 0.1879,
|
| 338 |
-
"step": 196
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 4.09,
|
| 342 |
-
"learning_rate": 6.523076923076923e-06,
|
| 343 |
-
"loss": 0.1636,
|
| 344 |
-
"step": 200
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 4.09,
|
| 348 |
-
"eval_loss": 0.5289868712425232,
|
| 349 |
-
"eval_runtime": 95.5188,
|
| 350 |
-
"eval_samples_per_second": 4.125,
|
| 351 |
-
"eval_steps_per_second": 0.136,
|
| 352 |
-
"eval_wer": 17.66426916981592,
|
| 353 |
-
"step": 200
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 4.1,
|
| 357 |
-
"learning_rate": 6.4000000000000006e-06,
|
| 358 |
-
"loss": 0.1767,
|
| 359 |
-
"step": 204
|
| 360 |
-
},
|
| 361 |
-
{
|
| 362 |
-
"epoch": 5.01,
|
| 363 |
-
"learning_rate": 6.276923076923077e-06,
|
| 364 |
-
"loss": 0.1657,
|
| 365 |
-
"step": 208
|
| 366 |
-
},
|
| 367 |
-
{
|
| 368 |
-
"epoch": 5.02,
|
| 369 |
-
"learning_rate": 6.153846153846155e-06,
|
| 370 |
-
"loss": 0.1607,
|
| 371 |
-
"step": 212
|
| 372 |
-
},
|
| 373 |
-
{
|
| 374 |
-
"epoch": 5.03,
|
| 375 |
-
"learning_rate": 6.030769230769231e-06,
|
| 376 |
-
"loss": 0.1458,
|
| 377 |
-
"step": 216
|
| 378 |
-
},
|
| 379 |
-
{
|
| 380 |
-
"epoch": 5.04,
|
| 381 |
-
"learning_rate": 5.907692307692308e-06,
|
| 382 |
-
"loss": 0.1541,
|
| 383 |
-
"step": 220
|
| 384 |
-
},
|
| 385 |
-
{
|
| 386 |
-
"epoch": 5.05,
|
| 387 |
-
"learning_rate": 5.784615384615385e-06,
|
| 388 |
-
"loss": 0.1494,
|
| 389 |
-
"step": 224
|
| 390 |
-
},
|
| 391 |
-
{
|
| 392 |
-
"epoch": 5.06,
|
| 393 |
-
"learning_rate": 5.661538461538462e-06,
|
| 394 |
-
"loss": 0.144,
|
| 395 |
-
"step": 228
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 5.07,
|
| 399 |
-
"learning_rate": 5.538461538461539e-06,
|
| 400 |
-
"loss": 0.1311,
|
| 401 |
-
"step": 232
|
| 402 |
-
},
|
| 403 |
-
{
|
| 404 |
-
"epoch": 5.08,
|
| 405 |
-
"learning_rate": 5.415384615384615e-06,
|
| 406 |
-
"loss": 0.1411,
|
| 407 |
-
"step": 236
|
| 408 |
-
},
|
| 409 |
-
{
|
| 410 |
-
"epoch": 5.09,
|
| 411 |
-
"learning_rate": 5.292307692307693e-06,
|
| 412 |
-
"loss": 0.1322,
|
| 413 |
-
"step": 240
|
| 414 |
-
},
|
| 415 |
-
{
|
| 416 |
-
"epoch": 5.09,
|
| 417 |
-
"eval_loss": 0.5350630283355713,
|
| 418 |
-
"eval_runtime": 92.5111,
|
| 419 |
-
"eval_samples_per_second": 4.259,
|
| 420 |
-
"eval_steps_per_second": 0.141,
|
| 421 |
-
"eval_wer": 18.2128489576984,
|
| 422 |
-
"step": 240
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 5.1,
|
| 426 |
-
"learning_rate": 5.16923076923077e-06,
|
| 427 |
-
"loss": 0.1436,
|
| 428 |
-
"step": 244
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 6.0,
|
| 432 |
-
"learning_rate": 5.046153846153846e-06,
|
| 433 |
-
"loss": 0.1375,
|
| 434 |
-
"step": 248
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 6.01,
|
| 438 |
-
"learning_rate": 4.923076923076924e-06,
|
| 439 |
-
"loss": 0.1361,
|
| 440 |
-
"step": 252
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 6.02,
|
| 444 |
-
"learning_rate": 4.800000000000001e-06,
|
| 445 |
-
"loss": 0.129,
|
| 446 |
-
"step": 256
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 6.03,
|
| 450 |
-
"learning_rate": 4.676923076923077e-06,
|
| 451 |
-
"loss": 0.1127,
|
| 452 |
-
"step": 260
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 6.04,
|
| 456 |
-
"learning_rate": 4.553846153846154e-06,
|
| 457 |
-
"loss": 0.1266,
|
| 458 |
-
"step": 264
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 6.05,
|
| 462 |
-
"learning_rate": 4.430769230769232e-06,
|
| 463 |
-
"loss": 0.1193,
|
| 464 |
-
"step": 268
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 6.06,
|
| 468 |
-
"learning_rate": 4.307692307692308e-06,
|
| 469 |
-
"loss": 0.1127,
|
| 470 |
-
"step": 272
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 6.07,
|
| 474 |
-
"learning_rate": 4.184615384615385e-06,
|
| 475 |
-
"loss": 0.1064,
|
| 476 |
-
"step": 276
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 6.08,
|
| 480 |
-
"learning_rate": 4.061538461538462e-06,
|
| 481 |
-
"loss": 0.123,
|
| 482 |
-
"step": 280
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 6.08,
|
| 486 |
-
"eval_loss": 0.5429388284683228,
|
| 487 |
-
"eval_runtime": 91.5818,
|
| 488 |
-
"eval_samples_per_second": 4.302,
|
| 489 |
-
"eval_steps_per_second": 0.142,
|
| 490 |
-
"eval_wer": 18.907716689016212,
|
| 491 |
-
"step": 280
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 6.09,
|
| 495 |
-
"learning_rate": 3.938461538461539e-06,
|
| 496 |
-
"loss": 0.1057,
|
| 497 |
-
"step": 284
|
| 498 |
-
},
|
| 499 |
-
{
|
| 500 |
-
"epoch": 7.0,
|
| 501 |
-
"learning_rate": 3.815384615384616e-06,
|
| 502 |
-
"loss": 0.1258,
|
| 503 |
-
"step": 288
|
| 504 |
-
},
|
| 505 |
-
{
|
| 506 |
-
"epoch": 7.01,
|
| 507 |
-
"learning_rate": 3.692307692307693e-06,
|
| 508 |
-
"loss": 0.1108,
|
| 509 |
-
"step": 292
|
| 510 |
-
},
|
| 511 |
-
{
|
| 512 |
-
"epoch": 7.02,
|
| 513 |
-
"learning_rate": 3.5692307692307692e-06,
|
| 514 |
-
"loss": 0.1115,
|
| 515 |
-
"step": 296
|
| 516 |
-
},
|
| 517 |
-
{
|
| 518 |
-
"epoch": 7.03,
|
| 519 |
-
"learning_rate": 3.4461538461538464e-06,
|
| 520 |
-
"loss": 0.0998,
|
| 521 |
-
"step": 300
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 7.04,
|
| 525 |
-
"learning_rate": 3.323076923076923e-06,
|
| 526 |
-
"loss": 0.1106,
|
| 527 |
-
"step": 304
|
| 528 |
-
},
|
| 529 |
-
{
|
| 530 |
-
"epoch": 7.05,
|
| 531 |
-
"learning_rate": 3.2000000000000003e-06,
|
| 532 |
-
"loss": 0.1045,
|
| 533 |
-
"step": 308
|
| 534 |
-
},
|
| 535 |
-
{
|
| 536 |
-
"epoch": 7.06,
|
| 537 |
-
"learning_rate": 3.0769230769230774e-06,
|
| 538 |
-
"loss": 0.0908,
|
| 539 |
-
"step": 312
|
| 540 |
-
},
|
| 541 |
-
{
|
| 542 |
-
"epoch": 7.07,
|
| 543 |
-
"learning_rate": 2.953846153846154e-06,
|
| 544 |
-
"loss": 0.0931,
|
| 545 |
-
"step": 316
|
| 546 |
-
},
|
| 547 |
-
{
|
| 548 |
-
"epoch": 7.08,
|
| 549 |
-
"learning_rate": 2.830769230769231e-06,
|
| 550 |
-
"loss": 0.1074,
|
| 551 |
-
"step": 320
|
| 552 |
-
},
|
| 553 |
-
{
|
| 554 |
-
"epoch": 7.08,
|
| 555 |
-
"eval_loss": 0.5500437021255493,
|
| 556 |
-
"eval_runtime": 104.0907,
|
| 557 |
-
"eval_samples_per_second": 3.785,
|
| 558 |
-
"eval_steps_per_second": 0.125,
|
| 559 |
-
"eval_wer": 19.054004632451544,
|
| 560 |
-
"step": 320
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 7.09,
|
| 564 |
-
"learning_rate": 2.7076923076923076e-06,
|
| 565 |
-
"loss": 0.0937,
|
| 566 |
-
"step": 324
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 7.1,
|
| 570 |
-
"learning_rate": 2.584615384615385e-06,
|
| 571 |
-
"loss": 0.1091,
|
| 572 |
-
"step": 328
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 8.01,
|
| 576 |
-
"learning_rate": 2.461538461538462e-06,
|
| 577 |
-
"loss": 0.0951,
|
| 578 |
-
"step": 332
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 8.02,
|
| 582 |
-
"learning_rate": 2.3384615384615387e-06,
|
| 583 |
-
"loss": 0.1003,
|
| 584 |
-
"step": 336
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 8.03,
|
| 588 |
-
"learning_rate": 2.215384615384616e-06,
|
| 589 |
-
"loss": 0.0836,
|
| 590 |
-
"step": 340
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 8.04,
|
| 594 |
-
"learning_rate": 2.0923076923076926e-06,
|
| 595 |
-
"loss": 0.0907,
|
| 596 |
-
"step": 344
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 8.05,
|
| 600 |
-
"learning_rate": 1.9692307692307693e-06,
|
| 601 |
-
"loss": 0.1013,
|
| 602 |
-
"step": 348
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 8.06,
|
| 606 |
-
"learning_rate": 1.8461538461538465e-06,
|
| 607 |
-
"loss": 0.0891,
|
| 608 |
-
"step": 352
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 8.07,
|
| 612 |
-
"learning_rate": 1.7230769230769232e-06,
|
| 613 |
-
"loss": 0.077,
|
| 614 |
-
"step": 356
|
| 615 |
-
},
|
| 616 |
-
{
|
| 617 |
-
"epoch": 8.08,
|
| 618 |
-
"learning_rate": 1.6000000000000001e-06,
|
| 619 |
-
"loss": 0.1007,
|
| 620 |
-
"step": 360
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 8.08,
|
| 624 |
-
"eval_loss": 0.5552565455436707,
|
| 625 |
-
"eval_runtime": 88.458,
|
| 626 |
-
"eval_samples_per_second": 4.454,
|
| 627 |
-
"eval_steps_per_second": 0.147,
|
| 628 |
-
"eval_wer": 19.310008533463368,
|
| 629 |
-
"step": 360
|
| 630 |
-
}
|
| 631 |
-
],
|
| 632 |
-
"max_steps": 407,
|
| 633 |
-
"num_train_epochs": 9223372036854775807,
|
| 634 |
-
"total_flos": 5.6288618938368e+17,
|
| 635 |
-
"trial_name": null,
|
| 636 |
-
"trial_params": null
|
| 637 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-360/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:fcbca0d141969bcb1c3cd0ef5a009221139334753b899d88e4d5003bd23f4b5f
|
| 3 |
-
size 3579
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/config.json
DELETED
|
@@ -1,142 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "openai/whisper-tiny",
|
| 3 |
-
"activation_dropout": 0.0,
|
| 4 |
-
"activation_function": "gelu",
|
| 5 |
-
"architectures": [
|
| 6 |
-
"WhisperForConditionalGeneration"
|
| 7 |
-
],
|
| 8 |
-
"attention_dropout": 0.0,
|
| 9 |
-
"begin_suppress_tokens": [
|
| 10 |
-
220,
|
| 11 |
-
50257
|
| 12 |
-
],
|
| 13 |
-
"bos_token_id": 50257,
|
| 14 |
-
"d_model": 384,
|
| 15 |
-
"decoder_attention_heads": 6,
|
| 16 |
-
"decoder_ffn_dim": 1536,
|
| 17 |
-
"decoder_layerdrop": 0.0,
|
| 18 |
-
"decoder_layers": 4,
|
| 19 |
-
"decoder_start_token_id": 50258,
|
| 20 |
-
"dropout": 0.0,
|
| 21 |
-
"encoder_attention_heads": 6,
|
| 22 |
-
"encoder_ffn_dim": 1536,
|
| 23 |
-
"encoder_layerdrop": 0.0,
|
| 24 |
-
"encoder_layers": 4,
|
| 25 |
-
"eos_token_id": 50257,
|
| 26 |
-
"forced_decoder_ids": [
|
| 27 |
-
[
|
| 28 |
-
1,
|
| 29 |
-
50259
|
| 30 |
-
],
|
| 31 |
-
[
|
| 32 |
-
2,
|
| 33 |
-
50359
|
| 34 |
-
],
|
| 35 |
-
[
|
| 36 |
-
3,
|
| 37 |
-
50363
|
| 38 |
-
]
|
| 39 |
-
],
|
| 40 |
-
"init_std": 0.02,
|
| 41 |
-
"is_encoder_decoder": true,
|
| 42 |
-
"max_length": 448,
|
| 43 |
-
"max_source_positions": 1500,
|
| 44 |
-
"max_target_positions": 448,
|
| 45 |
-
"model_type": "whisper",
|
| 46 |
-
"num_hidden_layers": 4,
|
| 47 |
-
"num_mel_bins": 80,
|
| 48 |
-
"pad_token_id": 50257,
|
| 49 |
-
"scale_embedding": false,
|
| 50 |
-
"suppress_tokens": [
|
| 51 |
-
1,
|
| 52 |
-
2,
|
| 53 |
-
7,
|
| 54 |
-
8,
|
| 55 |
-
9,
|
| 56 |
-
10,
|
| 57 |
-
14,
|
| 58 |
-
25,
|
| 59 |
-
26,
|
| 60 |
-
27,
|
| 61 |
-
28,
|
| 62 |
-
29,
|
| 63 |
-
31,
|
| 64 |
-
58,
|
| 65 |
-
59,
|
| 66 |
-
60,
|
| 67 |
-
61,
|
| 68 |
-
62,
|
| 69 |
-
63,
|
| 70 |
-
90,
|
| 71 |
-
91,
|
| 72 |
-
92,
|
| 73 |
-
93,
|
| 74 |
-
359,
|
| 75 |
-
503,
|
| 76 |
-
522,
|
| 77 |
-
542,
|
| 78 |
-
873,
|
| 79 |
-
893,
|
| 80 |
-
902,
|
| 81 |
-
918,
|
| 82 |
-
922,
|
| 83 |
-
931,
|
| 84 |
-
1350,
|
| 85 |
-
1853,
|
| 86 |
-
1982,
|
| 87 |
-
2460,
|
| 88 |
-
2627,
|
| 89 |
-
3246,
|
| 90 |
-
3253,
|
| 91 |
-
3268,
|
| 92 |
-
3536,
|
| 93 |
-
3846,
|
| 94 |
-
3961,
|
| 95 |
-
4183,
|
| 96 |
-
4667,
|
| 97 |
-
6585,
|
| 98 |
-
6647,
|
| 99 |
-
7273,
|
| 100 |
-
9061,
|
| 101 |
-
9383,
|
| 102 |
-
10428,
|
| 103 |
-
10929,
|
| 104 |
-
11938,
|
| 105 |
-
12033,
|
| 106 |
-
12331,
|
| 107 |
-
12562,
|
| 108 |
-
13793,
|
| 109 |
-
14157,
|
| 110 |
-
14635,
|
| 111 |
-
15265,
|
| 112 |
-
15618,
|
| 113 |
-
16553,
|
| 114 |
-
16604,
|
| 115 |
-
18362,
|
| 116 |
-
18956,
|
| 117 |
-
20075,
|
| 118 |
-
21675,
|
| 119 |
-
22520,
|
| 120 |
-
26130,
|
| 121 |
-
26161,
|
| 122 |
-
26435,
|
| 123 |
-
28279,
|
| 124 |
-
29464,
|
| 125 |
-
31650,
|
| 126 |
-
32302,
|
| 127 |
-
32470,
|
| 128 |
-
36865,
|
| 129 |
-
42863,
|
| 130 |
-
47425,
|
| 131 |
-
49870,
|
| 132 |
-
50254,
|
| 133 |
-
50258,
|
| 134 |
-
50360,
|
| 135 |
-
50361,
|
| 136 |
-
50362
|
| 137 |
-
],
|
| 138 |
-
"torch_dtype": "float32",
|
| 139 |
-
"transformers_version": "4.26.0.dev0",
|
| 140 |
-
"use_cache": true,
|
| 141 |
-
"vocab_size": 51865
|
| 142 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:ca2f216746e89a3c2dc592e56702d375bf97996f49afe1d761ee97223c74e35c
|
| 3 |
-
size 302183173
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/preprocessor_config.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-400/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:a2f00e8aebd51836c62a6367fa1a3bad01938ccd285ac6cc2c2dd7b6e9755793
|
| 3 |
-
size 151098921
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/rng_state.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:e0fee234ab8151645a0895ac4e9559fbd6bec4f70f802b8c94db562d283ad737
|
| 3 |
-
size 14639
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:37d85a7f00fa7bae4774c70bac351a030ccefea202dbec056f5a4d44e50b132c
|
| 3 |
-
size 557
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:1cf1f3c596de0ecc246fc5b02e9720244273de83ebc2f79d153609594e679a82
|
| 3 |
-
size 627
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/trainer_state.json
DELETED
|
@@ -1,706 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 9.076167076167076,
|
| 5 |
-
"global_step": 400,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.01,
|
| 12 |
-
"learning_rate": 0.0,
|
| 13 |
-
"loss": 1.8118,
|
| 14 |
-
"step": 4
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.02,
|
| 18 |
-
"learning_rate": 3.6585365853658536e-07,
|
| 19 |
-
"loss": 1.8122,
|
| 20 |
-
"step": 8
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.03,
|
| 24 |
-
"learning_rate": 8.53658536585366e-07,
|
| 25 |
-
"loss": 1.8174,
|
| 26 |
-
"step": 12
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.04,
|
| 30 |
-
"learning_rate": 1.3414634146341465e-06,
|
| 31 |
-
"loss": 1.7616,
|
| 32 |
-
"step": 16
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.05,
|
| 36 |
-
"learning_rate": 1.8292682926829268e-06,
|
| 37 |
-
"loss": 1.6875,
|
| 38 |
-
"step": 20
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.06,
|
| 42 |
-
"learning_rate": 2.317073170731708e-06,
|
| 43 |
-
"loss": 1.5201,
|
| 44 |
-
"step": 24
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.07,
|
| 48 |
-
"learning_rate": 2.8048780487804884e-06,
|
| 49 |
-
"loss": 1.3982,
|
| 50 |
-
"step": 28
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.08,
|
| 54 |
-
"learning_rate": 3.292682926829269e-06,
|
| 55 |
-
"loss": 1.3541,
|
| 56 |
-
"step": 32
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.09,
|
| 60 |
-
"learning_rate": 3.780487804878049e-06,
|
| 61 |
-
"loss": 1.2092,
|
| 62 |
-
"step": 36
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.1,
|
| 66 |
-
"learning_rate": 4.268292682926829e-06,
|
| 67 |
-
"loss": 1.1599,
|
| 68 |
-
"step": 40
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.1,
|
| 72 |
-
"eval_loss": 1.142654299736023,
|
| 73 |
-
"eval_runtime": 101.9854,
|
| 74 |
-
"eval_samples_per_second": 3.863,
|
| 75 |
-
"eval_steps_per_second": 0.127,
|
| 76 |
-
"eval_wer": 15.213946117274169,
|
| 77 |
-
"step": 40
|
| 78 |
-
},
|
| 79 |
-
{
|
| 80 |
-
"epoch": 1.01,
|
| 81 |
-
"learning_rate": 4.75609756097561e-06,
|
| 82 |
-
"loss": 1.0124,
|
| 83 |
-
"step": 44
|
| 84 |
-
},
|
| 85 |
-
{
|
| 86 |
-
"epoch": 1.02,
|
| 87 |
-
"learning_rate": 5.243902439024391e-06,
|
| 88 |
-
"loss": 0.9171,
|
| 89 |
-
"step": 48
|
| 90 |
-
},
|
| 91 |
-
{
|
| 92 |
-
"epoch": 1.03,
|
| 93 |
-
"learning_rate": 5.731707317073171e-06,
|
| 94 |
-
"loss": 0.8027,
|
| 95 |
-
"step": 52
|
| 96 |
-
},
|
| 97 |
-
{
|
| 98 |
-
"epoch": 1.04,
|
| 99 |
-
"learning_rate": 6.219512195121951e-06,
|
| 100 |
-
"loss": 0.7284,
|
| 101 |
-
"step": 56
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 1.05,
|
| 105 |
-
"learning_rate": 6.707317073170733e-06,
|
| 106 |
-
"loss": 0.6185,
|
| 107 |
-
"step": 60
|
| 108 |
-
},
|
| 109 |
-
{
|
| 110 |
-
"epoch": 1.06,
|
| 111 |
-
"learning_rate": 7.1951219512195125e-06,
|
| 112 |
-
"loss": 0.57,
|
| 113 |
-
"step": 64
|
| 114 |
-
},
|
| 115 |
-
{
|
| 116 |
-
"epoch": 1.07,
|
| 117 |
-
"learning_rate": 7.682926829268293e-06,
|
| 118 |
-
"loss": 0.4985,
|
| 119 |
-
"step": 68
|
| 120 |
-
},
|
| 121 |
-
{
|
| 122 |
-
"epoch": 1.08,
|
| 123 |
-
"learning_rate": 8.170731707317073e-06,
|
| 124 |
-
"loss": 0.488,
|
| 125 |
-
"step": 72
|
| 126 |
-
},
|
| 127 |
-
{
|
| 128 |
-
"epoch": 1.09,
|
| 129 |
-
"learning_rate": 8.658536585365854e-06,
|
| 130 |
-
"loss": 0.4569,
|
| 131 |
-
"step": 76
|
| 132 |
-
},
|
| 133 |
-
{
|
| 134 |
-
"epoch": 1.1,
|
| 135 |
-
"learning_rate": 9.146341463414635e-06,
|
| 136 |
-
"loss": 0.4655,
|
| 137 |
-
"step": 80
|
| 138 |
-
},
|
| 139 |
-
{
|
| 140 |
-
"epoch": 1.1,
|
| 141 |
-
"eval_loss": 0.5613037943840027,
|
| 142 |
-
"eval_runtime": 91.9697,
|
| 143 |
-
"eval_samples_per_second": 4.284,
|
| 144 |
-
"eval_steps_per_second": 0.141,
|
| 145 |
-
"eval_wer": 17.591125198098258,
|
| 146 |
-
"step": 80
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 2.0,
|
| 150 |
-
"learning_rate": 9.634146341463415e-06,
|
| 151 |
-
"loss": 0.425,
|
| 152 |
-
"step": 84
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 2.01,
|
| 156 |
-
"learning_rate": 9.96923076923077e-06,
|
| 157 |
-
"loss": 0.4162,
|
| 158 |
-
"step": 88
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 2.02,
|
| 162 |
-
"learning_rate": 9.846153846153848e-06,
|
| 163 |
-
"loss": 0.3809,
|
| 164 |
-
"step": 92
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 2.03,
|
| 168 |
-
"learning_rate": 9.723076923076924e-06,
|
| 169 |
-
"loss": 0.3533,
|
| 170 |
-
"step": 96
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 2.04,
|
| 174 |
-
"learning_rate": 9.600000000000001e-06,
|
| 175 |
-
"loss": 0.3511,
|
| 176 |
-
"step": 100
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 2.05,
|
| 180 |
-
"learning_rate": 9.476923076923079e-06,
|
| 181 |
-
"loss": 0.3475,
|
| 182 |
-
"step": 104
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 2.06,
|
| 186 |
-
"learning_rate": 9.353846153846155e-06,
|
| 187 |
-
"loss": 0.321,
|
| 188 |
-
"step": 108
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 2.07,
|
| 192 |
-
"learning_rate": 9.230769230769232e-06,
|
| 193 |
-
"loss": 0.2859,
|
| 194 |
-
"step": 112
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 2.08,
|
| 198 |
-
"learning_rate": 9.107692307692308e-06,
|
| 199 |
-
"loss": 0.3191,
|
| 200 |
-
"step": 116
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 2.09,
|
| 204 |
-
"learning_rate": 8.984615384615386e-06,
|
| 205 |
-
"loss": 0.2753,
|
| 206 |
-
"step": 120
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 2.09,
|
| 210 |
-
"eval_loss": 0.5241264700889587,
|
| 211 |
-
"eval_runtime": 88.0526,
|
| 212 |
-
"eval_samples_per_second": 4.475,
|
| 213 |
-
"eval_steps_per_second": 0.148,
|
| 214 |
-
"eval_wer": 17.21321467755699,
|
| 215 |
-
"step": 120
|
| 216 |
-
},
|
| 217 |
-
{
|
| 218 |
-
"epoch": 3.0,
|
| 219 |
-
"learning_rate": 8.861538461538463e-06,
|
| 220 |
-
"loss": 0.3104,
|
| 221 |
-
"step": 124
|
| 222 |
-
},
|
| 223 |
-
{
|
| 224 |
-
"epoch": 3.01,
|
| 225 |
-
"learning_rate": 8.73846153846154e-06,
|
| 226 |
-
"loss": 0.2734,
|
| 227 |
-
"step": 128
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 3.02,
|
| 231 |
-
"learning_rate": 8.615384615384617e-06,
|
| 232 |
-
"loss": 0.2608,
|
| 233 |
-
"step": 132
|
| 234 |
-
},
|
| 235 |
-
{
|
| 236 |
-
"epoch": 3.03,
|
| 237 |
-
"learning_rate": 8.492307692307693e-06,
|
| 238 |
-
"loss": 0.2509,
|
| 239 |
-
"step": 136
|
| 240 |
-
},
|
| 241 |
-
{
|
| 242 |
-
"epoch": 3.04,
|
| 243 |
-
"learning_rate": 8.36923076923077e-06,
|
| 244 |
-
"loss": 0.2548,
|
| 245 |
-
"step": 140
|
| 246 |
-
},
|
| 247 |
-
{
|
| 248 |
-
"epoch": 3.05,
|
| 249 |
-
"learning_rate": 8.246153846153848e-06,
|
| 250 |
-
"loss": 0.2469,
|
| 251 |
-
"step": 144
|
| 252 |
-
},
|
| 253 |
-
{
|
| 254 |
-
"epoch": 3.06,
|
| 255 |
-
"learning_rate": 8.123076923076924e-06,
|
| 256 |
-
"loss": 0.2231,
|
| 257 |
-
"step": 148
|
| 258 |
-
},
|
| 259 |
-
{
|
| 260 |
-
"epoch": 3.07,
|
| 261 |
-
"learning_rate": 8.000000000000001e-06,
|
| 262 |
-
"loss": 0.2138,
|
| 263 |
-
"step": 152
|
| 264 |
-
},
|
| 265 |
-
{
|
| 266 |
-
"epoch": 3.08,
|
| 267 |
-
"learning_rate": 7.876923076923077e-06,
|
| 268 |
-
"loss": 0.2349,
|
| 269 |
-
"step": 156
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 3.09,
|
| 273 |
-
"learning_rate": 7.753846153846155e-06,
|
| 274 |
-
"loss": 0.2077,
|
| 275 |
-
"step": 160
|
| 276 |
-
},
|
| 277 |
-
{
|
| 278 |
-
"epoch": 3.09,
|
| 279 |
-
"eval_loss": 0.5241798758506775,
|
| 280 |
-
"eval_runtime": 88.5317,
|
| 281 |
-
"eval_samples_per_second": 4.45,
|
| 282 |
-
"eval_steps_per_second": 0.147,
|
| 283 |
-
"eval_wer": 17.26197732536877,
|
| 284 |
-
"step": 160
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 3.1,
|
| 288 |
-
"learning_rate": 7.630769230769232e-06,
|
| 289 |
-
"loss": 0.2322,
|
| 290 |
-
"step": 164
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 4.01,
|
| 294 |
-
"learning_rate": 7.507692307692308e-06,
|
| 295 |
-
"loss": 0.2036,
|
| 296 |
-
"step": 168
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 4.02,
|
| 300 |
-
"learning_rate": 7.384615384615386e-06,
|
| 301 |
-
"loss": 0.2058,
|
| 302 |
-
"step": 172
|
| 303 |
-
},
|
| 304 |
-
{
|
| 305 |
-
"epoch": 4.03,
|
| 306 |
-
"learning_rate": 7.261538461538462e-06,
|
| 307 |
-
"loss": 0.1797,
|
| 308 |
-
"step": 176
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 4.04,
|
| 312 |
-
"learning_rate": 7.1384615384615385e-06,
|
| 313 |
-
"loss": 0.186,
|
| 314 |
-
"step": 180
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 4.05,
|
| 318 |
-
"learning_rate": 7.015384615384616e-06,
|
| 319 |
-
"loss": 0.2035,
|
| 320 |
-
"step": 184
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 4.06,
|
| 324 |
-
"learning_rate": 6.892307692307693e-06,
|
| 325 |
-
"loss": 0.1794,
|
| 326 |
-
"step": 188
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 4.07,
|
| 330 |
-
"learning_rate": 6.76923076923077e-06,
|
| 331 |
-
"loss": 0.1589,
|
| 332 |
-
"step": 192
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 4.08,
|
| 336 |
-
"learning_rate": 6.646153846153846e-06,
|
| 337 |
-
"loss": 0.1879,
|
| 338 |
-
"step": 196
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 4.09,
|
| 342 |
-
"learning_rate": 6.523076923076923e-06,
|
| 343 |
-
"loss": 0.1636,
|
| 344 |
-
"step": 200
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 4.09,
|
| 348 |
-
"eval_loss": 0.5289868712425232,
|
| 349 |
-
"eval_runtime": 95.5188,
|
| 350 |
-
"eval_samples_per_second": 4.125,
|
| 351 |
-
"eval_steps_per_second": 0.136,
|
| 352 |
-
"eval_wer": 17.66426916981592,
|
| 353 |
-
"step": 200
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 4.1,
|
| 357 |
-
"learning_rate": 6.4000000000000006e-06,
|
| 358 |
-
"loss": 0.1767,
|
| 359 |
-
"step": 204
|
| 360 |
-
},
|
| 361 |
-
{
|
| 362 |
-
"epoch": 5.01,
|
| 363 |
-
"learning_rate": 6.276923076923077e-06,
|
| 364 |
-
"loss": 0.1657,
|
| 365 |
-
"step": 208
|
| 366 |
-
},
|
| 367 |
-
{
|
| 368 |
-
"epoch": 5.02,
|
| 369 |
-
"learning_rate": 6.153846153846155e-06,
|
| 370 |
-
"loss": 0.1607,
|
| 371 |
-
"step": 212
|
| 372 |
-
},
|
| 373 |
-
{
|
| 374 |
-
"epoch": 5.03,
|
| 375 |
-
"learning_rate": 6.030769230769231e-06,
|
| 376 |
-
"loss": 0.1458,
|
| 377 |
-
"step": 216
|
| 378 |
-
},
|
| 379 |
-
{
|
| 380 |
-
"epoch": 5.04,
|
| 381 |
-
"learning_rate": 5.907692307692308e-06,
|
| 382 |
-
"loss": 0.1541,
|
| 383 |
-
"step": 220
|
| 384 |
-
},
|
| 385 |
-
{
|
| 386 |
-
"epoch": 5.05,
|
| 387 |
-
"learning_rate": 5.784615384615385e-06,
|
| 388 |
-
"loss": 0.1494,
|
| 389 |
-
"step": 224
|
| 390 |
-
},
|
| 391 |
-
{
|
| 392 |
-
"epoch": 5.06,
|
| 393 |
-
"learning_rate": 5.661538461538462e-06,
|
| 394 |
-
"loss": 0.144,
|
| 395 |
-
"step": 228
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 5.07,
|
| 399 |
-
"learning_rate": 5.538461538461539e-06,
|
| 400 |
-
"loss": 0.1311,
|
| 401 |
-
"step": 232
|
| 402 |
-
},
|
| 403 |
-
{
|
| 404 |
-
"epoch": 5.08,
|
| 405 |
-
"learning_rate": 5.415384615384615e-06,
|
| 406 |
-
"loss": 0.1411,
|
| 407 |
-
"step": 236
|
| 408 |
-
},
|
| 409 |
-
{
|
| 410 |
-
"epoch": 5.09,
|
| 411 |
-
"learning_rate": 5.292307692307693e-06,
|
| 412 |
-
"loss": 0.1322,
|
| 413 |
-
"step": 240
|
| 414 |
-
},
|
| 415 |
-
{
|
| 416 |
-
"epoch": 5.09,
|
| 417 |
-
"eval_loss": 0.5350630283355713,
|
| 418 |
-
"eval_runtime": 92.5111,
|
| 419 |
-
"eval_samples_per_second": 4.259,
|
| 420 |
-
"eval_steps_per_second": 0.141,
|
| 421 |
-
"eval_wer": 18.2128489576984,
|
| 422 |
-
"step": 240
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 5.1,
|
| 426 |
-
"learning_rate": 5.16923076923077e-06,
|
| 427 |
-
"loss": 0.1436,
|
| 428 |
-
"step": 244
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 6.0,
|
| 432 |
-
"learning_rate": 5.046153846153846e-06,
|
| 433 |
-
"loss": 0.1375,
|
| 434 |
-
"step": 248
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 6.01,
|
| 438 |
-
"learning_rate": 4.923076923076924e-06,
|
| 439 |
-
"loss": 0.1361,
|
| 440 |
-
"step": 252
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 6.02,
|
| 444 |
-
"learning_rate": 4.800000000000001e-06,
|
| 445 |
-
"loss": 0.129,
|
| 446 |
-
"step": 256
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 6.03,
|
| 450 |
-
"learning_rate": 4.676923076923077e-06,
|
| 451 |
-
"loss": 0.1127,
|
| 452 |
-
"step": 260
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 6.04,
|
| 456 |
-
"learning_rate": 4.553846153846154e-06,
|
| 457 |
-
"loss": 0.1266,
|
| 458 |
-
"step": 264
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 6.05,
|
| 462 |
-
"learning_rate": 4.430769230769232e-06,
|
| 463 |
-
"loss": 0.1193,
|
| 464 |
-
"step": 268
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 6.06,
|
| 468 |
-
"learning_rate": 4.307692307692308e-06,
|
| 469 |
-
"loss": 0.1127,
|
| 470 |
-
"step": 272
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 6.07,
|
| 474 |
-
"learning_rate": 4.184615384615385e-06,
|
| 475 |
-
"loss": 0.1064,
|
| 476 |
-
"step": 276
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 6.08,
|
| 480 |
-
"learning_rate": 4.061538461538462e-06,
|
| 481 |
-
"loss": 0.123,
|
| 482 |
-
"step": 280
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 6.08,
|
| 486 |
-
"eval_loss": 0.5429388284683228,
|
| 487 |
-
"eval_runtime": 91.5818,
|
| 488 |
-
"eval_samples_per_second": 4.302,
|
| 489 |
-
"eval_steps_per_second": 0.142,
|
| 490 |
-
"eval_wer": 18.907716689016212,
|
| 491 |
-
"step": 280
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 6.09,
|
| 495 |
-
"learning_rate": 3.938461538461539e-06,
|
| 496 |
-
"loss": 0.1057,
|
| 497 |
-
"step": 284
|
| 498 |
-
},
|
| 499 |
-
{
|
| 500 |
-
"epoch": 7.0,
|
| 501 |
-
"learning_rate": 3.815384615384616e-06,
|
| 502 |
-
"loss": 0.1258,
|
| 503 |
-
"step": 288
|
| 504 |
-
},
|
| 505 |
-
{
|
| 506 |
-
"epoch": 7.01,
|
| 507 |
-
"learning_rate": 3.692307692307693e-06,
|
| 508 |
-
"loss": 0.1108,
|
| 509 |
-
"step": 292
|
| 510 |
-
},
|
| 511 |
-
{
|
| 512 |
-
"epoch": 7.02,
|
| 513 |
-
"learning_rate": 3.5692307692307692e-06,
|
| 514 |
-
"loss": 0.1115,
|
| 515 |
-
"step": 296
|
| 516 |
-
},
|
| 517 |
-
{
|
| 518 |
-
"epoch": 7.03,
|
| 519 |
-
"learning_rate": 3.4461538461538464e-06,
|
| 520 |
-
"loss": 0.0998,
|
| 521 |
-
"step": 300
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 7.04,
|
| 525 |
-
"learning_rate": 3.323076923076923e-06,
|
| 526 |
-
"loss": 0.1106,
|
| 527 |
-
"step": 304
|
| 528 |
-
},
|
| 529 |
-
{
|
| 530 |
-
"epoch": 7.05,
|
| 531 |
-
"learning_rate": 3.2000000000000003e-06,
|
| 532 |
-
"loss": 0.1045,
|
| 533 |
-
"step": 308
|
| 534 |
-
},
|
| 535 |
-
{
|
| 536 |
-
"epoch": 7.06,
|
| 537 |
-
"learning_rate": 3.0769230769230774e-06,
|
| 538 |
-
"loss": 0.0908,
|
| 539 |
-
"step": 312
|
| 540 |
-
},
|
| 541 |
-
{
|
| 542 |
-
"epoch": 7.07,
|
| 543 |
-
"learning_rate": 2.953846153846154e-06,
|
| 544 |
-
"loss": 0.0931,
|
| 545 |
-
"step": 316
|
| 546 |
-
},
|
| 547 |
-
{
|
| 548 |
-
"epoch": 7.08,
|
| 549 |
-
"learning_rate": 2.830769230769231e-06,
|
| 550 |
-
"loss": 0.1074,
|
| 551 |
-
"step": 320
|
| 552 |
-
},
|
| 553 |
-
{
|
| 554 |
-
"epoch": 7.08,
|
| 555 |
-
"eval_loss": 0.5500437021255493,
|
| 556 |
-
"eval_runtime": 104.0907,
|
| 557 |
-
"eval_samples_per_second": 3.785,
|
| 558 |
-
"eval_steps_per_second": 0.125,
|
| 559 |
-
"eval_wer": 19.054004632451544,
|
| 560 |
-
"step": 320
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 7.09,
|
| 564 |
-
"learning_rate": 2.7076923076923076e-06,
|
| 565 |
-
"loss": 0.0937,
|
| 566 |
-
"step": 324
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 7.1,
|
| 570 |
-
"learning_rate": 2.584615384615385e-06,
|
| 571 |
-
"loss": 0.1091,
|
| 572 |
-
"step": 328
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 8.01,
|
| 576 |
-
"learning_rate": 2.461538461538462e-06,
|
| 577 |
-
"loss": 0.0951,
|
| 578 |
-
"step": 332
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 8.02,
|
| 582 |
-
"learning_rate": 2.3384615384615387e-06,
|
| 583 |
-
"loss": 0.1003,
|
| 584 |
-
"step": 336
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 8.03,
|
| 588 |
-
"learning_rate": 2.215384615384616e-06,
|
| 589 |
-
"loss": 0.0836,
|
| 590 |
-
"step": 340
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 8.04,
|
| 594 |
-
"learning_rate": 2.0923076923076926e-06,
|
| 595 |
-
"loss": 0.0907,
|
| 596 |
-
"step": 344
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 8.05,
|
| 600 |
-
"learning_rate": 1.9692307692307693e-06,
|
| 601 |
-
"loss": 0.1013,
|
| 602 |
-
"step": 348
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 8.06,
|
| 606 |
-
"learning_rate": 1.8461538461538465e-06,
|
| 607 |
-
"loss": 0.0891,
|
| 608 |
-
"step": 352
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 8.07,
|
| 612 |
-
"learning_rate": 1.7230769230769232e-06,
|
| 613 |
-
"loss": 0.077,
|
| 614 |
-
"step": 356
|
| 615 |
-
},
|
| 616 |
-
{
|
| 617 |
-
"epoch": 8.08,
|
| 618 |
-
"learning_rate": 1.6000000000000001e-06,
|
| 619 |
-
"loss": 0.1007,
|
| 620 |
-
"step": 360
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 8.08,
|
| 624 |
-
"eval_loss": 0.5552565455436707,
|
| 625 |
-
"eval_runtime": 88.458,
|
| 626 |
-
"eval_samples_per_second": 4.454,
|
| 627 |
-
"eval_steps_per_second": 0.147,
|
| 628 |
-
"eval_wer": 19.310008533463368,
|
| 629 |
-
"step": 360
|
| 630 |
-
},
|
| 631 |
-
{
|
| 632 |
-
"epoch": 8.09,
|
| 633 |
-
"learning_rate": 1.476923076923077e-06,
|
| 634 |
-
"loss": 0.0849,
|
| 635 |
-
"step": 364
|
| 636 |
-
},
|
| 637 |
-
{
|
| 638 |
-
"epoch": 8.1,
|
| 639 |
-
"learning_rate": 1.3538461538461538e-06,
|
| 640 |
-
"loss": 0.0971,
|
| 641 |
-
"step": 368
|
| 642 |
-
},
|
| 643 |
-
{
|
| 644 |
-
"epoch": 9.01,
|
| 645 |
-
"learning_rate": 1.230769230769231e-06,
|
| 646 |
-
"loss": 0.0876,
|
| 647 |
-
"step": 372
|
| 648 |
-
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 9.02,
|
| 651 |
-
"learning_rate": 1.107692307692308e-06,
|
| 652 |
-
"loss": 0.0879,
|
| 653 |
-
"step": 376
|
| 654 |
-
},
|
| 655 |
-
{
|
| 656 |
-
"epoch": 9.03,
|
| 657 |
-
"learning_rate": 9.846153846153847e-07,
|
| 658 |
-
"loss": 0.0805,
|
| 659 |
-
"step": 380
|
| 660 |
-
},
|
| 661 |
-
{
|
| 662 |
-
"epoch": 9.04,
|
| 663 |
-
"learning_rate": 8.615384615384616e-07,
|
| 664 |
-
"loss": 0.0888,
|
| 665 |
-
"step": 384
|
| 666 |
-
},
|
| 667 |
-
{
|
| 668 |
-
"epoch": 9.05,
|
| 669 |
-
"learning_rate": 7.384615384615385e-07,
|
| 670 |
-
"loss": 0.0858,
|
| 671 |
-
"step": 388
|
| 672 |
-
},
|
| 673 |
-
{
|
| 674 |
-
"epoch": 9.06,
|
| 675 |
-
"learning_rate": 6.153846153846155e-07,
|
| 676 |
-
"loss": 0.0825,
|
| 677 |
-
"step": 392
|
| 678 |
-
},
|
| 679 |
-
{
|
| 680 |
-
"epoch": 9.07,
|
| 681 |
-
"learning_rate": 4.923076923076923e-07,
|
| 682 |
-
"loss": 0.0748,
|
| 683 |
-
"step": 396
|
| 684 |
-
},
|
| 685 |
-
{
|
| 686 |
-
"epoch": 9.08,
|
| 687 |
-
"learning_rate": 3.6923076923076927e-07,
|
| 688 |
-
"loss": 0.0876,
|
| 689 |
-
"step": 400
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 9.08,
|
| 693 |
-
"eval_loss": 0.5568162202835083,
|
| 694 |
-
"eval_runtime": 89.7223,
|
| 695 |
-
"eval_samples_per_second": 4.391,
|
| 696 |
-
"eval_steps_per_second": 0.145,
|
| 697 |
-
"eval_wer": 19.3465805193222,
|
| 698 |
-
"step": 400
|
| 699 |
-
}
|
| 700 |
-
],
|
| 701 |
-
"max_steps": 407,
|
| 702 |
-
"num_train_epochs": 9223372036854775807,
|
| 703 |
-
"total_flos": 6.2536891981824e+17,
|
| 704 |
-
"trial_name": null,
|
| 705 |
-
"trial_params": null
|
| 706 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-400/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:fcbca0d141969bcb1c3cd0ef5a009221139334753b899d88e4d5003bd23f4b5f
|
| 3 |
-
size 3579
|
|
|
|
|
|
|
|
|
|
|
|
huggingface_training.py
CHANGED
|
@@ -1,3 +1,5 @@
|
|
|
|
|
|
|
|
| 1 |
import os # used to create output directory
|
| 2 |
from dataclasses import dataclass # used to define data collator
|
| 3 |
from math import ceil # used to round up decimals
|
|
@@ -27,13 +29,13 @@ config = AutoConfig.from_pretrained(model_id)
|
|
| 27 |
model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id)
|
| 28 |
|
| 29 |
dataset_id = "google/fleurs"
|
| 30 |
-
dataset_language_code = "
|
| 31 |
dataset = load_dataset(dataset_id, dataset_language_code, streaming=True)
|
| 32 |
|
| 33 |
"""The first time you run this code, make sure everything works fine using a small sample and low number of training steps. Just uncomment the next cell and run it. One note: since the dataset is loaded in streaming mode, the instruction will not be executed immediately. Instead, the dataset will be subsampled only when data will be needed during training."""
|
| 34 |
|
| 35 |
-
|
| 36 |
-
test_script = False
|
| 37 |
|
| 38 |
## Sample dataset for testing
|
| 39 |
if test_script is True:
|
|
@@ -186,11 +188,11 @@ class ShuffleCallback(TrainerCallback):
|
|
| 186 |
In our specific case, we could skip this step since English transcription is the default behaviour. Still, this is how you would do if you were in a multilingual setting.
|
| 187 |
"""
|
| 188 |
|
| 189 |
-
processor.tokenizer.set_prefix_tokens(language="en", task="transcribe")
|
| 190 |
|
| 191 |
## If you wanted to transcribe in Swedish
|
| 192 |
## (Of course, you'd need a Swedish dataset)
|
| 193 |
-
|
| 194 |
|
| 195 |
## If you wanted to get an English transcription from Swedish audio
|
| 196 |
# processor.tokenizer.set_prefix_tokens(language="sv", task="translate")
|
|
@@ -245,10 +247,6 @@ wandb.login()
|
|
| 245 |
wandb.init(project="whisper-training-post")
|
| 246 |
report_to = "wandb"
|
| 247 |
|
| 248 |
-
# Define (and create, if missing) output directory
|
| 249 |
-
output_dir = "."
|
| 250 |
-
# os.makedirs(output_dir, exist_ok=True)
|
| 251 |
-
|
| 252 |
# Check if we have a GPU.
|
| 253 |
# In case, we will use mixed precision
|
| 254 |
# to reduce memory footprint with
|
|
@@ -257,23 +255,24 @@ device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
| 257 |
use_fp16 = (device == "cuda")
|
| 258 |
|
| 259 |
# Let's first define the batch sizes
|
| 260 |
-
#
|
| 261 |
train_bs = 4 if test_script is True else 64
|
| 262 |
eval_bs = 2 if test_script is True else 32
|
| 263 |
|
| 264 |
# Then we infer the number of steps
|
| 265 |
# TODO: how did I find it?
|
| 266 |
-
num_training_samples =
|
| 267 |
-
num_epochs =
|
| 268 |
max_steps_full_training = ceil(num_training_samples * num_epochs / train_bs)
|
| 269 |
max_steps = 2 if test_script is True else max_steps_full_training
|
| 270 |
|
| 271 |
# We don't want to evaluate too often since it slows down training a lot
|
|
|
|
| 272 |
eval_steps = 1 if test_script is True else int(max_steps / 10)
|
| 273 |
logging_steps = 1 if test_script is True else int(max_steps / 100)
|
| 274 |
|
| 275 |
training_args = Seq2SeqTrainingArguments(
|
| 276 |
-
output_dir=
|
| 277 |
do_train=True,
|
| 278 |
do_eval=True,
|
| 279 |
max_steps=max_steps,
|
|
@@ -293,7 +292,7 @@ training_args = Seq2SeqTrainingArguments(
|
|
| 293 |
predict_with_generate=True,
|
| 294 |
generation_num_beams=1,
|
| 295 |
# track experiment
|
| 296 |
-
report_to=report_to
|
| 297 |
)
|
| 298 |
|
| 299 |
"""Now we can provide the trainer with the model, tokenizer (important: use the one you set language and task to! In this example, it is `processor.tokenizer`), training arguments, datasets, data collator, callback, and the method to compute metrics during evaluation.
|
|
@@ -369,4 +368,6 @@ trainer.log_metrics("test", final_metrics)
|
|
| 369 |
trainer.save_metrics("test", final_metrics)
|
| 370 |
print(final_metrics)
|
| 371 |
|
|
|
|
|
|
|
| 372 |
trainer.push_to_hub()
|
|
|
|
| 1 |
+
""" Whisper training script using Hugging Face Transformers. """
|
| 2 |
+
|
| 3 |
import os # used to create output directory
|
| 4 |
from dataclasses import dataclass # used to define data collator
|
| 5 |
from math import ceil # used to round up decimals
|
|
|
|
| 29 |
model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id)
|
| 30 |
|
| 31 |
dataset_id = "google/fleurs"
|
| 32 |
+
dataset_language_code = "sv_se"
|
| 33 |
dataset = load_dataset(dataset_id, dataset_language_code, streaming=True)
|
| 34 |
|
| 35 |
"""The first time you run this code, make sure everything works fine using a small sample and low number of training steps. Just uncomment the next cell and run it. One note: since the dataset is loaded in streaming mode, the instruction will not be executed immediately. Instead, the dataset will be subsampled only when data will be needed during training."""
|
| 36 |
|
| 37 |
+
test_script = True
|
| 38 |
+
# test_script = False
|
| 39 |
|
| 40 |
## Sample dataset for testing
|
| 41 |
if test_script is True:
|
|
|
|
| 188 |
In our specific case, we could skip this step since English transcription is the default behaviour. Still, this is how you would do if you were in a multilingual setting.
|
| 189 |
"""
|
| 190 |
|
| 191 |
+
# processor.tokenizer.set_prefix_tokens(language="en", task="transcribe")
|
| 192 |
|
| 193 |
## If you wanted to transcribe in Swedish
|
| 194 |
## (Of course, you'd need a Swedish dataset)
|
| 195 |
+
processor.tokenizer.set_prefix_tokens(language="sv", task="transcribe")
|
| 196 |
|
| 197 |
## If you wanted to get an English transcription from Swedish audio
|
| 198 |
# processor.tokenizer.set_prefix_tokens(language="sv", task="translate")
|
|
|
|
| 247 |
wandb.init(project="whisper-training-post")
|
| 248 |
report_to = "wandb"
|
| 249 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 250 |
# Check if we have a GPU.
|
| 251 |
# In case, we will use mixed precision
|
| 252 |
# to reduce memory footprint with
|
|
|
|
| 255 |
use_fp16 = (device == "cuda")
|
| 256 |
|
| 257 |
# Let's first define the batch sizes
|
| 258 |
+
# Adapt it to your hardware
|
| 259 |
train_bs = 4 if test_script is True else 64
|
| 260 |
eval_bs = 2 if test_script is True else 32
|
| 261 |
|
| 262 |
# Then we infer the number of steps
|
| 263 |
# TODO: how did I find it?
|
| 264 |
+
num_training_samples = 2385
|
| 265 |
+
num_epochs = 3
|
| 266 |
max_steps_full_training = ceil(num_training_samples * num_epochs / train_bs)
|
| 267 |
max_steps = 2 if test_script is True else max_steps_full_training
|
| 268 |
|
| 269 |
# We don't want to evaluate too often since it slows down training a lot
|
| 270 |
+
# but neither too little, since we want to see how the model is training
|
| 271 |
eval_steps = 1 if test_script is True else int(max_steps / 10)
|
| 272 |
logging_steps = 1 if test_script is True else int(max_steps / 100)
|
| 273 |
|
| 274 |
training_args = Seq2SeqTrainingArguments(
|
| 275 |
+
output_dir=".",
|
| 276 |
do_train=True,
|
| 277 |
do_eval=True,
|
| 278 |
max_steps=max_steps,
|
|
|
|
| 292 |
predict_with_generate=True,
|
| 293 |
generation_num_beams=1,
|
| 294 |
# track experiment
|
| 295 |
+
report_to=report_to
|
| 296 |
)
|
| 297 |
|
| 298 |
"""Now we can provide the trainer with the model, tokenizer (important: use the one you set language and task to! In this example, it is `processor.tokenizer`), training arguments, datasets, data collator, callback, and the method to compute metrics during evaluation.
|
|
|
|
| 368 |
trainer.save_metrics("test", final_metrics)
|
| 369 |
print(final_metrics)
|
| 370 |
|
| 371 |
+
# Pushing to hub during training slows down training
|
| 372 |
+
# so we push it only in the end.
|
| 373 |
trainer.push_to_hub()
|