Training in progress, step 250
Browse files- config.json +46 -0
- generation_config.json +249 -0
- model.safetensors +3 -0
- preprocessor_config.json +14 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_dropout": 0.0,
|
| 3 |
+
"activation_function": "gelu",
|
| 4 |
+
"apply_spec_augment": false,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"WhisperForConditionalGeneration"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 50257,
|
| 10 |
+
"classifier_proj_size": 256,
|
| 11 |
+
"d_model": 512,
|
| 12 |
+
"decoder_attention_heads": 8,
|
| 13 |
+
"decoder_ffn_dim": 2048,
|
| 14 |
+
"decoder_layerdrop": 0.0,
|
| 15 |
+
"decoder_layers": 6,
|
| 16 |
+
"decoder_start_token_id": 50258,
|
| 17 |
+
"dropout": 0.0,
|
| 18 |
+
"dtype": "float32",
|
| 19 |
+
"encoder_attention_heads": 8,
|
| 20 |
+
"encoder_ffn_dim": 2048,
|
| 21 |
+
"encoder_layerdrop": 0.0,
|
| 22 |
+
"encoder_layers": 6,
|
| 23 |
+
"eos_token_id": 50257,
|
| 24 |
+
"forced_decoder_ids": null,
|
| 25 |
+
"init_std": 0.02,
|
| 26 |
+
"is_encoder_decoder": true,
|
| 27 |
+
"mask_feature_length": 10,
|
| 28 |
+
"mask_feature_min_masks": 0,
|
| 29 |
+
"mask_feature_prob": 0.0,
|
| 30 |
+
"mask_time_length": 10,
|
| 31 |
+
"mask_time_min_masks": 2,
|
| 32 |
+
"mask_time_prob": 0.05,
|
| 33 |
+
"max_source_positions": 1500,
|
| 34 |
+
"max_target_positions": 448,
|
| 35 |
+
"median_filter_width": 7,
|
| 36 |
+
"model_type": "whisper",
|
| 37 |
+
"num_hidden_layers": 6,
|
| 38 |
+
"num_mel_bins": 80,
|
| 39 |
+
"pad_token_id": 50257,
|
| 40 |
+
"scale_embedding": false,
|
| 41 |
+
"tie_word_embeddings": true,
|
| 42 |
+
"transformers_version": "5.0.0",
|
| 43 |
+
"use_cache": false,
|
| 44 |
+
"use_weighted_layer_sum": false,
|
| 45 |
+
"vocab_size": 51865
|
| 46 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,249 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": false,
|
| 3 |
+
"assistant_confidence_threshold": 0.4,
|
| 4 |
+
"assistant_lookbehind": 10,
|
| 5 |
+
"begin_suppress_tokens": [
|
| 6 |
+
220,
|
| 7 |
+
50257
|
| 8 |
+
],
|
| 9 |
+
"bos_token_id": 50257,
|
| 10 |
+
"decoder_start_token_id": 50258,
|
| 11 |
+
"diversity_penalty": 0.0,
|
| 12 |
+
"do_sample": false,
|
| 13 |
+
"early_stopping": false,
|
| 14 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 15 |
+
"encoder_repetition_penalty": 1.0,
|
| 16 |
+
"eos_token_id": 50257,
|
| 17 |
+
"epsilon_cutoff": 0.0,
|
| 18 |
+
"eta_cutoff": 0.0,
|
| 19 |
+
"forced_decoder_ids": null,
|
| 20 |
+
"id_to_lang": {
|
| 21 |
+
"50259": "<|en|>",
|
| 22 |
+
"50260": "<|zh|>",
|
| 23 |
+
"50261": "<|de|>",
|
| 24 |
+
"50262": "<|es|>",
|
| 25 |
+
"50263": "<|ru|>",
|
| 26 |
+
"50264": "<|ko|>",
|
| 27 |
+
"50265": "<|fr|>",
|
| 28 |
+
"50266": "<|ja|>",
|
| 29 |
+
"50267": "<|pt|>",
|
| 30 |
+
"50268": "<|tr|>",
|
| 31 |
+
"50269": "<|pl|>",
|
| 32 |
+
"50270": "<|ca|>",
|
| 33 |
+
"50271": "<|nl|>",
|
| 34 |
+
"50272": "<|ar|>",
|
| 35 |
+
"50273": "<|sv|>",
|
| 36 |
+
"50274": "<|it|>",
|
| 37 |
+
"50275": "<|id|>",
|
| 38 |
+
"50276": "<|hi|>",
|
| 39 |
+
"50277": "<|fi|>",
|
| 40 |
+
"50278": "<|vi|>",
|
| 41 |
+
"50279": "<|he|>",
|
| 42 |
+
"50280": "<|uk|>",
|
| 43 |
+
"50281": "<|el|>",
|
| 44 |
+
"50282": "<|ms|>",
|
| 45 |
+
"50283": "<|cs|>",
|
| 46 |
+
"50284": "<|ro|>",
|
| 47 |
+
"50285": "<|da|>",
|
| 48 |
+
"50286": "<|hu|>",
|
| 49 |
+
"50287": "<|ta|>",
|
| 50 |
+
"50288": "<|no|>",
|
| 51 |
+
"50289": "<|th|>",
|
| 52 |
+
"50290": "<|ur|>",
|
| 53 |
+
"50291": "<|hr|>",
|
| 54 |
+
"50292": "<|bg|>",
|
| 55 |
+
"50293": "<|lt|>",
|
| 56 |
+
"50294": "<|la|>",
|
| 57 |
+
"50295": "<|mi|>",
|
| 58 |
+
"50296": "<|ml|>",
|
| 59 |
+
"50297": "<|cy|>",
|
| 60 |
+
"50298": "<|sk|>",
|
| 61 |
+
"50299": "<|te|>",
|
| 62 |
+
"50300": "<|fa|>",
|
| 63 |
+
"50301": "<|lv|>",
|
| 64 |
+
"50302": "<|bn|>",
|
| 65 |
+
"50303": "<|sr|>",
|
| 66 |
+
"50304": "<|az|>",
|
| 67 |
+
"50305": "<|sl|>",
|
| 68 |
+
"50306": "<|kn|>",
|
| 69 |
+
"50307": "<|et|>",
|
| 70 |
+
"50308": "<|mk|>",
|
| 71 |
+
"50309": "<|br|>",
|
| 72 |
+
"50310": "<|eu|>",
|
| 73 |
+
"50311": "<|is|>",
|
| 74 |
+
"50312": "<|hy|>",
|
| 75 |
+
"50313": "<|ne|>",
|
| 76 |
+
"50314": "<|mn|>",
|
| 77 |
+
"50315": "<|bs|>",
|
| 78 |
+
"50316": "<|kk|>",
|
| 79 |
+
"50317": "<|sq|>",
|
| 80 |
+
"50318": "<|sw|>",
|
| 81 |
+
"50319": "<|gl|>",
|
| 82 |
+
"50320": "<|mr|>",
|
| 83 |
+
"50321": "<|pa|>",
|
| 84 |
+
"50322": "<|si|>",
|
| 85 |
+
"50323": "<|km|>",
|
| 86 |
+
"50324": "<|sn|>",
|
| 87 |
+
"50325": "<|yo|>",
|
| 88 |
+
"50326": "<|so|>",
|
| 89 |
+
"50327": "<|af|>",
|
| 90 |
+
"50328": "<|oc|>",
|
| 91 |
+
"50329": "<|ka|>",
|
| 92 |
+
"50330": "<|be|>",
|
| 93 |
+
"50331": "<|tg|>",
|
| 94 |
+
"50332": "<|sd|>",
|
| 95 |
+
"50333": "<|gu|>",
|
| 96 |
+
"50334": "<|am|>",
|
| 97 |
+
"50335": "<|yi|>",
|
| 98 |
+
"50336": "<|lo|>",
|
| 99 |
+
"50337": "<|uz|>",
|
| 100 |
+
"50338": "<|fo|>",
|
| 101 |
+
"50339": "<|ht|>",
|
| 102 |
+
"50340": "<|ps|>",
|
| 103 |
+
"50341": "<|tk|>",
|
| 104 |
+
"50342": "<|nn|>",
|
| 105 |
+
"50343": "<|mt|>",
|
| 106 |
+
"50344": "<|sa|>",
|
| 107 |
+
"50345": "<|lb|>",
|
| 108 |
+
"50346": "<|my|>",
|
| 109 |
+
"50347": "<|bo|>",
|
| 110 |
+
"50348": "<|tl|>",
|
| 111 |
+
"50349": "<|mg|>",
|
| 112 |
+
"50350": "<|as|>",
|
| 113 |
+
"50351": "<|tt|>",
|
| 114 |
+
"50352": "<|haw|>",
|
| 115 |
+
"50353": "<|ln|>",
|
| 116 |
+
"50354": "<|ha|>",
|
| 117 |
+
"50355": "<|ba|>",
|
| 118 |
+
"50356": "<|jw|>",
|
| 119 |
+
"50357": "<|su|>"
|
| 120 |
+
},
|
| 121 |
+
"lang_to_id": {
|
| 122 |
+
"<|af|>": 50327,
|
| 123 |
+
"<|am|>": 50334,
|
| 124 |
+
"<|ar|>": 50272,
|
| 125 |
+
"<|as|>": 50350,
|
| 126 |
+
"<|az|>": 50304,
|
| 127 |
+
"<|ba|>": 50355,
|
| 128 |
+
"<|be|>": 50330,
|
| 129 |
+
"<|bg|>": 50292,
|
| 130 |
+
"<|bn|>": 50302,
|
| 131 |
+
"<|bo|>": 50347,
|
| 132 |
+
"<|br|>": 50309,
|
| 133 |
+
"<|bs|>": 50315,
|
| 134 |
+
"<|ca|>": 50270,
|
| 135 |
+
"<|cs|>": 50283,
|
| 136 |
+
"<|cy|>": 50297,
|
| 137 |
+
"<|da|>": 50285,
|
| 138 |
+
"<|de|>": 50261,
|
| 139 |
+
"<|el|>": 50281,
|
| 140 |
+
"<|en|>": 50259,
|
| 141 |
+
"<|es|>": 50262,
|
| 142 |
+
"<|et|>": 50307,
|
| 143 |
+
"<|eu|>": 50310,
|
| 144 |
+
"<|fa|>": 50300,
|
| 145 |
+
"<|fi|>": 50277,
|
| 146 |
+
"<|fo|>": 50338,
|
| 147 |
+
"<|fr|>": 50265,
|
| 148 |
+
"<|gl|>": 50319,
|
| 149 |
+
"<|gu|>": 50333,
|
| 150 |
+
"<|haw|>": 50352,
|
| 151 |
+
"<|ha|>": 50354,
|
| 152 |
+
"<|he|>": 50279,
|
| 153 |
+
"<|hi|>": 50276,
|
| 154 |
+
"<|hr|>": 50291,
|
| 155 |
+
"<|ht|>": 50339,
|
| 156 |
+
"<|hu|>": 50286,
|
| 157 |
+
"<|hy|>": 50312,
|
| 158 |
+
"<|id|>": 50275,
|
| 159 |
+
"<|is|>": 50311,
|
| 160 |
+
"<|it|>": 50274,
|
| 161 |
+
"<|ja|>": 50266,
|
| 162 |
+
"<|jw|>": 50356,
|
| 163 |
+
"<|ka|>": 50329,
|
| 164 |
+
"<|kk|>": 50316,
|
| 165 |
+
"<|km|>": 50323,
|
| 166 |
+
"<|kn|>": 50306,
|
| 167 |
+
"<|ko|>": 50264,
|
| 168 |
+
"<|la|>": 50294,
|
| 169 |
+
"<|lb|>": 50345,
|
| 170 |
+
"<|ln|>": 50353,
|
| 171 |
+
"<|lo|>": 50336,
|
| 172 |
+
"<|lt|>": 50293,
|
| 173 |
+
"<|lv|>": 50301,
|
| 174 |
+
"<|mg|>": 50349,
|
| 175 |
+
"<|mi|>": 50295,
|
| 176 |
+
"<|mk|>": 50308,
|
| 177 |
+
"<|ml|>": 50296,
|
| 178 |
+
"<|mn|>": 50314,
|
| 179 |
+
"<|mr|>": 50320,
|
| 180 |
+
"<|ms|>": 50282,
|
| 181 |
+
"<|mt|>": 50343,
|
| 182 |
+
"<|my|>": 50346,
|
| 183 |
+
"<|ne|>": 50313,
|
| 184 |
+
"<|nl|>": 50271,
|
| 185 |
+
"<|nn|>": 50342,
|
| 186 |
+
"<|no|>": 50288,
|
| 187 |
+
"<|oc|>": 50328,
|
| 188 |
+
"<|pa|>": 50321,
|
| 189 |
+
"<|pl|>": 50269,
|
| 190 |
+
"<|ps|>": 50340,
|
| 191 |
+
"<|pt|>": 50267,
|
| 192 |
+
"<|ro|>": 50284,
|
| 193 |
+
"<|ru|>": 50263,
|
| 194 |
+
"<|sa|>": 50344,
|
| 195 |
+
"<|sd|>": 50332,
|
| 196 |
+
"<|si|>": 50322,
|
| 197 |
+
"<|sk|>": 50298,
|
| 198 |
+
"<|sl|>": 50305,
|
| 199 |
+
"<|sn|>": 50324,
|
| 200 |
+
"<|so|>": 50326,
|
| 201 |
+
"<|sq|>": 50317,
|
| 202 |
+
"<|sr|>": 50303,
|
| 203 |
+
"<|su|>": 50357,
|
| 204 |
+
"<|sv|>": 50273,
|
| 205 |
+
"<|sw|>": 50318,
|
| 206 |
+
"<|ta|>": 50287,
|
| 207 |
+
"<|te|>": 50299,
|
| 208 |
+
"<|tg|>": 50331,
|
| 209 |
+
"<|th|>": 50289,
|
| 210 |
+
"<|tk|>": 50341,
|
| 211 |
+
"<|tl|>": 50348,
|
| 212 |
+
"<|tr|>": 50268,
|
| 213 |
+
"<|tt|>": 50351,
|
| 214 |
+
"<|uk|>": 50280,
|
| 215 |
+
"<|ur|>": 50290,
|
| 216 |
+
"<|uz|>": 50337,
|
| 217 |
+
"<|vi|>": 50278,
|
| 218 |
+
"<|yi|>": 50335,
|
| 219 |
+
"<|yo|>": 50325,
|
| 220 |
+
"<|zh|>": 50260
|
| 221 |
+
},
|
| 222 |
+
"language": "arabic",
|
| 223 |
+
"length_penalty": 1.0,
|
| 224 |
+
"max_length": 1024,
|
| 225 |
+
"min_length": 0,
|
| 226 |
+
"no_repeat_ngram_size": 0,
|
| 227 |
+
"num_assistant_tokens": 20,
|
| 228 |
+
"num_assistant_tokens_schedule": "constant",
|
| 229 |
+
"num_beam_groups": 1,
|
| 230 |
+
"num_beams": 1,
|
| 231 |
+
"num_return_sequences": 1,
|
| 232 |
+
"output_scores": false,
|
| 233 |
+
"pad_token_id": 50257,
|
| 234 |
+
"remove_invalid_values": false,
|
| 235 |
+
"repetition_penalty": 1.0,
|
| 236 |
+
"return_dict_in_generate": false,
|
| 237 |
+
"target_lookbehind": 10,
|
| 238 |
+
"task": "transcribe",
|
| 239 |
+
"task_to_id": {
|
| 240 |
+
"transcribe": 50359,
|
| 241 |
+
"translate": 50358
|
| 242 |
+
},
|
| 243 |
+
"temperature": 1.0,
|
| 244 |
+
"top_k": 50,
|
| 245 |
+
"top_p": 1.0,
|
| 246 |
+
"transformers_version": "5.0.0",
|
| 247 |
+
"typical_p": 1.0,
|
| 248 |
+
"use_cache": false
|
| 249 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:89630fd66211506e60a806a008bb677ff5d04cf9ff4867403ed2e980523ea761
|
| 3 |
+
size 290403936
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chunk_length": 30,
|
| 3 |
+
"dither": 0.0,
|
| 4 |
+
"feature_extractor_type": "WhisperFeatureExtractor",
|
| 5 |
+
"feature_size": 80,
|
| 6 |
+
"hop_length": 160,
|
| 7 |
+
"n_fft": 400,
|
| 8 |
+
"n_samples": 480000,
|
| 9 |
+
"nb_max_frames": 3000,
|
| 10 |
+
"padding_side": "right",
|
| 11 |
+
"padding_value": 0.0,
|
| 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 sha256:4562d7bf5b8f4c8374cc816a6f9417c832b621c8ecaccdc4fff4216ff063642c
|
| 3 |
+
size 5393
|