Upload folder using huggingface_hub
Browse files- chinese/checkpoint-7322/config.json +65 -0
- chinese/checkpoint-7322/generation_config.json +260 -0
- chinese/checkpoint-7322/model-00001-of-00002.safetensors +3 -0
- chinese/checkpoint-7322/model-00002-of-00002.safetensors +3 -0
- chinese/checkpoint-7322/model.safetensors.index.json +0 -0
- chinese/checkpoint-7322/optimizer.pt +3 -0
- chinese/checkpoint-7322/preprocessor_config.json +14 -0
- chinese/checkpoint-7322/rng_state.pth +3 -0
- chinese/checkpoint-7322/scheduler.pt +3 -0
- chinese/checkpoint-7322/trainer_state.json +1791 -0
- chinese/checkpoint-7322/training_args.bin +3 -0
- pinyin/config.json +61 -0
- pinyin/generation_config.json +260 -0
- pinyin/model-00001-of-00002.safetensors +3 -0
- pinyin/model-00002-of-00002.safetensors +3 -0
- pinyin/model.safetensors.index.json +0 -0
- pinyin/optimizer.pt +3 -0
- pinyin/preprocessor_config.json +14 -0
- pinyin/rng_state.pth +3 -0
- pinyin/scheduler.pt +3 -0
- pinyin/trainer_state.json +1122 -0
- pinyin/training_args.bin +3 -0
chinese/checkpoint-7322/config.json
ADDED
|
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "openai/whisper-large",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "gelu",
|
| 5 |
+
"apply_spec_augment": false,
|
| 6 |
+
"architectures": [
|
| 7 |
+
"WhisperForConditionalGeneration"
|
| 8 |
+
],
|
| 9 |
+
"attention_dropout": 0.0,
|
| 10 |
+
"begin_suppress_tokens": [
|
| 11 |
+
220,
|
| 12 |
+
50257
|
| 13 |
+
],
|
| 14 |
+
"bos_token_id": 50257,
|
| 15 |
+
"classifier_proj_size": 256,
|
| 16 |
+
"d_model": 1280,
|
| 17 |
+
"decoder_attention_heads": 20,
|
| 18 |
+
"decoder_ffn_dim": 5120,
|
| 19 |
+
"decoder_layerdrop": 0.0,
|
| 20 |
+
"decoder_layers": 32,
|
| 21 |
+
"decoder_start_token_id": 50258,
|
| 22 |
+
"dropout": 0.0,
|
| 23 |
+
"encoder_attention_heads": 20,
|
| 24 |
+
"encoder_ffn_dim": 5120,
|
| 25 |
+
"encoder_layerdrop": 0.0,
|
| 26 |
+
"encoder_layers": 32,
|
| 27 |
+
"eos_token_id": 50257,
|
| 28 |
+
"forced_decoder_ids": [
|
| 29 |
+
[
|
| 30 |
+
1,
|
| 31 |
+
50260
|
| 32 |
+
],
|
| 33 |
+
[
|
| 34 |
+
2,
|
| 35 |
+
50359
|
| 36 |
+
],
|
| 37 |
+
[
|
| 38 |
+
3,
|
| 39 |
+
50363
|
| 40 |
+
]
|
| 41 |
+
],
|
| 42 |
+
"init_std": 0.02,
|
| 43 |
+
"is_encoder_decoder": true,
|
| 44 |
+
"mask_feature_length": 10,
|
| 45 |
+
"mask_feature_min_masks": 0,
|
| 46 |
+
"mask_feature_prob": 0.0,
|
| 47 |
+
"mask_time_length": 10,
|
| 48 |
+
"mask_time_min_masks": 2,
|
| 49 |
+
"mask_time_prob": 0.05,
|
| 50 |
+
"max_length": 448,
|
| 51 |
+
"max_source_positions": 1500,
|
| 52 |
+
"max_target_positions": 448,
|
| 53 |
+
"median_filter_width": 7,
|
| 54 |
+
"model_type": "whisper",
|
| 55 |
+
"num_hidden_layers": 32,
|
| 56 |
+
"num_mel_bins": 80,
|
| 57 |
+
"pad_token_id": 50257,
|
| 58 |
+
"scale_embedding": false,
|
| 59 |
+
"suppress_tokens": [],
|
| 60 |
+
"torch_dtype": "float32",
|
| 61 |
+
"transformers_version": "4.36.2",
|
| 62 |
+
"use_cache": true,
|
| 63 |
+
"use_weighted_layer_sum": false,
|
| 64 |
+
"vocab_size": 51865
|
| 65 |
+
}
|
chinese/checkpoint-7322/generation_config.json
ADDED
|
@@ -0,0 +1,260 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alignment_heads": [
|
| 3 |
+
[
|
| 4 |
+
9,
|
| 5 |
+
19
|
| 6 |
+
],
|
| 7 |
+
[
|
| 8 |
+
11,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
[
|
| 12 |
+
11,
|
| 13 |
+
4
|
| 14 |
+
],
|
| 15 |
+
[
|
| 16 |
+
11,
|
| 17 |
+
17
|
| 18 |
+
],
|
| 19 |
+
[
|
| 20 |
+
22,
|
| 21 |
+
7
|
| 22 |
+
],
|
| 23 |
+
[
|
| 24 |
+
22,
|
| 25 |
+
11
|
| 26 |
+
],
|
| 27 |
+
[
|
| 28 |
+
22,
|
| 29 |
+
17
|
| 30 |
+
],
|
| 31 |
+
[
|
| 32 |
+
23,
|
| 33 |
+
2
|
| 34 |
+
],
|
| 35 |
+
[
|
| 36 |
+
23,
|
| 37 |
+
15
|
| 38 |
+
]
|
| 39 |
+
],
|
| 40 |
+
"begin_suppress_tokens": [
|
| 41 |
+
220,
|
| 42 |
+
50257
|
| 43 |
+
],
|
| 44 |
+
"bos_token_id": 50257,
|
| 45 |
+
"decoder_start_token_id": 50258,
|
| 46 |
+
"eos_token_id": 50257,
|
| 47 |
+
"forced_decoder_ids": [
|
| 48 |
+
[
|
| 49 |
+
1,
|
| 50 |
+
null
|
| 51 |
+
],
|
| 52 |
+
[
|
| 53 |
+
2,
|
| 54 |
+
50359
|
| 55 |
+
]
|
| 56 |
+
],
|
| 57 |
+
"is_multilingual": true,
|
| 58 |
+
"lang_to_id": {
|
| 59 |
+
"<|af|>": 50327,
|
| 60 |
+
"<|am|>": 50334,
|
| 61 |
+
"<|ar|>": 50272,
|
| 62 |
+
"<|as|>": 50350,
|
| 63 |
+
"<|az|>": 50304,
|
| 64 |
+
"<|ba|>": 50355,
|
| 65 |
+
"<|be|>": 50330,
|
| 66 |
+
"<|bg|>": 50292,
|
| 67 |
+
"<|bn|>": 50302,
|
| 68 |
+
"<|bo|>": 50347,
|
| 69 |
+
"<|br|>": 50309,
|
| 70 |
+
"<|bs|>": 50315,
|
| 71 |
+
"<|ca|>": 50270,
|
| 72 |
+
"<|cs|>": 50283,
|
| 73 |
+
"<|cy|>": 50297,
|
| 74 |
+
"<|da|>": 50285,
|
| 75 |
+
"<|de|>": 50261,
|
| 76 |
+
"<|el|>": 50281,
|
| 77 |
+
"<|en|>": 50259,
|
| 78 |
+
"<|es|>": 50262,
|
| 79 |
+
"<|et|>": 50307,
|
| 80 |
+
"<|eu|>": 50310,
|
| 81 |
+
"<|fa|>": 50300,
|
| 82 |
+
"<|fi|>": 50277,
|
| 83 |
+
"<|fo|>": 50338,
|
| 84 |
+
"<|fr|>": 50265,
|
| 85 |
+
"<|gl|>": 50319,
|
| 86 |
+
"<|gu|>": 50333,
|
| 87 |
+
"<|haw|>": 50352,
|
| 88 |
+
"<|ha|>": 50354,
|
| 89 |
+
"<|he|>": 50279,
|
| 90 |
+
"<|hi|>": 50276,
|
| 91 |
+
"<|hr|>": 50291,
|
| 92 |
+
"<|ht|>": 50339,
|
| 93 |
+
"<|hu|>": 50286,
|
| 94 |
+
"<|hy|>": 50312,
|
| 95 |
+
"<|id|>": 50275,
|
| 96 |
+
"<|is|>": 50311,
|
| 97 |
+
"<|it|>": 50274,
|
| 98 |
+
"<|ja|>": 50266,
|
| 99 |
+
"<|jw|>": 50356,
|
| 100 |
+
"<|ka|>": 50329,
|
| 101 |
+
"<|kk|>": 50316,
|
| 102 |
+
"<|km|>": 50323,
|
| 103 |
+
"<|kn|>": 50306,
|
| 104 |
+
"<|ko|>": 50264,
|
| 105 |
+
"<|la|>": 50294,
|
| 106 |
+
"<|lb|>": 50345,
|
| 107 |
+
"<|ln|>": 50353,
|
| 108 |
+
"<|lo|>": 50336,
|
| 109 |
+
"<|lt|>": 50293,
|
| 110 |
+
"<|lv|>": 50301,
|
| 111 |
+
"<|mg|>": 50349,
|
| 112 |
+
"<|mi|>": 50295,
|
| 113 |
+
"<|mk|>": 50308,
|
| 114 |
+
"<|ml|>": 50296,
|
| 115 |
+
"<|mn|>": 50314,
|
| 116 |
+
"<|mr|>": 50320,
|
| 117 |
+
"<|ms|>": 50282,
|
| 118 |
+
"<|mt|>": 50343,
|
| 119 |
+
"<|my|>": 50346,
|
| 120 |
+
"<|ne|>": 50313,
|
| 121 |
+
"<|nl|>": 50271,
|
| 122 |
+
"<|nn|>": 50342,
|
| 123 |
+
"<|no|>": 50288,
|
| 124 |
+
"<|oc|>": 50328,
|
| 125 |
+
"<|pa|>": 50321,
|
| 126 |
+
"<|pl|>": 50269,
|
| 127 |
+
"<|ps|>": 50340,
|
| 128 |
+
"<|pt|>": 50267,
|
| 129 |
+
"<|ro|>": 50284,
|
| 130 |
+
"<|ru|>": 50263,
|
| 131 |
+
"<|sa|>": 50344,
|
| 132 |
+
"<|sd|>": 50332,
|
| 133 |
+
"<|si|>": 50322,
|
| 134 |
+
"<|sk|>": 50298,
|
| 135 |
+
"<|sl|>": 50305,
|
| 136 |
+
"<|sn|>": 50324,
|
| 137 |
+
"<|so|>": 50326,
|
| 138 |
+
"<|sq|>": 50317,
|
| 139 |
+
"<|sr|>": 50303,
|
| 140 |
+
"<|su|>": 50357,
|
| 141 |
+
"<|sv|>": 50273,
|
| 142 |
+
"<|sw|>": 50318,
|
| 143 |
+
"<|ta|>": 50287,
|
| 144 |
+
"<|te|>": 50299,
|
| 145 |
+
"<|tg|>": 50331,
|
| 146 |
+
"<|th|>": 50289,
|
| 147 |
+
"<|tk|>": 50341,
|
| 148 |
+
"<|tl|>": 50348,
|
| 149 |
+
"<|tr|>": 50268,
|
| 150 |
+
"<|tt|>": 50351,
|
| 151 |
+
"<|uk|>": 50280,
|
| 152 |
+
"<|ur|>": 50290,
|
| 153 |
+
"<|uz|>": 50337,
|
| 154 |
+
"<|vi|>": 50278,
|
| 155 |
+
"<|yi|>": 50335,
|
| 156 |
+
"<|yo|>": 50325,
|
| 157 |
+
"<|zh|>": 50260
|
| 158 |
+
},
|
| 159 |
+
"max_initial_timestamp_index": 50,
|
| 160 |
+
"max_length": 448,
|
| 161 |
+
"no_timestamps_token_id": 50363,
|
| 162 |
+
"pad_token_id": 50257,
|
| 163 |
+
"prev_sot_token_id": 50361,
|
| 164 |
+
"return_timestamps": false,
|
| 165 |
+
"suppress_tokens": [
|
| 166 |
+
1,
|
| 167 |
+
2,
|
| 168 |
+
7,
|
| 169 |
+
8,
|
| 170 |
+
9,
|
| 171 |
+
10,
|
| 172 |
+
14,
|
| 173 |
+
25,
|
| 174 |
+
26,
|
| 175 |
+
27,
|
| 176 |
+
28,
|
| 177 |
+
29,
|
| 178 |
+
31,
|
| 179 |
+
58,
|
| 180 |
+
59,
|
| 181 |
+
60,
|
| 182 |
+
61,
|
| 183 |
+
62,
|
| 184 |
+
63,
|
| 185 |
+
90,
|
| 186 |
+
91,
|
| 187 |
+
92,
|
| 188 |
+
93,
|
| 189 |
+
359,
|
| 190 |
+
503,
|
| 191 |
+
522,
|
| 192 |
+
542,
|
| 193 |
+
873,
|
| 194 |
+
893,
|
| 195 |
+
902,
|
| 196 |
+
918,
|
| 197 |
+
922,
|
| 198 |
+
931,
|
| 199 |
+
1350,
|
| 200 |
+
1853,
|
| 201 |
+
1982,
|
| 202 |
+
2460,
|
| 203 |
+
2627,
|
| 204 |
+
3246,
|
| 205 |
+
3253,
|
| 206 |
+
3268,
|
| 207 |
+
3536,
|
| 208 |
+
3846,
|
| 209 |
+
3961,
|
| 210 |
+
4183,
|
| 211 |
+
4667,
|
| 212 |
+
6585,
|
| 213 |
+
6647,
|
| 214 |
+
7273,
|
| 215 |
+
9061,
|
| 216 |
+
9383,
|
| 217 |
+
10428,
|
| 218 |
+
10929,
|
| 219 |
+
11938,
|
| 220 |
+
12033,
|
| 221 |
+
12331,
|
| 222 |
+
12562,
|
| 223 |
+
13793,
|
| 224 |
+
14157,
|
| 225 |
+
14635,
|
| 226 |
+
15265,
|
| 227 |
+
15618,
|
| 228 |
+
16553,
|
| 229 |
+
16604,
|
| 230 |
+
18362,
|
| 231 |
+
18956,
|
| 232 |
+
20075,
|
| 233 |
+
21675,
|
| 234 |
+
22520,
|
| 235 |
+
26130,
|
| 236 |
+
26161,
|
| 237 |
+
26435,
|
| 238 |
+
28279,
|
| 239 |
+
29464,
|
| 240 |
+
31650,
|
| 241 |
+
32302,
|
| 242 |
+
32470,
|
| 243 |
+
36865,
|
| 244 |
+
42863,
|
| 245 |
+
47425,
|
| 246 |
+
49870,
|
| 247 |
+
50254,
|
| 248 |
+
50258,
|
| 249 |
+
50358,
|
| 250 |
+
50359,
|
| 251 |
+
50360,
|
| 252 |
+
50361,
|
| 253 |
+
50362
|
| 254 |
+
],
|
| 255 |
+
"task_to_id": {
|
| 256 |
+
"transcribe": 50359,
|
| 257 |
+
"translate": 50358
|
| 258 |
+
},
|
| 259 |
+
"transformers_version": "4.36.2"
|
| 260 |
+
}
|
chinese/checkpoint-7322/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e895867a140110c3814cae7b5c3a2c9cd76ec13b03b4732d263f11af967691b3
|
| 3 |
+
size 4992706480
|
chinese/checkpoint-7322/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a8a5f8ee63c3abeb986ff88234e489e00c1e734cf03394e2fb039b71da7a236
|
| 3 |
+
size 1180663192
|
chinese/checkpoint-7322/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
chinese/checkpoint-7322/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7aa130efd22e3a33a0c4dc0f1f206933b77e9cc604c74def8ea6b264678b87f4
|
| 3 |
+
size 12332175139
|
chinese/checkpoint-7322/preprocessor_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chunk_length": 30,
|
| 3 |
+
"feature_extractor_type": "WhisperFeatureExtractor",
|
| 4 |
+
"feature_size": 80,
|
| 5 |
+
"hop_length": 160,
|
| 6 |
+
"n_fft": 400,
|
| 7 |
+
"n_samples": 480000,
|
| 8 |
+
"nb_max_frames": 3000,
|
| 9 |
+
"padding_side": "right",
|
| 10 |
+
"padding_value": 0.0,
|
| 11 |
+
"processor_class": "WhisperProcessor",
|
| 12 |
+
"return_attention_mask": false,
|
| 13 |
+
"sampling_rate": 16000
|
| 14 |
+
}
|
chinese/checkpoint-7322/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1003140837975a56aeed1034c9f0994376ae022772ef48cdf3175f3cab27d63e
|
| 3 |
+
size 14575
|
chinese/checkpoint-7322/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:281f84621e9b6a006c36a55b8e933ce68619f2227ec31941dfdc8b8f5d4ed02f
|
| 3 |
+
size 627
|
chinese/checkpoint-7322/trainer_state.json
ADDED
|
@@ -0,0 +1,1791 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 56.830823497490165,
|
| 3 |
+
"best_model_checkpoint": "/app/data/hieunq/whisper_asr_train/whisper-large-chinese/checkpoint-7322",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 7322,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01,
|
| 13 |
+
"learning_rate": 6.1715596831466814e-06,
|
| 14 |
+
"loss": 1.2908,
|
| 15 |
+
"step": 25
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 6.163119366293362e-06,
|
| 20 |
+
"loss": 0.4083,
|
| 21 |
+
"step": 50
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.02,
|
| 25 |
+
"learning_rate": 6.154679049440044e-06,
|
| 26 |
+
"loss": 0.374,
|
| 27 |
+
"step": 75
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.03,
|
| 31 |
+
"learning_rate": 6.146238732586725e-06,
|
| 32 |
+
"loss": 0.3357,
|
| 33 |
+
"step": 100
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.03,
|
| 37 |
+
"learning_rate": 6.137798415733407e-06,
|
| 38 |
+
"loss": 0.3155,
|
| 39 |
+
"step": 125
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.04,
|
| 43 |
+
"learning_rate": 6.129358098880087e-06,
|
| 44 |
+
"loss": 0.3666,
|
| 45 |
+
"step": 150
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.05,
|
| 49 |
+
"learning_rate": 6.120917782026768e-06,
|
| 50 |
+
"loss": 0.3536,
|
| 51 |
+
"step": 175
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.05,
|
| 55 |
+
"learning_rate": 6.11247746517345e-06,
|
| 56 |
+
"loss": 0.3199,
|
| 57 |
+
"step": 200
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.06,
|
| 61 |
+
"learning_rate": 6.104037148320131e-06,
|
| 62 |
+
"loss": 0.319,
|
| 63 |
+
"step": 225
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.07,
|
| 67 |
+
"learning_rate": 6.095596831466812e-06,
|
| 68 |
+
"loss": 0.3515,
|
| 69 |
+
"step": 250
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.08,
|
| 73 |
+
"learning_rate": 6.087156514613494e-06,
|
| 74 |
+
"loss": 0.307,
|
| 75 |
+
"step": 275
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.08,
|
| 79 |
+
"learning_rate": 6.078716197760175e-06,
|
| 80 |
+
"loss": 0.3225,
|
| 81 |
+
"step": 300
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.09,
|
| 85 |
+
"learning_rate": 6.070275880906856e-06,
|
| 86 |
+
"loss": 0.3027,
|
| 87 |
+
"step": 325
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.1,
|
| 91 |
+
"learning_rate": 6.061835564053537e-06,
|
| 92 |
+
"loss": 0.3104,
|
| 93 |
+
"step": 350
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1,
|
| 97 |
+
"learning_rate": 6.053395247200219e-06,
|
| 98 |
+
"loss": 0.2731,
|
| 99 |
+
"step": 375
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.11,
|
| 103 |
+
"learning_rate": 6.0449549303469e-06,
|
| 104 |
+
"loss": 0.2996,
|
| 105 |
+
"step": 400
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.12,
|
| 109 |
+
"learning_rate": 6.0365146134935814e-06,
|
| 110 |
+
"loss": 0.3431,
|
| 111 |
+
"step": 425
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.12,
|
| 115 |
+
"learning_rate": 6.028074296640262e-06,
|
| 116 |
+
"loss": 0.3011,
|
| 117 |
+
"step": 450
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.13,
|
| 121 |
+
"learning_rate": 6.019633979786943e-06,
|
| 122 |
+
"loss": 0.3444,
|
| 123 |
+
"step": 475
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.14,
|
| 127 |
+
"learning_rate": 6.0111936629336245e-06,
|
| 128 |
+
"loss": 0.3152,
|
| 129 |
+
"step": 500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.14,
|
| 133 |
+
"learning_rate": 6.002753346080307e-06,
|
| 134 |
+
"loss": 0.3099,
|
| 135 |
+
"step": 525
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.15,
|
| 139 |
+
"learning_rate": 5.994313029226987e-06,
|
| 140 |
+
"loss": 0.2888,
|
| 141 |
+
"step": 550
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.16,
|
| 145 |
+
"learning_rate": 5.9858727123736684e-06,
|
| 146 |
+
"loss": 0.3475,
|
| 147 |
+
"step": 575
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.16,
|
| 151 |
+
"learning_rate": 5.97743239552035e-06,
|
| 152 |
+
"loss": 0.3177,
|
| 153 |
+
"step": 600
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.17,
|
| 157 |
+
"learning_rate": 5.968992078667031e-06,
|
| 158 |
+
"loss": 0.3131,
|
| 159 |
+
"step": 625
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.18,
|
| 163 |
+
"learning_rate": 5.9605517618137115e-06,
|
| 164 |
+
"loss": 0.285,
|
| 165 |
+
"step": 650
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.18,
|
| 169 |
+
"learning_rate": 5.952111444960394e-06,
|
| 170 |
+
"loss": 0.2687,
|
| 171 |
+
"step": 675
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.19,
|
| 175 |
+
"learning_rate": 5.943671128107075e-06,
|
| 176 |
+
"loss": 0.2852,
|
| 177 |
+
"step": 700
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.2,
|
| 181 |
+
"learning_rate": 5.935230811253756e-06,
|
| 182 |
+
"loss": 0.2886,
|
| 183 |
+
"step": 725
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.2,
|
| 187 |
+
"learning_rate": 5.926790494400437e-06,
|
| 188 |
+
"loss": 0.2863,
|
| 189 |
+
"step": 750
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.21,
|
| 193 |
+
"learning_rate": 5.918350177547118e-06,
|
| 194 |
+
"loss": 0.3231,
|
| 195 |
+
"step": 775
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.22,
|
| 199 |
+
"learning_rate": 5.909909860693799e-06,
|
| 200 |
+
"loss": 0.2638,
|
| 201 |
+
"step": 800
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.23,
|
| 205 |
+
"learning_rate": 5.9014695438404815e-06,
|
| 206 |
+
"loss": 0.2892,
|
| 207 |
+
"step": 825
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.23,
|
| 211 |
+
"learning_rate": 5.893029226987162e-06,
|
| 212 |
+
"loss": 0.2969,
|
| 213 |
+
"step": 850
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.24,
|
| 217 |
+
"learning_rate": 5.884588910133843e-06,
|
| 218 |
+
"loss": 0.3163,
|
| 219 |
+
"step": 875
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.25,
|
| 223 |
+
"learning_rate": 5.8761485932805245e-06,
|
| 224 |
+
"loss": 0.2862,
|
| 225 |
+
"step": 900
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.25,
|
| 229 |
+
"learning_rate": 5.867708276427206e-06,
|
| 230 |
+
"loss": 0.2791,
|
| 231 |
+
"step": 925
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.26,
|
| 235 |
+
"learning_rate": 5.859267959573887e-06,
|
| 236 |
+
"loss": 0.2662,
|
| 237 |
+
"step": 950
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.27,
|
| 241 |
+
"learning_rate": 5.8508276427205685e-06,
|
| 242 |
+
"loss": 0.3065,
|
| 243 |
+
"step": 975
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.27,
|
| 247 |
+
"learning_rate": 5.84238732586725e-06,
|
| 248 |
+
"loss": 0.2954,
|
| 249 |
+
"step": 1000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.28,
|
| 253 |
+
"learning_rate": 5.833947009013931e-06,
|
| 254 |
+
"loss": 0.3087,
|
| 255 |
+
"step": 1025
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.29,
|
| 259 |
+
"learning_rate": 5.8255066921606115e-06,
|
| 260 |
+
"loss": 0.2659,
|
| 261 |
+
"step": 1050
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.29,
|
| 265 |
+
"learning_rate": 5.817066375307293e-06,
|
| 266 |
+
"loss": 0.2946,
|
| 267 |
+
"step": 1075
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.3,
|
| 271 |
+
"learning_rate": 5.808626058453974e-06,
|
| 272 |
+
"loss": 0.2661,
|
| 273 |
+
"step": 1100
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.31,
|
| 277 |
+
"learning_rate": 5.800185741600656e-06,
|
| 278 |
+
"loss": 0.2881,
|
| 279 |
+
"step": 1125
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.31,
|
| 283 |
+
"learning_rate": 5.791745424747337e-06,
|
| 284 |
+
"loss": 0.2819,
|
| 285 |
+
"step": 1150
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.32,
|
| 289 |
+
"learning_rate": 5.783305107894018e-06,
|
| 290 |
+
"loss": 0.32,
|
| 291 |
+
"step": 1175
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.33,
|
| 295 |
+
"learning_rate": 5.774864791040699e-06,
|
| 296 |
+
"loss": 0.3108,
|
| 297 |
+
"step": 1200
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.33,
|
| 301 |
+
"learning_rate": 5.766424474187381e-06,
|
| 302 |
+
"loss": 0.2866,
|
| 303 |
+
"step": 1225
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.34,
|
| 307 |
+
"learning_rate": 5.757984157334062e-06,
|
| 308 |
+
"loss": 0.3029,
|
| 309 |
+
"step": 1250
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.35,
|
| 313 |
+
"learning_rate": 5.749543840480743e-06,
|
| 314 |
+
"loss": 0.2577,
|
| 315 |
+
"step": 1275
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.36,
|
| 319 |
+
"learning_rate": 5.7411035236274246e-06,
|
| 320 |
+
"loss": 0.2835,
|
| 321 |
+
"step": 1300
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.36,
|
| 325 |
+
"learning_rate": 5.732663206774106e-06,
|
| 326 |
+
"loss": 0.2623,
|
| 327 |
+
"step": 1325
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.37,
|
| 331 |
+
"learning_rate": 5.724222889920786e-06,
|
| 332 |
+
"loss": 0.3214,
|
| 333 |
+
"step": 1350
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.38,
|
| 337 |
+
"learning_rate": 5.715782573067468e-06,
|
| 338 |
+
"loss": 0.2773,
|
| 339 |
+
"step": 1375
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.38,
|
| 343 |
+
"learning_rate": 5.707342256214149e-06,
|
| 344 |
+
"loss": 0.2758,
|
| 345 |
+
"step": 1400
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.39,
|
| 349 |
+
"learning_rate": 5.698901939360831e-06,
|
| 350 |
+
"loss": 0.2813,
|
| 351 |
+
"step": 1425
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.4,
|
| 355 |
+
"learning_rate": 5.6904616225075115e-06,
|
| 356 |
+
"loss": 0.2874,
|
| 357 |
+
"step": 1450
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.4,
|
| 361 |
+
"learning_rate": 5.682021305654193e-06,
|
| 362 |
+
"loss": 0.261,
|
| 363 |
+
"step": 1475
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.41,
|
| 367 |
+
"learning_rate": 5.673580988800874e-06,
|
| 368 |
+
"loss": 0.2798,
|
| 369 |
+
"step": 1500
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.42,
|
| 373 |
+
"learning_rate": 5.6651406719475555e-06,
|
| 374 |
+
"loss": 0.2965,
|
| 375 |
+
"step": 1525
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.42,
|
| 379 |
+
"learning_rate": 5.656700355094237e-06,
|
| 380 |
+
"loss": 0.2431,
|
| 381 |
+
"step": 1550
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.43,
|
| 385 |
+
"learning_rate": 5.648260038240918e-06,
|
| 386 |
+
"loss": 0.2916,
|
| 387 |
+
"step": 1575
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.44,
|
| 391 |
+
"learning_rate": 5.639819721387599e-06,
|
| 392 |
+
"loss": 0.3695,
|
| 393 |
+
"step": 1600
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.44,
|
| 397 |
+
"learning_rate": 5.631379404534281e-06,
|
| 398 |
+
"loss": 0.2764,
|
| 399 |
+
"step": 1625
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.45,
|
| 403 |
+
"learning_rate": 5.622939087680961e-06,
|
| 404 |
+
"loss": 0.2653,
|
| 405 |
+
"step": 1650
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.46,
|
| 409 |
+
"learning_rate": 5.6144987708276424e-06,
|
| 410 |
+
"loss": 0.2866,
|
| 411 |
+
"step": 1675
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.46,
|
| 415 |
+
"learning_rate": 5.606058453974325e-06,
|
| 416 |
+
"loss": 0.3056,
|
| 417 |
+
"step": 1700
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.47,
|
| 421 |
+
"learning_rate": 5.597618137121006e-06,
|
| 422 |
+
"loss": 0.238,
|
| 423 |
+
"step": 1725
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.48,
|
| 427 |
+
"learning_rate": 5.589177820267686e-06,
|
| 428 |
+
"loss": 0.2842,
|
| 429 |
+
"step": 1750
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.48,
|
| 433 |
+
"learning_rate": 5.580737503414368e-06,
|
| 434 |
+
"loss": 0.2917,
|
| 435 |
+
"step": 1775
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.49,
|
| 439 |
+
"learning_rate": 5.572297186561049e-06,
|
| 440 |
+
"loss": 0.2697,
|
| 441 |
+
"step": 1800
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.5,
|
| 445 |
+
"learning_rate": 5.56385686970773e-06,
|
| 446 |
+
"loss": 0.2741,
|
| 447 |
+
"step": 1825
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.51,
|
| 451 |
+
"learning_rate": 5.5554165528544116e-06,
|
| 452 |
+
"loss": 0.2716,
|
| 453 |
+
"step": 1850
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.51,
|
| 457 |
+
"learning_rate": 5.546976236001093e-06,
|
| 458 |
+
"loss": 0.2946,
|
| 459 |
+
"step": 1875
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.52,
|
| 463 |
+
"learning_rate": 5.538535919147774e-06,
|
| 464 |
+
"loss": 0.3018,
|
| 465 |
+
"step": 1900
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.53,
|
| 469 |
+
"learning_rate": 5.5300956022944555e-06,
|
| 470 |
+
"loss": 0.2724,
|
| 471 |
+
"step": 1925
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.53,
|
| 475 |
+
"learning_rate": 5.521655285441136e-06,
|
| 476 |
+
"loss": 0.266,
|
| 477 |
+
"step": 1950
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.54,
|
| 481 |
+
"learning_rate": 5.513214968587817e-06,
|
| 482 |
+
"loss": 0.279,
|
| 483 |
+
"step": 1975
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.55,
|
| 487 |
+
"learning_rate": 5.504774651734499e-06,
|
| 488 |
+
"loss": 0.2936,
|
| 489 |
+
"step": 2000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.55,
|
| 493 |
+
"learning_rate": 5.496334334881181e-06,
|
| 494 |
+
"loss": 0.3085,
|
| 495 |
+
"step": 2025
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.56,
|
| 499 |
+
"learning_rate": 5.487894018027861e-06,
|
| 500 |
+
"loss": 0.2742,
|
| 501 |
+
"step": 2050
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.57,
|
| 505 |
+
"learning_rate": 5.4794537011745425e-06,
|
| 506 |
+
"loss": 0.2908,
|
| 507 |
+
"step": 2075
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.57,
|
| 511 |
+
"learning_rate": 5.471013384321224e-06,
|
| 512 |
+
"loss": 0.2731,
|
| 513 |
+
"step": 2100
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.58,
|
| 517 |
+
"learning_rate": 5.462573067467905e-06,
|
| 518 |
+
"loss": 0.2506,
|
| 519 |
+
"step": 2125
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.59,
|
| 523 |
+
"learning_rate": 5.454132750614586e-06,
|
| 524 |
+
"loss": 0.3134,
|
| 525 |
+
"step": 2150
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.59,
|
| 529 |
+
"learning_rate": 5.445692433761268e-06,
|
| 530 |
+
"loss": 0.3139,
|
| 531 |
+
"step": 2175
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.6,
|
| 535 |
+
"learning_rate": 5.437252116907949e-06,
|
| 536 |
+
"loss": 0.2645,
|
| 537 |
+
"step": 2200
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.61,
|
| 541 |
+
"learning_rate": 5.42881180005463e-06,
|
| 542 |
+
"loss": 0.2586,
|
| 543 |
+
"step": 2225
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.61,
|
| 547 |
+
"learning_rate": 5.420371483201311e-06,
|
| 548 |
+
"loss": 0.271,
|
| 549 |
+
"step": 2250
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.62,
|
| 553 |
+
"learning_rate": 5.411931166347992e-06,
|
| 554 |
+
"loss": 0.2609,
|
| 555 |
+
"step": 2275
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.63,
|
| 559 |
+
"learning_rate": 5.403490849494674e-06,
|
| 560 |
+
"loss": 0.2556,
|
| 561 |
+
"step": 2300
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.64,
|
| 565 |
+
"learning_rate": 5.3950505326413555e-06,
|
| 566 |
+
"loss": 0.2484,
|
| 567 |
+
"step": 2325
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.64,
|
| 571 |
+
"learning_rate": 5.386610215788036e-06,
|
| 572 |
+
"loss": 0.2852,
|
| 573 |
+
"step": 2350
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.65,
|
| 577 |
+
"learning_rate": 5.378169898934717e-06,
|
| 578 |
+
"loss": 0.316,
|
| 579 |
+
"step": 2375
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.66,
|
| 583 |
+
"learning_rate": 5.3697295820813986e-06,
|
| 584 |
+
"loss": 0.2798,
|
| 585 |
+
"step": 2400
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.66,
|
| 589 |
+
"learning_rate": 5.36128926522808e-06,
|
| 590 |
+
"loss": 0.2677,
|
| 591 |
+
"step": 2425
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.67,
|
| 595 |
+
"learning_rate": 5.352848948374761e-06,
|
| 596 |
+
"loss": 0.2962,
|
| 597 |
+
"step": 2450
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.68,
|
| 601 |
+
"learning_rate": 5.3444086315214425e-06,
|
| 602 |
+
"loss": 0.2392,
|
| 603 |
+
"step": 2475
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.68,
|
| 607 |
+
"learning_rate": 5.335968314668124e-06,
|
| 608 |
+
"loss": 0.2874,
|
| 609 |
+
"step": 2500
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.69,
|
| 613 |
+
"learning_rate": 5.327527997814805e-06,
|
| 614 |
+
"loss": 0.2796,
|
| 615 |
+
"step": 2525
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.7,
|
| 619 |
+
"learning_rate": 5.3190876809614855e-06,
|
| 620 |
+
"loss": 0.2947,
|
| 621 |
+
"step": 2550
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.7,
|
| 625 |
+
"learning_rate": 5.310647364108167e-06,
|
| 626 |
+
"loss": 0.2904,
|
| 627 |
+
"step": 2575
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.71,
|
| 631 |
+
"learning_rate": 5.302207047254849e-06,
|
| 632 |
+
"loss": 0.2777,
|
| 633 |
+
"step": 2600
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.72,
|
| 637 |
+
"learning_rate": 5.29376673040153e-06,
|
| 638 |
+
"loss": 0.2468,
|
| 639 |
+
"step": 2625
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.72,
|
| 643 |
+
"learning_rate": 5.285326413548211e-06,
|
| 644 |
+
"loss": 0.2745,
|
| 645 |
+
"step": 2650
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.73,
|
| 649 |
+
"learning_rate": 5.276886096694892e-06,
|
| 650 |
+
"loss": 0.2793,
|
| 651 |
+
"step": 2675
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.74,
|
| 655 |
+
"learning_rate": 5.268445779841573e-06,
|
| 656 |
+
"loss": 0.2873,
|
| 657 |
+
"step": 2700
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.74,
|
| 661 |
+
"learning_rate": 5.260005462988255e-06,
|
| 662 |
+
"loss": 0.2586,
|
| 663 |
+
"step": 2725
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.75,
|
| 667 |
+
"learning_rate": 5.251565146134936e-06,
|
| 668 |
+
"loss": 0.2637,
|
| 669 |
+
"step": 2750
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.76,
|
| 673 |
+
"learning_rate": 5.243124829281617e-06,
|
| 674 |
+
"loss": 0.2922,
|
| 675 |
+
"step": 2775
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.76,
|
| 679 |
+
"learning_rate": 5.234684512428299e-06,
|
| 680 |
+
"loss": 0.2675,
|
| 681 |
+
"step": 2800
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.77,
|
| 685 |
+
"learning_rate": 5.22624419557498e-06,
|
| 686 |
+
"loss": 0.2953,
|
| 687 |
+
"step": 2825
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.78,
|
| 691 |
+
"learning_rate": 5.21780387872166e-06,
|
| 692 |
+
"loss": 0.2293,
|
| 693 |
+
"step": 2850
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.79,
|
| 697 |
+
"learning_rate": 5.2093635618683425e-06,
|
| 698 |
+
"loss": 0.2422,
|
| 699 |
+
"step": 2875
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.79,
|
| 703 |
+
"learning_rate": 5.200923245015024e-06,
|
| 704 |
+
"loss": 0.292,
|
| 705 |
+
"step": 2900
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.8,
|
| 709 |
+
"learning_rate": 5.192482928161705e-06,
|
| 710 |
+
"loss": 0.2779,
|
| 711 |
+
"step": 2925
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.81,
|
| 715 |
+
"learning_rate": 5.1840426113083856e-06,
|
| 716 |
+
"loss": 0.3213,
|
| 717 |
+
"step": 2950
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.81,
|
| 721 |
+
"learning_rate": 5.175602294455067e-06,
|
| 722 |
+
"loss": 0.2653,
|
| 723 |
+
"step": 2975
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.82,
|
| 727 |
+
"learning_rate": 5.167161977601748e-06,
|
| 728 |
+
"loss": 0.2165,
|
| 729 |
+
"step": 3000
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.83,
|
| 733 |
+
"learning_rate": 5.1587216607484295e-06,
|
| 734 |
+
"loss": 0.2913,
|
| 735 |
+
"step": 3025
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.83,
|
| 739 |
+
"learning_rate": 5.150281343895111e-06,
|
| 740 |
+
"loss": 0.2627,
|
| 741 |
+
"step": 3050
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.84,
|
| 745 |
+
"learning_rate": 5.141841027041792e-06,
|
| 746 |
+
"loss": 0.2562,
|
| 747 |
+
"step": 3075
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.85,
|
| 751 |
+
"learning_rate": 5.133400710188473e-06,
|
| 752 |
+
"loss": 0.2611,
|
| 753 |
+
"step": 3100
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.85,
|
| 757 |
+
"learning_rate": 5.124960393335154e-06,
|
| 758 |
+
"loss": 0.2752,
|
| 759 |
+
"step": 3125
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.86,
|
| 763 |
+
"learning_rate": 5.116520076481835e-06,
|
| 764 |
+
"loss": 0.3039,
|
| 765 |
+
"step": 3150
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.87,
|
| 769 |
+
"learning_rate": 5.108079759628517e-06,
|
| 770 |
+
"loss": 0.2758,
|
| 771 |
+
"step": 3175
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 0.87,
|
| 775 |
+
"learning_rate": 5.099639442775199e-06,
|
| 776 |
+
"loss": 0.2495,
|
| 777 |
+
"step": 3200
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.88,
|
| 781 |
+
"learning_rate": 5.091199125921879e-06,
|
| 782 |
+
"loss": 0.2435,
|
| 783 |
+
"step": 3225
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.89,
|
| 787 |
+
"learning_rate": 5.08275880906856e-06,
|
| 788 |
+
"loss": 0.2637,
|
| 789 |
+
"step": 3250
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.89,
|
| 793 |
+
"learning_rate": 5.074318492215242e-06,
|
| 794 |
+
"loss": 0.3487,
|
| 795 |
+
"step": 3275
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.9,
|
| 799 |
+
"learning_rate": 5.065878175361923e-06,
|
| 800 |
+
"loss": 0.2781,
|
| 801 |
+
"step": 3300
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.91,
|
| 805 |
+
"learning_rate": 5.057437858508604e-06,
|
| 806 |
+
"loss": 0.2242,
|
| 807 |
+
"step": 3325
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.92,
|
| 811 |
+
"learning_rate": 5.048997541655286e-06,
|
| 812 |
+
"loss": 0.2589,
|
| 813 |
+
"step": 3350
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.92,
|
| 817 |
+
"learning_rate": 5.040557224801967e-06,
|
| 818 |
+
"loss": 0.2365,
|
| 819 |
+
"step": 3375
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.93,
|
| 823 |
+
"learning_rate": 5.032116907948648e-06,
|
| 824 |
+
"loss": 0.2659,
|
| 825 |
+
"step": 3400
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.94,
|
| 829 |
+
"learning_rate": 5.023676591095329e-06,
|
| 830 |
+
"loss": 0.2468,
|
| 831 |
+
"step": 3425
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.94,
|
| 835 |
+
"learning_rate": 5.01523627424201e-06,
|
| 836 |
+
"loss": 0.2516,
|
| 837 |
+
"step": 3450
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.95,
|
| 841 |
+
"learning_rate": 5.006795957388692e-06,
|
| 842 |
+
"loss": 0.2809,
|
| 843 |
+
"step": 3475
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.96,
|
| 847 |
+
"learning_rate": 4.998355640535373e-06,
|
| 848 |
+
"loss": 0.3185,
|
| 849 |
+
"step": 3500
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.96,
|
| 853 |
+
"learning_rate": 4.989915323682054e-06,
|
| 854 |
+
"loss": 0.2825,
|
| 855 |
+
"step": 3525
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.97,
|
| 859 |
+
"learning_rate": 4.981475006828735e-06,
|
| 860 |
+
"loss": 0.2818,
|
| 861 |
+
"step": 3550
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.98,
|
| 865 |
+
"learning_rate": 4.9730346899754165e-06,
|
| 866 |
+
"loss": 0.2833,
|
| 867 |
+
"step": 3575
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.98,
|
| 871 |
+
"learning_rate": 4.964594373122098e-06,
|
| 872 |
+
"loss": 0.2716,
|
| 873 |
+
"step": 3600
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.99,
|
| 877 |
+
"learning_rate": 4.956154056268779e-06,
|
| 878 |
+
"loss": 0.3022,
|
| 879 |
+
"step": 3625
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.0,
|
| 883 |
+
"learning_rate": 4.94771373941546e-06,
|
| 884 |
+
"loss": 0.2745,
|
| 885 |
+
"step": 3650
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1.0,
|
| 889 |
+
"eval_loss": 0.27263349294662476,
|
| 890 |
+
"eval_runtime": 5336.6361,
|
| 891 |
+
"eval_samples_per_second": 1.372,
|
| 892 |
+
"eval_steps_per_second": 0.172,
|
| 893 |
+
"eval_wer": 57.088590421923755,
|
| 894 |
+
"step": 3661
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 1.0,
|
| 898 |
+
"learning_rate": 4.939273422562142e-06,
|
| 899 |
+
"loss": 0.1763,
|
| 900 |
+
"step": 3675
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 1.01,
|
| 904 |
+
"learning_rate": 4.930833105708823e-06,
|
| 905 |
+
"loss": 0.137,
|
| 906 |
+
"step": 3700
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.02,
|
| 910 |
+
"learning_rate": 4.9223927888555035e-06,
|
| 911 |
+
"loss": 0.1398,
|
| 912 |
+
"step": 3725
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.02,
|
| 916 |
+
"learning_rate": 4.913952472002185e-06,
|
| 917 |
+
"loss": 0.1329,
|
| 918 |
+
"step": 3750
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.03,
|
| 922 |
+
"learning_rate": 4.905512155148867e-06,
|
| 923 |
+
"loss": 0.1263,
|
| 924 |
+
"step": 3775
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 1.04,
|
| 928 |
+
"learning_rate": 4.897071838295548e-06,
|
| 929 |
+
"loss": 0.1056,
|
| 930 |
+
"step": 3800
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.04,
|
| 934 |
+
"learning_rate": 4.888631521442229e-06,
|
| 935 |
+
"loss": 0.1095,
|
| 936 |
+
"step": 3825
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 1.05,
|
| 940 |
+
"learning_rate": 4.88019120458891e-06,
|
| 941 |
+
"loss": 0.1457,
|
| 942 |
+
"step": 3850
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 1.06,
|
| 946 |
+
"learning_rate": 4.871750887735591e-06,
|
| 947 |
+
"loss": 0.1464,
|
| 948 |
+
"step": 3875
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.07,
|
| 952 |
+
"learning_rate": 4.863310570882273e-06,
|
| 953 |
+
"loss": 0.1194,
|
| 954 |
+
"step": 3900
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.07,
|
| 958 |
+
"learning_rate": 4.854870254028954e-06,
|
| 959 |
+
"loss": 0.1312,
|
| 960 |
+
"step": 3925
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.08,
|
| 964 |
+
"learning_rate": 4.846429937175635e-06,
|
| 965 |
+
"loss": 0.1279,
|
| 966 |
+
"step": 3950
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 1.09,
|
| 970 |
+
"learning_rate": 4.8379896203223165e-06,
|
| 971 |
+
"loss": 0.1328,
|
| 972 |
+
"step": 3975
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 1.09,
|
| 976 |
+
"learning_rate": 4.829549303468998e-06,
|
| 977 |
+
"loss": 0.1252,
|
| 978 |
+
"step": 4000
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.1,
|
| 982 |
+
"learning_rate": 4.821108986615678e-06,
|
| 983 |
+
"loss": 0.1125,
|
| 984 |
+
"step": 4025
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 1.11,
|
| 988 |
+
"learning_rate": 4.81266866976236e-06,
|
| 989 |
+
"loss": 0.1354,
|
| 990 |
+
"step": 4050
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.11,
|
| 994 |
+
"learning_rate": 4.804228352909042e-06,
|
| 995 |
+
"loss": 0.1301,
|
| 996 |
+
"step": 4075
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.12,
|
| 1000 |
+
"learning_rate": 4.795788036055723e-06,
|
| 1001 |
+
"loss": 0.116,
|
| 1002 |
+
"step": 4100
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.13,
|
| 1006 |
+
"learning_rate": 4.7873477192024035e-06,
|
| 1007 |
+
"loss": 0.1535,
|
| 1008 |
+
"step": 4125
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 1.13,
|
| 1012 |
+
"learning_rate": 4.778907402349085e-06,
|
| 1013 |
+
"loss": 0.1318,
|
| 1014 |
+
"step": 4150
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.14,
|
| 1018 |
+
"learning_rate": 4.770467085495766e-06,
|
| 1019 |
+
"loss": 0.1302,
|
| 1020 |
+
"step": 4175
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.15,
|
| 1024 |
+
"learning_rate": 4.762026768642447e-06,
|
| 1025 |
+
"loss": 0.1093,
|
| 1026 |
+
"step": 4200
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.15,
|
| 1030 |
+
"learning_rate": 4.753586451789129e-06,
|
| 1031 |
+
"loss": 0.1315,
|
| 1032 |
+
"step": 4225
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 1.16,
|
| 1036 |
+
"learning_rate": 4.74514613493581e-06,
|
| 1037 |
+
"loss": 0.1154,
|
| 1038 |
+
"step": 4250
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 1.17,
|
| 1042 |
+
"learning_rate": 4.736705818082491e-06,
|
| 1043 |
+
"loss": 0.1433,
|
| 1044 |
+
"step": 4275
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.17,
|
| 1048 |
+
"learning_rate": 4.728265501229173e-06,
|
| 1049 |
+
"loss": 0.1177,
|
| 1050 |
+
"step": 4300
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 1.18,
|
| 1054 |
+
"learning_rate": 4.719825184375853e-06,
|
| 1055 |
+
"loss": 0.1203,
|
| 1056 |
+
"step": 4325
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 1.19,
|
| 1060 |
+
"learning_rate": 4.711384867522535e-06,
|
| 1061 |
+
"loss": 0.1563,
|
| 1062 |
+
"step": 4350
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 1.2,
|
| 1066 |
+
"learning_rate": 4.7029445506692165e-06,
|
| 1067 |
+
"loss": 0.1426,
|
| 1068 |
+
"step": 4375
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 1.2,
|
| 1072 |
+
"learning_rate": 4.694504233815898e-06,
|
| 1073 |
+
"loss": 0.1487,
|
| 1074 |
+
"step": 4400
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.21,
|
| 1078 |
+
"learning_rate": 4.686063916962578e-06,
|
| 1079 |
+
"loss": 0.1437,
|
| 1080 |
+
"step": 4425
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.22,
|
| 1084 |
+
"learning_rate": 4.67762360010926e-06,
|
| 1085 |
+
"loss": 0.1164,
|
| 1086 |
+
"step": 4450
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.22,
|
| 1090 |
+
"learning_rate": 4.669183283255941e-06,
|
| 1091 |
+
"loss": 0.1192,
|
| 1092 |
+
"step": 4475
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 1.23,
|
| 1096 |
+
"learning_rate": 4.660742966402622e-06,
|
| 1097 |
+
"loss": 0.1271,
|
| 1098 |
+
"step": 4500
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 1.24,
|
| 1102 |
+
"learning_rate": 4.6523026495493035e-06,
|
| 1103 |
+
"loss": 0.1229,
|
| 1104 |
+
"step": 4525
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 1.24,
|
| 1108 |
+
"learning_rate": 4.643862332695985e-06,
|
| 1109 |
+
"loss": 0.1305,
|
| 1110 |
+
"step": 4550
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 1.25,
|
| 1114 |
+
"learning_rate": 4.635422015842666e-06,
|
| 1115 |
+
"loss": 0.1124,
|
| 1116 |
+
"step": 4575
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 1.26,
|
| 1120 |
+
"learning_rate": 4.626981698989347e-06,
|
| 1121 |
+
"loss": 0.1316,
|
| 1122 |
+
"step": 4600
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.26,
|
| 1126 |
+
"learning_rate": 4.618541382136028e-06,
|
| 1127 |
+
"loss": 0.1331,
|
| 1128 |
+
"step": 4625
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 1.27,
|
| 1132 |
+
"learning_rate": 4.61010106528271e-06,
|
| 1133 |
+
"loss": 0.1217,
|
| 1134 |
+
"step": 4650
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.28,
|
| 1138 |
+
"learning_rate": 4.601660748429391e-06,
|
| 1139 |
+
"loss": 0.1217,
|
| 1140 |
+
"step": 4675
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 1.28,
|
| 1144 |
+
"learning_rate": 4.593220431576073e-06,
|
| 1145 |
+
"loss": 0.1398,
|
| 1146 |
+
"step": 4700
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 1.29,
|
| 1150 |
+
"learning_rate": 4.584780114722753e-06,
|
| 1151 |
+
"loss": 0.1288,
|
| 1152 |
+
"step": 4725
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 1.3,
|
| 1156 |
+
"learning_rate": 4.576339797869434e-06,
|
| 1157 |
+
"loss": 0.1283,
|
| 1158 |
+
"step": 4750
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 1.3,
|
| 1162 |
+
"learning_rate": 4.567899481016116e-06,
|
| 1163 |
+
"loss": 0.1065,
|
| 1164 |
+
"step": 4775
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.31,
|
| 1168 |
+
"learning_rate": 4.559459164162797e-06,
|
| 1169 |
+
"loss": 0.1309,
|
| 1170 |
+
"step": 4800
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.32,
|
| 1174 |
+
"learning_rate": 4.551018847309478e-06,
|
| 1175 |
+
"loss": 0.1238,
|
| 1176 |
+
"step": 4825
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 1.32,
|
| 1180 |
+
"learning_rate": 4.54257853045616e-06,
|
| 1181 |
+
"loss": 0.1488,
|
| 1182 |
+
"step": 4850
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 1.33,
|
| 1186 |
+
"learning_rate": 4.534138213602841e-06,
|
| 1187 |
+
"loss": 0.1172,
|
| 1188 |
+
"step": 4875
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 1.34,
|
| 1192 |
+
"learning_rate": 4.525697896749522e-06,
|
| 1193 |
+
"loss": 0.1319,
|
| 1194 |
+
"step": 4900
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.35,
|
| 1198 |
+
"learning_rate": 4.517257579896203e-06,
|
| 1199 |
+
"loss": 0.1304,
|
| 1200 |
+
"step": 4925
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.35,
|
| 1204 |
+
"learning_rate": 4.508817263042885e-06,
|
| 1205 |
+
"loss": 0.1398,
|
| 1206 |
+
"step": 4950
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.36,
|
| 1210 |
+
"learning_rate": 4.500376946189566e-06,
|
| 1211 |
+
"loss": 0.1367,
|
| 1212 |
+
"step": 4975
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 1.37,
|
| 1216 |
+
"learning_rate": 4.4919366293362474e-06,
|
| 1217 |
+
"loss": 0.1398,
|
| 1218 |
+
"step": 5000
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 1.37,
|
| 1222 |
+
"learning_rate": 4.483496312482928e-06,
|
| 1223 |
+
"loss": 0.1251,
|
| 1224 |
+
"step": 5025
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.38,
|
| 1228 |
+
"learning_rate": 4.475055995629609e-06,
|
| 1229 |
+
"loss": 0.1568,
|
| 1230 |
+
"step": 5050
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 1.39,
|
| 1234 |
+
"learning_rate": 4.4666156787762905e-06,
|
| 1235 |
+
"loss": 0.1268,
|
| 1236 |
+
"step": 5075
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 1.39,
|
| 1240 |
+
"learning_rate": 4.458175361922973e-06,
|
| 1241 |
+
"loss": 0.1233,
|
| 1242 |
+
"step": 5100
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.4,
|
| 1246 |
+
"learning_rate": 4.449735045069653e-06,
|
| 1247 |
+
"loss": 0.1307,
|
| 1248 |
+
"step": 5125
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.41,
|
| 1252 |
+
"learning_rate": 4.441294728216334e-06,
|
| 1253 |
+
"loss": 0.13,
|
| 1254 |
+
"step": 5150
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.41,
|
| 1258 |
+
"learning_rate": 4.432854411363016e-06,
|
| 1259 |
+
"loss": 0.1542,
|
| 1260 |
+
"step": 5175
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 1.42,
|
| 1264 |
+
"learning_rate": 4.424414094509697e-06,
|
| 1265 |
+
"loss": 0.1318,
|
| 1266 |
+
"step": 5200
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 1.43,
|
| 1270 |
+
"learning_rate": 4.415973777656378e-06,
|
| 1271 |
+
"loss": 0.1492,
|
| 1272 |
+
"step": 5225
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 1.43,
|
| 1276 |
+
"learning_rate": 4.40753346080306e-06,
|
| 1277 |
+
"loss": 0.1296,
|
| 1278 |
+
"step": 5250
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 1.44,
|
| 1282 |
+
"learning_rate": 4.399093143949741e-06,
|
| 1283 |
+
"loss": 0.1341,
|
| 1284 |
+
"step": 5275
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.45,
|
| 1288 |
+
"learning_rate": 4.390652827096422e-06,
|
| 1289 |
+
"loss": 0.1146,
|
| 1290 |
+
"step": 5300
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.45,
|
| 1294 |
+
"learning_rate": 4.382212510243103e-06,
|
| 1295 |
+
"loss": 0.1466,
|
| 1296 |
+
"step": 5325
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.46,
|
| 1300 |
+
"learning_rate": 4.373772193389784e-06,
|
| 1301 |
+
"loss": 0.1182,
|
| 1302 |
+
"step": 5350
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 1.47,
|
| 1306 |
+
"learning_rate": 4.365331876536465e-06,
|
| 1307 |
+
"loss": 0.1388,
|
| 1308 |
+
"step": 5375
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 1.48,
|
| 1312 |
+
"learning_rate": 4.3568915596831475e-06,
|
| 1313 |
+
"loss": 0.1346,
|
| 1314 |
+
"step": 5400
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 1.48,
|
| 1318 |
+
"learning_rate": 4.348451242829828e-06,
|
| 1319 |
+
"loss": 0.1169,
|
| 1320 |
+
"step": 5425
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 1.49,
|
| 1324 |
+
"learning_rate": 4.340010925976509e-06,
|
| 1325 |
+
"loss": 0.1112,
|
| 1326 |
+
"step": 5450
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 1.5,
|
| 1330 |
+
"learning_rate": 4.3315706091231905e-06,
|
| 1331 |
+
"loss": 0.1454,
|
| 1332 |
+
"step": 5475
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.5,
|
| 1336 |
+
"learning_rate": 4.323130292269872e-06,
|
| 1337 |
+
"loss": 0.1231,
|
| 1338 |
+
"step": 5500
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.51,
|
| 1342 |
+
"learning_rate": 4.314689975416553e-06,
|
| 1343 |
+
"loss": 0.1218,
|
| 1344 |
+
"step": 5525
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 1.52,
|
| 1348 |
+
"learning_rate": 4.3062496585632344e-06,
|
| 1349 |
+
"loss": 0.1412,
|
| 1350 |
+
"step": 5550
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 1.52,
|
| 1354 |
+
"learning_rate": 4.297809341709916e-06,
|
| 1355 |
+
"loss": 0.1393,
|
| 1356 |
+
"step": 5575
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 1.53,
|
| 1360 |
+
"learning_rate": 4.289369024856597e-06,
|
| 1361 |
+
"loss": 0.126,
|
| 1362 |
+
"step": 5600
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 1.54,
|
| 1366 |
+
"learning_rate": 4.2809287080032775e-06,
|
| 1367 |
+
"loss": 0.1399,
|
| 1368 |
+
"step": 5625
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 1.54,
|
| 1372 |
+
"learning_rate": 4.272488391149959e-06,
|
| 1373 |
+
"loss": 0.1509,
|
| 1374 |
+
"step": 5650
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.55,
|
| 1378 |
+
"learning_rate": 4.26404807429664e-06,
|
| 1379 |
+
"loss": 0.1508,
|
| 1380 |
+
"step": 5675
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 1.56,
|
| 1384 |
+
"learning_rate": 4.255607757443322e-06,
|
| 1385 |
+
"loss": 0.1013,
|
| 1386 |
+
"step": 5700
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 1.56,
|
| 1390 |
+
"learning_rate": 4.247167440590003e-06,
|
| 1391 |
+
"loss": 0.1416,
|
| 1392 |
+
"step": 5725
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 1.57,
|
| 1396 |
+
"learning_rate": 4.238727123736684e-06,
|
| 1397 |
+
"loss": 0.1611,
|
| 1398 |
+
"step": 5750
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 1.58,
|
| 1402 |
+
"learning_rate": 4.230286806883365e-06,
|
| 1403 |
+
"loss": 0.1448,
|
| 1404 |
+
"step": 5775
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 1.58,
|
| 1408 |
+
"learning_rate": 4.221846490030047e-06,
|
| 1409 |
+
"loss": 0.1402,
|
| 1410 |
+
"step": 5800
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.59,
|
| 1414 |
+
"learning_rate": 4.213406173176728e-06,
|
| 1415 |
+
"loss": 0.1075,
|
| 1416 |
+
"step": 5825
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 1.6,
|
| 1420 |
+
"learning_rate": 4.204965856323409e-06,
|
| 1421 |
+
"loss": 0.1546,
|
| 1422 |
+
"step": 5850
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.6,
|
| 1426 |
+
"learning_rate": 4.1965255394700905e-06,
|
| 1427 |
+
"loss": 0.1454,
|
| 1428 |
+
"step": 5875
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 1.61,
|
| 1432 |
+
"learning_rate": 4.188085222616772e-06,
|
| 1433 |
+
"loss": 0.124,
|
| 1434 |
+
"step": 5900
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 1.62,
|
| 1438 |
+
"learning_rate": 4.179644905763452e-06,
|
| 1439 |
+
"loss": 0.1219,
|
| 1440 |
+
"step": 5925
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 1.63,
|
| 1444 |
+
"learning_rate": 4.171204588910134e-06,
|
| 1445 |
+
"loss": 0.1218,
|
| 1446 |
+
"step": 5950
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 1.63,
|
| 1450 |
+
"learning_rate": 4.162764272056815e-06,
|
| 1451 |
+
"loss": 0.1452,
|
| 1452 |
+
"step": 5975
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 1.64,
|
| 1456 |
+
"learning_rate": 4.154323955203496e-06,
|
| 1457 |
+
"loss": 0.145,
|
| 1458 |
+
"step": 6000
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 1.65,
|
| 1462 |
+
"learning_rate": 4.1458836383501775e-06,
|
| 1463 |
+
"loss": 0.1274,
|
| 1464 |
+
"step": 6025
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.65,
|
| 1468 |
+
"learning_rate": 4.137443321496859e-06,
|
| 1469 |
+
"loss": 0.1271,
|
| 1470 |
+
"step": 6050
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 1.66,
|
| 1474 |
+
"learning_rate": 4.12900300464354e-06,
|
| 1475 |
+
"loss": 0.1401,
|
| 1476 |
+
"step": 6075
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 1.67,
|
| 1480 |
+
"learning_rate": 4.120562687790221e-06,
|
| 1481 |
+
"loss": 0.1151,
|
| 1482 |
+
"step": 6100
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 1.67,
|
| 1486 |
+
"learning_rate": 4.112122370936903e-06,
|
| 1487 |
+
"loss": 0.125,
|
| 1488 |
+
"step": 6125
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 1.68,
|
| 1492 |
+
"learning_rate": 4.103682054083584e-06,
|
| 1493 |
+
"loss": 0.1576,
|
| 1494 |
+
"step": 6150
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 1.69,
|
| 1498 |
+
"learning_rate": 4.095241737230265e-06,
|
| 1499 |
+
"loss": 0.1314,
|
| 1500 |
+
"step": 6175
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 1.69,
|
| 1504 |
+
"learning_rate": 4.086801420376946e-06,
|
| 1505 |
+
"loss": 0.1287,
|
| 1506 |
+
"step": 6200
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.7,
|
| 1510 |
+
"learning_rate": 4.078361103523627e-06,
|
| 1511 |
+
"loss": 0.1163,
|
| 1512 |
+
"step": 6225
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 1.71,
|
| 1516 |
+
"learning_rate": 4.069920786670308e-06,
|
| 1517 |
+
"loss": 0.1209,
|
| 1518 |
+
"step": 6250
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 1.71,
|
| 1522 |
+
"learning_rate": 4.0614804698169906e-06,
|
| 1523 |
+
"loss": 0.1271,
|
| 1524 |
+
"step": 6275
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 1.72,
|
| 1528 |
+
"learning_rate": 4.053040152963671e-06,
|
| 1529 |
+
"loss": 0.1339,
|
| 1530 |
+
"step": 6300
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 1.73,
|
| 1534 |
+
"learning_rate": 4.044599836110352e-06,
|
| 1535 |
+
"loss": 0.1283,
|
| 1536 |
+
"step": 6325
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 1.73,
|
| 1540 |
+
"learning_rate": 4.036159519257034e-06,
|
| 1541 |
+
"loss": 0.1269,
|
| 1542 |
+
"step": 6350
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 1.74,
|
| 1546 |
+
"learning_rate": 4.027719202403715e-06,
|
| 1547 |
+
"loss": 0.1416,
|
| 1548 |
+
"step": 6375
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.75,
|
| 1552 |
+
"learning_rate": 4.019278885550396e-06,
|
| 1553 |
+
"loss": 0.1126,
|
| 1554 |
+
"step": 6400
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 1.75,
|
| 1558 |
+
"learning_rate": 4.0108385686970776e-06,
|
| 1559 |
+
"loss": 0.1204,
|
| 1560 |
+
"step": 6425
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 1.76,
|
| 1564 |
+
"learning_rate": 4.002398251843759e-06,
|
| 1565 |
+
"loss": 0.1189,
|
| 1566 |
+
"step": 6450
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 1.77,
|
| 1570 |
+
"learning_rate": 3.99395793499044e-06,
|
| 1571 |
+
"loss": 0.1479,
|
| 1572 |
+
"step": 6475
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 1.78,
|
| 1576 |
+
"learning_rate": 3.985517618137121e-06,
|
| 1577 |
+
"loss": 0.1274,
|
| 1578 |
+
"step": 6500
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 1.78,
|
| 1582 |
+
"learning_rate": 3.977077301283802e-06,
|
| 1583 |
+
"loss": 0.1273,
|
| 1584 |
+
"step": 6525
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 1.79,
|
| 1588 |
+
"learning_rate": 3.968636984430483e-06,
|
| 1589 |
+
"loss": 0.1071,
|
| 1590 |
+
"step": 6550
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.8,
|
| 1594 |
+
"learning_rate": 3.960196667577165e-06,
|
| 1595 |
+
"loss": 0.1358,
|
| 1596 |
+
"step": 6575
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"epoch": 1.8,
|
| 1600 |
+
"learning_rate": 3.951756350723846e-06,
|
| 1601 |
+
"loss": 0.1314,
|
| 1602 |
+
"step": 6600
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 1.81,
|
| 1606 |
+
"learning_rate": 3.943316033870527e-06,
|
| 1607 |
+
"loss": 0.1223,
|
| 1608 |
+
"step": 6625
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 1.82,
|
| 1612 |
+
"learning_rate": 3.9348757170172084e-06,
|
| 1613 |
+
"loss": 0.1145,
|
| 1614 |
+
"step": 6650
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 1.82,
|
| 1618 |
+
"learning_rate": 3.92643540016389e-06,
|
| 1619 |
+
"loss": 0.1331,
|
| 1620 |
+
"step": 6675
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 1.83,
|
| 1624 |
+
"learning_rate": 3.917995083310571e-06,
|
| 1625 |
+
"loss": 0.1198,
|
| 1626 |
+
"step": 6700
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 1.84,
|
| 1630 |
+
"learning_rate": 3.909554766457252e-06,
|
| 1631 |
+
"loss": 0.1517,
|
| 1632 |
+
"step": 6725
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.84,
|
| 1636 |
+
"learning_rate": 3.901114449603934e-06,
|
| 1637 |
+
"loss": 0.1477,
|
| 1638 |
+
"step": 6750
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 1.85,
|
| 1642 |
+
"learning_rate": 3.892674132750615e-06,
|
| 1643 |
+
"loss": 0.1187,
|
| 1644 |
+
"step": 6775
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 1.86,
|
| 1648 |
+
"learning_rate": 3.884233815897295e-06,
|
| 1649 |
+
"loss": 0.1293,
|
| 1650 |
+
"step": 6800
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 1.86,
|
| 1654 |
+
"learning_rate": 3.875793499043977e-06,
|
| 1655 |
+
"loss": 0.1228,
|
| 1656 |
+
"step": 6825
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 1.87,
|
| 1660 |
+
"learning_rate": 3.867353182190658e-06,
|
| 1661 |
+
"loss": 0.1554,
|
| 1662 |
+
"step": 6850
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 1.88,
|
| 1666 |
+
"learning_rate": 3.85891286533734e-06,
|
| 1667 |
+
"loss": 0.1236,
|
| 1668 |
+
"step": 6875
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 1.88,
|
| 1672 |
+
"learning_rate": 3.850472548484021e-06,
|
| 1673 |
+
"loss": 0.1146,
|
| 1674 |
+
"step": 6900
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.89,
|
| 1678 |
+
"learning_rate": 3.842032231630702e-06,
|
| 1679 |
+
"loss": 0.1364,
|
| 1680 |
+
"step": 6925
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 1.9,
|
| 1684 |
+
"learning_rate": 3.833591914777383e-06,
|
| 1685 |
+
"loss": 0.1254,
|
| 1686 |
+
"step": 6950
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 1.91,
|
| 1690 |
+
"learning_rate": 3.8251515979240646e-06,
|
| 1691 |
+
"loss": 0.1209,
|
| 1692 |
+
"step": 6975
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 1.91,
|
| 1696 |
+
"learning_rate": 3.816711281070746e-06,
|
| 1697 |
+
"loss": 0.1159,
|
| 1698 |
+
"step": 7000
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 1.92,
|
| 1702 |
+
"learning_rate": 3.8082709642174267e-06,
|
| 1703 |
+
"loss": 0.1266,
|
| 1704 |
+
"step": 7025
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 1.93,
|
| 1708 |
+
"learning_rate": 3.7998306473641085e-06,
|
| 1709 |
+
"loss": 0.1497,
|
| 1710 |
+
"step": 7050
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 1.93,
|
| 1714 |
+
"learning_rate": 3.7913903305107898e-06,
|
| 1715 |
+
"loss": 0.1159,
|
| 1716 |
+
"step": 7075
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.94,
|
| 1720 |
+
"learning_rate": 3.7829500136574702e-06,
|
| 1721 |
+
"loss": 0.1423,
|
| 1722 |
+
"step": 7100
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 1.95,
|
| 1726 |
+
"learning_rate": 3.774509696804152e-06,
|
| 1727 |
+
"loss": 0.1307,
|
| 1728 |
+
"step": 7125
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 1.95,
|
| 1732 |
+
"learning_rate": 3.7660693799508333e-06,
|
| 1733 |
+
"loss": 0.1195,
|
| 1734 |
+
"step": 7150
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 1.96,
|
| 1738 |
+
"learning_rate": 3.7576290630975146e-06,
|
| 1739 |
+
"loss": 0.1396,
|
| 1740 |
+
"step": 7175
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 1.97,
|
| 1744 |
+
"learning_rate": 3.7491887462441954e-06,
|
| 1745 |
+
"loss": 0.135,
|
| 1746 |
+
"step": 7200
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 1.97,
|
| 1750 |
+
"learning_rate": 3.7407484293908767e-06,
|
| 1751 |
+
"loss": 0.141,
|
| 1752 |
+
"step": 7225
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 1.98,
|
| 1756 |
+
"learning_rate": 3.732308112537558e-06,
|
| 1757 |
+
"loss": 0.1353,
|
| 1758 |
+
"step": 7250
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 1.99,
|
| 1762 |
+
"learning_rate": 3.7238677956842398e-06,
|
| 1763 |
+
"loss": 0.149,
|
| 1764 |
+
"step": 7275
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 1.99,
|
| 1768 |
+
"learning_rate": 3.7154274788309202e-06,
|
| 1769 |
+
"loss": 0.1259,
|
| 1770 |
+
"step": 7300
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.0,
|
| 1774 |
+
"eval_loss": 0.27949637174606323,
|
| 1775 |
+
"eval_runtime": 5445.6948,
|
| 1776 |
+
"eval_samples_per_second": 1.345,
|
| 1777 |
+
"eval_steps_per_second": 0.168,
|
| 1778 |
+
"eval_wer": 56.830823497490165,
|
| 1779 |
+
"step": 7322
|
| 1780 |
+
}
|
| 1781 |
+
],
|
| 1782 |
+
"logging_steps": 25,
|
| 1783 |
+
"max_steps": 18305,
|
| 1784 |
+
"num_input_tokens_seen": 0,
|
| 1785 |
+
"num_train_epochs": 5,
|
| 1786 |
+
"save_steps": 500,
|
| 1787 |
+
"total_flos": 1.243626193723392e+20,
|
| 1788 |
+
"train_batch_size": 8,
|
| 1789 |
+
"trial_name": null,
|
| 1790 |
+
"trial_params": null
|
| 1791 |
+
}
|
chinese/checkpoint-7322/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:142e723e6d372563550e9f77d73f452dad5a788dba33d0766b0938c7b2327f4d
|
| 3 |
+
size 4475
|
pinyin/config.json
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "openai/whisper-large",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "gelu",
|
| 5 |
+
"apply_spec_augment": false,
|
| 6 |
+
"architectures": [
|
| 7 |
+
"WhisperForConditionalGeneration"
|
| 8 |
+
],
|
| 9 |
+
"attention_dropout": 0.0,
|
| 10 |
+
"begin_suppress_tokens": [
|
| 11 |
+
220,
|
| 12 |
+
50257
|
| 13 |
+
],
|
| 14 |
+
"bos_token_id": 50257,
|
| 15 |
+
"classifier_proj_size": 256,
|
| 16 |
+
"d_model": 1280,
|
| 17 |
+
"decoder_attention_heads": 20,
|
| 18 |
+
"decoder_ffn_dim": 5120,
|
| 19 |
+
"decoder_layerdrop": 0.0,
|
| 20 |
+
"decoder_layers": 32,
|
| 21 |
+
"decoder_start_token_id": 50258,
|
| 22 |
+
"dropout": 0.0,
|
| 23 |
+
"encoder_attention_heads": 20,
|
| 24 |
+
"encoder_ffn_dim": 5120,
|
| 25 |
+
"encoder_layerdrop": 0.0,
|
| 26 |
+
"encoder_layers": 32,
|
| 27 |
+
"eos_token_id": 50257,
|
| 28 |
+
"forced_decoder_ids": [
|
| 29 |
+
[
|
| 30 |
+
1,
|
| 31 |
+
50359
|
| 32 |
+
],
|
| 33 |
+
[
|
| 34 |
+
2,
|
| 35 |
+
50363
|
| 36 |
+
]
|
| 37 |
+
],
|
| 38 |
+
"init_std": 0.02,
|
| 39 |
+
"is_encoder_decoder": true,
|
| 40 |
+
"mask_feature_length": 10,
|
| 41 |
+
"mask_feature_min_masks": 0,
|
| 42 |
+
"mask_feature_prob": 0.0,
|
| 43 |
+
"mask_time_length": 10,
|
| 44 |
+
"mask_time_min_masks": 2,
|
| 45 |
+
"mask_time_prob": 0.05,
|
| 46 |
+
"max_length": 448,
|
| 47 |
+
"max_source_positions": 1500,
|
| 48 |
+
"max_target_positions": 448,
|
| 49 |
+
"median_filter_width": 7,
|
| 50 |
+
"model_type": "whisper",
|
| 51 |
+
"num_hidden_layers": 32,
|
| 52 |
+
"num_mel_bins": 80,
|
| 53 |
+
"pad_token_id": 50257,
|
| 54 |
+
"scale_embedding": false,
|
| 55 |
+
"suppress_tokens": [],
|
| 56 |
+
"torch_dtype": "float32",
|
| 57 |
+
"transformers_version": "4.36.2",
|
| 58 |
+
"use_cache": true,
|
| 59 |
+
"use_weighted_layer_sum": false,
|
| 60 |
+
"vocab_size": 51865
|
| 61 |
+
}
|
pinyin/generation_config.json
ADDED
|
@@ -0,0 +1,260 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alignment_heads": [
|
| 3 |
+
[
|
| 4 |
+
9,
|
| 5 |
+
19
|
| 6 |
+
],
|
| 7 |
+
[
|
| 8 |
+
11,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
[
|
| 12 |
+
11,
|
| 13 |
+
4
|
| 14 |
+
],
|
| 15 |
+
[
|
| 16 |
+
11,
|
| 17 |
+
17
|
| 18 |
+
],
|
| 19 |
+
[
|
| 20 |
+
22,
|
| 21 |
+
7
|
| 22 |
+
],
|
| 23 |
+
[
|
| 24 |
+
22,
|
| 25 |
+
11
|
| 26 |
+
],
|
| 27 |
+
[
|
| 28 |
+
22,
|
| 29 |
+
17
|
| 30 |
+
],
|
| 31 |
+
[
|
| 32 |
+
23,
|
| 33 |
+
2
|
| 34 |
+
],
|
| 35 |
+
[
|
| 36 |
+
23,
|
| 37 |
+
15
|
| 38 |
+
]
|
| 39 |
+
],
|
| 40 |
+
"begin_suppress_tokens": [
|
| 41 |
+
220,
|
| 42 |
+
50257
|
| 43 |
+
],
|
| 44 |
+
"bos_token_id": 50257,
|
| 45 |
+
"decoder_start_token_id": 50258,
|
| 46 |
+
"eos_token_id": 50257,
|
| 47 |
+
"forced_decoder_ids": [
|
| 48 |
+
[
|
| 49 |
+
1,
|
| 50 |
+
null
|
| 51 |
+
],
|
| 52 |
+
[
|
| 53 |
+
2,
|
| 54 |
+
50359
|
| 55 |
+
]
|
| 56 |
+
],
|
| 57 |
+
"is_multilingual": true,
|
| 58 |
+
"lang_to_id": {
|
| 59 |
+
"<|af|>": 50327,
|
| 60 |
+
"<|am|>": 50334,
|
| 61 |
+
"<|ar|>": 50272,
|
| 62 |
+
"<|as|>": 50350,
|
| 63 |
+
"<|az|>": 50304,
|
| 64 |
+
"<|ba|>": 50355,
|
| 65 |
+
"<|be|>": 50330,
|
| 66 |
+
"<|bg|>": 50292,
|
| 67 |
+
"<|bn|>": 50302,
|
| 68 |
+
"<|bo|>": 50347,
|
| 69 |
+
"<|br|>": 50309,
|
| 70 |
+
"<|bs|>": 50315,
|
| 71 |
+
"<|ca|>": 50270,
|
| 72 |
+
"<|cs|>": 50283,
|
| 73 |
+
"<|cy|>": 50297,
|
| 74 |
+
"<|da|>": 50285,
|
| 75 |
+
"<|de|>": 50261,
|
| 76 |
+
"<|el|>": 50281,
|
| 77 |
+
"<|en|>": 50259,
|
| 78 |
+
"<|es|>": 50262,
|
| 79 |
+
"<|et|>": 50307,
|
| 80 |
+
"<|eu|>": 50310,
|
| 81 |
+
"<|fa|>": 50300,
|
| 82 |
+
"<|fi|>": 50277,
|
| 83 |
+
"<|fo|>": 50338,
|
| 84 |
+
"<|fr|>": 50265,
|
| 85 |
+
"<|gl|>": 50319,
|
| 86 |
+
"<|gu|>": 50333,
|
| 87 |
+
"<|haw|>": 50352,
|
| 88 |
+
"<|ha|>": 50354,
|
| 89 |
+
"<|he|>": 50279,
|
| 90 |
+
"<|hi|>": 50276,
|
| 91 |
+
"<|hr|>": 50291,
|
| 92 |
+
"<|ht|>": 50339,
|
| 93 |
+
"<|hu|>": 50286,
|
| 94 |
+
"<|hy|>": 50312,
|
| 95 |
+
"<|id|>": 50275,
|
| 96 |
+
"<|is|>": 50311,
|
| 97 |
+
"<|it|>": 50274,
|
| 98 |
+
"<|ja|>": 50266,
|
| 99 |
+
"<|jw|>": 50356,
|
| 100 |
+
"<|ka|>": 50329,
|
| 101 |
+
"<|kk|>": 50316,
|
| 102 |
+
"<|km|>": 50323,
|
| 103 |
+
"<|kn|>": 50306,
|
| 104 |
+
"<|ko|>": 50264,
|
| 105 |
+
"<|la|>": 50294,
|
| 106 |
+
"<|lb|>": 50345,
|
| 107 |
+
"<|ln|>": 50353,
|
| 108 |
+
"<|lo|>": 50336,
|
| 109 |
+
"<|lt|>": 50293,
|
| 110 |
+
"<|lv|>": 50301,
|
| 111 |
+
"<|mg|>": 50349,
|
| 112 |
+
"<|mi|>": 50295,
|
| 113 |
+
"<|mk|>": 50308,
|
| 114 |
+
"<|ml|>": 50296,
|
| 115 |
+
"<|mn|>": 50314,
|
| 116 |
+
"<|mr|>": 50320,
|
| 117 |
+
"<|ms|>": 50282,
|
| 118 |
+
"<|mt|>": 50343,
|
| 119 |
+
"<|my|>": 50346,
|
| 120 |
+
"<|ne|>": 50313,
|
| 121 |
+
"<|nl|>": 50271,
|
| 122 |
+
"<|nn|>": 50342,
|
| 123 |
+
"<|no|>": 50288,
|
| 124 |
+
"<|oc|>": 50328,
|
| 125 |
+
"<|pa|>": 50321,
|
| 126 |
+
"<|pl|>": 50269,
|
| 127 |
+
"<|ps|>": 50340,
|
| 128 |
+
"<|pt|>": 50267,
|
| 129 |
+
"<|ro|>": 50284,
|
| 130 |
+
"<|ru|>": 50263,
|
| 131 |
+
"<|sa|>": 50344,
|
| 132 |
+
"<|sd|>": 50332,
|
| 133 |
+
"<|si|>": 50322,
|
| 134 |
+
"<|sk|>": 50298,
|
| 135 |
+
"<|sl|>": 50305,
|
| 136 |
+
"<|sn|>": 50324,
|
| 137 |
+
"<|so|>": 50326,
|
| 138 |
+
"<|sq|>": 50317,
|
| 139 |
+
"<|sr|>": 50303,
|
| 140 |
+
"<|su|>": 50357,
|
| 141 |
+
"<|sv|>": 50273,
|
| 142 |
+
"<|sw|>": 50318,
|
| 143 |
+
"<|ta|>": 50287,
|
| 144 |
+
"<|te|>": 50299,
|
| 145 |
+
"<|tg|>": 50331,
|
| 146 |
+
"<|th|>": 50289,
|
| 147 |
+
"<|tk|>": 50341,
|
| 148 |
+
"<|tl|>": 50348,
|
| 149 |
+
"<|tr|>": 50268,
|
| 150 |
+
"<|tt|>": 50351,
|
| 151 |
+
"<|uk|>": 50280,
|
| 152 |
+
"<|ur|>": 50290,
|
| 153 |
+
"<|uz|>": 50337,
|
| 154 |
+
"<|vi|>": 50278,
|
| 155 |
+
"<|yi|>": 50335,
|
| 156 |
+
"<|yo|>": 50325,
|
| 157 |
+
"<|zh|>": 50260
|
| 158 |
+
},
|
| 159 |
+
"max_initial_timestamp_index": 50,
|
| 160 |
+
"max_length": 448,
|
| 161 |
+
"no_timestamps_token_id": 50363,
|
| 162 |
+
"pad_token_id": 50257,
|
| 163 |
+
"prev_sot_token_id": 50361,
|
| 164 |
+
"return_timestamps": false,
|
| 165 |
+
"suppress_tokens": [
|
| 166 |
+
1,
|
| 167 |
+
2,
|
| 168 |
+
7,
|
| 169 |
+
8,
|
| 170 |
+
9,
|
| 171 |
+
10,
|
| 172 |
+
14,
|
| 173 |
+
25,
|
| 174 |
+
26,
|
| 175 |
+
27,
|
| 176 |
+
28,
|
| 177 |
+
29,
|
| 178 |
+
31,
|
| 179 |
+
58,
|
| 180 |
+
59,
|
| 181 |
+
60,
|
| 182 |
+
61,
|
| 183 |
+
62,
|
| 184 |
+
63,
|
| 185 |
+
90,
|
| 186 |
+
91,
|
| 187 |
+
92,
|
| 188 |
+
93,
|
| 189 |
+
359,
|
| 190 |
+
503,
|
| 191 |
+
522,
|
| 192 |
+
542,
|
| 193 |
+
873,
|
| 194 |
+
893,
|
| 195 |
+
902,
|
| 196 |
+
918,
|
| 197 |
+
922,
|
| 198 |
+
931,
|
| 199 |
+
1350,
|
| 200 |
+
1853,
|
| 201 |
+
1982,
|
| 202 |
+
2460,
|
| 203 |
+
2627,
|
| 204 |
+
3246,
|
| 205 |
+
3253,
|
| 206 |
+
3268,
|
| 207 |
+
3536,
|
| 208 |
+
3846,
|
| 209 |
+
3961,
|
| 210 |
+
4183,
|
| 211 |
+
4667,
|
| 212 |
+
6585,
|
| 213 |
+
6647,
|
| 214 |
+
7273,
|
| 215 |
+
9061,
|
| 216 |
+
9383,
|
| 217 |
+
10428,
|
| 218 |
+
10929,
|
| 219 |
+
11938,
|
| 220 |
+
12033,
|
| 221 |
+
12331,
|
| 222 |
+
12562,
|
| 223 |
+
13793,
|
| 224 |
+
14157,
|
| 225 |
+
14635,
|
| 226 |
+
15265,
|
| 227 |
+
15618,
|
| 228 |
+
16553,
|
| 229 |
+
16604,
|
| 230 |
+
18362,
|
| 231 |
+
18956,
|
| 232 |
+
20075,
|
| 233 |
+
21675,
|
| 234 |
+
22520,
|
| 235 |
+
26130,
|
| 236 |
+
26161,
|
| 237 |
+
26435,
|
| 238 |
+
28279,
|
| 239 |
+
29464,
|
| 240 |
+
31650,
|
| 241 |
+
32302,
|
| 242 |
+
32470,
|
| 243 |
+
36865,
|
| 244 |
+
42863,
|
| 245 |
+
47425,
|
| 246 |
+
49870,
|
| 247 |
+
50254,
|
| 248 |
+
50258,
|
| 249 |
+
50358,
|
| 250 |
+
50359,
|
| 251 |
+
50360,
|
| 252 |
+
50361,
|
| 253 |
+
50362
|
| 254 |
+
],
|
| 255 |
+
"task_to_id": {
|
| 256 |
+
"transcribe": 50359,
|
| 257 |
+
"translate": 50358
|
| 258 |
+
},
|
| 259 |
+
"transformers_version": "4.36.2"
|
| 260 |
+
}
|
pinyin/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4475b27906313fbda39edfd90adc75bbb98f52768327d389d1c9e27e68c2bc7
|
| 3 |
+
size 4992706480
|
pinyin/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:69c4823076b68fc350a0b31415dd995c7ae5257a1596a5b08d73d193c9f9e73f
|
| 3 |
+
size 1180663192
|
pinyin/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
pinyin/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e81c31e37293ca544b55df5386cc1d1e92054a409ad820c2cd06fc1c878a232
|
| 3 |
+
size 12332175139
|
pinyin/preprocessor_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chunk_length": 30,
|
| 3 |
+
"feature_extractor_type": "WhisperFeatureExtractor",
|
| 4 |
+
"feature_size": 80,
|
| 5 |
+
"hop_length": 160,
|
| 6 |
+
"n_fft": 400,
|
| 7 |
+
"n_samples": 480000,
|
| 8 |
+
"nb_max_frames": 3000,
|
| 9 |
+
"padding_side": "right",
|
| 10 |
+
"padding_value": 0.0,
|
| 11 |
+
"processor_class": "WhisperProcessor",
|
| 12 |
+
"return_attention_mask": false,
|
| 13 |
+
"sampling_rate": 16000
|
| 14 |
+
}
|
pinyin/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84d1f8afd11da187329cb9379a83fb4b4cad0edc9894da2cb7b10465cd1b1b7c
|
| 3 |
+
size 14575
|
pinyin/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be0b7627fed8c6ccefdfa70b1e77dc6072803bbe470acaff660313276beafa7d
|
| 3 |
+
size 627
|
pinyin/trainer_state.json
ADDED
|
@@ -0,0 +1,1122 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 23.738796414852754,
|
| 3 |
+
"best_model_checkpoint": "/app/data/hieunq/whisper_asr_train/whisper-large-chinese-pipyin/checkpoint-4410",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4410,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03,
|
| 13 |
+
"learning_rate": 6.144965986394558e-06,
|
| 14 |
+
"loss": 2.1824,
|
| 15 |
+
"step": 25
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.06,
|
| 19 |
+
"learning_rate": 6.109931972789116e-06,
|
| 20 |
+
"loss": 0.8831,
|
| 21 |
+
"step": 50
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.09,
|
| 25 |
+
"learning_rate": 6.074897959183673e-06,
|
| 26 |
+
"loss": 0.6375,
|
| 27 |
+
"step": 75
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.11,
|
| 31 |
+
"learning_rate": 6.039863945578232e-06,
|
| 32 |
+
"loss": 0.5378,
|
| 33 |
+
"step": 100
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.14,
|
| 37 |
+
"learning_rate": 6.004829931972789e-06,
|
| 38 |
+
"loss": 0.4835,
|
| 39 |
+
"step": 125
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.17,
|
| 43 |
+
"learning_rate": 5.969795918367347e-06,
|
| 44 |
+
"loss": 0.5321,
|
| 45 |
+
"step": 150
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.2,
|
| 49 |
+
"learning_rate": 5.934761904761905e-06,
|
| 50 |
+
"loss": 0.5099,
|
| 51 |
+
"step": 175
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.23,
|
| 55 |
+
"learning_rate": 5.899727891156462e-06,
|
| 56 |
+
"loss": 0.5304,
|
| 57 |
+
"step": 200
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.26,
|
| 61 |
+
"learning_rate": 5.864693877551021e-06,
|
| 62 |
+
"loss": 0.4958,
|
| 63 |
+
"step": 225
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.28,
|
| 67 |
+
"learning_rate": 5.8296598639455785e-06,
|
| 68 |
+
"loss": 0.4695,
|
| 69 |
+
"step": 250
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.31,
|
| 73 |
+
"learning_rate": 5.794625850340136e-06,
|
| 74 |
+
"loss": 0.4351,
|
| 75 |
+
"step": 275
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.34,
|
| 79 |
+
"learning_rate": 5.759591836734694e-06,
|
| 80 |
+
"loss": 0.4394,
|
| 81 |
+
"step": 300
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.37,
|
| 85 |
+
"learning_rate": 5.7245578231292515e-06,
|
| 86 |
+
"loss": 0.414,
|
| 87 |
+
"step": 325
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.4,
|
| 91 |
+
"learning_rate": 5.689523809523809e-06,
|
| 92 |
+
"loss": 0.4521,
|
| 93 |
+
"step": 350
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.43,
|
| 97 |
+
"learning_rate": 5.654489795918368e-06,
|
| 98 |
+
"loss": 0.4334,
|
| 99 |
+
"step": 375
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.45,
|
| 103 |
+
"learning_rate": 5.619455782312925e-06,
|
| 104 |
+
"loss": 0.4217,
|
| 105 |
+
"step": 400
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.48,
|
| 109 |
+
"learning_rate": 5.584421768707483e-06,
|
| 110 |
+
"loss": 0.4108,
|
| 111 |
+
"step": 425
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.51,
|
| 115 |
+
"learning_rate": 5.549387755102041e-06,
|
| 116 |
+
"loss": 0.4005,
|
| 117 |
+
"step": 450
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.54,
|
| 121 |
+
"learning_rate": 5.514353741496598e-06,
|
| 122 |
+
"loss": 0.4097,
|
| 123 |
+
"step": 475
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.57,
|
| 127 |
+
"learning_rate": 5.479319727891157e-06,
|
| 128 |
+
"loss": 0.3855,
|
| 129 |
+
"step": 500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.6,
|
| 133 |
+
"learning_rate": 5.4442857142857146e-06,
|
| 134 |
+
"loss": 0.4167,
|
| 135 |
+
"step": 525
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.62,
|
| 139 |
+
"learning_rate": 5.409251700680272e-06,
|
| 140 |
+
"loss": 0.3667,
|
| 141 |
+
"step": 550
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.65,
|
| 145 |
+
"learning_rate": 5.37421768707483e-06,
|
| 146 |
+
"loss": 0.4307,
|
| 147 |
+
"step": 575
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.68,
|
| 151 |
+
"learning_rate": 5.3391836734693876e-06,
|
| 152 |
+
"loss": 0.3693,
|
| 153 |
+
"step": 600
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.71,
|
| 157 |
+
"learning_rate": 5.304149659863946e-06,
|
| 158 |
+
"loss": 0.3089,
|
| 159 |
+
"step": 625
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.74,
|
| 163 |
+
"learning_rate": 5.269115646258504e-06,
|
| 164 |
+
"loss": 0.3185,
|
| 165 |
+
"step": 650
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.77,
|
| 169 |
+
"learning_rate": 5.234081632653061e-06,
|
| 170 |
+
"loss": 0.3783,
|
| 171 |
+
"step": 675
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.79,
|
| 175 |
+
"learning_rate": 5.199047619047619e-06,
|
| 176 |
+
"loss": 0.3681,
|
| 177 |
+
"step": 700
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.82,
|
| 181 |
+
"learning_rate": 5.164013605442177e-06,
|
| 182 |
+
"loss": 0.4317,
|
| 183 |
+
"step": 725
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.85,
|
| 187 |
+
"learning_rate": 5.128979591836735e-06,
|
| 188 |
+
"loss": 0.3815,
|
| 189 |
+
"step": 750
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.88,
|
| 193 |
+
"learning_rate": 5.093945578231293e-06,
|
| 194 |
+
"loss": 0.3618,
|
| 195 |
+
"step": 775
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.91,
|
| 199 |
+
"learning_rate": 5.05891156462585e-06,
|
| 200 |
+
"loss": 0.3376,
|
| 201 |
+
"step": 800
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.94,
|
| 205 |
+
"learning_rate": 5.023877551020408e-06,
|
| 206 |
+
"loss": 0.3485,
|
| 207 |
+
"step": 825
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.96,
|
| 211 |
+
"learning_rate": 4.988843537414966e-06,
|
| 212 |
+
"loss": 0.3397,
|
| 213 |
+
"step": 850
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.99,
|
| 217 |
+
"learning_rate": 4.9538095238095245e-06,
|
| 218 |
+
"loss": 0.3836,
|
| 219 |
+
"step": 875
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 1.0,
|
| 223 |
+
"eval_loss": 0.342790424823761,
|
| 224 |
+
"eval_runtime": 1384.9795,
|
| 225 |
+
"eval_samples_per_second": 1.274,
|
| 226 |
+
"eval_steps_per_second": 0.16,
|
| 227 |
+
"eval_wer": 27.76781903542467,
|
| 228 |
+
"step": 882
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.02,
|
| 232 |
+
"learning_rate": 4.918775510204081e-06,
|
| 233 |
+
"loss": 0.2476,
|
| 234 |
+
"step": 900
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 1.05,
|
| 238 |
+
"learning_rate": 4.883741496598639e-06,
|
| 239 |
+
"loss": 0.1817,
|
| 240 |
+
"step": 925
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 1.08,
|
| 244 |
+
"learning_rate": 4.8487074829931975e-06,
|
| 245 |
+
"loss": 0.2218,
|
| 246 |
+
"step": 950
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 1.11,
|
| 250 |
+
"learning_rate": 4.813673469387755e-06,
|
| 251 |
+
"loss": 0.1971,
|
| 252 |
+
"step": 975
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 1.13,
|
| 256 |
+
"learning_rate": 4.778639455782314e-06,
|
| 257 |
+
"loss": 0.2224,
|
| 258 |
+
"step": 1000
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 1.16,
|
| 262 |
+
"learning_rate": 4.7436054421768705e-06,
|
| 263 |
+
"loss": 0.1781,
|
| 264 |
+
"step": 1025
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.19,
|
| 268 |
+
"learning_rate": 4.708571428571428e-06,
|
| 269 |
+
"loss": 0.2359,
|
| 270 |
+
"step": 1050
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.22,
|
| 274 |
+
"learning_rate": 4.673537414965987e-06,
|
| 275 |
+
"loss": 0.1955,
|
| 276 |
+
"step": 1075
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 1.25,
|
| 280 |
+
"learning_rate": 4.638503401360544e-06,
|
| 281 |
+
"loss": 0.2005,
|
| 282 |
+
"step": 1100
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.28,
|
| 286 |
+
"learning_rate": 4.603469387755103e-06,
|
| 287 |
+
"loss": 0.2131,
|
| 288 |
+
"step": 1125
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.3,
|
| 292 |
+
"learning_rate": 4.56843537414966e-06,
|
| 293 |
+
"loss": 0.179,
|
| 294 |
+
"step": 1150
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 1.33,
|
| 298 |
+
"learning_rate": 4.533401360544217e-06,
|
| 299 |
+
"loss": 0.2156,
|
| 300 |
+
"step": 1175
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 1.36,
|
| 304 |
+
"learning_rate": 4.498367346938776e-06,
|
| 305 |
+
"loss": 0.186,
|
| 306 |
+
"step": 1200
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 1.39,
|
| 310 |
+
"learning_rate": 4.4633333333333335e-06,
|
| 311 |
+
"loss": 0.2004,
|
| 312 |
+
"step": 1225
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.42,
|
| 316 |
+
"learning_rate": 4.428299319727891e-06,
|
| 317 |
+
"loss": 0.2206,
|
| 318 |
+
"step": 1250
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.45,
|
| 322 |
+
"learning_rate": 4.393265306122449e-06,
|
| 323 |
+
"loss": 0.2029,
|
| 324 |
+
"step": 1275
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.47,
|
| 328 |
+
"learning_rate": 4.3582312925170065e-06,
|
| 329 |
+
"loss": 0.1952,
|
| 330 |
+
"step": 1300
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.5,
|
| 334 |
+
"learning_rate": 4.323197278911565e-06,
|
| 335 |
+
"loss": 0.2412,
|
| 336 |
+
"step": 1325
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.53,
|
| 340 |
+
"learning_rate": 4.288163265306123e-06,
|
| 341 |
+
"loss": 0.2011,
|
| 342 |
+
"step": 1350
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.56,
|
| 346 |
+
"learning_rate": 4.25312925170068e-06,
|
| 347 |
+
"loss": 0.1771,
|
| 348 |
+
"step": 1375
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 1.59,
|
| 352 |
+
"learning_rate": 4.218095238095238e-06,
|
| 353 |
+
"loss": 0.2092,
|
| 354 |
+
"step": 1400
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.62,
|
| 358 |
+
"learning_rate": 4.183061224489796e-06,
|
| 359 |
+
"loss": 0.1916,
|
| 360 |
+
"step": 1425
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.64,
|
| 364 |
+
"learning_rate": 4.148027210884354e-06,
|
| 365 |
+
"loss": 0.191,
|
| 366 |
+
"step": 1450
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.67,
|
| 370 |
+
"learning_rate": 4.112993197278912e-06,
|
| 371 |
+
"loss": 0.1907,
|
| 372 |
+
"step": 1475
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.7,
|
| 376 |
+
"learning_rate": 4.0779591836734696e-06,
|
| 377 |
+
"loss": 0.1984,
|
| 378 |
+
"step": 1500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.73,
|
| 382 |
+
"learning_rate": 4.042925170068027e-06,
|
| 383 |
+
"loss": 0.1986,
|
| 384 |
+
"step": 1525
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.76,
|
| 388 |
+
"learning_rate": 4.007891156462585e-06,
|
| 389 |
+
"loss": 0.196,
|
| 390 |
+
"step": 1550
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.79,
|
| 394 |
+
"learning_rate": 3.972857142857143e-06,
|
| 395 |
+
"loss": 0.1685,
|
| 396 |
+
"step": 1575
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 1.81,
|
| 400 |
+
"learning_rate": 3.937823129251701e-06,
|
| 401 |
+
"loss": 0.2104,
|
| 402 |
+
"step": 1600
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.84,
|
| 406 |
+
"learning_rate": 3.902789115646259e-06,
|
| 407 |
+
"loss": 0.2079,
|
| 408 |
+
"step": 1625
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.87,
|
| 412 |
+
"learning_rate": 3.8677551020408164e-06,
|
| 413 |
+
"loss": 0.2119,
|
| 414 |
+
"step": 1650
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.9,
|
| 418 |
+
"learning_rate": 3.832721088435374e-06,
|
| 419 |
+
"loss": 0.2,
|
| 420 |
+
"step": 1675
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.93,
|
| 424 |
+
"learning_rate": 3.797687074829932e-06,
|
| 425 |
+
"loss": 0.1765,
|
| 426 |
+
"step": 1700
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 1.96,
|
| 430 |
+
"learning_rate": 3.76265306122449e-06,
|
| 431 |
+
"loss": 0.2026,
|
| 432 |
+
"step": 1725
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.98,
|
| 436 |
+
"learning_rate": 3.7276190476190475e-06,
|
| 437 |
+
"loss": 0.1914,
|
| 438 |
+
"step": 1750
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 2.0,
|
| 442 |
+
"eval_loss": 0.32358798384666443,
|
| 443 |
+
"eval_runtime": 1373.0201,
|
| 444 |
+
"eval_samples_per_second": 1.285,
|
| 445 |
+
"eval_steps_per_second": 0.161,
|
| 446 |
+
"eval_wer": 25.761843790012804,
|
| 447 |
+
"step": 1764
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 2.01,
|
| 451 |
+
"learning_rate": 3.6925850340136056e-06,
|
| 452 |
+
"loss": 0.1853,
|
| 453 |
+
"step": 1775
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 2.04,
|
| 457 |
+
"learning_rate": 3.6575510204081633e-06,
|
| 458 |
+
"loss": 0.104,
|
| 459 |
+
"step": 1800
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 2.07,
|
| 463 |
+
"learning_rate": 3.622517006802721e-06,
|
| 464 |
+
"loss": 0.1111,
|
| 465 |
+
"step": 1825
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 2.1,
|
| 469 |
+
"learning_rate": 3.587482993197279e-06,
|
| 470 |
+
"loss": 0.1044,
|
| 471 |
+
"step": 1850
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 2.13,
|
| 475 |
+
"learning_rate": 3.5524489795918367e-06,
|
| 476 |
+
"loss": 0.0938,
|
| 477 |
+
"step": 1875
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 2.15,
|
| 481 |
+
"learning_rate": 3.517414965986395e-06,
|
| 482 |
+
"loss": 0.1344,
|
| 483 |
+
"step": 1900
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 2.18,
|
| 487 |
+
"learning_rate": 3.4823809523809525e-06,
|
| 488 |
+
"loss": 0.1153,
|
| 489 |
+
"step": 1925
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 2.21,
|
| 493 |
+
"learning_rate": 3.44734693877551e-06,
|
| 494 |
+
"loss": 0.0989,
|
| 495 |
+
"step": 1950
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 2.24,
|
| 499 |
+
"learning_rate": 3.4123129251700682e-06,
|
| 500 |
+
"loss": 0.1086,
|
| 501 |
+
"step": 1975
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 2.27,
|
| 505 |
+
"learning_rate": 3.377278911564626e-06,
|
| 506 |
+
"loss": 0.1255,
|
| 507 |
+
"step": 2000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 2.3,
|
| 511 |
+
"learning_rate": 3.342244897959184e-06,
|
| 512 |
+
"loss": 0.1093,
|
| 513 |
+
"step": 2025
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.32,
|
| 517 |
+
"learning_rate": 3.3072108843537417e-06,
|
| 518 |
+
"loss": 0.0984,
|
| 519 |
+
"step": 2050
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 2.35,
|
| 523 |
+
"learning_rate": 3.2721768707482993e-06,
|
| 524 |
+
"loss": 0.0984,
|
| 525 |
+
"step": 2075
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 2.38,
|
| 529 |
+
"learning_rate": 3.2371428571428574e-06,
|
| 530 |
+
"loss": 0.1139,
|
| 531 |
+
"step": 2100
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 2.41,
|
| 535 |
+
"learning_rate": 3.202108843537415e-06,
|
| 536 |
+
"loss": 0.1043,
|
| 537 |
+
"step": 2125
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 2.44,
|
| 541 |
+
"learning_rate": 3.167074829931973e-06,
|
| 542 |
+
"loss": 0.0983,
|
| 543 |
+
"step": 2150
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 2.47,
|
| 547 |
+
"learning_rate": 3.132040816326531e-06,
|
| 548 |
+
"loss": 0.1194,
|
| 549 |
+
"step": 2175
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 2.49,
|
| 553 |
+
"learning_rate": 3.097006802721088e-06,
|
| 554 |
+
"loss": 0.0958,
|
| 555 |
+
"step": 2200
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 2.52,
|
| 559 |
+
"learning_rate": 3.061972789115646e-06,
|
| 560 |
+
"loss": 0.0857,
|
| 561 |
+
"step": 2225
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 2.55,
|
| 565 |
+
"learning_rate": 3.026938775510204e-06,
|
| 566 |
+
"loss": 0.1022,
|
| 567 |
+
"step": 2250
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 2.58,
|
| 571 |
+
"learning_rate": 2.991904761904762e-06,
|
| 572 |
+
"loss": 0.1118,
|
| 573 |
+
"step": 2275
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 2.61,
|
| 577 |
+
"learning_rate": 2.95687074829932e-06,
|
| 578 |
+
"loss": 0.092,
|
| 579 |
+
"step": 2300
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 2.64,
|
| 583 |
+
"learning_rate": 2.9218367346938777e-06,
|
| 584 |
+
"loss": 0.1063,
|
| 585 |
+
"step": 2325
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 2.66,
|
| 589 |
+
"learning_rate": 2.8868027210884354e-06,
|
| 590 |
+
"loss": 0.1028,
|
| 591 |
+
"step": 2350
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.69,
|
| 595 |
+
"learning_rate": 2.851768707482993e-06,
|
| 596 |
+
"loss": 0.1077,
|
| 597 |
+
"step": 2375
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 2.72,
|
| 601 |
+
"learning_rate": 2.816734693877551e-06,
|
| 602 |
+
"loss": 0.0974,
|
| 603 |
+
"step": 2400
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 2.75,
|
| 607 |
+
"learning_rate": 2.781700680272109e-06,
|
| 608 |
+
"loss": 0.108,
|
| 609 |
+
"step": 2425
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 2.78,
|
| 613 |
+
"learning_rate": 2.7466666666666665e-06,
|
| 614 |
+
"loss": 0.0852,
|
| 615 |
+
"step": 2450
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 2.81,
|
| 619 |
+
"learning_rate": 2.7116326530612246e-06,
|
| 620 |
+
"loss": 0.1009,
|
| 621 |
+
"step": 2475
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 2.83,
|
| 625 |
+
"learning_rate": 2.6765986394557822e-06,
|
| 626 |
+
"loss": 0.1127,
|
| 627 |
+
"step": 2500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.86,
|
| 631 |
+
"learning_rate": 2.6415646258503403e-06,
|
| 632 |
+
"loss": 0.1072,
|
| 633 |
+
"step": 2525
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.89,
|
| 637 |
+
"learning_rate": 2.606530612244898e-06,
|
| 638 |
+
"loss": 0.0893,
|
| 639 |
+
"step": 2550
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.92,
|
| 643 |
+
"learning_rate": 2.5714965986394557e-06,
|
| 644 |
+
"loss": 0.0945,
|
| 645 |
+
"step": 2575
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 2.95,
|
| 649 |
+
"learning_rate": 2.5364625850340138e-06,
|
| 650 |
+
"loss": 0.1149,
|
| 651 |
+
"step": 2600
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 2.98,
|
| 655 |
+
"learning_rate": 2.5014285714285714e-06,
|
| 656 |
+
"loss": 0.0844,
|
| 657 |
+
"step": 2625
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.0,
|
| 661 |
+
"eval_loss": 0.33547088503837585,
|
| 662 |
+
"eval_runtime": 1375.3795,
|
| 663 |
+
"eval_samples_per_second": 1.283,
|
| 664 |
+
"eval_steps_per_second": 0.161,
|
| 665 |
+
"eval_wer": 24.609475032010245,
|
| 666 |
+
"step": 2646
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 3.0,
|
| 670 |
+
"learning_rate": 2.4663945578231295e-06,
|
| 671 |
+
"loss": 0.0889,
|
| 672 |
+
"step": 2650
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 3.03,
|
| 676 |
+
"learning_rate": 2.4313605442176868e-06,
|
| 677 |
+
"loss": 0.0382,
|
| 678 |
+
"step": 2675
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 3.06,
|
| 682 |
+
"learning_rate": 2.396326530612245e-06,
|
| 683 |
+
"loss": 0.0627,
|
| 684 |
+
"step": 2700
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.09,
|
| 688 |
+
"learning_rate": 2.361292517006803e-06,
|
| 689 |
+
"loss": 0.0562,
|
| 690 |
+
"step": 2725
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 3.12,
|
| 694 |
+
"learning_rate": 2.3262585034013606e-06,
|
| 695 |
+
"loss": 0.049,
|
| 696 |
+
"step": 2750
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 3.15,
|
| 700 |
+
"learning_rate": 2.2912244897959187e-06,
|
| 701 |
+
"loss": 0.0432,
|
| 702 |
+
"step": 2775
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 3.17,
|
| 706 |
+
"learning_rate": 2.256190476190476e-06,
|
| 707 |
+
"loss": 0.0594,
|
| 708 |
+
"step": 2800
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 3.2,
|
| 712 |
+
"learning_rate": 2.221156462585034e-06,
|
| 713 |
+
"loss": 0.0415,
|
| 714 |
+
"step": 2825
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 3.23,
|
| 718 |
+
"learning_rate": 2.1861224489795917e-06,
|
| 719 |
+
"loss": 0.0451,
|
| 720 |
+
"step": 2850
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 3.26,
|
| 724 |
+
"learning_rate": 2.15108843537415e-06,
|
| 725 |
+
"loss": 0.045,
|
| 726 |
+
"step": 2875
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 3.29,
|
| 730 |
+
"learning_rate": 2.1160544217687075e-06,
|
| 731 |
+
"loss": 0.0503,
|
| 732 |
+
"step": 2900
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 3.32,
|
| 736 |
+
"learning_rate": 2.081020408163265e-06,
|
| 737 |
+
"loss": 0.0368,
|
| 738 |
+
"step": 2925
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 3.34,
|
| 742 |
+
"learning_rate": 2.0459863945578232e-06,
|
| 743 |
+
"loss": 0.0447,
|
| 744 |
+
"step": 2950
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 3.37,
|
| 748 |
+
"learning_rate": 2.010952380952381e-06,
|
| 749 |
+
"loss": 0.041,
|
| 750 |
+
"step": 2975
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.4,
|
| 754 |
+
"learning_rate": 1.975918367346939e-06,
|
| 755 |
+
"loss": 0.054,
|
| 756 |
+
"step": 3000
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 3.43,
|
| 760 |
+
"learning_rate": 1.9408843537414967e-06,
|
| 761 |
+
"loss": 0.0409,
|
| 762 |
+
"step": 3025
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.46,
|
| 766 |
+
"learning_rate": 1.9058503401360543e-06,
|
| 767 |
+
"loss": 0.0423,
|
| 768 |
+
"step": 3050
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.49,
|
| 772 |
+
"learning_rate": 1.8708163265306122e-06,
|
| 773 |
+
"loss": 0.0485,
|
| 774 |
+
"step": 3075
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.51,
|
| 778 |
+
"learning_rate": 1.8357823129251701e-06,
|
| 779 |
+
"loss": 0.048,
|
| 780 |
+
"step": 3100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 3.54,
|
| 784 |
+
"learning_rate": 1.8007482993197282e-06,
|
| 785 |
+
"loss": 0.0509,
|
| 786 |
+
"step": 3125
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 3.57,
|
| 790 |
+
"learning_rate": 1.7657142857142857e-06,
|
| 791 |
+
"loss": 0.0525,
|
| 792 |
+
"step": 3150
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 3.6,
|
| 796 |
+
"learning_rate": 1.7306802721088435e-06,
|
| 797 |
+
"loss": 0.0539,
|
| 798 |
+
"step": 3175
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 3.63,
|
| 802 |
+
"learning_rate": 1.6956462585034014e-06,
|
| 803 |
+
"loss": 0.0417,
|
| 804 |
+
"step": 3200
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.66,
|
| 808 |
+
"learning_rate": 1.6606122448979593e-06,
|
| 809 |
+
"loss": 0.0415,
|
| 810 |
+
"step": 3225
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 3.68,
|
| 814 |
+
"learning_rate": 1.625578231292517e-06,
|
| 815 |
+
"loss": 0.0557,
|
| 816 |
+
"step": 3250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 3.71,
|
| 820 |
+
"learning_rate": 1.5905442176870748e-06,
|
| 821 |
+
"loss": 0.0458,
|
| 822 |
+
"step": 3275
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 3.74,
|
| 826 |
+
"learning_rate": 1.5555102040816327e-06,
|
| 827 |
+
"loss": 0.0425,
|
| 828 |
+
"step": 3300
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 3.77,
|
| 832 |
+
"learning_rate": 1.5204761904761904e-06,
|
| 833 |
+
"loss": 0.0505,
|
| 834 |
+
"step": 3325
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 3.8,
|
| 838 |
+
"learning_rate": 1.4854421768707483e-06,
|
| 839 |
+
"loss": 0.0505,
|
| 840 |
+
"step": 3350
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 3.83,
|
| 844 |
+
"learning_rate": 1.4504081632653062e-06,
|
| 845 |
+
"loss": 0.0445,
|
| 846 |
+
"step": 3375
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 3.85,
|
| 850 |
+
"learning_rate": 1.415374149659864e-06,
|
| 851 |
+
"loss": 0.0462,
|
| 852 |
+
"step": 3400
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 3.88,
|
| 856 |
+
"learning_rate": 1.380340136054422e-06,
|
| 857 |
+
"loss": 0.0556,
|
| 858 |
+
"step": 3425
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 3.91,
|
| 862 |
+
"learning_rate": 1.3453061224489796e-06,
|
| 863 |
+
"loss": 0.0459,
|
| 864 |
+
"step": 3450
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 3.94,
|
| 868 |
+
"learning_rate": 1.3102721088435375e-06,
|
| 869 |
+
"loss": 0.0439,
|
| 870 |
+
"step": 3475
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 3.97,
|
| 874 |
+
"learning_rate": 1.2752380952380951e-06,
|
| 875 |
+
"loss": 0.0586,
|
| 876 |
+
"step": 3500
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 4.0,
|
| 880 |
+
"learning_rate": 1.240204081632653e-06,
|
| 881 |
+
"loss": 0.0418,
|
| 882 |
+
"step": 3525
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 4.0,
|
| 886 |
+
"eval_loss": 0.37163934111595154,
|
| 887 |
+
"eval_runtime": 1376.5549,
|
| 888 |
+
"eval_samples_per_second": 1.281,
|
| 889 |
+
"eval_steps_per_second": 0.161,
|
| 890 |
+
"eval_wer": 23.781476739223216,
|
| 891 |
+
"step": 3528
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 4.02,
|
| 895 |
+
"learning_rate": 1.205170068027211e-06,
|
| 896 |
+
"loss": 0.0216,
|
| 897 |
+
"step": 3550
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 4.05,
|
| 901 |
+
"learning_rate": 1.1701360544217688e-06,
|
| 902 |
+
"loss": 0.0167,
|
| 903 |
+
"step": 3575
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 4.08,
|
| 907 |
+
"learning_rate": 1.1351020408163267e-06,
|
| 908 |
+
"loss": 0.0179,
|
| 909 |
+
"step": 3600
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 4.11,
|
| 913 |
+
"learning_rate": 1.1000680272108843e-06,
|
| 914 |
+
"loss": 0.016,
|
| 915 |
+
"step": 3625
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 4.14,
|
| 919 |
+
"learning_rate": 1.0650340136054422e-06,
|
| 920 |
+
"loss": 0.0157,
|
| 921 |
+
"step": 3650
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 4.17,
|
| 925 |
+
"learning_rate": 1.0299999999999999e-06,
|
| 926 |
+
"loss": 0.019,
|
| 927 |
+
"step": 3675
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 4.2,
|
| 931 |
+
"learning_rate": 9.949659863945578e-07,
|
| 932 |
+
"loss": 0.017,
|
| 933 |
+
"step": 3700
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 4.22,
|
| 937 |
+
"learning_rate": 9.599319727891156e-07,
|
| 938 |
+
"loss": 0.0193,
|
| 939 |
+
"step": 3725
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 4.25,
|
| 943 |
+
"learning_rate": 9.248979591836735e-07,
|
| 944 |
+
"loss": 0.0144,
|
| 945 |
+
"step": 3750
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 4.28,
|
| 949 |
+
"learning_rate": 8.898639455782314e-07,
|
| 950 |
+
"loss": 0.0161,
|
| 951 |
+
"step": 3775
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 4.31,
|
| 955 |
+
"learning_rate": 8.548299319727891e-07,
|
| 956 |
+
"loss": 0.018,
|
| 957 |
+
"step": 3800
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 4.34,
|
| 961 |
+
"learning_rate": 8.197959183673471e-07,
|
| 962 |
+
"loss": 0.0173,
|
| 963 |
+
"step": 3825
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 4.37,
|
| 967 |
+
"learning_rate": 7.847619047619047e-07,
|
| 968 |
+
"loss": 0.013,
|
| 969 |
+
"step": 3850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 4.39,
|
| 973 |
+
"learning_rate": 7.497278911564626e-07,
|
| 974 |
+
"loss": 0.018,
|
| 975 |
+
"step": 3875
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 4.42,
|
| 979 |
+
"learning_rate": 7.146938775510205e-07,
|
| 980 |
+
"loss": 0.012,
|
| 981 |
+
"step": 3900
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 4.45,
|
| 985 |
+
"learning_rate": 6.796598639455783e-07,
|
| 986 |
+
"loss": 0.0224,
|
| 987 |
+
"step": 3925
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 4.48,
|
| 991 |
+
"learning_rate": 6.44625850340136e-07,
|
| 992 |
+
"loss": 0.0142,
|
| 993 |
+
"step": 3950
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 4.51,
|
| 997 |
+
"learning_rate": 6.095918367346939e-07,
|
| 998 |
+
"loss": 0.0167,
|
| 999 |
+
"step": 3975
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 4.54,
|
| 1003 |
+
"learning_rate": 5.745578231292517e-07,
|
| 1004 |
+
"loss": 0.0136,
|
| 1005 |
+
"step": 4000
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 4.56,
|
| 1009 |
+
"learning_rate": 5.395238095238095e-07,
|
| 1010 |
+
"loss": 0.0184,
|
| 1011 |
+
"step": 4025
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 4.59,
|
| 1015 |
+
"learning_rate": 5.044897959183673e-07,
|
| 1016 |
+
"loss": 0.0168,
|
| 1017 |
+
"step": 4050
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 4.62,
|
| 1021 |
+
"learning_rate": 4.6945578231292523e-07,
|
| 1022 |
+
"loss": 0.0137,
|
| 1023 |
+
"step": 4075
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 4.65,
|
| 1027 |
+
"learning_rate": 4.34421768707483e-07,
|
| 1028 |
+
"loss": 0.0133,
|
| 1029 |
+
"step": 4100
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 4.68,
|
| 1033 |
+
"learning_rate": 3.9938775510204083e-07,
|
| 1034 |
+
"loss": 0.0141,
|
| 1035 |
+
"step": 4125
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 4.71,
|
| 1039 |
+
"learning_rate": 3.6435374149659866e-07,
|
| 1040 |
+
"loss": 0.0125,
|
| 1041 |
+
"step": 4150
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 4.73,
|
| 1045 |
+
"learning_rate": 3.293197278911565e-07,
|
| 1046 |
+
"loss": 0.0261,
|
| 1047 |
+
"step": 4175
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 4.76,
|
| 1051 |
+
"learning_rate": 2.9428571428571426e-07,
|
| 1052 |
+
"loss": 0.0153,
|
| 1053 |
+
"step": 4200
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 4.79,
|
| 1057 |
+
"learning_rate": 2.5925170068027214e-07,
|
| 1058 |
+
"loss": 0.0115,
|
| 1059 |
+
"step": 4225
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 4.82,
|
| 1063 |
+
"learning_rate": 2.2421768707482994e-07,
|
| 1064 |
+
"loss": 0.0155,
|
| 1065 |
+
"step": 4250
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 4.85,
|
| 1069 |
+
"learning_rate": 1.8918367346938774e-07,
|
| 1070 |
+
"loss": 0.0126,
|
| 1071 |
+
"step": 4275
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 4.88,
|
| 1075 |
+
"learning_rate": 1.541496598639456e-07,
|
| 1076 |
+
"loss": 0.0146,
|
| 1077 |
+
"step": 4300
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 4.9,
|
| 1081 |
+
"learning_rate": 1.191156462585034e-07,
|
| 1082 |
+
"loss": 0.0199,
|
| 1083 |
+
"step": 4325
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 4.93,
|
| 1087 |
+
"learning_rate": 8.408163265306122e-08,
|
| 1088 |
+
"loss": 0.0149,
|
| 1089 |
+
"step": 4350
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.96,
|
| 1093 |
+
"learning_rate": 4.9047619047619045e-08,
|
| 1094 |
+
"loss": 0.0104,
|
| 1095 |
+
"step": 4375
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 4.99,
|
| 1099 |
+
"learning_rate": 1.4013605442176871e-08,
|
| 1100 |
+
"loss": 0.0121,
|
| 1101 |
+
"step": 4400
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 5.0,
|
| 1105 |
+
"eval_loss": 0.4422626793384552,
|
| 1106 |
+
"eval_runtime": 1364.6517,
|
| 1107 |
+
"eval_samples_per_second": 1.293,
|
| 1108 |
+
"eval_steps_per_second": 0.162,
|
| 1109 |
+
"eval_wer": 23.738796414852754,
|
| 1110 |
+
"step": 4410
|
| 1111 |
+
}
|
| 1112 |
+
],
|
| 1113 |
+
"logging_steps": 25,
|
| 1114 |
+
"max_steps": 4410,
|
| 1115 |
+
"num_input_tokens_seen": 0,
|
| 1116 |
+
"num_train_epochs": 5,
|
| 1117 |
+
"save_steps": 500,
|
| 1118 |
+
"total_flos": 7.4863010193408e+19,
|
| 1119 |
+
"train_batch_size": 8,
|
| 1120 |
+
"trial_name": null,
|
| 1121 |
+
"trial_params": null
|
| 1122 |
+
}
|
pinyin/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7fc0127f3f5d081caf316d6cfc7160c2506932a8ffd4ac87a4b83bd1770da8be
|
| 3 |
+
size 4539
|