Upload folder using huggingface_hub
Browse files- checkpoint-9804/config.json +66 -0
- checkpoint-9804/generation_config.json +317 -0
- checkpoint-9804/model-00001-of-00002.safetensors +3 -0
- checkpoint-9804/model-00002-of-00002.safetensors +3 -0
- checkpoint-9804/model.safetensors.index.json +0 -0
- checkpoint-9804/optimizer.pt +3 -0
- checkpoint-9804/preprocessor_config.json +14 -0
- checkpoint-9804/rng_state.pth +3 -0
- checkpoint-9804/scheduler.pt +3 -0
- checkpoint-9804/trainer_state.json +1857 -0
- checkpoint-9804/training_args.bin +3 -0
- config.json +66 -0
- generation_config.json +317 -0
- model-00001-of-00002.safetensors +3 -0
- model-00002-of-00002.safetensors +3 -0
- model.safetensors.index.json +0 -0
- preprocessor_config.json +14 -0
- training_args.bin +3 -0
checkpoint-9804/config.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "openai/whisper-large-v2",
|
| 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 |
+
50259
|
| 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 |
+
"language": "English",
|
| 45 |
+
"mask_feature_length": 10,
|
| 46 |
+
"mask_feature_min_masks": 0,
|
| 47 |
+
"mask_feature_prob": 0.0,
|
| 48 |
+
"mask_time_length": 10,
|
| 49 |
+
"mask_time_min_masks": 2,
|
| 50 |
+
"mask_time_prob": 0.05,
|
| 51 |
+
"max_length": 448,
|
| 52 |
+
"max_source_positions": 1500,
|
| 53 |
+
"max_target_positions": 448,
|
| 54 |
+
"median_filter_width": 7,
|
| 55 |
+
"model_type": "whisper",
|
| 56 |
+
"num_hidden_layers": 32,
|
| 57 |
+
"num_mel_bins": 80,
|
| 58 |
+
"pad_token_id": 50257,
|
| 59 |
+
"scale_embedding": false,
|
| 60 |
+
"suppress_tokens": [],
|
| 61 |
+
"torch_dtype": "float32",
|
| 62 |
+
"transformers_version": "4.37.2",
|
| 63 |
+
"use_cache": true,
|
| 64 |
+
"use_weighted_layer_sum": false,
|
| 65 |
+
"vocab_size": 51865
|
| 66 |
+
}
|
checkpoint-9804/generation_config.json
ADDED
|
@@ -0,0 +1,317 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alignment_heads": [
|
| 3 |
+
[
|
| 4 |
+
10,
|
| 5 |
+
12
|
| 6 |
+
],
|
| 7 |
+
[
|
| 8 |
+
13,
|
| 9 |
+
17
|
| 10 |
+
],
|
| 11 |
+
[
|
| 12 |
+
16,
|
| 13 |
+
11
|
| 14 |
+
],
|
| 15 |
+
[
|
| 16 |
+
16,
|
| 17 |
+
12
|
| 18 |
+
],
|
| 19 |
+
[
|
| 20 |
+
16,
|
| 21 |
+
13
|
| 22 |
+
],
|
| 23 |
+
[
|
| 24 |
+
17,
|
| 25 |
+
15
|
| 26 |
+
],
|
| 27 |
+
[
|
| 28 |
+
17,
|
| 29 |
+
16
|
| 30 |
+
],
|
| 31 |
+
[
|
| 32 |
+
18,
|
| 33 |
+
4
|
| 34 |
+
],
|
| 35 |
+
[
|
| 36 |
+
18,
|
| 37 |
+
11
|
| 38 |
+
],
|
| 39 |
+
[
|
| 40 |
+
18,
|
| 41 |
+
19
|
| 42 |
+
],
|
| 43 |
+
[
|
| 44 |
+
19,
|
| 45 |
+
11
|
| 46 |
+
],
|
| 47 |
+
[
|
| 48 |
+
21,
|
| 49 |
+
2
|
| 50 |
+
],
|
| 51 |
+
[
|
| 52 |
+
21,
|
| 53 |
+
3
|
| 54 |
+
],
|
| 55 |
+
[
|
| 56 |
+
22,
|
| 57 |
+
3
|
| 58 |
+
],
|
| 59 |
+
[
|
| 60 |
+
22,
|
| 61 |
+
9
|
| 62 |
+
],
|
| 63 |
+
[
|
| 64 |
+
22,
|
| 65 |
+
12
|
| 66 |
+
],
|
| 67 |
+
[
|
| 68 |
+
23,
|
| 69 |
+
5
|
| 70 |
+
],
|
| 71 |
+
[
|
| 72 |
+
23,
|
| 73 |
+
7
|
| 74 |
+
],
|
| 75 |
+
[
|
| 76 |
+
23,
|
| 77 |
+
13
|
| 78 |
+
],
|
| 79 |
+
[
|
| 80 |
+
25,
|
| 81 |
+
5
|
| 82 |
+
],
|
| 83 |
+
[
|
| 84 |
+
26,
|
| 85 |
+
1
|
| 86 |
+
],
|
| 87 |
+
[
|
| 88 |
+
26,
|
| 89 |
+
12
|
| 90 |
+
],
|
| 91 |
+
[
|
| 92 |
+
27,
|
| 93 |
+
15
|
| 94 |
+
]
|
| 95 |
+
],
|
| 96 |
+
"begin_suppress_tokens": [
|
| 97 |
+
220,
|
| 98 |
+
50257
|
| 99 |
+
],
|
| 100 |
+
"bos_token_id": 50257,
|
| 101 |
+
"decoder_start_token_id": 50258,
|
| 102 |
+
"eos_token_id": 50257,
|
| 103 |
+
"forced_decoder_ids": [
|
| 104 |
+
[
|
| 105 |
+
1,
|
| 106 |
+
null
|
| 107 |
+
],
|
| 108 |
+
[
|
| 109 |
+
2,
|
| 110 |
+
50359
|
| 111 |
+
]
|
| 112 |
+
],
|
| 113 |
+
"is_multilingual": true,
|
| 114 |
+
"lang_to_id": {
|
| 115 |
+
"<|af|>": 50327,
|
| 116 |
+
"<|am|>": 50334,
|
| 117 |
+
"<|ar|>": 50272,
|
| 118 |
+
"<|as|>": 50350,
|
| 119 |
+
"<|az|>": 50304,
|
| 120 |
+
"<|ba|>": 50355,
|
| 121 |
+
"<|be|>": 50330,
|
| 122 |
+
"<|bg|>": 50292,
|
| 123 |
+
"<|bn|>": 50302,
|
| 124 |
+
"<|bo|>": 50347,
|
| 125 |
+
"<|br|>": 50309,
|
| 126 |
+
"<|bs|>": 50315,
|
| 127 |
+
"<|ca|>": 50270,
|
| 128 |
+
"<|cs|>": 50283,
|
| 129 |
+
"<|cy|>": 50297,
|
| 130 |
+
"<|da|>": 50285,
|
| 131 |
+
"<|de|>": 50261,
|
| 132 |
+
"<|el|>": 50281,
|
| 133 |
+
"<|en|>": 50259,
|
| 134 |
+
"<|es|>": 50262,
|
| 135 |
+
"<|et|>": 50307,
|
| 136 |
+
"<|eu|>": 50310,
|
| 137 |
+
"<|fa|>": 50300,
|
| 138 |
+
"<|fi|>": 50277,
|
| 139 |
+
"<|fo|>": 50338,
|
| 140 |
+
"<|fr|>": 50265,
|
| 141 |
+
"<|gl|>": 50319,
|
| 142 |
+
"<|gu|>": 50333,
|
| 143 |
+
"<|haw|>": 50352,
|
| 144 |
+
"<|ha|>": 50354,
|
| 145 |
+
"<|he|>": 50279,
|
| 146 |
+
"<|hi|>": 50276,
|
| 147 |
+
"<|hr|>": 50291,
|
| 148 |
+
"<|ht|>": 50339,
|
| 149 |
+
"<|hu|>": 50286,
|
| 150 |
+
"<|hy|>": 50312,
|
| 151 |
+
"<|id|>": 50275,
|
| 152 |
+
"<|is|>": 50311,
|
| 153 |
+
"<|it|>": 50274,
|
| 154 |
+
"<|ja|>": 50266,
|
| 155 |
+
"<|jw|>": 50356,
|
| 156 |
+
"<|ka|>": 50329,
|
| 157 |
+
"<|kk|>": 50316,
|
| 158 |
+
"<|km|>": 50323,
|
| 159 |
+
"<|kn|>": 50306,
|
| 160 |
+
"<|ko|>": 50264,
|
| 161 |
+
"<|la|>": 50294,
|
| 162 |
+
"<|lb|>": 50345,
|
| 163 |
+
"<|ln|>": 50353,
|
| 164 |
+
"<|lo|>": 50336,
|
| 165 |
+
"<|lt|>": 50293,
|
| 166 |
+
"<|lv|>": 50301,
|
| 167 |
+
"<|mg|>": 50349,
|
| 168 |
+
"<|mi|>": 50295,
|
| 169 |
+
"<|mk|>": 50308,
|
| 170 |
+
"<|ml|>": 50296,
|
| 171 |
+
"<|mn|>": 50314,
|
| 172 |
+
"<|mr|>": 50320,
|
| 173 |
+
"<|ms|>": 50282,
|
| 174 |
+
"<|mt|>": 50343,
|
| 175 |
+
"<|my|>": 50346,
|
| 176 |
+
"<|ne|>": 50313,
|
| 177 |
+
"<|nl|>": 50271,
|
| 178 |
+
"<|nn|>": 50342,
|
| 179 |
+
"<|no|>": 50288,
|
| 180 |
+
"<|oc|>": 50328,
|
| 181 |
+
"<|pa|>": 50321,
|
| 182 |
+
"<|pl|>": 50269,
|
| 183 |
+
"<|ps|>": 50340,
|
| 184 |
+
"<|pt|>": 50267,
|
| 185 |
+
"<|ro|>": 50284,
|
| 186 |
+
"<|ru|>": 50263,
|
| 187 |
+
"<|sa|>": 50344,
|
| 188 |
+
"<|sd|>": 50332,
|
| 189 |
+
"<|si|>": 50322,
|
| 190 |
+
"<|sk|>": 50298,
|
| 191 |
+
"<|sl|>": 50305,
|
| 192 |
+
"<|sn|>": 50324,
|
| 193 |
+
"<|so|>": 50326,
|
| 194 |
+
"<|sq|>": 50317,
|
| 195 |
+
"<|sr|>": 50303,
|
| 196 |
+
"<|su|>": 50357,
|
| 197 |
+
"<|sv|>": 50273,
|
| 198 |
+
"<|sw|>": 50318,
|
| 199 |
+
"<|ta|>": 50287,
|
| 200 |
+
"<|te|>": 50299,
|
| 201 |
+
"<|tg|>": 50331,
|
| 202 |
+
"<|th|>": 50289,
|
| 203 |
+
"<|tk|>": 50341,
|
| 204 |
+
"<|tl|>": 50348,
|
| 205 |
+
"<|tr|>": 50268,
|
| 206 |
+
"<|tt|>": 50351,
|
| 207 |
+
"<|uk|>": 50280,
|
| 208 |
+
"<|ur|>": 50290,
|
| 209 |
+
"<|uz|>": 50337,
|
| 210 |
+
"<|vi|>": 50278,
|
| 211 |
+
"<|yi|>": 50335,
|
| 212 |
+
"<|yo|>": 50325,
|
| 213 |
+
"<|zh|>": 50260
|
| 214 |
+
},
|
| 215 |
+
"max_initial_timestamp_index": 50,
|
| 216 |
+
"max_length": 448,
|
| 217 |
+
"no_timestamps_token_id": 50363,
|
| 218 |
+
"pad_token_id": 50257,
|
| 219 |
+
"prev_sot_token_id": 50361,
|
| 220 |
+
"return_timestamps": false,
|
| 221 |
+
"suppress_tokens": [
|
| 222 |
+
1,
|
| 223 |
+
2,
|
| 224 |
+
7,
|
| 225 |
+
8,
|
| 226 |
+
9,
|
| 227 |
+
10,
|
| 228 |
+
14,
|
| 229 |
+
25,
|
| 230 |
+
26,
|
| 231 |
+
27,
|
| 232 |
+
28,
|
| 233 |
+
29,
|
| 234 |
+
31,
|
| 235 |
+
58,
|
| 236 |
+
59,
|
| 237 |
+
60,
|
| 238 |
+
61,
|
| 239 |
+
62,
|
| 240 |
+
63,
|
| 241 |
+
90,
|
| 242 |
+
91,
|
| 243 |
+
92,
|
| 244 |
+
93,
|
| 245 |
+
359,
|
| 246 |
+
503,
|
| 247 |
+
522,
|
| 248 |
+
542,
|
| 249 |
+
873,
|
| 250 |
+
893,
|
| 251 |
+
902,
|
| 252 |
+
918,
|
| 253 |
+
922,
|
| 254 |
+
931,
|
| 255 |
+
1350,
|
| 256 |
+
1853,
|
| 257 |
+
1982,
|
| 258 |
+
2460,
|
| 259 |
+
2627,
|
| 260 |
+
3246,
|
| 261 |
+
3253,
|
| 262 |
+
3268,
|
| 263 |
+
3536,
|
| 264 |
+
3846,
|
| 265 |
+
3961,
|
| 266 |
+
4183,
|
| 267 |
+
4667,
|
| 268 |
+
6585,
|
| 269 |
+
6647,
|
| 270 |
+
7273,
|
| 271 |
+
9061,
|
| 272 |
+
9383,
|
| 273 |
+
10428,
|
| 274 |
+
10929,
|
| 275 |
+
11938,
|
| 276 |
+
12033,
|
| 277 |
+
12331,
|
| 278 |
+
12562,
|
| 279 |
+
13793,
|
| 280 |
+
14157,
|
| 281 |
+
14635,
|
| 282 |
+
15265,
|
| 283 |
+
15618,
|
| 284 |
+
16553,
|
| 285 |
+
16604,
|
| 286 |
+
18362,
|
| 287 |
+
18956,
|
| 288 |
+
20075,
|
| 289 |
+
21675,
|
| 290 |
+
22520,
|
| 291 |
+
26130,
|
| 292 |
+
26161,
|
| 293 |
+
26435,
|
| 294 |
+
28279,
|
| 295 |
+
29464,
|
| 296 |
+
31650,
|
| 297 |
+
32302,
|
| 298 |
+
32470,
|
| 299 |
+
36865,
|
| 300 |
+
42863,
|
| 301 |
+
47425,
|
| 302 |
+
49870,
|
| 303 |
+
50254,
|
| 304 |
+
50258,
|
| 305 |
+
50358,
|
| 306 |
+
50359,
|
| 307 |
+
50360,
|
| 308 |
+
50361,
|
| 309 |
+
50362
|
| 310 |
+
],
|
| 311 |
+
"task": "transcribe",
|
| 312 |
+
"task_to_id": {
|
| 313 |
+
"transcribe": 50359,
|
| 314 |
+
"translate": 50358
|
| 315 |
+
},
|
| 316 |
+
"transformers_version": "4.37.2"
|
| 317 |
+
}
|
checkpoint-9804/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05d488c81e0554d4efc94652f9e0dfd65d736032a01842bc2ce447171b31ee17
|
| 3 |
+
size 4992706480
|
checkpoint-9804/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d344b9f2bc3ce4a1da1b3d07f380b0f549c7675d1594799816ff5a9324c06b0c
|
| 3 |
+
size 1180663192
|
checkpoint-9804/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-9804/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7bf765708606472b28f77f402bcbdde604ca20cc73131df8d95b12b1604a5223
|
| 3 |
+
size 3095074288
|
checkpoint-9804/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 |
+
}
|
checkpoint-9804/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:57e48ac3185dcda490c773dc9402ac6a7c4b95dd2410627616a809bda905e181
|
| 3 |
+
size 14244
|
checkpoint-9804/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8425189420313da11fdbfea3e9bca0ef7099c3f5988e8310af7c843e16e961a
|
| 3 |
+
size 1064
|
checkpoint-9804/trainer_state.json
ADDED
|
@@ -0,0 +1,1857 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.999898011218766,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 9804,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 4.7619047619047613e-08,
|
| 14 |
+
"loss": 5.1575,
|
| 15 |
+
"step": 32
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 1.0204081632653061e-07,
|
| 20 |
+
"loss": 4.7952,
|
| 21 |
+
"step": 64
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 1.564625850340136e-07,
|
| 26 |
+
"loss": 4.5513,
|
| 27 |
+
"step": 96
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01,
|
| 31 |
+
"learning_rate": 2.0918367346938776e-07,
|
| 32 |
+
"loss": 3.9771,
|
| 33 |
+
"step": 128
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 2.619047619047619e-07,
|
| 38 |
+
"loss": 3.2318,
|
| 39 |
+
"step": 160
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02,
|
| 43 |
+
"learning_rate": 3.163265306122449e-07,
|
| 44 |
+
"loss": 2.508,
|
| 45 |
+
"step": 192
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.02,
|
| 49 |
+
"learning_rate": 3.707482993197279e-07,
|
| 50 |
+
"loss": 2.0866,
|
| 51 |
+
"step": 224
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 4.2517006802721085e-07,
|
| 56 |
+
"loss": 1.8228,
|
| 57 |
+
"step": 256
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.03,
|
| 61 |
+
"learning_rate": 4.795918367346938e-07,
|
| 62 |
+
"loss": 1.6854,
|
| 63 |
+
"step": 288
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.03,
|
| 67 |
+
"learning_rate": 5.340136054421769e-07,
|
| 68 |
+
"loss": 1.4892,
|
| 69 |
+
"step": 320
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04,
|
| 73 |
+
"learning_rate": 5.884353741496599e-07,
|
| 74 |
+
"loss": 1.2407,
|
| 75 |
+
"step": 352
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.04,
|
| 79 |
+
"learning_rate": 6.428571428571429e-07,
|
| 80 |
+
"loss": 1.0248,
|
| 81 |
+
"step": 384
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.04,
|
| 85 |
+
"learning_rate": 6.95578231292517e-07,
|
| 86 |
+
"loss": 0.577,
|
| 87 |
+
"step": 416
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.05,
|
| 91 |
+
"learning_rate": 7.5e-07,
|
| 92 |
+
"loss": 0.401,
|
| 93 |
+
"step": 448
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.05,
|
| 97 |
+
"learning_rate": 8.04421768707483e-07,
|
| 98 |
+
"loss": 0.3855,
|
| 99 |
+
"step": 480
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.05,
|
| 103 |
+
"learning_rate": 8.58843537414966e-07,
|
| 104 |
+
"loss": 0.3437,
|
| 105 |
+
"step": 512
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.06,
|
| 109 |
+
"learning_rate": 9.132653061224489e-07,
|
| 110 |
+
"loss": 0.3528,
|
| 111 |
+
"step": 544
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.06,
|
| 115 |
+
"learning_rate": 9.67687074829932e-07,
|
| 116 |
+
"loss": 0.4408,
|
| 117 |
+
"step": 576
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.06,
|
| 121 |
+
"learning_rate": 9.985895627644569e-07,
|
| 122 |
+
"loss": 0.4262,
|
| 123 |
+
"step": 608
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.07,
|
| 127 |
+
"learning_rate": 9.951177172615818e-07,
|
| 128 |
+
"loss": 0.3917,
|
| 129 |
+
"step": 640
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.07,
|
| 133 |
+
"learning_rate": 9.916458717587067e-07,
|
| 134 |
+
"loss": 0.4139,
|
| 135 |
+
"step": 672
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.07,
|
| 139 |
+
"learning_rate": 9.881740262558316e-07,
|
| 140 |
+
"loss": 0.3101,
|
| 141 |
+
"step": 704
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.08,
|
| 145 |
+
"learning_rate": 9.847021807529563e-07,
|
| 146 |
+
"loss": 0.4566,
|
| 147 |
+
"step": 736
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.08,
|
| 151 |
+
"learning_rate": 9.812303352500815e-07,
|
| 152 |
+
"loss": 0.2914,
|
| 153 |
+
"step": 768
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.08,
|
| 157 |
+
"learning_rate": 9.777584897472062e-07,
|
| 158 |
+
"loss": 0.4062,
|
| 159 |
+
"step": 800
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.08,
|
| 163 |
+
"learning_rate": 9.74286644244331e-07,
|
| 164 |
+
"loss": 0.3887,
|
| 165 |
+
"step": 832
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.09,
|
| 169 |
+
"learning_rate": 9.70814798741456e-07,
|
| 170 |
+
"loss": 0.3655,
|
| 171 |
+
"step": 864
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.09,
|
| 175 |
+
"learning_rate": 9.67342953238581e-07,
|
| 176 |
+
"loss": 0.3668,
|
| 177 |
+
"step": 896
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.09,
|
| 181 |
+
"learning_rate": 9.638711077357056e-07,
|
| 182 |
+
"loss": 0.3761,
|
| 183 |
+
"step": 928
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.1,
|
| 187 |
+
"learning_rate": 9.603992622328305e-07,
|
| 188 |
+
"loss": 0.3528,
|
| 189 |
+
"step": 960
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.1,
|
| 193 |
+
"learning_rate": 9.569274167299554e-07,
|
| 194 |
+
"loss": 0.2583,
|
| 195 |
+
"step": 992
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.1,
|
| 199 |
+
"learning_rate": 9.534555712270804e-07,
|
| 200 |
+
"loss": 0.3145,
|
| 201 |
+
"step": 1024
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.11,
|
| 205 |
+
"learning_rate": 9.499837257242053e-07,
|
| 206 |
+
"loss": 0.2842,
|
| 207 |
+
"step": 1056
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.11,
|
| 211 |
+
"learning_rate": 9.465118802213301e-07,
|
| 212 |
+
"loss": 0.2259,
|
| 213 |
+
"step": 1088
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.11,
|
| 217 |
+
"learning_rate": 9.430400347184549e-07,
|
| 218 |
+
"loss": 0.3493,
|
| 219 |
+
"step": 1120
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.12,
|
| 223 |
+
"learning_rate": 9.395681892155799e-07,
|
| 224 |
+
"loss": 0.3421,
|
| 225 |
+
"step": 1152
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.12,
|
| 229 |
+
"learning_rate": 9.360963437127047e-07,
|
| 230 |
+
"loss": 0.4288,
|
| 231 |
+
"step": 1184
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.12,
|
| 235 |
+
"learning_rate": 9.326244982098295e-07,
|
| 236 |
+
"loss": 0.2881,
|
| 237 |
+
"step": 1216
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.13,
|
| 241 |
+
"learning_rate": 9.291526527069546e-07,
|
| 242 |
+
"loss": 0.3361,
|
| 243 |
+
"step": 1248
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.13,
|
| 247 |
+
"learning_rate": 9.256808072040794e-07,
|
| 248 |
+
"loss": 0.2797,
|
| 249 |
+
"step": 1280
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.13,
|
| 253 |
+
"learning_rate": 9.222089617012042e-07,
|
| 254 |
+
"loss": 0.2531,
|
| 255 |
+
"step": 1312
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.14,
|
| 259 |
+
"learning_rate": 9.187371161983291e-07,
|
| 260 |
+
"loss": 0.282,
|
| 261 |
+
"step": 1344
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.14,
|
| 265 |
+
"learning_rate": 9.15265270695454e-07,
|
| 266 |
+
"loss": 0.3408,
|
| 267 |
+
"step": 1376
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.14,
|
| 271 |
+
"learning_rate": 9.117934251925788e-07,
|
| 272 |
+
"loss": 0.2354,
|
| 273 |
+
"step": 1408
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.15,
|
| 277 |
+
"learning_rate": 9.083215796897038e-07,
|
| 278 |
+
"loss": 0.3474,
|
| 279 |
+
"step": 1440
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.15,
|
| 283 |
+
"learning_rate": 9.048497341868287e-07,
|
| 284 |
+
"loss": 0.3546,
|
| 285 |
+
"step": 1472
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.15,
|
| 289 |
+
"learning_rate": 9.013778886839535e-07,
|
| 290 |
+
"loss": 0.3966,
|
| 291 |
+
"step": 1504
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.16,
|
| 295 |
+
"learning_rate": 8.979060431810784e-07,
|
| 296 |
+
"loss": 0.3376,
|
| 297 |
+
"step": 1536
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.16,
|
| 301 |
+
"learning_rate": 8.944341976782033e-07,
|
| 302 |
+
"loss": 0.3938,
|
| 303 |
+
"step": 1568
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.16,
|
| 307 |
+
"learning_rate": 8.909623521753281e-07,
|
| 308 |
+
"loss": 0.3343,
|
| 309 |
+
"step": 1600
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.17,
|
| 313 |
+
"learning_rate": 8.87490506672453e-07,
|
| 314 |
+
"loss": 0.3399,
|
| 315 |
+
"step": 1632
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.17,
|
| 319 |
+
"learning_rate": 8.840186611695779e-07,
|
| 320 |
+
"loss": 0.3072,
|
| 321 |
+
"step": 1664
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.17,
|
| 325 |
+
"learning_rate": 8.805468156667028e-07,
|
| 326 |
+
"loss": 0.262,
|
| 327 |
+
"step": 1696
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.18,
|
| 331 |
+
"learning_rate": 8.770749701638277e-07,
|
| 332 |
+
"loss": 0.2995,
|
| 333 |
+
"step": 1728
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.18,
|
| 337 |
+
"learning_rate": 8.736031246609525e-07,
|
| 338 |
+
"loss": 0.2804,
|
| 339 |
+
"step": 1760
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.18,
|
| 343 |
+
"learning_rate": 8.701312791580774e-07,
|
| 344 |
+
"loss": 0.3693,
|
| 345 |
+
"step": 1792
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.19,
|
| 349 |
+
"learning_rate": 8.666594336552023e-07,
|
| 350 |
+
"loss": 0.2589,
|
| 351 |
+
"step": 1824
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.19,
|
| 355 |
+
"learning_rate": 8.631875881523272e-07,
|
| 356 |
+
"loss": 0.2638,
|
| 357 |
+
"step": 1856
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.19,
|
| 361 |
+
"learning_rate": 8.59715742649452e-07,
|
| 362 |
+
"loss": 0.3516,
|
| 363 |
+
"step": 1888
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.2,
|
| 367 |
+
"learning_rate": 8.56243897146577e-07,
|
| 368 |
+
"loss": 0.369,
|
| 369 |
+
"step": 1920
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.2,
|
| 373 |
+
"learning_rate": 8.527720516437018e-07,
|
| 374 |
+
"loss": 0.3453,
|
| 375 |
+
"step": 1952
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.2,
|
| 379 |
+
"learning_rate": 8.493002061408266e-07,
|
| 380 |
+
"loss": 0.3813,
|
| 381 |
+
"step": 1984
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.21,
|
| 385 |
+
"learning_rate": 8.458283606379516e-07,
|
| 386 |
+
"loss": 0.2657,
|
| 387 |
+
"step": 2016
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.21,
|
| 391 |
+
"learning_rate": 8.423565151350764e-07,
|
| 392 |
+
"loss": 0.3514,
|
| 393 |
+
"step": 2048
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.21,
|
| 397 |
+
"learning_rate": 8.388846696322013e-07,
|
| 398 |
+
"loss": 0.2214,
|
| 399 |
+
"step": 2080
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.22,
|
| 403 |
+
"learning_rate": 8.354128241293263e-07,
|
| 404 |
+
"loss": 0.4351,
|
| 405 |
+
"step": 2112
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.22,
|
| 409 |
+
"learning_rate": 8.319409786264511e-07,
|
| 410 |
+
"loss": 0.2986,
|
| 411 |
+
"step": 2144
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.22,
|
| 415 |
+
"learning_rate": 8.284691331235759e-07,
|
| 416 |
+
"loss": 0.2211,
|
| 417 |
+
"step": 2176
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.23,
|
| 421 |
+
"learning_rate": 8.249972876207008e-07,
|
| 422 |
+
"loss": 0.2291,
|
| 423 |
+
"step": 2208
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.23,
|
| 427 |
+
"learning_rate": 8.215254421178257e-07,
|
| 428 |
+
"loss": 0.3048,
|
| 429 |
+
"step": 2240
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.23,
|
| 433 |
+
"learning_rate": 8.180535966149505e-07,
|
| 434 |
+
"loss": 0.3369,
|
| 435 |
+
"step": 2272
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.23,
|
| 439 |
+
"learning_rate": 8.145817511120755e-07,
|
| 440 |
+
"loss": 0.2413,
|
| 441 |
+
"step": 2304
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.24,
|
| 445 |
+
"learning_rate": 8.111099056092004e-07,
|
| 446 |
+
"loss": 0.2656,
|
| 447 |
+
"step": 2336
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.24,
|
| 451 |
+
"learning_rate": 8.076380601063252e-07,
|
| 452 |
+
"loss": 0.2661,
|
| 453 |
+
"step": 2368
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.24,
|
| 457 |
+
"learning_rate": 8.041662146034501e-07,
|
| 458 |
+
"loss": 0.3069,
|
| 459 |
+
"step": 2400
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.25,
|
| 463 |
+
"learning_rate": 8.00694369100575e-07,
|
| 464 |
+
"loss": 0.2357,
|
| 465 |
+
"step": 2432
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.25,
|
| 469 |
+
"learning_rate": 7.972225235976998e-07,
|
| 470 |
+
"loss": 0.3374,
|
| 471 |
+
"step": 2464
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.25,
|
| 475 |
+
"learning_rate": 7.937506780948248e-07,
|
| 476 |
+
"loss": 0.2678,
|
| 477 |
+
"step": 2496
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.26,
|
| 481 |
+
"learning_rate": 7.902788325919496e-07,
|
| 482 |
+
"loss": 0.2813,
|
| 483 |
+
"step": 2528
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.26,
|
| 487 |
+
"learning_rate": 7.868069870890745e-07,
|
| 488 |
+
"loss": 0.386,
|
| 489 |
+
"step": 2560
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.26,
|
| 493 |
+
"learning_rate": 7.833351415861994e-07,
|
| 494 |
+
"loss": 0.2247,
|
| 495 |
+
"step": 2592
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.27,
|
| 499 |
+
"learning_rate": 7.798632960833242e-07,
|
| 500 |
+
"loss": 0.3458,
|
| 501 |
+
"step": 2624
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.27,
|
| 505 |
+
"learning_rate": 7.763914505804491e-07,
|
| 506 |
+
"loss": 0.2967,
|
| 507 |
+
"step": 2656
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.27,
|
| 511 |
+
"learning_rate": 7.72919605077574e-07,
|
| 512 |
+
"loss": 0.277,
|
| 513 |
+
"step": 2688
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.28,
|
| 517 |
+
"learning_rate": 7.694477595746989e-07,
|
| 518 |
+
"loss": 0.2149,
|
| 519 |
+
"step": 2720
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.28,
|
| 523 |
+
"learning_rate": 7.659759140718237e-07,
|
| 524 |
+
"loss": 0.3624,
|
| 525 |
+
"step": 2752
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.28,
|
| 529 |
+
"learning_rate": 7.625040685689487e-07,
|
| 530 |
+
"loss": 0.3259,
|
| 531 |
+
"step": 2784
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.29,
|
| 535 |
+
"learning_rate": 7.590322230660735e-07,
|
| 536 |
+
"loss": 0.2905,
|
| 537 |
+
"step": 2816
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.29,
|
| 541 |
+
"learning_rate": 7.555603775631984e-07,
|
| 542 |
+
"loss": 0.2397,
|
| 543 |
+
"step": 2848
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.29,
|
| 547 |
+
"learning_rate": 7.520885320603233e-07,
|
| 548 |
+
"loss": 0.4047,
|
| 549 |
+
"step": 2880
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.3,
|
| 553 |
+
"learning_rate": 7.486166865574481e-07,
|
| 554 |
+
"loss": 0.258,
|
| 555 |
+
"step": 2912
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.3,
|
| 559 |
+
"learning_rate": 7.451448410545731e-07,
|
| 560 |
+
"loss": 0.3667,
|
| 561 |
+
"step": 2944
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.3,
|
| 565 |
+
"learning_rate": 7.41672995551698e-07,
|
| 566 |
+
"loss": 0.3022,
|
| 567 |
+
"step": 2976
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.31,
|
| 571 |
+
"learning_rate": 7.382011500488228e-07,
|
| 572 |
+
"loss": 0.2435,
|
| 573 |
+
"step": 3008
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.31,
|
| 577 |
+
"learning_rate": 7.347293045459477e-07,
|
| 578 |
+
"loss": 0.3491,
|
| 579 |
+
"step": 3040
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.31,
|
| 583 |
+
"learning_rate": 7.312574590430725e-07,
|
| 584 |
+
"loss": 0.2885,
|
| 585 |
+
"step": 3072
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.32,
|
| 589 |
+
"learning_rate": 7.277856135401974e-07,
|
| 590 |
+
"loss": 0.3822,
|
| 591 |
+
"step": 3104
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.32,
|
| 595 |
+
"learning_rate": 7.243137680373224e-07,
|
| 596 |
+
"loss": 0.3265,
|
| 597 |
+
"step": 3136
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.32,
|
| 601 |
+
"learning_rate": 7.208419225344472e-07,
|
| 602 |
+
"loss": 0.2879,
|
| 603 |
+
"step": 3168
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.33,
|
| 607 |
+
"learning_rate": 7.173700770315721e-07,
|
| 608 |
+
"loss": 0.3678,
|
| 609 |
+
"step": 3200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.33,
|
| 613 |
+
"learning_rate": 7.13898231528697e-07,
|
| 614 |
+
"loss": 0.2696,
|
| 615 |
+
"step": 3232
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.33,
|
| 619 |
+
"learning_rate": 7.104263860258218e-07,
|
| 620 |
+
"loss": 0.3353,
|
| 621 |
+
"step": 3264
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.34,
|
| 625 |
+
"learning_rate": 7.069545405229467e-07,
|
| 626 |
+
"loss": 0.3552,
|
| 627 |
+
"step": 3296
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.34,
|
| 631 |
+
"learning_rate": 7.034826950200716e-07,
|
| 632 |
+
"loss": 0.2246,
|
| 633 |
+
"step": 3328
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.34,
|
| 637 |
+
"learning_rate": 7.000108495171965e-07,
|
| 638 |
+
"loss": 0.2622,
|
| 639 |
+
"step": 3360
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.35,
|
| 643 |
+
"learning_rate": 6.965390040143213e-07,
|
| 644 |
+
"loss": 0.286,
|
| 645 |
+
"step": 3392
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.35,
|
| 649 |
+
"learning_rate": 6.930671585114463e-07,
|
| 650 |
+
"loss": 0.2755,
|
| 651 |
+
"step": 3424
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.35,
|
| 655 |
+
"learning_rate": 6.895953130085711e-07,
|
| 656 |
+
"loss": 0.3111,
|
| 657 |
+
"step": 3456
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.36,
|
| 661 |
+
"learning_rate": 6.861234675056959e-07,
|
| 662 |
+
"loss": 0.2431,
|
| 663 |
+
"step": 3488
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.36,
|
| 667 |
+
"learning_rate": 6.826516220028209e-07,
|
| 668 |
+
"loss": 0.2097,
|
| 669 |
+
"step": 3520
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.36,
|
| 673 |
+
"learning_rate": 6.791797764999457e-07,
|
| 674 |
+
"loss": 0.3134,
|
| 675 |
+
"step": 3552
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.37,
|
| 679 |
+
"learning_rate": 6.757079309970706e-07,
|
| 680 |
+
"loss": 0.2535,
|
| 681 |
+
"step": 3584
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.37,
|
| 685 |
+
"learning_rate": 6.722360854941956e-07,
|
| 686 |
+
"loss": 0.3329,
|
| 687 |
+
"step": 3616
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.37,
|
| 691 |
+
"learning_rate": 6.687642399913204e-07,
|
| 692 |
+
"loss": 0.2857,
|
| 693 |
+
"step": 3648
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.38,
|
| 697 |
+
"learning_rate": 6.652923944884452e-07,
|
| 698 |
+
"loss": 0.2697,
|
| 699 |
+
"step": 3680
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.38,
|
| 703 |
+
"learning_rate": 6.618205489855701e-07,
|
| 704 |
+
"loss": 0.2815,
|
| 705 |
+
"step": 3712
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.38,
|
| 709 |
+
"learning_rate": 6.58348703482695e-07,
|
| 710 |
+
"loss": 0.2856,
|
| 711 |
+
"step": 3744
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.39,
|
| 715 |
+
"learning_rate": 6.548768579798198e-07,
|
| 716 |
+
"loss": 0.2912,
|
| 717 |
+
"step": 3776
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.39,
|
| 721 |
+
"learning_rate": 6.514050124769448e-07,
|
| 722 |
+
"loss": 0.2916,
|
| 723 |
+
"step": 3808
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.39,
|
| 727 |
+
"learning_rate": 6.479331669740697e-07,
|
| 728 |
+
"loss": 0.308,
|
| 729 |
+
"step": 3840
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.39,
|
| 733 |
+
"learning_rate": 6.444613214711945e-07,
|
| 734 |
+
"loss": 0.2934,
|
| 735 |
+
"step": 3872
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.4,
|
| 739 |
+
"learning_rate": 6.409894759683194e-07,
|
| 740 |
+
"loss": 0.2869,
|
| 741 |
+
"step": 3904
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.4,
|
| 745 |
+
"learning_rate": 6.375176304654442e-07,
|
| 746 |
+
"loss": 0.3232,
|
| 747 |
+
"step": 3936
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.4,
|
| 751 |
+
"learning_rate": 6.340457849625691e-07,
|
| 752 |
+
"loss": 0.2486,
|
| 753 |
+
"step": 3968
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.41,
|
| 757 |
+
"learning_rate": 6.305739394596941e-07,
|
| 758 |
+
"loss": 0.3123,
|
| 759 |
+
"step": 4000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.41,
|
| 763 |
+
"learning_rate": 6.271020939568189e-07,
|
| 764 |
+
"loss": 0.3183,
|
| 765 |
+
"step": 4032
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.41,
|
| 769 |
+
"learning_rate": 6.236302484539438e-07,
|
| 770 |
+
"loss": 0.3688,
|
| 771 |
+
"step": 4064
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 0.42,
|
| 775 |
+
"learning_rate": 6.201584029510687e-07,
|
| 776 |
+
"loss": 0.3236,
|
| 777 |
+
"step": 4096
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.42,
|
| 781 |
+
"learning_rate": 6.166865574481935e-07,
|
| 782 |
+
"loss": 0.2206,
|
| 783 |
+
"step": 4128
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.42,
|
| 787 |
+
"learning_rate": 6.132147119453183e-07,
|
| 788 |
+
"loss": 0.2519,
|
| 789 |
+
"step": 4160
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.43,
|
| 793 |
+
"learning_rate": 6.097428664424433e-07,
|
| 794 |
+
"loss": 0.2467,
|
| 795 |
+
"step": 4192
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.43,
|
| 799 |
+
"learning_rate": 6.062710209395682e-07,
|
| 800 |
+
"loss": 0.2555,
|
| 801 |
+
"step": 4224
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.43,
|
| 805 |
+
"learning_rate": 6.02799175436693e-07,
|
| 806 |
+
"loss": 0.2391,
|
| 807 |
+
"step": 4256
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.44,
|
| 811 |
+
"learning_rate": 5.99327329933818e-07,
|
| 812 |
+
"loss": 0.3167,
|
| 813 |
+
"step": 4288
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.44,
|
| 817 |
+
"learning_rate": 5.958554844309428e-07,
|
| 818 |
+
"loss": 0.2342,
|
| 819 |
+
"step": 4320
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.44,
|
| 823 |
+
"learning_rate": 5.923836389280676e-07,
|
| 824 |
+
"loss": 0.1986,
|
| 825 |
+
"step": 4352
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.45,
|
| 829 |
+
"learning_rate": 5.889117934251926e-07,
|
| 830 |
+
"loss": 0.2359,
|
| 831 |
+
"step": 4384
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.45,
|
| 835 |
+
"learning_rate": 5.854399479223174e-07,
|
| 836 |
+
"loss": 0.2453,
|
| 837 |
+
"step": 4416
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.45,
|
| 841 |
+
"learning_rate": 5.819681024194423e-07,
|
| 842 |
+
"loss": 0.3077,
|
| 843 |
+
"step": 4448
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.46,
|
| 847 |
+
"learning_rate": 5.784962569165672e-07,
|
| 848 |
+
"loss": 0.2052,
|
| 849 |
+
"step": 4480
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.46,
|
| 853 |
+
"learning_rate": 5.750244114136921e-07,
|
| 854 |
+
"loss": 0.2313,
|
| 855 |
+
"step": 4512
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.46,
|
| 859 |
+
"learning_rate": 5.715525659108169e-07,
|
| 860 |
+
"loss": 0.3616,
|
| 861 |
+
"step": 4544
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.47,
|
| 865 |
+
"learning_rate": 5.680807204079418e-07,
|
| 866 |
+
"loss": 0.2848,
|
| 867 |
+
"step": 4576
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.47,
|
| 871 |
+
"learning_rate": 5.646088749050667e-07,
|
| 872 |
+
"loss": 0.2016,
|
| 873 |
+
"step": 4608
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.47,
|
| 877 |
+
"learning_rate": 5.611370294021915e-07,
|
| 878 |
+
"loss": 0.2467,
|
| 879 |
+
"step": 4640
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.48,
|
| 883 |
+
"learning_rate": 5.576651838993165e-07,
|
| 884 |
+
"loss": 0.3043,
|
| 885 |
+
"step": 4672
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.48,
|
| 889 |
+
"learning_rate": 5.541933383964414e-07,
|
| 890 |
+
"loss": 0.2283,
|
| 891 |
+
"step": 4704
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.48,
|
| 895 |
+
"learning_rate": 5.507214928935662e-07,
|
| 896 |
+
"loss": 0.1914,
|
| 897 |
+
"step": 4736
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.49,
|
| 901 |
+
"learning_rate": 5.472496473906911e-07,
|
| 902 |
+
"loss": 0.1776,
|
| 903 |
+
"step": 4768
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.49,
|
| 907 |
+
"learning_rate": 5.437778018878159e-07,
|
| 908 |
+
"loss": 0.2395,
|
| 909 |
+
"step": 4800
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.49,
|
| 913 |
+
"learning_rate": 5.403059563849408e-07,
|
| 914 |
+
"loss": 0.3197,
|
| 915 |
+
"step": 4832
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.5,
|
| 919 |
+
"learning_rate": 5.368341108820658e-07,
|
| 920 |
+
"loss": 0.2135,
|
| 921 |
+
"step": 4864
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.5,
|
| 925 |
+
"learning_rate": 5.333622653791906e-07,
|
| 926 |
+
"loss": 0.2104,
|
| 927 |
+
"step": 4896
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.5,
|
| 931 |
+
"learning_rate": 5.298904198763155e-07,
|
| 932 |
+
"loss": 0.2355,
|
| 933 |
+
"step": 4928
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.51,
|
| 937 |
+
"learning_rate": 5.264185743734404e-07,
|
| 938 |
+
"loss": 0.34,
|
| 939 |
+
"step": 4960
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.51,
|
| 943 |
+
"learning_rate": 5.229467288705652e-07,
|
| 944 |
+
"loss": 0.3004,
|
| 945 |
+
"step": 4992
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.51,
|
| 949 |
+
"learning_rate": 5.1947488336769e-07,
|
| 950 |
+
"loss": 0.3025,
|
| 951 |
+
"step": 5024
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 0.52,
|
| 955 |
+
"learning_rate": 5.16003037864815e-07,
|
| 956 |
+
"loss": 0.2561,
|
| 957 |
+
"step": 5056
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 0.52,
|
| 961 |
+
"learning_rate": 5.125311923619399e-07,
|
| 962 |
+
"loss": 0.2891,
|
| 963 |
+
"step": 5088
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.52,
|
| 967 |
+
"learning_rate": 5.090593468590647e-07,
|
| 968 |
+
"loss": 0.2668,
|
| 969 |
+
"step": 5120
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.53,
|
| 973 |
+
"learning_rate": 5.055875013561897e-07,
|
| 974 |
+
"loss": 0.2539,
|
| 975 |
+
"step": 5152
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.53,
|
| 979 |
+
"learning_rate": 5.021156558533145e-07,
|
| 980 |
+
"loss": 0.2294,
|
| 981 |
+
"step": 5184
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.53,
|
| 985 |
+
"learning_rate": 4.986438103504394e-07,
|
| 986 |
+
"loss": 0.2286,
|
| 987 |
+
"step": 5216
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.54,
|
| 991 |
+
"learning_rate": 4.951719648475642e-07,
|
| 992 |
+
"loss": 0.2987,
|
| 993 |
+
"step": 5248
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.54,
|
| 997 |
+
"learning_rate": 4.917001193446891e-07,
|
| 998 |
+
"loss": 0.2871,
|
| 999 |
+
"step": 5280
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.54,
|
| 1003 |
+
"learning_rate": 4.882282738418141e-07,
|
| 1004 |
+
"loss": 0.3253,
|
| 1005 |
+
"step": 5312
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 0.55,
|
| 1009 |
+
"learning_rate": 4.847564283389389e-07,
|
| 1010 |
+
"loss": 0.2106,
|
| 1011 |
+
"step": 5344
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.55,
|
| 1015 |
+
"learning_rate": 4.812845828360638e-07,
|
| 1016 |
+
"loss": 0.2428,
|
| 1017 |
+
"step": 5376
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.55,
|
| 1021 |
+
"learning_rate": 4.778127373331887e-07,
|
| 1022 |
+
"loss": 0.1568,
|
| 1023 |
+
"step": 5408
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.55,
|
| 1027 |
+
"learning_rate": 4.743408918303135e-07,
|
| 1028 |
+
"loss": 0.2879,
|
| 1029 |
+
"step": 5440
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.56,
|
| 1033 |
+
"learning_rate": 4.708690463274384e-07,
|
| 1034 |
+
"loss": 0.2221,
|
| 1035 |
+
"step": 5472
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.56,
|
| 1039 |
+
"learning_rate": 4.673972008245633e-07,
|
| 1040 |
+
"loss": 0.3336,
|
| 1041 |
+
"step": 5504
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.56,
|
| 1045 |
+
"learning_rate": 4.639253553216881e-07,
|
| 1046 |
+
"loss": 0.2265,
|
| 1047 |
+
"step": 5536
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 0.57,
|
| 1051 |
+
"learning_rate": 4.6045350981881303e-07,
|
| 1052 |
+
"loss": 0.2408,
|
| 1053 |
+
"step": 5568
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.57,
|
| 1057 |
+
"learning_rate": 4.5698166431593795e-07,
|
| 1058 |
+
"loss": 0.3562,
|
| 1059 |
+
"step": 5600
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.57,
|
| 1063 |
+
"learning_rate": 4.5350981881306276e-07,
|
| 1064 |
+
"loss": 0.1839,
|
| 1065 |
+
"step": 5632
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.58,
|
| 1069 |
+
"learning_rate": 4.500379733101877e-07,
|
| 1070 |
+
"loss": 0.2341,
|
| 1071 |
+
"step": 5664
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.58,
|
| 1075 |
+
"learning_rate": 4.4656612780731254e-07,
|
| 1076 |
+
"loss": 0.2683,
|
| 1077 |
+
"step": 5696
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 0.58,
|
| 1081 |
+
"learning_rate": 4.430942823044374e-07,
|
| 1082 |
+
"loss": 0.3978,
|
| 1083 |
+
"step": 5728
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 0.59,
|
| 1087 |
+
"learning_rate": 4.396224368015623e-07,
|
| 1088 |
+
"loss": 0.2493,
|
| 1089 |
+
"step": 5760
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.59,
|
| 1093 |
+
"learning_rate": 4.361505912986872e-07,
|
| 1094 |
+
"loss": 0.28,
|
| 1095 |
+
"step": 5792
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.59,
|
| 1099 |
+
"learning_rate": 4.3267874579581205e-07,
|
| 1100 |
+
"loss": 0.2263,
|
| 1101 |
+
"step": 5824
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.6,
|
| 1105 |
+
"learning_rate": 4.292069002929369e-07,
|
| 1106 |
+
"loss": 0.3391,
|
| 1107 |
+
"step": 5856
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.6,
|
| 1111 |
+
"learning_rate": 4.2573505479006183e-07,
|
| 1112 |
+
"loss": 0.3003,
|
| 1113 |
+
"step": 5888
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.6,
|
| 1117 |
+
"learning_rate": 4.2226320928718675e-07,
|
| 1118 |
+
"loss": 0.2635,
|
| 1119 |
+
"step": 5920
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.61,
|
| 1123 |
+
"learning_rate": 4.1879136378431156e-07,
|
| 1124 |
+
"loss": 0.2997,
|
| 1125 |
+
"step": 5952
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.61,
|
| 1129 |
+
"learning_rate": 4.153195182814365e-07,
|
| 1130 |
+
"loss": 0.2135,
|
| 1131 |
+
"step": 5984
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.61,
|
| 1135 |
+
"learning_rate": 4.1184767277856134e-07,
|
| 1136 |
+
"loss": 0.2651,
|
| 1137 |
+
"step": 6016
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.62,
|
| 1141 |
+
"learning_rate": 4.083758272756862e-07,
|
| 1142 |
+
"loss": 0.1867,
|
| 1143 |
+
"step": 6048
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.62,
|
| 1147 |
+
"learning_rate": 4.049039817728111e-07,
|
| 1148 |
+
"loss": 0.1986,
|
| 1149 |
+
"step": 6080
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.62,
|
| 1153 |
+
"learning_rate": 4.01432136269936e-07,
|
| 1154 |
+
"loss": 0.2905,
|
| 1155 |
+
"step": 6112
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.63,
|
| 1159 |
+
"learning_rate": 3.9796029076706085e-07,
|
| 1160 |
+
"loss": 0.2226,
|
| 1161 |
+
"step": 6144
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.63,
|
| 1165 |
+
"learning_rate": 3.944884452641857e-07,
|
| 1166 |
+
"loss": 0.2472,
|
| 1167 |
+
"step": 6176
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.63,
|
| 1171 |
+
"learning_rate": 3.9101659976131063e-07,
|
| 1172 |
+
"loss": 0.2103,
|
| 1173 |
+
"step": 6208
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.64,
|
| 1177 |
+
"learning_rate": 3.8754475425843544e-07,
|
| 1178 |
+
"loss": 0.2976,
|
| 1179 |
+
"step": 6240
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.64,
|
| 1183 |
+
"learning_rate": 3.8407290875556036e-07,
|
| 1184 |
+
"loss": 0.2777,
|
| 1185 |
+
"step": 6272
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.64,
|
| 1189 |
+
"learning_rate": 3.806010632526853e-07,
|
| 1190 |
+
"loss": 0.2452,
|
| 1191 |
+
"step": 6304
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.65,
|
| 1195 |
+
"learning_rate": 3.771292177498101e-07,
|
| 1196 |
+
"loss": 0.2243,
|
| 1197 |
+
"step": 6336
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.65,
|
| 1201 |
+
"learning_rate": 3.73657372246935e-07,
|
| 1202 |
+
"loss": 0.3171,
|
| 1203 |
+
"step": 6368
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.65,
|
| 1207 |
+
"learning_rate": 3.7018552674405987e-07,
|
| 1208 |
+
"loss": 0.2472,
|
| 1209 |
+
"step": 6400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.66,
|
| 1213 |
+
"learning_rate": 3.6671368124118474e-07,
|
| 1214 |
+
"loss": 0.2778,
|
| 1215 |
+
"step": 6432
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.66,
|
| 1219 |
+
"learning_rate": 3.6335033091027445e-07,
|
| 1220 |
+
"loss": 0.2834,
|
| 1221 |
+
"step": 6464
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.66,
|
| 1225 |
+
"learning_rate": 3.5987848540739937e-07,
|
| 1226 |
+
"loss": 0.23,
|
| 1227 |
+
"step": 6496
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.67,
|
| 1231 |
+
"learning_rate": 3.5640663990452423e-07,
|
| 1232 |
+
"loss": 0.2185,
|
| 1233 |
+
"step": 6528
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.67,
|
| 1237 |
+
"learning_rate": 3.529347944016491e-07,
|
| 1238 |
+
"loss": 0.282,
|
| 1239 |
+
"step": 6560
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.67,
|
| 1243 |
+
"learning_rate": 3.4946294889877396e-07,
|
| 1244 |
+
"loss": 0.1885,
|
| 1245 |
+
"step": 6592
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.68,
|
| 1249 |
+
"learning_rate": 3.459911033958989e-07,
|
| 1250 |
+
"loss": 0.263,
|
| 1251 |
+
"step": 6624
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.68,
|
| 1255 |
+
"learning_rate": 3.425192578930238e-07,
|
| 1256 |
+
"loss": 0.2431,
|
| 1257 |
+
"step": 6656
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.68,
|
| 1261 |
+
"learning_rate": 3.390474123901486e-07,
|
| 1262 |
+
"loss": 0.2472,
|
| 1263 |
+
"step": 6688
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.69,
|
| 1267 |
+
"learning_rate": 3.355755668872735e-07,
|
| 1268 |
+
"loss": 0.284,
|
| 1269 |
+
"step": 6720
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.69,
|
| 1273 |
+
"learning_rate": 3.321037213843984e-07,
|
| 1274 |
+
"loss": 0.2947,
|
| 1275 |
+
"step": 6752
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.69,
|
| 1279 |
+
"learning_rate": 3.2863187588152325e-07,
|
| 1280 |
+
"loss": 0.2448,
|
| 1281 |
+
"step": 6784
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.7,
|
| 1285 |
+
"learning_rate": 3.251600303786481e-07,
|
| 1286 |
+
"loss": 0.3035,
|
| 1287 |
+
"step": 6816
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 0.7,
|
| 1291 |
+
"learning_rate": 3.2168818487577303e-07,
|
| 1292 |
+
"loss": 0.3378,
|
| 1293 |
+
"step": 6848
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 0.7,
|
| 1297 |
+
"learning_rate": 3.182163393728979e-07,
|
| 1298 |
+
"loss": 0.2878,
|
| 1299 |
+
"step": 6880
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.7,
|
| 1303 |
+
"learning_rate": 3.1474449387002276e-07,
|
| 1304 |
+
"loss": 0.1969,
|
| 1305 |
+
"step": 6912
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.71,
|
| 1309 |
+
"learning_rate": 3.112726483671477e-07,
|
| 1310 |
+
"loss": 0.2282,
|
| 1311 |
+
"step": 6944
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.71,
|
| 1315 |
+
"learning_rate": 3.078008028642725e-07,
|
| 1316 |
+
"loss": 0.2472,
|
| 1317 |
+
"step": 6976
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 0.71,
|
| 1321 |
+
"learning_rate": 3.043289573613974e-07,
|
| 1322 |
+
"loss": 0.3127,
|
| 1323 |
+
"step": 7008
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 0.72,
|
| 1327 |
+
"learning_rate": 3.008571118585223e-07,
|
| 1328 |
+
"loss": 0.3757,
|
| 1329 |
+
"step": 7040
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 0.72,
|
| 1333 |
+
"learning_rate": 2.9738526635564714e-07,
|
| 1334 |
+
"loss": 0.2801,
|
| 1335 |
+
"step": 7072
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 0.72,
|
| 1339 |
+
"learning_rate": 2.9402191602473685e-07,
|
| 1340 |
+
"loss": 0.3094,
|
| 1341 |
+
"step": 7104
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.73,
|
| 1345 |
+
"learning_rate": 2.9055007052186177e-07,
|
| 1346 |
+
"loss": 0.3006,
|
| 1347 |
+
"step": 7136
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.73,
|
| 1351 |
+
"learning_rate": 2.8707822501898663e-07,
|
| 1352 |
+
"loss": 0.2431,
|
| 1353 |
+
"step": 7168
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.73,
|
| 1357 |
+
"learning_rate": 2.836063795161115e-07,
|
| 1358 |
+
"loss": 0.217,
|
| 1359 |
+
"step": 7200
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.74,
|
| 1363 |
+
"learning_rate": 2.801345340132364e-07,
|
| 1364 |
+
"loss": 0.1981,
|
| 1365 |
+
"step": 7232
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 0.74,
|
| 1369 |
+
"learning_rate": 2.766626885103613e-07,
|
| 1370 |
+
"loss": 0.242,
|
| 1371 |
+
"step": 7264
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 0.74,
|
| 1375 |
+
"learning_rate": 2.7319084300748614e-07,
|
| 1376 |
+
"loss": 0.2426,
|
| 1377 |
+
"step": 7296
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 0.75,
|
| 1381 |
+
"learning_rate": 2.69718997504611e-07,
|
| 1382 |
+
"loss": 0.2771,
|
| 1383 |
+
"step": 7328
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.75,
|
| 1387 |
+
"learning_rate": 2.662471520017359e-07,
|
| 1388 |
+
"loss": 0.2508,
|
| 1389 |
+
"step": 7360
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.75,
|
| 1393 |
+
"learning_rate": 2.6277530649886084e-07,
|
| 1394 |
+
"loss": 0.2495,
|
| 1395 |
+
"step": 7392
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.76,
|
| 1399 |
+
"learning_rate": 2.5930346099598565e-07,
|
| 1400 |
+
"loss": 0.2353,
|
| 1401 |
+
"step": 7424
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 0.76,
|
| 1405 |
+
"learning_rate": 2.5583161549311057e-07,
|
| 1406 |
+
"loss": 0.2047,
|
| 1407 |
+
"step": 7456
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 0.76,
|
| 1411 |
+
"learning_rate": 2.5235976999023543e-07,
|
| 1412 |
+
"loss": 0.36,
|
| 1413 |
+
"step": 7488
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 0.77,
|
| 1417 |
+
"learning_rate": 2.488879244873603e-07,
|
| 1418 |
+
"loss": 0.2685,
|
| 1419 |
+
"step": 7520
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.77,
|
| 1423 |
+
"learning_rate": 2.4541607898448516e-07,
|
| 1424 |
+
"loss": 0.2505,
|
| 1425 |
+
"step": 7552
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.77,
|
| 1429 |
+
"learning_rate": 2.419442334816101e-07,
|
| 1430 |
+
"loss": 0.2223,
|
| 1431 |
+
"step": 7584
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.78,
|
| 1435 |
+
"learning_rate": 2.3847238797873494e-07,
|
| 1436 |
+
"loss": 0.2466,
|
| 1437 |
+
"step": 7616
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.78,
|
| 1441 |
+
"learning_rate": 2.350005424758598e-07,
|
| 1442 |
+
"loss": 0.2872,
|
| 1443 |
+
"step": 7648
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 0.78,
|
| 1447 |
+
"learning_rate": 2.315286969729847e-07,
|
| 1448 |
+
"loss": 0.2257,
|
| 1449 |
+
"step": 7680
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.79,
|
| 1453 |
+
"learning_rate": 2.2805685147010956e-07,
|
| 1454 |
+
"loss": 0.2488,
|
| 1455 |
+
"step": 7712
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 0.79,
|
| 1459 |
+
"learning_rate": 2.2458500596723445e-07,
|
| 1460 |
+
"loss": 0.3667,
|
| 1461 |
+
"step": 7744
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 0.79,
|
| 1465 |
+
"learning_rate": 2.2111316046435932e-07,
|
| 1466 |
+
"loss": 0.2617,
|
| 1467 |
+
"step": 7776
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.8,
|
| 1471 |
+
"learning_rate": 2.1764131496148418e-07,
|
| 1472 |
+
"loss": 0.1686,
|
| 1473 |
+
"step": 7808
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.8,
|
| 1477 |
+
"learning_rate": 2.141694694586091e-07,
|
| 1478 |
+
"loss": 0.2216,
|
| 1479 |
+
"step": 7840
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.8,
|
| 1483 |
+
"learning_rate": 2.1069762395573396e-07,
|
| 1484 |
+
"loss": 0.2418,
|
| 1485 |
+
"step": 7872
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 0.81,
|
| 1489 |
+
"learning_rate": 2.0722577845285885e-07,
|
| 1490 |
+
"loss": 0.2023,
|
| 1491 |
+
"step": 7904
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 0.81,
|
| 1495 |
+
"learning_rate": 2.0375393294998372e-07,
|
| 1496 |
+
"loss": 0.3171,
|
| 1497 |
+
"step": 7936
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 0.81,
|
| 1501 |
+
"learning_rate": 2.0028208744710858e-07,
|
| 1502 |
+
"loss": 0.2518,
|
| 1503 |
+
"step": 7968
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 0.82,
|
| 1507 |
+
"learning_rate": 1.9681024194423347e-07,
|
| 1508 |
+
"loss": 0.2388,
|
| 1509 |
+
"step": 8000
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 0.82,
|
| 1513 |
+
"learning_rate": 1.9333839644135836e-07,
|
| 1514 |
+
"loss": 0.2612,
|
| 1515 |
+
"step": 8032
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.82,
|
| 1519 |
+
"learning_rate": 1.8986655093848323e-07,
|
| 1520 |
+
"loss": 0.3058,
|
| 1521 |
+
"step": 8064
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.83,
|
| 1525 |
+
"learning_rate": 1.8639470543560812e-07,
|
| 1526 |
+
"loss": 0.2292,
|
| 1527 |
+
"step": 8096
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 0.83,
|
| 1531 |
+
"learning_rate": 1.8292285993273298e-07,
|
| 1532 |
+
"loss": 0.3214,
|
| 1533 |
+
"step": 8128
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.83,
|
| 1537 |
+
"learning_rate": 1.7945101442985785e-07,
|
| 1538 |
+
"loss": 0.2081,
|
| 1539 |
+
"step": 8160
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.84,
|
| 1543 |
+
"learning_rate": 1.7597916892698276e-07,
|
| 1544 |
+
"loss": 0.3473,
|
| 1545 |
+
"step": 8192
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 0.84,
|
| 1549 |
+
"learning_rate": 1.7250732342410763e-07,
|
| 1550 |
+
"loss": 0.2174,
|
| 1551 |
+
"step": 8224
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.84,
|
| 1555 |
+
"learning_rate": 1.690354779212325e-07,
|
| 1556 |
+
"loss": 0.2021,
|
| 1557 |
+
"step": 8256
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.85,
|
| 1561 |
+
"learning_rate": 1.6556363241835738e-07,
|
| 1562 |
+
"loss": 0.2808,
|
| 1563 |
+
"step": 8288
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.85,
|
| 1567 |
+
"learning_rate": 1.6209178691548225e-07,
|
| 1568 |
+
"loss": 0.2753,
|
| 1569 |
+
"step": 8320
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.85,
|
| 1573 |
+
"learning_rate": 1.586199414126071e-07,
|
| 1574 |
+
"loss": 0.1965,
|
| 1575 |
+
"step": 8352
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.86,
|
| 1579 |
+
"learning_rate": 1.5514809590973203e-07,
|
| 1580 |
+
"loss": 0.2076,
|
| 1581 |
+
"step": 8384
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 0.86,
|
| 1585 |
+
"learning_rate": 1.516762504068569e-07,
|
| 1586 |
+
"loss": 0.2856,
|
| 1587 |
+
"step": 8416
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 0.86,
|
| 1591 |
+
"learning_rate": 1.4820440490398176e-07,
|
| 1592 |
+
"loss": 0.2812,
|
| 1593 |
+
"step": 8448
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.86,
|
| 1597 |
+
"learning_rate": 1.4473255940110665e-07,
|
| 1598 |
+
"loss": 0.28,
|
| 1599 |
+
"step": 8480
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.87,
|
| 1603 |
+
"learning_rate": 1.412607138982315e-07,
|
| 1604 |
+
"loss": 0.2871,
|
| 1605 |
+
"step": 8512
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.87,
|
| 1609 |
+
"learning_rate": 1.377888683953564e-07,
|
| 1610 |
+
"loss": 0.2063,
|
| 1611 |
+
"step": 8544
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.87,
|
| 1615 |
+
"learning_rate": 1.343170228924813e-07,
|
| 1616 |
+
"loss": 0.2929,
|
| 1617 |
+
"step": 8576
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.88,
|
| 1621 |
+
"learning_rate": 1.3084517738960616e-07,
|
| 1622 |
+
"loss": 0.2784,
|
| 1623 |
+
"step": 8608
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.88,
|
| 1627 |
+
"learning_rate": 1.2737333188673102e-07,
|
| 1628 |
+
"loss": 0.2889,
|
| 1629 |
+
"step": 8640
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.88,
|
| 1633 |
+
"learning_rate": 1.239014863838559e-07,
|
| 1634 |
+
"loss": 0.2367,
|
| 1635 |
+
"step": 8672
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.89,
|
| 1639 |
+
"learning_rate": 1.2042964088098078e-07,
|
| 1640 |
+
"loss": 0.2572,
|
| 1641 |
+
"step": 8704
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.89,
|
| 1645 |
+
"learning_rate": 1.1695779537810567e-07,
|
| 1646 |
+
"loss": 0.2167,
|
| 1647 |
+
"step": 8736
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.89,
|
| 1651 |
+
"learning_rate": 1.1348594987523056e-07,
|
| 1652 |
+
"loss": 0.2448,
|
| 1653 |
+
"step": 8768
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.9,
|
| 1657 |
+
"learning_rate": 1.1001410437235542e-07,
|
| 1658 |
+
"loss": 0.2384,
|
| 1659 |
+
"step": 8800
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.9,
|
| 1663 |
+
"learning_rate": 1.065422588694803e-07,
|
| 1664 |
+
"loss": 0.2742,
|
| 1665 |
+
"step": 8832
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.9,
|
| 1669 |
+
"learning_rate": 1.0307041336660519e-07,
|
| 1670 |
+
"loss": 0.2615,
|
| 1671 |
+
"step": 8864
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 0.91,
|
| 1675 |
+
"learning_rate": 9.959856786373005e-08,
|
| 1676 |
+
"loss": 0.2511,
|
| 1677 |
+
"step": 8896
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.91,
|
| 1681 |
+
"learning_rate": 9.612672236085493e-08,
|
| 1682 |
+
"loss": 0.218,
|
| 1683 |
+
"step": 8928
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.91,
|
| 1687 |
+
"learning_rate": 9.265487685797982e-08,
|
| 1688 |
+
"loss": 0.1846,
|
| 1689 |
+
"step": 8960
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.92,
|
| 1693 |
+
"learning_rate": 8.918303135510468e-08,
|
| 1694 |
+
"loss": 0.2521,
|
| 1695 |
+
"step": 8992
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 0.92,
|
| 1699 |
+
"learning_rate": 8.571118585222958e-08,
|
| 1700 |
+
"loss": 0.1836,
|
| 1701 |
+
"step": 9024
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 0.92,
|
| 1705 |
+
"learning_rate": 8.223934034935445e-08,
|
| 1706 |
+
"loss": 0.2421,
|
| 1707 |
+
"step": 9056
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.93,
|
| 1711 |
+
"learning_rate": 7.876749484647933e-08,
|
| 1712 |
+
"loss": 0.2631,
|
| 1713 |
+
"step": 9088
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 0.93,
|
| 1717 |
+
"learning_rate": 7.529564934360421e-08,
|
| 1718 |
+
"loss": 0.2602,
|
| 1719 |
+
"step": 9120
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.93,
|
| 1723 |
+
"learning_rate": 7.182380384072909e-08,
|
| 1724 |
+
"loss": 0.2896,
|
| 1725 |
+
"step": 9152
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.94,
|
| 1729 |
+
"learning_rate": 6.835195833785396e-08,
|
| 1730 |
+
"loss": 0.2901,
|
| 1731 |
+
"step": 9184
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.94,
|
| 1735 |
+
"learning_rate": 6.488011283497884e-08,
|
| 1736 |
+
"loss": 0.2488,
|
| 1737 |
+
"step": 9216
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 0.94,
|
| 1741 |
+
"learning_rate": 6.140826733210372e-08,
|
| 1742 |
+
"loss": 0.2209,
|
| 1743 |
+
"step": 9248
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 0.95,
|
| 1747 |
+
"learning_rate": 5.7936421829228595e-08,
|
| 1748 |
+
"loss": 0.2494,
|
| 1749 |
+
"step": 9280
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 0.95,
|
| 1753 |
+
"learning_rate": 5.446457632635348e-08,
|
| 1754 |
+
"loss": 0.251,
|
| 1755 |
+
"step": 9312
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 0.95,
|
| 1759 |
+
"learning_rate": 5.099273082347835e-08,
|
| 1760 |
+
"loss": 0.2243,
|
| 1761 |
+
"step": 9344
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 0.96,
|
| 1765 |
+
"learning_rate": 4.752088532060323e-08,
|
| 1766 |
+
"loss": 0.259,
|
| 1767 |
+
"step": 9376
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.96,
|
| 1771 |
+
"learning_rate": 4.404903981772811e-08,
|
| 1772 |
+
"loss": 0.267,
|
| 1773 |
+
"step": 9408
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.96,
|
| 1777 |
+
"learning_rate": 4.057719431485299e-08,
|
| 1778 |
+
"loss": 0.2674,
|
| 1779 |
+
"step": 9440
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 0.97,
|
| 1783 |
+
"learning_rate": 3.7105348811977866e-08,
|
| 1784 |
+
"loss": 0.2383,
|
| 1785 |
+
"step": 9472
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 0.97,
|
| 1789 |
+
"learning_rate": 3.363350330910274e-08,
|
| 1790 |
+
"loss": 0.2545,
|
| 1791 |
+
"step": 9504
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 0.97,
|
| 1795 |
+
"learning_rate": 3.016165780622762e-08,
|
| 1796 |
+
"loss": 0.2904,
|
| 1797 |
+
"step": 9536
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 0.98,
|
| 1801 |
+
"learning_rate": 2.6689812303352498e-08,
|
| 1802 |
+
"loss": 0.2211,
|
| 1803 |
+
"step": 9568
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 0.98,
|
| 1807 |
+
"learning_rate": 2.321796680047738e-08,
|
| 1808 |
+
"loss": 0.2802,
|
| 1809 |
+
"step": 9600
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.98,
|
| 1813 |
+
"learning_rate": 1.9746121297602256e-08,
|
| 1814 |
+
"loss": 0.2471,
|
| 1815 |
+
"step": 9632
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.99,
|
| 1819 |
+
"learning_rate": 1.6274275794727136e-08,
|
| 1820 |
+
"loss": 0.2593,
|
| 1821 |
+
"step": 9664
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 0.99,
|
| 1825 |
+
"learning_rate": 1.2802430291852012e-08,
|
| 1826 |
+
"loss": 0.1877,
|
| 1827 |
+
"step": 9696
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 0.99,
|
| 1831 |
+
"learning_rate": 9.33058478897689e-09,
|
| 1832 |
+
"loss": 0.2619,
|
| 1833 |
+
"step": 9728
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 1.0,
|
| 1837 |
+
"learning_rate": 5.858739286101768e-09,
|
| 1838 |
+
"loss": 0.1832,
|
| 1839 |
+
"step": 9760
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 1.0,
|
| 1843 |
+
"learning_rate": 2.3868937832266464e-09,
|
| 1844 |
+
"loss": 0.1479,
|
| 1845 |
+
"step": 9792
|
| 1846 |
+
}
|
| 1847 |
+
],
|
| 1848 |
+
"logging_steps": 32,
|
| 1849 |
+
"max_steps": 9805,
|
| 1850 |
+
"num_input_tokens_seen": 0,
|
| 1851 |
+
"num_train_epochs": 1,
|
| 1852 |
+
"save_steps": 9804,
|
| 1853 |
+
"total_flos": 4.16311373758464e+19,
|
| 1854 |
+
"train_batch_size": 2,
|
| 1855 |
+
"trial_name": null,
|
| 1856 |
+
"trial_params": null
|
| 1857 |
+
}
|
checkpoint-9804/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9dceb78a7286f69a51092265062bdc1281c52106c7586697930b99c2fc44aa03
|
| 3 |
+
size 5048
|
config.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "openai/whisper-large-v2",
|
| 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 |
+
50259
|
| 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 |
+
"language": "English",
|
| 45 |
+
"mask_feature_length": 10,
|
| 46 |
+
"mask_feature_min_masks": 0,
|
| 47 |
+
"mask_feature_prob": 0.0,
|
| 48 |
+
"mask_time_length": 10,
|
| 49 |
+
"mask_time_min_masks": 2,
|
| 50 |
+
"mask_time_prob": 0.05,
|
| 51 |
+
"max_length": 448,
|
| 52 |
+
"max_source_positions": 1500,
|
| 53 |
+
"max_target_positions": 448,
|
| 54 |
+
"median_filter_width": 7,
|
| 55 |
+
"model_type": "whisper",
|
| 56 |
+
"num_hidden_layers": 32,
|
| 57 |
+
"num_mel_bins": 80,
|
| 58 |
+
"pad_token_id": 50257,
|
| 59 |
+
"scale_embedding": false,
|
| 60 |
+
"suppress_tokens": [],
|
| 61 |
+
"torch_dtype": "float32",
|
| 62 |
+
"transformers_version": "4.37.2",
|
| 63 |
+
"use_cache": true,
|
| 64 |
+
"use_weighted_layer_sum": false,
|
| 65 |
+
"vocab_size": 51865
|
| 66 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,317 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alignment_heads": [
|
| 3 |
+
[
|
| 4 |
+
10,
|
| 5 |
+
12
|
| 6 |
+
],
|
| 7 |
+
[
|
| 8 |
+
13,
|
| 9 |
+
17
|
| 10 |
+
],
|
| 11 |
+
[
|
| 12 |
+
16,
|
| 13 |
+
11
|
| 14 |
+
],
|
| 15 |
+
[
|
| 16 |
+
16,
|
| 17 |
+
12
|
| 18 |
+
],
|
| 19 |
+
[
|
| 20 |
+
16,
|
| 21 |
+
13
|
| 22 |
+
],
|
| 23 |
+
[
|
| 24 |
+
17,
|
| 25 |
+
15
|
| 26 |
+
],
|
| 27 |
+
[
|
| 28 |
+
17,
|
| 29 |
+
16
|
| 30 |
+
],
|
| 31 |
+
[
|
| 32 |
+
18,
|
| 33 |
+
4
|
| 34 |
+
],
|
| 35 |
+
[
|
| 36 |
+
18,
|
| 37 |
+
11
|
| 38 |
+
],
|
| 39 |
+
[
|
| 40 |
+
18,
|
| 41 |
+
19
|
| 42 |
+
],
|
| 43 |
+
[
|
| 44 |
+
19,
|
| 45 |
+
11
|
| 46 |
+
],
|
| 47 |
+
[
|
| 48 |
+
21,
|
| 49 |
+
2
|
| 50 |
+
],
|
| 51 |
+
[
|
| 52 |
+
21,
|
| 53 |
+
3
|
| 54 |
+
],
|
| 55 |
+
[
|
| 56 |
+
22,
|
| 57 |
+
3
|
| 58 |
+
],
|
| 59 |
+
[
|
| 60 |
+
22,
|
| 61 |
+
9
|
| 62 |
+
],
|
| 63 |
+
[
|
| 64 |
+
22,
|
| 65 |
+
12
|
| 66 |
+
],
|
| 67 |
+
[
|
| 68 |
+
23,
|
| 69 |
+
5
|
| 70 |
+
],
|
| 71 |
+
[
|
| 72 |
+
23,
|
| 73 |
+
7
|
| 74 |
+
],
|
| 75 |
+
[
|
| 76 |
+
23,
|
| 77 |
+
13
|
| 78 |
+
],
|
| 79 |
+
[
|
| 80 |
+
25,
|
| 81 |
+
5
|
| 82 |
+
],
|
| 83 |
+
[
|
| 84 |
+
26,
|
| 85 |
+
1
|
| 86 |
+
],
|
| 87 |
+
[
|
| 88 |
+
26,
|
| 89 |
+
12
|
| 90 |
+
],
|
| 91 |
+
[
|
| 92 |
+
27,
|
| 93 |
+
15
|
| 94 |
+
]
|
| 95 |
+
],
|
| 96 |
+
"begin_suppress_tokens": [
|
| 97 |
+
220,
|
| 98 |
+
50257
|
| 99 |
+
],
|
| 100 |
+
"bos_token_id": 50257,
|
| 101 |
+
"decoder_start_token_id": 50258,
|
| 102 |
+
"eos_token_id": 50257,
|
| 103 |
+
"forced_decoder_ids": [
|
| 104 |
+
[
|
| 105 |
+
1,
|
| 106 |
+
null
|
| 107 |
+
],
|
| 108 |
+
[
|
| 109 |
+
2,
|
| 110 |
+
50359
|
| 111 |
+
]
|
| 112 |
+
],
|
| 113 |
+
"is_multilingual": true,
|
| 114 |
+
"lang_to_id": {
|
| 115 |
+
"<|af|>": 50327,
|
| 116 |
+
"<|am|>": 50334,
|
| 117 |
+
"<|ar|>": 50272,
|
| 118 |
+
"<|as|>": 50350,
|
| 119 |
+
"<|az|>": 50304,
|
| 120 |
+
"<|ba|>": 50355,
|
| 121 |
+
"<|be|>": 50330,
|
| 122 |
+
"<|bg|>": 50292,
|
| 123 |
+
"<|bn|>": 50302,
|
| 124 |
+
"<|bo|>": 50347,
|
| 125 |
+
"<|br|>": 50309,
|
| 126 |
+
"<|bs|>": 50315,
|
| 127 |
+
"<|ca|>": 50270,
|
| 128 |
+
"<|cs|>": 50283,
|
| 129 |
+
"<|cy|>": 50297,
|
| 130 |
+
"<|da|>": 50285,
|
| 131 |
+
"<|de|>": 50261,
|
| 132 |
+
"<|el|>": 50281,
|
| 133 |
+
"<|en|>": 50259,
|
| 134 |
+
"<|es|>": 50262,
|
| 135 |
+
"<|et|>": 50307,
|
| 136 |
+
"<|eu|>": 50310,
|
| 137 |
+
"<|fa|>": 50300,
|
| 138 |
+
"<|fi|>": 50277,
|
| 139 |
+
"<|fo|>": 50338,
|
| 140 |
+
"<|fr|>": 50265,
|
| 141 |
+
"<|gl|>": 50319,
|
| 142 |
+
"<|gu|>": 50333,
|
| 143 |
+
"<|haw|>": 50352,
|
| 144 |
+
"<|ha|>": 50354,
|
| 145 |
+
"<|he|>": 50279,
|
| 146 |
+
"<|hi|>": 50276,
|
| 147 |
+
"<|hr|>": 50291,
|
| 148 |
+
"<|ht|>": 50339,
|
| 149 |
+
"<|hu|>": 50286,
|
| 150 |
+
"<|hy|>": 50312,
|
| 151 |
+
"<|id|>": 50275,
|
| 152 |
+
"<|is|>": 50311,
|
| 153 |
+
"<|it|>": 50274,
|
| 154 |
+
"<|ja|>": 50266,
|
| 155 |
+
"<|jw|>": 50356,
|
| 156 |
+
"<|ka|>": 50329,
|
| 157 |
+
"<|kk|>": 50316,
|
| 158 |
+
"<|km|>": 50323,
|
| 159 |
+
"<|kn|>": 50306,
|
| 160 |
+
"<|ko|>": 50264,
|
| 161 |
+
"<|la|>": 50294,
|
| 162 |
+
"<|lb|>": 50345,
|
| 163 |
+
"<|ln|>": 50353,
|
| 164 |
+
"<|lo|>": 50336,
|
| 165 |
+
"<|lt|>": 50293,
|
| 166 |
+
"<|lv|>": 50301,
|
| 167 |
+
"<|mg|>": 50349,
|
| 168 |
+
"<|mi|>": 50295,
|
| 169 |
+
"<|mk|>": 50308,
|
| 170 |
+
"<|ml|>": 50296,
|
| 171 |
+
"<|mn|>": 50314,
|
| 172 |
+
"<|mr|>": 50320,
|
| 173 |
+
"<|ms|>": 50282,
|
| 174 |
+
"<|mt|>": 50343,
|
| 175 |
+
"<|my|>": 50346,
|
| 176 |
+
"<|ne|>": 50313,
|
| 177 |
+
"<|nl|>": 50271,
|
| 178 |
+
"<|nn|>": 50342,
|
| 179 |
+
"<|no|>": 50288,
|
| 180 |
+
"<|oc|>": 50328,
|
| 181 |
+
"<|pa|>": 50321,
|
| 182 |
+
"<|pl|>": 50269,
|
| 183 |
+
"<|ps|>": 50340,
|
| 184 |
+
"<|pt|>": 50267,
|
| 185 |
+
"<|ro|>": 50284,
|
| 186 |
+
"<|ru|>": 50263,
|
| 187 |
+
"<|sa|>": 50344,
|
| 188 |
+
"<|sd|>": 50332,
|
| 189 |
+
"<|si|>": 50322,
|
| 190 |
+
"<|sk|>": 50298,
|
| 191 |
+
"<|sl|>": 50305,
|
| 192 |
+
"<|sn|>": 50324,
|
| 193 |
+
"<|so|>": 50326,
|
| 194 |
+
"<|sq|>": 50317,
|
| 195 |
+
"<|sr|>": 50303,
|
| 196 |
+
"<|su|>": 50357,
|
| 197 |
+
"<|sv|>": 50273,
|
| 198 |
+
"<|sw|>": 50318,
|
| 199 |
+
"<|ta|>": 50287,
|
| 200 |
+
"<|te|>": 50299,
|
| 201 |
+
"<|tg|>": 50331,
|
| 202 |
+
"<|th|>": 50289,
|
| 203 |
+
"<|tk|>": 50341,
|
| 204 |
+
"<|tl|>": 50348,
|
| 205 |
+
"<|tr|>": 50268,
|
| 206 |
+
"<|tt|>": 50351,
|
| 207 |
+
"<|uk|>": 50280,
|
| 208 |
+
"<|ur|>": 50290,
|
| 209 |
+
"<|uz|>": 50337,
|
| 210 |
+
"<|vi|>": 50278,
|
| 211 |
+
"<|yi|>": 50335,
|
| 212 |
+
"<|yo|>": 50325,
|
| 213 |
+
"<|zh|>": 50260
|
| 214 |
+
},
|
| 215 |
+
"max_initial_timestamp_index": 50,
|
| 216 |
+
"max_length": 448,
|
| 217 |
+
"no_timestamps_token_id": 50363,
|
| 218 |
+
"pad_token_id": 50257,
|
| 219 |
+
"prev_sot_token_id": 50361,
|
| 220 |
+
"return_timestamps": false,
|
| 221 |
+
"suppress_tokens": [
|
| 222 |
+
1,
|
| 223 |
+
2,
|
| 224 |
+
7,
|
| 225 |
+
8,
|
| 226 |
+
9,
|
| 227 |
+
10,
|
| 228 |
+
14,
|
| 229 |
+
25,
|
| 230 |
+
26,
|
| 231 |
+
27,
|
| 232 |
+
28,
|
| 233 |
+
29,
|
| 234 |
+
31,
|
| 235 |
+
58,
|
| 236 |
+
59,
|
| 237 |
+
60,
|
| 238 |
+
61,
|
| 239 |
+
62,
|
| 240 |
+
63,
|
| 241 |
+
90,
|
| 242 |
+
91,
|
| 243 |
+
92,
|
| 244 |
+
93,
|
| 245 |
+
359,
|
| 246 |
+
503,
|
| 247 |
+
522,
|
| 248 |
+
542,
|
| 249 |
+
873,
|
| 250 |
+
893,
|
| 251 |
+
902,
|
| 252 |
+
918,
|
| 253 |
+
922,
|
| 254 |
+
931,
|
| 255 |
+
1350,
|
| 256 |
+
1853,
|
| 257 |
+
1982,
|
| 258 |
+
2460,
|
| 259 |
+
2627,
|
| 260 |
+
3246,
|
| 261 |
+
3253,
|
| 262 |
+
3268,
|
| 263 |
+
3536,
|
| 264 |
+
3846,
|
| 265 |
+
3961,
|
| 266 |
+
4183,
|
| 267 |
+
4667,
|
| 268 |
+
6585,
|
| 269 |
+
6647,
|
| 270 |
+
7273,
|
| 271 |
+
9061,
|
| 272 |
+
9383,
|
| 273 |
+
10428,
|
| 274 |
+
10929,
|
| 275 |
+
11938,
|
| 276 |
+
12033,
|
| 277 |
+
12331,
|
| 278 |
+
12562,
|
| 279 |
+
13793,
|
| 280 |
+
14157,
|
| 281 |
+
14635,
|
| 282 |
+
15265,
|
| 283 |
+
15618,
|
| 284 |
+
16553,
|
| 285 |
+
16604,
|
| 286 |
+
18362,
|
| 287 |
+
18956,
|
| 288 |
+
20075,
|
| 289 |
+
21675,
|
| 290 |
+
22520,
|
| 291 |
+
26130,
|
| 292 |
+
26161,
|
| 293 |
+
26435,
|
| 294 |
+
28279,
|
| 295 |
+
29464,
|
| 296 |
+
31650,
|
| 297 |
+
32302,
|
| 298 |
+
32470,
|
| 299 |
+
36865,
|
| 300 |
+
42863,
|
| 301 |
+
47425,
|
| 302 |
+
49870,
|
| 303 |
+
50254,
|
| 304 |
+
50258,
|
| 305 |
+
50358,
|
| 306 |
+
50359,
|
| 307 |
+
50360,
|
| 308 |
+
50361,
|
| 309 |
+
50362
|
| 310 |
+
],
|
| 311 |
+
"task": "transcribe",
|
| 312 |
+
"task_to_id": {
|
| 313 |
+
"transcribe": 50359,
|
| 314 |
+
"translate": 50358
|
| 315 |
+
},
|
| 316 |
+
"transformers_version": "4.37.2"
|
| 317 |
+
}
|
model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:32724c310fffb23aaad5dabb379092be841a6947b870d0e1290be5d8eb9debf7
|
| 3 |
+
size 4992706480
|
model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb4d0467cf7aa5703d273de128e4fbb31af23f41154db0fead9c829d2130c81e
|
| 3 |
+
size 1180663192
|
model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
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 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9dceb78a7286f69a51092265062bdc1281c52106c7586697930b99c2fc44aa03
|
| 3 |
+
size 5048
|