Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- checkpoint-10075/config.json +66 -0
- checkpoint-10075/generation_config.json +317 -0
- checkpoint-10075/model-00001-of-00002.safetensors +3 -0
- checkpoint-10075/model-00002-of-00002.safetensors +3 -0
- checkpoint-10075/model.safetensors.index.json +0 -0
- checkpoint-10075/optimizer.pt +3 -0
- checkpoint-10075/preprocessor_config.json +14 -0
- checkpoint-10075/rng_state.pth +3 -0
- checkpoint-10075/scheduler.pt +3 -0
- checkpoint-10075/trainer_state.json +645 -0
- checkpoint-10075/training_args.bin +3 -0
- checkpoint-13433/config.json +66 -0
- checkpoint-13433/generation_config.json +317 -0
- checkpoint-13433/model-00001-of-00002.safetensors +3 -0
- checkpoint-13433/model-00002-of-00002.safetensors +3 -0
- checkpoint-13433/model.safetensors.index.json +0 -0
- checkpoint-13433/optimizer.pt +3 -0
- checkpoint-13433/preprocessor_config.json +14 -0
- checkpoint-13433/rng_state.pth +3 -0
- checkpoint-13433/scheduler.pt +3 -0
- checkpoint-13433/trainer_state.json +1275 -0
- checkpoint-13433/training_args.bin +3 -0
- checkpoint-16791/config.json +66 -0
- checkpoint-16791/generation_config.json +317 -0
- checkpoint-16791/model-00001-of-00002.safetensors +3 -0
- checkpoint-16791/model-00002-of-00002.safetensors +3 -0
- checkpoint-16791/model.safetensors.index.json +0 -0
- checkpoint-16791/optimizer.pt +3 -0
- checkpoint-16791/preprocessor_config.json +14 -0
- checkpoint-16791/rng_state.pth +3 -0
- checkpoint-16791/scheduler.pt +3 -0
- checkpoint-16791/trainer_state.json +1905 -0
- checkpoint-16791/training_args.bin +3 -0
- checkpoint-20149/config.json +66 -0
- checkpoint-20149/generation_config.json +317 -0
- checkpoint-20149/model-00001-of-00002.safetensors +3 -0
- checkpoint-20149/model-00002-of-00002.safetensors +3 -0
- checkpoint-20149/model.safetensors.index.json +0 -0
- checkpoint-20149/optimizer.pt +3 -0
- checkpoint-20149/preprocessor_config.json +14 -0
- checkpoint-20149/rng_state.pth +3 -0
- checkpoint-20149/scheduler.pt +3 -0
- checkpoint-20149/trainer_state.json +2535 -0
- checkpoint-20149/training_args.bin +3 -0
- checkpoint-6717/config.json +66 -0
- checkpoint-6717/generation_config.json +317 -0
- checkpoint-6717/model-00001-of-00002.safetensors +3 -0
- checkpoint-6717/model-00002-of-00002.safetensors +3 -0
- checkpoint-6717/model.safetensors.index.json +0 -0
- checkpoint-6717/optimizer.pt +3 -0
checkpoint-10075/config.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "models/openai/whisper-large-v2/finetune/jacob_filter/checkpoint-6717",
|
| 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-10075/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-10075/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f3ff75421447334fd0a2ead84a356ec07b488e43b6492c807caeff59f8c7641
|
| 3 |
+
size 4992706480
|
checkpoint-10075/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29e8bfbe4544338e59c49c157627ff9fd563af558246fe0940e874b402a9e880
|
| 3 |
+
size 1180663192
|
checkpoint-10075/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-10075/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e094cf541543b4e1a8038f87df9518b3b8a73e713aee1760587b0f404e45739
|
| 3 |
+
size 3095074288
|
checkpoint-10075/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-10075/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a5fd5470791ad73efaef8471cd3e2f2da4ef4c2cff54ddd562898211ac51825
|
| 3 |
+
size 14244
|
checkpoint-10075/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d17eded206da68210dfc655022b88a19b27a6dc1deea237f15ca64d17aae0dc8
|
| 3 |
+
size 1064
|
checkpoint-10075/trainer_state.json
ADDED
|
@@ -0,0 +1,645 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.4999255620068483,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 3358,
|
| 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": 3.722084367245657e-08,
|
| 14 |
+
"loss": 0.2836,
|
| 15 |
+
"step": 32
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 7.692307692307692e-08,
|
| 20 |
+
"loss": 0.2111,
|
| 21 |
+
"step": 64
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 1.1662531017369727e-07,
|
| 26 |
+
"loss": 0.2145,
|
| 27 |
+
"step": 96
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.02,
|
| 31 |
+
"learning_rate": 1.563275434243176e-07,
|
| 32 |
+
"loss": 0.2375,
|
| 33 |
+
"step": 128
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 1.9602977667493795e-07,
|
| 38 |
+
"loss": 0.1839,
|
| 39 |
+
"step": 160
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03,
|
| 43 |
+
"learning_rate": 2.3573200992555832e-07,
|
| 44 |
+
"loss": 0.2864,
|
| 45 |
+
"step": 192
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03,
|
| 49 |
+
"learning_rate": 2.7543424317617863e-07,
|
| 50 |
+
"loss": 0.1799,
|
| 51 |
+
"step": 224
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.04,
|
| 55 |
+
"learning_rate": 3.1513647642679897e-07,
|
| 56 |
+
"loss": 0.1609,
|
| 57 |
+
"step": 256
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.04,
|
| 61 |
+
"learning_rate": 3.5483870967741936e-07,
|
| 62 |
+
"loss": 0.1434,
|
| 63 |
+
"step": 288
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.05,
|
| 67 |
+
"learning_rate": 3.945409429280397e-07,
|
| 68 |
+
"loss": 0.1699,
|
| 69 |
+
"step": 320
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.05,
|
| 73 |
+
"learning_rate": 4.3424317617866004e-07,
|
| 74 |
+
"loss": 0.1898,
|
| 75 |
+
"step": 352
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.06,
|
| 79 |
+
"learning_rate": 4.739454094292804e-07,
|
| 80 |
+
"loss": 0.1665,
|
| 81 |
+
"step": 384
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.06,
|
| 85 |
+
"learning_rate": 5.136476426799007e-07,
|
| 86 |
+
"loss": 0.1358,
|
| 87 |
+
"step": 416
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.07,
|
| 91 |
+
"learning_rate": 5.533498759305211e-07,
|
| 92 |
+
"loss": 0.1713,
|
| 93 |
+
"step": 448
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07,
|
| 97 |
+
"learning_rate": 5.930521091811415e-07,
|
| 98 |
+
"loss": 0.1817,
|
| 99 |
+
"step": 480
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.08,
|
| 103 |
+
"learning_rate": 6.327543424317618e-07,
|
| 104 |
+
"loss": 0.149,
|
| 105 |
+
"step": 512
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.08,
|
| 109 |
+
"learning_rate": 6.724565756823821e-07,
|
| 110 |
+
"loss": 0.165,
|
| 111 |
+
"step": 544
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.09,
|
| 115 |
+
"learning_rate": 7.121588089330024e-07,
|
| 116 |
+
"loss": 0.2282,
|
| 117 |
+
"step": 576
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.09,
|
| 121 |
+
"learning_rate": 7.518610421836227e-07,
|
| 122 |
+
"loss": 0.1666,
|
| 123 |
+
"step": 608
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.1,
|
| 127 |
+
"learning_rate": 7.915632754342431e-07,
|
| 128 |
+
"loss": 0.1229,
|
| 129 |
+
"step": 640
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1,
|
| 133 |
+
"learning_rate": 8.312655086848634e-07,
|
| 134 |
+
"loss": 0.2118,
|
| 135 |
+
"step": 672
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1,
|
| 139 |
+
"learning_rate": 8.709677419354838e-07,
|
| 140 |
+
"loss": 0.1869,
|
| 141 |
+
"step": 704
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.11,
|
| 145 |
+
"learning_rate": 9.106699751861042e-07,
|
| 146 |
+
"loss": 0.2196,
|
| 147 |
+
"step": 736
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.11,
|
| 151 |
+
"learning_rate": 9.503722084367245e-07,
|
| 152 |
+
"loss": 0.0936,
|
| 153 |
+
"step": 768
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.12,
|
| 157 |
+
"learning_rate": 9.90074441687345e-07,
|
| 158 |
+
"loss": 0.1839,
|
| 159 |
+
"step": 800
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.12,
|
| 163 |
+
"learning_rate": 9.980994615140957e-07,
|
| 164 |
+
"loss": 0.1675,
|
| 165 |
+
"step": 832
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.13,
|
| 169 |
+
"learning_rate": 9.955654101995564e-07,
|
| 170 |
+
"loss": 0.1595,
|
| 171 |
+
"step": 864
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13,
|
| 175 |
+
"learning_rate": 9.930313588850174e-07,
|
| 176 |
+
"loss": 0.1556,
|
| 177 |
+
"step": 896
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.14,
|
| 181 |
+
"learning_rate": 9.905764966740576e-07,
|
| 182 |
+
"loss": 0.1711,
|
| 183 |
+
"step": 928
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.14,
|
| 187 |
+
"learning_rate": 9.880424453595185e-07,
|
| 188 |
+
"loss": 0.1777,
|
| 189 |
+
"step": 960
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.15,
|
| 193 |
+
"learning_rate": 9.855083940449792e-07,
|
| 194 |
+
"loss": 0.2031,
|
| 195 |
+
"step": 992
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.15,
|
| 199 |
+
"learning_rate": 9.829743427304402e-07,
|
| 200 |
+
"loss": 0.1529,
|
| 201 |
+
"step": 1024
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.16,
|
| 205 |
+
"learning_rate": 9.804402914159011e-07,
|
| 206 |
+
"loss": 0.1365,
|
| 207 |
+
"step": 1056
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.16,
|
| 211 |
+
"learning_rate": 9.77906240101362e-07,
|
| 212 |
+
"loss": 0.1922,
|
| 213 |
+
"step": 1088
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.17,
|
| 217 |
+
"learning_rate": 9.75372188786823e-07,
|
| 218 |
+
"loss": 0.133,
|
| 219 |
+
"step": 1120
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.17,
|
| 223 |
+
"learning_rate": 9.728381374722837e-07,
|
| 224 |
+
"loss": 0.1692,
|
| 225 |
+
"step": 1152
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.18,
|
| 229 |
+
"learning_rate": 9.703040861577447e-07,
|
| 230 |
+
"loss": 0.1022,
|
| 231 |
+
"step": 1184
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.18,
|
| 235 |
+
"learning_rate": 9.677700348432054e-07,
|
| 236 |
+
"loss": 0.2052,
|
| 237 |
+
"step": 1216
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.19,
|
| 241 |
+
"learning_rate": 9.652359835286664e-07,
|
| 242 |
+
"loss": 0.1546,
|
| 243 |
+
"step": 1248
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.19,
|
| 247 |
+
"learning_rate": 9.627019322141273e-07,
|
| 248 |
+
"loss": 0.149,
|
| 249 |
+
"step": 1280
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.2,
|
| 253 |
+
"learning_rate": 9.601678808995883e-07,
|
| 254 |
+
"loss": 0.1281,
|
| 255 |
+
"step": 1312
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.2,
|
| 259 |
+
"learning_rate": 9.57633829585049e-07,
|
| 260 |
+
"loss": 0.1437,
|
| 261 |
+
"step": 1344
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.2,
|
| 265 |
+
"learning_rate": 9.5509977827051e-07,
|
| 266 |
+
"loss": 0.2097,
|
| 267 |
+
"step": 1376
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.21,
|
| 271 |
+
"learning_rate": 9.525657269559708e-07,
|
| 272 |
+
"loss": 0.1308,
|
| 273 |
+
"step": 1408
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.21,
|
| 277 |
+
"learning_rate": 9.500316756414317e-07,
|
| 278 |
+
"loss": 0.1691,
|
| 279 |
+
"step": 1440
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.22,
|
| 283 |
+
"learning_rate": 9.474976243268927e-07,
|
| 284 |
+
"loss": 0.2319,
|
| 285 |
+
"step": 1472
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.22,
|
| 289 |
+
"learning_rate": 9.449635730123534e-07,
|
| 290 |
+
"loss": 0.2226,
|
| 291 |
+
"step": 1504
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.23,
|
| 295 |
+
"learning_rate": 9.424295216978143e-07,
|
| 296 |
+
"loss": 0.1789,
|
| 297 |
+
"step": 1536
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.23,
|
| 301 |
+
"learning_rate": 9.398954703832752e-07,
|
| 302 |
+
"loss": 0.1932,
|
| 303 |
+
"step": 1568
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.24,
|
| 307 |
+
"learning_rate": 9.373614190687361e-07,
|
| 308 |
+
"loss": 0.1718,
|
| 309 |
+
"step": 1600
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.24,
|
| 313 |
+
"learning_rate": 9.34827367754197e-07,
|
| 314 |
+
"loss": 0.156,
|
| 315 |
+
"step": 1632
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.25,
|
| 319 |
+
"learning_rate": 9.322933164396578e-07,
|
| 320 |
+
"loss": 0.1512,
|
| 321 |
+
"step": 1664
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.25,
|
| 325 |
+
"learning_rate": 9.297592651251187e-07,
|
| 326 |
+
"loss": 0.0968,
|
| 327 |
+
"step": 1696
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.26,
|
| 331 |
+
"learning_rate": 9.272252138105796e-07,
|
| 332 |
+
"loss": 0.0932,
|
| 333 |
+
"step": 1728
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.26,
|
| 337 |
+
"learning_rate": 9.246911624960405e-07,
|
| 338 |
+
"loss": 0.2464,
|
| 339 |
+
"step": 1760
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27,
|
| 343 |
+
"learning_rate": 9.221571111815014e-07,
|
| 344 |
+
"loss": 0.2036,
|
| 345 |
+
"step": 1792
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.27,
|
| 349 |
+
"learning_rate": 9.196230598669623e-07,
|
| 350 |
+
"loss": 0.1245,
|
| 351 |
+
"step": 1824
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.28,
|
| 355 |
+
"learning_rate": 9.170890085524232e-07,
|
| 356 |
+
"loss": 0.1097,
|
| 357 |
+
"step": 1856
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.28,
|
| 361 |
+
"learning_rate": 9.14554957237884e-07,
|
| 362 |
+
"loss": 0.1844,
|
| 363 |
+
"step": 1888
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.29,
|
| 367 |
+
"learning_rate": 9.120209059233449e-07,
|
| 368 |
+
"loss": 0.1114,
|
| 369 |
+
"step": 1920
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.29,
|
| 373 |
+
"learning_rate": 9.094868546088058e-07,
|
| 374 |
+
"loss": 0.1992,
|
| 375 |
+
"step": 1952
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.3,
|
| 379 |
+
"learning_rate": 9.069528032942667e-07,
|
| 380 |
+
"loss": 0.1721,
|
| 381 |
+
"step": 1984
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.3,
|
| 385 |
+
"learning_rate": 9.044187519797275e-07,
|
| 386 |
+
"loss": 0.1473,
|
| 387 |
+
"step": 2016
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.3,
|
| 391 |
+
"learning_rate": 9.018847006651884e-07,
|
| 392 |
+
"loss": 0.1865,
|
| 393 |
+
"step": 2048
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.31,
|
| 397 |
+
"learning_rate": 8.993506493506493e-07,
|
| 398 |
+
"loss": 0.1583,
|
| 399 |
+
"step": 2080
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.31,
|
| 403 |
+
"learning_rate": 8.968165980361102e-07,
|
| 404 |
+
"loss": 0.1866,
|
| 405 |
+
"step": 2112
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.32,
|
| 409 |
+
"learning_rate": 8.942825467215711e-07,
|
| 410 |
+
"loss": 0.1617,
|
| 411 |
+
"step": 2144
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.32,
|
| 415 |
+
"learning_rate": 8.917484954070319e-07,
|
| 416 |
+
"loss": 0.1189,
|
| 417 |
+
"step": 2176
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.33,
|
| 421 |
+
"learning_rate": 8.892144440924928e-07,
|
| 422 |
+
"loss": 0.148,
|
| 423 |
+
"step": 2208
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.33,
|
| 427 |
+
"learning_rate": 8.866803927779537e-07,
|
| 428 |
+
"loss": 0.131,
|
| 429 |
+
"step": 2240
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.34,
|
| 433 |
+
"learning_rate": 8.841463414634146e-07,
|
| 434 |
+
"loss": 0.2261,
|
| 435 |
+
"step": 2272
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.34,
|
| 439 |
+
"learning_rate": 8.816122901488755e-07,
|
| 440 |
+
"loss": 0.1742,
|
| 441 |
+
"step": 2304
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.35,
|
| 445 |
+
"learning_rate": 8.790782388343364e-07,
|
| 446 |
+
"loss": 0.164,
|
| 447 |
+
"step": 2336
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.35,
|
| 451 |
+
"learning_rate": 8.765441875197972e-07,
|
| 452 |
+
"loss": 0.1161,
|
| 453 |
+
"step": 2368
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.36,
|
| 457 |
+
"learning_rate": 8.74010136205258e-07,
|
| 458 |
+
"loss": 0.1636,
|
| 459 |
+
"step": 2400
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.36,
|
| 463 |
+
"learning_rate": 8.71476084890719e-07,
|
| 464 |
+
"loss": 0.2416,
|
| 465 |
+
"step": 2432
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.37,
|
| 469 |
+
"learning_rate": 8.689420335761799e-07,
|
| 470 |
+
"loss": 0.1632,
|
| 471 |
+
"step": 2464
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.37,
|
| 475 |
+
"learning_rate": 8.664079822616408e-07,
|
| 476 |
+
"loss": 0.1477,
|
| 477 |
+
"step": 2496
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.38,
|
| 481 |
+
"learning_rate": 8.638739309471016e-07,
|
| 482 |
+
"loss": 0.2083,
|
| 483 |
+
"step": 2528
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.38,
|
| 487 |
+
"learning_rate": 8.613398796325625e-07,
|
| 488 |
+
"loss": 0.1599,
|
| 489 |
+
"step": 2560
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.39,
|
| 493 |
+
"learning_rate": 8.588058283180234e-07,
|
| 494 |
+
"loss": 0.1817,
|
| 495 |
+
"step": 2592
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.39,
|
| 499 |
+
"learning_rate": 8.562717770034843e-07,
|
| 500 |
+
"loss": 0.1005,
|
| 501 |
+
"step": 2624
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.4,
|
| 505 |
+
"learning_rate": 8.537377256889452e-07,
|
| 506 |
+
"loss": 0.168,
|
| 507 |
+
"step": 2656
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.4,
|
| 511 |
+
"learning_rate": 8.51203674374406e-07,
|
| 512 |
+
"loss": 0.2418,
|
| 513 |
+
"step": 2688
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4,
|
| 517 |
+
"learning_rate": 8.486696230598669e-07,
|
| 518 |
+
"loss": 0.1881,
|
| 519 |
+
"step": 2720
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.41,
|
| 523 |
+
"learning_rate": 8.461355717453278e-07,
|
| 524 |
+
"loss": 0.1829,
|
| 525 |
+
"step": 2752
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.41,
|
| 529 |
+
"learning_rate": 8.436015204307887e-07,
|
| 530 |
+
"loss": 0.1073,
|
| 531 |
+
"step": 2784
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.42,
|
| 535 |
+
"learning_rate": 8.410674691162496e-07,
|
| 536 |
+
"loss": 0.1324,
|
| 537 |
+
"step": 2816
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.42,
|
| 541 |
+
"learning_rate": 8.385334178017105e-07,
|
| 542 |
+
"loss": 0.2077,
|
| 543 |
+
"step": 2848
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.43,
|
| 547 |
+
"learning_rate": 8.359993664871713e-07,
|
| 548 |
+
"loss": 0.2248,
|
| 549 |
+
"step": 2880
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.43,
|
| 553 |
+
"learning_rate": 8.334653151726322e-07,
|
| 554 |
+
"loss": 0.1337,
|
| 555 |
+
"step": 2912
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.44,
|
| 559 |
+
"learning_rate": 8.30931263858093e-07,
|
| 560 |
+
"loss": 0.1906,
|
| 561 |
+
"step": 2944
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.44,
|
| 565 |
+
"learning_rate": 8.28397212543554e-07,
|
| 566 |
+
"loss": 0.1893,
|
| 567 |
+
"step": 2976
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.45,
|
| 571 |
+
"learning_rate": 8.259423503325942e-07,
|
| 572 |
+
"loss": 0.2029,
|
| 573 |
+
"step": 3008
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.45,
|
| 577 |
+
"learning_rate": 8.234082990180551e-07,
|
| 578 |
+
"loss": 0.157,
|
| 579 |
+
"step": 3040
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.46,
|
| 583 |
+
"learning_rate": 8.208742477035159e-07,
|
| 584 |
+
"loss": 0.1433,
|
| 585 |
+
"step": 3072
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.46,
|
| 589 |
+
"learning_rate": 8.183401963889769e-07,
|
| 590 |
+
"loss": 0.1689,
|
| 591 |
+
"step": 3104
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.47,
|
| 595 |
+
"learning_rate": 8.158061450744377e-07,
|
| 596 |
+
"loss": 0.2012,
|
| 597 |
+
"step": 3136
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.47,
|
| 601 |
+
"learning_rate": 8.132720937598986e-07,
|
| 602 |
+
"loss": 0.175,
|
| 603 |
+
"step": 3168
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.48,
|
| 607 |
+
"learning_rate": 8.107380424453595e-07,
|
| 608 |
+
"loss": 0.1961,
|
| 609 |
+
"step": 3200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.48,
|
| 613 |
+
"learning_rate": 8.082039911308203e-07,
|
| 614 |
+
"loss": 0.2547,
|
| 615 |
+
"step": 3232
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.49,
|
| 619 |
+
"learning_rate": 8.056699398162813e-07,
|
| 620 |
+
"loss": 0.1935,
|
| 621 |
+
"step": 3264
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.49,
|
| 625 |
+
"learning_rate": 8.031358885017421e-07,
|
| 626 |
+
"loss": 0.2149,
|
| 627 |
+
"step": 3296
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.5,
|
| 631 |
+
"learning_rate": 8.00601837187203e-07,
|
| 632 |
+
"loss": 0.1809,
|
| 633 |
+
"step": 3328
|
| 634 |
+
}
|
| 635 |
+
],
|
| 636 |
+
"logging_steps": 32,
|
| 637 |
+
"max_steps": 13434,
|
| 638 |
+
"num_input_tokens_seen": 0,
|
| 639 |
+
"num_train_epochs": 2,
|
| 640 |
+
"save_steps": 3358,
|
| 641 |
+
"total_flos": 1.42592165756928e+19,
|
| 642 |
+
"train_batch_size": 2,
|
| 643 |
+
"trial_name": null,
|
| 644 |
+
"trial_params": null
|
| 645 |
+
}
|
checkpoint-10075/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e4db6f6fcaf2005bb3f583cfbd6e7afdd0dfdcdb7d4db7e107d5c22865fcf47
|
| 3 |
+
size 5048
|
checkpoint-13433/config.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "models/openai/whisper-large-v2/finetune/jacob_filter/checkpoint-6717",
|
| 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-13433/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-13433/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee34cf01664c4d09992b7898eb4d3550b0cf01fcc207114c2c92d32c1cfb3be5
|
| 3 |
+
size 4992706480
|
checkpoint-13433/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:959fce97317119d41ad9ef50f8a03f01d8c15f04bc22a6389145eb90e67c0253
|
| 3 |
+
size 1180663192
|
checkpoint-13433/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-13433/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9acbb2427a1e0414f01f93b19473b6320113c10042867d2a0bc0334e09e6cb73
|
| 3 |
+
size 3095074288
|
checkpoint-13433/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-13433/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e1cbb1a645a82041618b79e1039e0d9f582f753f274baed7d07ea4e6cedee30
|
| 3 |
+
size 14244
|
checkpoint-13433/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bb000772249d98077622e512b0edbd5ca3709d74fd76af0bc1b708185f853f8f
|
| 3 |
+
size 1064
|
checkpoint-13433/trainer_state.json
ADDED
|
@@ -0,0 +1,1275 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9998511240136966,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 6716,
|
| 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": 3.722084367245657e-08,
|
| 14 |
+
"loss": 0.2836,
|
| 15 |
+
"step": 32
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 7.692307692307692e-08,
|
| 20 |
+
"loss": 0.2111,
|
| 21 |
+
"step": 64
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 1.1662531017369727e-07,
|
| 26 |
+
"loss": 0.2145,
|
| 27 |
+
"step": 96
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.02,
|
| 31 |
+
"learning_rate": 1.563275434243176e-07,
|
| 32 |
+
"loss": 0.2375,
|
| 33 |
+
"step": 128
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 1.9602977667493795e-07,
|
| 38 |
+
"loss": 0.1839,
|
| 39 |
+
"step": 160
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03,
|
| 43 |
+
"learning_rate": 2.3573200992555832e-07,
|
| 44 |
+
"loss": 0.2864,
|
| 45 |
+
"step": 192
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03,
|
| 49 |
+
"learning_rate": 2.7543424317617863e-07,
|
| 50 |
+
"loss": 0.1799,
|
| 51 |
+
"step": 224
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.04,
|
| 55 |
+
"learning_rate": 3.1513647642679897e-07,
|
| 56 |
+
"loss": 0.1609,
|
| 57 |
+
"step": 256
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.04,
|
| 61 |
+
"learning_rate": 3.5483870967741936e-07,
|
| 62 |
+
"loss": 0.1434,
|
| 63 |
+
"step": 288
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.05,
|
| 67 |
+
"learning_rate": 3.945409429280397e-07,
|
| 68 |
+
"loss": 0.1699,
|
| 69 |
+
"step": 320
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.05,
|
| 73 |
+
"learning_rate": 4.3424317617866004e-07,
|
| 74 |
+
"loss": 0.1898,
|
| 75 |
+
"step": 352
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.06,
|
| 79 |
+
"learning_rate": 4.739454094292804e-07,
|
| 80 |
+
"loss": 0.1665,
|
| 81 |
+
"step": 384
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.06,
|
| 85 |
+
"learning_rate": 5.136476426799007e-07,
|
| 86 |
+
"loss": 0.1358,
|
| 87 |
+
"step": 416
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.07,
|
| 91 |
+
"learning_rate": 5.533498759305211e-07,
|
| 92 |
+
"loss": 0.1713,
|
| 93 |
+
"step": 448
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07,
|
| 97 |
+
"learning_rate": 5.930521091811415e-07,
|
| 98 |
+
"loss": 0.1817,
|
| 99 |
+
"step": 480
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.08,
|
| 103 |
+
"learning_rate": 6.327543424317618e-07,
|
| 104 |
+
"loss": 0.149,
|
| 105 |
+
"step": 512
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.08,
|
| 109 |
+
"learning_rate": 6.724565756823821e-07,
|
| 110 |
+
"loss": 0.165,
|
| 111 |
+
"step": 544
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.09,
|
| 115 |
+
"learning_rate": 7.121588089330024e-07,
|
| 116 |
+
"loss": 0.2282,
|
| 117 |
+
"step": 576
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.09,
|
| 121 |
+
"learning_rate": 7.518610421836227e-07,
|
| 122 |
+
"loss": 0.1666,
|
| 123 |
+
"step": 608
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.1,
|
| 127 |
+
"learning_rate": 7.915632754342431e-07,
|
| 128 |
+
"loss": 0.1229,
|
| 129 |
+
"step": 640
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1,
|
| 133 |
+
"learning_rate": 8.312655086848634e-07,
|
| 134 |
+
"loss": 0.2118,
|
| 135 |
+
"step": 672
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1,
|
| 139 |
+
"learning_rate": 8.709677419354838e-07,
|
| 140 |
+
"loss": 0.1869,
|
| 141 |
+
"step": 704
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.11,
|
| 145 |
+
"learning_rate": 9.106699751861042e-07,
|
| 146 |
+
"loss": 0.2196,
|
| 147 |
+
"step": 736
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.11,
|
| 151 |
+
"learning_rate": 9.503722084367245e-07,
|
| 152 |
+
"loss": 0.0936,
|
| 153 |
+
"step": 768
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.12,
|
| 157 |
+
"learning_rate": 9.90074441687345e-07,
|
| 158 |
+
"loss": 0.1839,
|
| 159 |
+
"step": 800
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.12,
|
| 163 |
+
"learning_rate": 9.980994615140957e-07,
|
| 164 |
+
"loss": 0.1675,
|
| 165 |
+
"step": 832
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.13,
|
| 169 |
+
"learning_rate": 9.955654101995564e-07,
|
| 170 |
+
"loss": 0.1595,
|
| 171 |
+
"step": 864
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13,
|
| 175 |
+
"learning_rate": 9.930313588850174e-07,
|
| 176 |
+
"loss": 0.1556,
|
| 177 |
+
"step": 896
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.14,
|
| 181 |
+
"learning_rate": 9.905764966740576e-07,
|
| 182 |
+
"loss": 0.1711,
|
| 183 |
+
"step": 928
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.14,
|
| 187 |
+
"learning_rate": 9.880424453595185e-07,
|
| 188 |
+
"loss": 0.1777,
|
| 189 |
+
"step": 960
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.15,
|
| 193 |
+
"learning_rate": 9.855083940449792e-07,
|
| 194 |
+
"loss": 0.2031,
|
| 195 |
+
"step": 992
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.15,
|
| 199 |
+
"learning_rate": 9.829743427304402e-07,
|
| 200 |
+
"loss": 0.1529,
|
| 201 |
+
"step": 1024
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.16,
|
| 205 |
+
"learning_rate": 9.804402914159011e-07,
|
| 206 |
+
"loss": 0.1365,
|
| 207 |
+
"step": 1056
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.16,
|
| 211 |
+
"learning_rate": 9.77906240101362e-07,
|
| 212 |
+
"loss": 0.1922,
|
| 213 |
+
"step": 1088
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.17,
|
| 217 |
+
"learning_rate": 9.75372188786823e-07,
|
| 218 |
+
"loss": 0.133,
|
| 219 |
+
"step": 1120
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.17,
|
| 223 |
+
"learning_rate": 9.728381374722837e-07,
|
| 224 |
+
"loss": 0.1692,
|
| 225 |
+
"step": 1152
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.18,
|
| 229 |
+
"learning_rate": 9.703040861577447e-07,
|
| 230 |
+
"loss": 0.1022,
|
| 231 |
+
"step": 1184
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.18,
|
| 235 |
+
"learning_rate": 9.677700348432054e-07,
|
| 236 |
+
"loss": 0.2052,
|
| 237 |
+
"step": 1216
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.19,
|
| 241 |
+
"learning_rate": 9.652359835286664e-07,
|
| 242 |
+
"loss": 0.1546,
|
| 243 |
+
"step": 1248
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.19,
|
| 247 |
+
"learning_rate": 9.627019322141273e-07,
|
| 248 |
+
"loss": 0.149,
|
| 249 |
+
"step": 1280
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.2,
|
| 253 |
+
"learning_rate": 9.601678808995883e-07,
|
| 254 |
+
"loss": 0.1281,
|
| 255 |
+
"step": 1312
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.2,
|
| 259 |
+
"learning_rate": 9.57633829585049e-07,
|
| 260 |
+
"loss": 0.1437,
|
| 261 |
+
"step": 1344
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.2,
|
| 265 |
+
"learning_rate": 9.5509977827051e-07,
|
| 266 |
+
"loss": 0.2097,
|
| 267 |
+
"step": 1376
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.21,
|
| 271 |
+
"learning_rate": 9.525657269559708e-07,
|
| 272 |
+
"loss": 0.1308,
|
| 273 |
+
"step": 1408
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.21,
|
| 277 |
+
"learning_rate": 9.500316756414317e-07,
|
| 278 |
+
"loss": 0.1691,
|
| 279 |
+
"step": 1440
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.22,
|
| 283 |
+
"learning_rate": 9.474976243268927e-07,
|
| 284 |
+
"loss": 0.2319,
|
| 285 |
+
"step": 1472
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.22,
|
| 289 |
+
"learning_rate": 9.449635730123534e-07,
|
| 290 |
+
"loss": 0.2226,
|
| 291 |
+
"step": 1504
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.23,
|
| 295 |
+
"learning_rate": 9.424295216978143e-07,
|
| 296 |
+
"loss": 0.1789,
|
| 297 |
+
"step": 1536
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.23,
|
| 301 |
+
"learning_rate": 9.398954703832752e-07,
|
| 302 |
+
"loss": 0.1932,
|
| 303 |
+
"step": 1568
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.24,
|
| 307 |
+
"learning_rate": 9.373614190687361e-07,
|
| 308 |
+
"loss": 0.1718,
|
| 309 |
+
"step": 1600
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.24,
|
| 313 |
+
"learning_rate": 9.34827367754197e-07,
|
| 314 |
+
"loss": 0.156,
|
| 315 |
+
"step": 1632
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.25,
|
| 319 |
+
"learning_rate": 9.322933164396578e-07,
|
| 320 |
+
"loss": 0.1512,
|
| 321 |
+
"step": 1664
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.25,
|
| 325 |
+
"learning_rate": 9.297592651251187e-07,
|
| 326 |
+
"loss": 0.0968,
|
| 327 |
+
"step": 1696
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.26,
|
| 331 |
+
"learning_rate": 9.272252138105796e-07,
|
| 332 |
+
"loss": 0.0932,
|
| 333 |
+
"step": 1728
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.26,
|
| 337 |
+
"learning_rate": 9.246911624960405e-07,
|
| 338 |
+
"loss": 0.2464,
|
| 339 |
+
"step": 1760
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27,
|
| 343 |
+
"learning_rate": 9.221571111815014e-07,
|
| 344 |
+
"loss": 0.2036,
|
| 345 |
+
"step": 1792
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.27,
|
| 349 |
+
"learning_rate": 9.196230598669623e-07,
|
| 350 |
+
"loss": 0.1245,
|
| 351 |
+
"step": 1824
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.28,
|
| 355 |
+
"learning_rate": 9.170890085524232e-07,
|
| 356 |
+
"loss": 0.1097,
|
| 357 |
+
"step": 1856
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.28,
|
| 361 |
+
"learning_rate": 9.14554957237884e-07,
|
| 362 |
+
"loss": 0.1844,
|
| 363 |
+
"step": 1888
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.29,
|
| 367 |
+
"learning_rate": 9.120209059233449e-07,
|
| 368 |
+
"loss": 0.1114,
|
| 369 |
+
"step": 1920
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.29,
|
| 373 |
+
"learning_rate": 9.094868546088058e-07,
|
| 374 |
+
"loss": 0.1992,
|
| 375 |
+
"step": 1952
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.3,
|
| 379 |
+
"learning_rate": 9.069528032942667e-07,
|
| 380 |
+
"loss": 0.1721,
|
| 381 |
+
"step": 1984
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.3,
|
| 385 |
+
"learning_rate": 9.044187519797275e-07,
|
| 386 |
+
"loss": 0.1473,
|
| 387 |
+
"step": 2016
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.3,
|
| 391 |
+
"learning_rate": 9.018847006651884e-07,
|
| 392 |
+
"loss": 0.1865,
|
| 393 |
+
"step": 2048
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.31,
|
| 397 |
+
"learning_rate": 8.993506493506493e-07,
|
| 398 |
+
"loss": 0.1583,
|
| 399 |
+
"step": 2080
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.31,
|
| 403 |
+
"learning_rate": 8.968165980361102e-07,
|
| 404 |
+
"loss": 0.1866,
|
| 405 |
+
"step": 2112
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.32,
|
| 409 |
+
"learning_rate": 8.942825467215711e-07,
|
| 410 |
+
"loss": 0.1617,
|
| 411 |
+
"step": 2144
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.32,
|
| 415 |
+
"learning_rate": 8.917484954070319e-07,
|
| 416 |
+
"loss": 0.1189,
|
| 417 |
+
"step": 2176
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.33,
|
| 421 |
+
"learning_rate": 8.892144440924928e-07,
|
| 422 |
+
"loss": 0.148,
|
| 423 |
+
"step": 2208
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.33,
|
| 427 |
+
"learning_rate": 8.866803927779537e-07,
|
| 428 |
+
"loss": 0.131,
|
| 429 |
+
"step": 2240
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.34,
|
| 433 |
+
"learning_rate": 8.841463414634146e-07,
|
| 434 |
+
"loss": 0.2261,
|
| 435 |
+
"step": 2272
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.34,
|
| 439 |
+
"learning_rate": 8.816122901488755e-07,
|
| 440 |
+
"loss": 0.1742,
|
| 441 |
+
"step": 2304
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.35,
|
| 445 |
+
"learning_rate": 8.790782388343364e-07,
|
| 446 |
+
"loss": 0.164,
|
| 447 |
+
"step": 2336
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.35,
|
| 451 |
+
"learning_rate": 8.765441875197972e-07,
|
| 452 |
+
"loss": 0.1161,
|
| 453 |
+
"step": 2368
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.36,
|
| 457 |
+
"learning_rate": 8.74010136205258e-07,
|
| 458 |
+
"loss": 0.1636,
|
| 459 |
+
"step": 2400
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.36,
|
| 463 |
+
"learning_rate": 8.71476084890719e-07,
|
| 464 |
+
"loss": 0.2416,
|
| 465 |
+
"step": 2432
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.37,
|
| 469 |
+
"learning_rate": 8.689420335761799e-07,
|
| 470 |
+
"loss": 0.1632,
|
| 471 |
+
"step": 2464
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.37,
|
| 475 |
+
"learning_rate": 8.664079822616408e-07,
|
| 476 |
+
"loss": 0.1477,
|
| 477 |
+
"step": 2496
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.38,
|
| 481 |
+
"learning_rate": 8.638739309471016e-07,
|
| 482 |
+
"loss": 0.2083,
|
| 483 |
+
"step": 2528
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.38,
|
| 487 |
+
"learning_rate": 8.613398796325625e-07,
|
| 488 |
+
"loss": 0.1599,
|
| 489 |
+
"step": 2560
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.39,
|
| 493 |
+
"learning_rate": 8.588058283180234e-07,
|
| 494 |
+
"loss": 0.1817,
|
| 495 |
+
"step": 2592
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.39,
|
| 499 |
+
"learning_rate": 8.562717770034843e-07,
|
| 500 |
+
"loss": 0.1005,
|
| 501 |
+
"step": 2624
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.4,
|
| 505 |
+
"learning_rate": 8.537377256889452e-07,
|
| 506 |
+
"loss": 0.168,
|
| 507 |
+
"step": 2656
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.4,
|
| 511 |
+
"learning_rate": 8.51203674374406e-07,
|
| 512 |
+
"loss": 0.2418,
|
| 513 |
+
"step": 2688
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4,
|
| 517 |
+
"learning_rate": 8.486696230598669e-07,
|
| 518 |
+
"loss": 0.1881,
|
| 519 |
+
"step": 2720
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.41,
|
| 523 |
+
"learning_rate": 8.461355717453278e-07,
|
| 524 |
+
"loss": 0.1829,
|
| 525 |
+
"step": 2752
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.41,
|
| 529 |
+
"learning_rate": 8.436015204307887e-07,
|
| 530 |
+
"loss": 0.1073,
|
| 531 |
+
"step": 2784
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.42,
|
| 535 |
+
"learning_rate": 8.410674691162496e-07,
|
| 536 |
+
"loss": 0.1324,
|
| 537 |
+
"step": 2816
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.42,
|
| 541 |
+
"learning_rate": 8.385334178017105e-07,
|
| 542 |
+
"loss": 0.2077,
|
| 543 |
+
"step": 2848
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.43,
|
| 547 |
+
"learning_rate": 8.359993664871713e-07,
|
| 548 |
+
"loss": 0.2248,
|
| 549 |
+
"step": 2880
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.43,
|
| 553 |
+
"learning_rate": 8.334653151726322e-07,
|
| 554 |
+
"loss": 0.1337,
|
| 555 |
+
"step": 2912
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.44,
|
| 559 |
+
"learning_rate": 8.30931263858093e-07,
|
| 560 |
+
"loss": 0.1906,
|
| 561 |
+
"step": 2944
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.44,
|
| 565 |
+
"learning_rate": 8.28397212543554e-07,
|
| 566 |
+
"loss": 0.1893,
|
| 567 |
+
"step": 2976
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.45,
|
| 571 |
+
"learning_rate": 8.259423503325942e-07,
|
| 572 |
+
"loss": 0.2029,
|
| 573 |
+
"step": 3008
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.45,
|
| 577 |
+
"learning_rate": 8.234082990180551e-07,
|
| 578 |
+
"loss": 0.157,
|
| 579 |
+
"step": 3040
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.46,
|
| 583 |
+
"learning_rate": 8.208742477035159e-07,
|
| 584 |
+
"loss": 0.1433,
|
| 585 |
+
"step": 3072
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.46,
|
| 589 |
+
"learning_rate": 8.183401963889769e-07,
|
| 590 |
+
"loss": 0.1689,
|
| 591 |
+
"step": 3104
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.47,
|
| 595 |
+
"learning_rate": 8.158061450744377e-07,
|
| 596 |
+
"loss": 0.2012,
|
| 597 |
+
"step": 3136
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.47,
|
| 601 |
+
"learning_rate": 8.132720937598986e-07,
|
| 602 |
+
"loss": 0.175,
|
| 603 |
+
"step": 3168
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.48,
|
| 607 |
+
"learning_rate": 8.107380424453595e-07,
|
| 608 |
+
"loss": 0.1961,
|
| 609 |
+
"step": 3200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.48,
|
| 613 |
+
"learning_rate": 8.082039911308203e-07,
|
| 614 |
+
"loss": 0.2547,
|
| 615 |
+
"step": 3232
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.49,
|
| 619 |
+
"learning_rate": 8.056699398162813e-07,
|
| 620 |
+
"loss": 0.1935,
|
| 621 |
+
"step": 3264
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.49,
|
| 625 |
+
"learning_rate": 8.031358885017421e-07,
|
| 626 |
+
"loss": 0.2149,
|
| 627 |
+
"step": 3296
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.5,
|
| 631 |
+
"learning_rate": 8.00601837187203e-07,
|
| 632 |
+
"loss": 0.1809,
|
| 633 |
+
"step": 3328
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5,
|
| 637 |
+
"learning_rate": 7.980677858726639e-07,
|
| 638 |
+
"loss": 0.2072,
|
| 639 |
+
"step": 3360
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5,
|
| 643 |
+
"learning_rate": 7.955337345581247e-07,
|
| 644 |
+
"loss": 0.2116,
|
| 645 |
+
"step": 3392
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.51,
|
| 649 |
+
"learning_rate": 7.929996832435857e-07,
|
| 650 |
+
"loss": 0.1737,
|
| 651 |
+
"step": 3424
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.51,
|
| 655 |
+
"learning_rate": 7.904656319290464e-07,
|
| 656 |
+
"loss": 0.2219,
|
| 657 |
+
"step": 3456
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.52,
|
| 661 |
+
"learning_rate": 7.879315806145074e-07,
|
| 662 |
+
"loss": 0.1849,
|
| 663 |
+
"step": 3488
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.52,
|
| 667 |
+
"learning_rate": 7.853975292999683e-07,
|
| 668 |
+
"loss": 0.1884,
|
| 669 |
+
"step": 3520
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.53,
|
| 673 |
+
"learning_rate": 7.828634779854292e-07,
|
| 674 |
+
"loss": 0.2192,
|
| 675 |
+
"step": 3552
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.53,
|
| 679 |
+
"learning_rate": 7.803294266708901e-07,
|
| 680 |
+
"loss": 0.1958,
|
| 681 |
+
"step": 3584
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.54,
|
| 685 |
+
"learning_rate": 7.777953753563509e-07,
|
| 686 |
+
"loss": 0.1433,
|
| 687 |
+
"step": 3616
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.54,
|
| 691 |
+
"learning_rate": 7.752613240418118e-07,
|
| 692 |
+
"loss": 0.2151,
|
| 693 |
+
"step": 3648
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.55,
|
| 697 |
+
"learning_rate": 7.727272727272727e-07,
|
| 698 |
+
"loss": 0.1675,
|
| 699 |
+
"step": 3680
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.55,
|
| 703 |
+
"learning_rate": 7.701932214127336e-07,
|
| 704 |
+
"loss": 0.1586,
|
| 705 |
+
"step": 3712
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.56,
|
| 709 |
+
"learning_rate": 7.676591700981945e-07,
|
| 710 |
+
"loss": 0.2881,
|
| 711 |
+
"step": 3744
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.56,
|
| 715 |
+
"learning_rate": 7.651251187836553e-07,
|
| 716 |
+
"loss": 0.196,
|
| 717 |
+
"step": 3776
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.57,
|
| 721 |
+
"learning_rate": 7.625910674691162e-07,
|
| 722 |
+
"loss": 0.1285,
|
| 723 |
+
"step": 3808
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.57,
|
| 727 |
+
"learning_rate": 7.60057016154577e-07,
|
| 728 |
+
"loss": 0.2262,
|
| 729 |
+
"step": 3840
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.58,
|
| 733 |
+
"learning_rate": 7.57522964840038e-07,
|
| 734 |
+
"loss": 0.2309,
|
| 735 |
+
"step": 3872
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.58,
|
| 739 |
+
"learning_rate": 7.549889135254989e-07,
|
| 740 |
+
"loss": 0.1533,
|
| 741 |
+
"step": 3904
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.59,
|
| 745 |
+
"learning_rate": 7.524548622109597e-07,
|
| 746 |
+
"loss": 0.1297,
|
| 747 |
+
"step": 3936
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.59,
|
| 751 |
+
"learning_rate": 7.499208108964206e-07,
|
| 752 |
+
"loss": 0.1808,
|
| 753 |
+
"step": 3968
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.6,
|
| 757 |
+
"learning_rate": 7.473867595818814e-07,
|
| 758 |
+
"loss": 0.2401,
|
| 759 |
+
"step": 4000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6,
|
| 763 |
+
"learning_rate": 7.448527082673424e-07,
|
| 764 |
+
"loss": 0.2507,
|
| 765 |
+
"step": 4032
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.61,
|
| 769 |
+
"learning_rate": 7.423186569528033e-07,
|
| 770 |
+
"loss": 0.1562,
|
| 771 |
+
"step": 4064
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 0.61,
|
| 775 |
+
"learning_rate": 7.397846056382642e-07,
|
| 776 |
+
"loss": 0.1912,
|
| 777 |
+
"step": 4096
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.61,
|
| 781 |
+
"learning_rate": 7.373297434273043e-07,
|
| 782 |
+
"loss": 0.1703,
|
| 783 |
+
"step": 4128
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.62,
|
| 787 |
+
"learning_rate": 7.347956921127653e-07,
|
| 788 |
+
"loss": 0.1471,
|
| 789 |
+
"step": 4160
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.62,
|
| 793 |
+
"learning_rate": 7.322616407982262e-07,
|
| 794 |
+
"loss": 0.1539,
|
| 795 |
+
"step": 4192
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.63,
|
| 799 |
+
"learning_rate": 7.297275894836869e-07,
|
| 800 |
+
"loss": 0.1521,
|
| 801 |
+
"step": 4224
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.63,
|
| 805 |
+
"learning_rate": 7.271935381691479e-07,
|
| 806 |
+
"loss": 0.2623,
|
| 807 |
+
"step": 4256
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.64,
|
| 811 |
+
"learning_rate": 7.246594868546087e-07,
|
| 812 |
+
"loss": 0.1753,
|
| 813 |
+
"step": 4288
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.64,
|
| 817 |
+
"learning_rate": 7.221254355400697e-07,
|
| 818 |
+
"loss": 0.1945,
|
| 819 |
+
"step": 4320
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.65,
|
| 823 |
+
"learning_rate": 7.195913842255306e-07,
|
| 824 |
+
"loss": 0.2153,
|
| 825 |
+
"step": 4352
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.65,
|
| 829 |
+
"learning_rate": 7.170573329109915e-07,
|
| 830 |
+
"loss": 0.2841,
|
| 831 |
+
"step": 4384
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.66,
|
| 835 |
+
"learning_rate": 7.145232815964523e-07,
|
| 836 |
+
"loss": 0.1759,
|
| 837 |
+
"step": 4416
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.66,
|
| 841 |
+
"learning_rate": 7.119892302819131e-07,
|
| 842 |
+
"loss": 0.2214,
|
| 843 |
+
"step": 4448
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.67,
|
| 847 |
+
"learning_rate": 7.094551789673741e-07,
|
| 848 |
+
"loss": 0.188,
|
| 849 |
+
"step": 4480
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.67,
|
| 853 |
+
"learning_rate": 7.069211276528349e-07,
|
| 854 |
+
"loss": 0.1579,
|
| 855 |
+
"step": 4512
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.68,
|
| 859 |
+
"learning_rate": 7.043870763382959e-07,
|
| 860 |
+
"loss": 0.2213,
|
| 861 |
+
"step": 4544
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.68,
|
| 865 |
+
"learning_rate": 7.018530250237567e-07,
|
| 866 |
+
"loss": 0.2042,
|
| 867 |
+
"step": 4576
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.69,
|
| 871 |
+
"learning_rate": 6.993189737092175e-07,
|
| 872 |
+
"loss": 0.1852,
|
| 873 |
+
"step": 4608
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.69,
|
| 877 |
+
"learning_rate": 6.967849223946785e-07,
|
| 878 |
+
"loss": 0.1716,
|
| 879 |
+
"step": 4640
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.7,
|
| 883 |
+
"learning_rate": 6.942508710801393e-07,
|
| 884 |
+
"loss": 0.1645,
|
| 885 |
+
"step": 4672
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.7,
|
| 889 |
+
"learning_rate": 6.917168197656003e-07,
|
| 890 |
+
"loss": 0.1986,
|
| 891 |
+
"step": 4704
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.71,
|
| 895 |
+
"learning_rate": 6.89182768451061e-07,
|
| 896 |
+
"loss": 0.2531,
|
| 897 |
+
"step": 4736
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.71,
|
| 901 |
+
"learning_rate": 6.866487171365219e-07,
|
| 902 |
+
"loss": 0.1792,
|
| 903 |
+
"step": 4768
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.71,
|
| 907 |
+
"learning_rate": 6.841146658219829e-07,
|
| 908 |
+
"loss": 0.1843,
|
| 909 |
+
"step": 4800
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.72,
|
| 913 |
+
"learning_rate": 6.815806145074437e-07,
|
| 914 |
+
"loss": 0.2175,
|
| 915 |
+
"step": 4832
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.72,
|
| 919 |
+
"learning_rate": 6.790465631929047e-07,
|
| 920 |
+
"loss": 0.2083,
|
| 921 |
+
"step": 4864
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.73,
|
| 925 |
+
"learning_rate": 6.765125118783655e-07,
|
| 926 |
+
"loss": 0.1729,
|
| 927 |
+
"step": 4896
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.73,
|
| 931 |
+
"learning_rate": 6.739784605638263e-07,
|
| 932 |
+
"loss": 0.1849,
|
| 933 |
+
"step": 4928
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.74,
|
| 937 |
+
"learning_rate": 6.714444092492873e-07,
|
| 938 |
+
"loss": 0.2374,
|
| 939 |
+
"step": 4960
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.74,
|
| 943 |
+
"learning_rate": 6.689103579347481e-07,
|
| 944 |
+
"loss": 0.241,
|
| 945 |
+
"step": 4992
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.75,
|
| 949 |
+
"learning_rate": 6.663763066202091e-07,
|
| 950 |
+
"loss": 0.1853,
|
| 951 |
+
"step": 5024
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 0.75,
|
| 955 |
+
"learning_rate": 6.638422553056699e-07,
|
| 956 |
+
"loss": 0.1957,
|
| 957 |
+
"step": 5056
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 0.76,
|
| 961 |
+
"learning_rate": 6.613082039911308e-07,
|
| 962 |
+
"loss": 0.2052,
|
| 963 |
+
"step": 5088
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.76,
|
| 967 |
+
"learning_rate": 6.587741526765917e-07,
|
| 968 |
+
"loss": 0.2321,
|
| 969 |
+
"step": 5120
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.77,
|
| 973 |
+
"learning_rate": 6.562401013620525e-07,
|
| 974 |
+
"loss": 0.1804,
|
| 975 |
+
"step": 5152
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.77,
|
| 979 |
+
"learning_rate": 6.537060500475135e-07,
|
| 980 |
+
"loss": 0.1842,
|
| 981 |
+
"step": 5184
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.78,
|
| 985 |
+
"learning_rate": 6.511719987329743e-07,
|
| 986 |
+
"loss": 0.2388,
|
| 987 |
+
"step": 5216
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.78,
|
| 991 |
+
"learning_rate": 6.486379474184352e-07,
|
| 992 |
+
"loss": 0.2417,
|
| 993 |
+
"step": 5248
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.79,
|
| 997 |
+
"learning_rate": 6.46103896103896e-07,
|
| 998 |
+
"loss": 0.2224,
|
| 999 |
+
"step": 5280
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.79,
|
| 1003 |
+
"learning_rate": 6.435698447893569e-07,
|
| 1004 |
+
"loss": 0.2029,
|
| 1005 |
+
"step": 5312
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 0.8,
|
| 1009 |
+
"learning_rate": 6.410357934748179e-07,
|
| 1010 |
+
"loss": 0.2807,
|
| 1011 |
+
"step": 5344
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8,
|
| 1015 |
+
"learning_rate": 6.385017421602787e-07,
|
| 1016 |
+
"loss": 0.192,
|
| 1017 |
+
"step": 5376
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.81,
|
| 1021 |
+
"learning_rate": 6.359676908457397e-07,
|
| 1022 |
+
"loss": 0.1848,
|
| 1023 |
+
"step": 5408
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.81,
|
| 1027 |
+
"learning_rate": 6.334336395312004e-07,
|
| 1028 |
+
"loss": 0.2143,
|
| 1029 |
+
"step": 5440
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.81,
|
| 1033 |
+
"learning_rate": 6.308995882166613e-07,
|
| 1034 |
+
"loss": 0.2421,
|
| 1035 |
+
"step": 5472
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.82,
|
| 1039 |
+
"learning_rate": 6.283655369021223e-07,
|
| 1040 |
+
"loss": 0.1724,
|
| 1041 |
+
"step": 5504
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.82,
|
| 1045 |
+
"learning_rate": 6.258314855875831e-07,
|
| 1046 |
+
"loss": 0.1207,
|
| 1047 |
+
"step": 5536
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 0.83,
|
| 1051 |
+
"learning_rate": 6.232974342730441e-07,
|
| 1052 |
+
"loss": 0.2259,
|
| 1053 |
+
"step": 5568
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.83,
|
| 1057 |
+
"learning_rate": 6.207633829585048e-07,
|
| 1058 |
+
"loss": 0.2504,
|
| 1059 |
+
"step": 5600
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.84,
|
| 1063 |
+
"learning_rate": 6.182293316439658e-07,
|
| 1064 |
+
"loss": 0.188,
|
| 1065 |
+
"step": 5632
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.84,
|
| 1069 |
+
"learning_rate": 6.156952803294266e-07,
|
| 1070 |
+
"loss": 0.1893,
|
| 1071 |
+
"step": 5664
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.85,
|
| 1075 |
+
"learning_rate": 6.131612290148875e-07,
|
| 1076 |
+
"loss": 0.1905,
|
| 1077 |
+
"step": 5696
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 0.85,
|
| 1081 |
+
"learning_rate": 6.106271777003485e-07,
|
| 1082 |
+
"loss": 0.2594,
|
| 1083 |
+
"step": 5728
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 0.86,
|
| 1087 |
+
"learning_rate": 6.080931263858092e-07,
|
| 1088 |
+
"loss": 0.3084,
|
| 1089 |
+
"step": 5760
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.86,
|
| 1093 |
+
"learning_rate": 6.055590750712702e-07,
|
| 1094 |
+
"loss": 0.1925,
|
| 1095 |
+
"step": 5792
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.87,
|
| 1099 |
+
"learning_rate": 6.03025023756731e-07,
|
| 1100 |
+
"loss": 0.186,
|
| 1101 |
+
"step": 5824
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.87,
|
| 1105 |
+
"learning_rate": 6.004909724421919e-07,
|
| 1106 |
+
"loss": 0.2302,
|
| 1107 |
+
"step": 5856
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.88,
|
| 1111 |
+
"learning_rate": 5.979569211276529e-07,
|
| 1112 |
+
"loss": 0.1371,
|
| 1113 |
+
"step": 5888
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.88,
|
| 1117 |
+
"learning_rate": 5.954228698131137e-07,
|
| 1118 |
+
"loss": 0.231,
|
| 1119 |
+
"step": 5920
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.89,
|
| 1123 |
+
"learning_rate": 5.928888184985746e-07,
|
| 1124 |
+
"loss": 0.2012,
|
| 1125 |
+
"step": 5952
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.89,
|
| 1129 |
+
"learning_rate": 5.903547671840354e-07,
|
| 1130 |
+
"loss": 0.2006,
|
| 1131 |
+
"step": 5984
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.9,
|
| 1135 |
+
"learning_rate": 5.878207158694963e-07,
|
| 1136 |
+
"loss": 0.215,
|
| 1137 |
+
"step": 6016
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9,
|
| 1141 |
+
"learning_rate": 5.852866645549572e-07,
|
| 1142 |
+
"loss": 0.1471,
|
| 1143 |
+
"step": 6048
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.91,
|
| 1147 |
+
"learning_rate": 5.827526132404181e-07,
|
| 1148 |
+
"loss": 0.2364,
|
| 1149 |
+
"step": 6080
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.91,
|
| 1153 |
+
"learning_rate": 5.80218561925879e-07,
|
| 1154 |
+
"loss": 0.2881,
|
| 1155 |
+
"step": 6112
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.91,
|
| 1159 |
+
"learning_rate": 5.776845106113398e-07,
|
| 1160 |
+
"loss": 0.1536,
|
| 1161 |
+
"step": 6144
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.92,
|
| 1165 |
+
"learning_rate": 5.751504592968008e-07,
|
| 1166 |
+
"loss": 0.2317,
|
| 1167 |
+
"step": 6176
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.92,
|
| 1171 |
+
"learning_rate": 5.726164079822616e-07,
|
| 1172 |
+
"loss": 0.1952,
|
| 1173 |
+
"step": 6208
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.93,
|
| 1177 |
+
"learning_rate": 5.700823566677225e-07,
|
| 1178 |
+
"loss": 0.1602,
|
| 1179 |
+
"step": 6240
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.93,
|
| 1183 |
+
"learning_rate": 5.675483053531834e-07,
|
| 1184 |
+
"loss": 0.212,
|
| 1185 |
+
"step": 6272
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.94,
|
| 1189 |
+
"learning_rate": 5.650142540386442e-07,
|
| 1190 |
+
"loss": 0.2401,
|
| 1191 |
+
"step": 6304
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.94,
|
| 1195 |
+
"learning_rate": 5.624802027241052e-07,
|
| 1196 |
+
"loss": 0.1992,
|
| 1197 |
+
"step": 6336
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.95,
|
| 1201 |
+
"learning_rate": 5.59946151409566e-07,
|
| 1202 |
+
"loss": 0.2616,
|
| 1203 |
+
"step": 6368
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.95,
|
| 1207 |
+
"learning_rate": 5.574121000950269e-07,
|
| 1208 |
+
"loss": 0.146,
|
| 1209 |
+
"step": 6400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.96,
|
| 1213 |
+
"learning_rate": 5.548780487804878e-07,
|
| 1214 |
+
"loss": 0.2081,
|
| 1215 |
+
"step": 6432
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.96,
|
| 1219 |
+
"learning_rate": 5.523439974659486e-07,
|
| 1220 |
+
"loss": 0.207,
|
| 1221 |
+
"step": 6464
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.97,
|
| 1225 |
+
"learning_rate": 5.498099461514096e-07,
|
| 1226 |
+
"loss": 0.2631,
|
| 1227 |
+
"step": 6496
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.97,
|
| 1231 |
+
"learning_rate": 5.472758948368704e-07,
|
| 1232 |
+
"loss": 0.1721,
|
| 1233 |
+
"step": 6528
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.98,
|
| 1237 |
+
"learning_rate": 5.447418435223313e-07,
|
| 1238 |
+
"loss": 0.1908,
|
| 1239 |
+
"step": 6560
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.98,
|
| 1243 |
+
"learning_rate": 5.422869813113715e-07,
|
| 1244 |
+
"loss": 0.2238,
|
| 1245 |
+
"step": 6592
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.99,
|
| 1249 |
+
"learning_rate": 5.397529299968325e-07,
|
| 1250 |
+
"loss": 0.2524,
|
| 1251 |
+
"step": 6624
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.99,
|
| 1255 |
+
"learning_rate": 5.372188786822933e-07,
|
| 1256 |
+
"loss": 0.1968,
|
| 1257 |
+
"step": 6656
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.0,
|
| 1261 |
+
"learning_rate": 5.346848273677542e-07,
|
| 1262 |
+
"loss": 0.2379,
|
| 1263 |
+
"step": 6688
|
| 1264 |
+
}
|
| 1265 |
+
],
|
| 1266 |
+
"logging_steps": 32,
|
| 1267 |
+
"max_steps": 13434,
|
| 1268 |
+
"num_input_tokens_seen": 0,
|
| 1269 |
+
"num_train_epochs": 2,
|
| 1270 |
+
"save_steps": 3358,
|
| 1271 |
+
"total_flos": 2.85184331513856e+19,
|
| 1272 |
+
"train_batch_size": 2,
|
| 1273 |
+
"trial_name": null,
|
| 1274 |
+
"trial_params": null
|
| 1275 |
+
}
|
checkpoint-13433/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e4db6f6fcaf2005bb3f583cfbd6e7afdd0dfdcdb7d4db7e107d5c22865fcf47
|
| 3 |
+
size 5048
|
checkpoint-16791/config.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "models/openai/whisper-large-v2/finetune/jacob_filter/checkpoint-6717",
|
| 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-16791/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-16791/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:095b5daf6da9dafe086366063c7aa0b1a9ed8311d1231882f77ca62bfd622ff7
|
| 3 |
+
size 4992706480
|
checkpoint-16791/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a5b9fb976d7d42bc07f0ba0dbacffaa0061ac0f0625724aba517b14008174711
|
| 3 |
+
size 1180663192
|
checkpoint-16791/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-16791/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2a436d2addaadda0fea3b2c8395715e532fa1206ea3a07af9201e12fb839ab6
|
| 3 |
+
size 3095074288
|
checkpoint-16791/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-16791/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a1d6627426ed3268f8693989ae5aaa270ce7dc271867c833e122fa4a892074de
|
| 3 |
+
size 14244
|
checkpoint-16791/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:86d89410b53de562dce1ae7961ba3c63eca690064d505613a551336065d4913e
|
| 3 |
+
size 1064
|
checkpoint-16791/trainer_state.json
ADDED
|
@@ -0,0 +1,1905 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.4997766860205448,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 10074,
|
| 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": 3.722084367245657e-08,
|
| 14 |
+
"loss": 0.2836,
|
| 15 |
+
"step": 32
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 7.692307692307692e-08,
|
| 20 |
+
"loss": 0.2111,
|
| 21 |
+
"step": 64
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 1.1662531017369727e-07,
|
| 26 |
+
"loss": 0.2145,
|
| 27 |
+
"step": 96
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.02,
|
| 31 |
+
"learning_rate": 1.563275434243176e-07,
|
| 32 |
+
"loss": 0.2375,
|
| 33 |
+
"step": 128
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 1.9602977667493795e-07,
|
| 38 |
+
"loss": 0.1839,
|
| 39 |
+
"step": 160
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03,
|
| 43 |
+
"learning_rate": 2.3573200992555832e-07,
|
| 44 |
+
"loss": 0.2864,
|
| 45 |
+
"step": 192
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03,
|
| 49 |
+
"learning_rate": 2.7543424317617863e-07,
|
| 50 |
+
"loss": 0.1799,
|
| 51 |
+
"step": 224
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.04,
|
| 55 |
+
"learning_rate": 3.1513647642679897e-07,
|
| 56 |
+
"loss": 0.1609,
|
| 57 |
+
"step": 256
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.04,
|
| 61 |
+
"learning_rate": 3.5483870967741936e-07,
|
| 62 |
+
"loss": 0.1434,
|
| 63 |
+
"step": 288
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.05,
|
| 67 |
+
"learning_rate": 3.945409429280397e-07,
|
| 68 |
+
"loss": 0.1699,
|
| 69 |
+
"step": 320
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.05,
|
| 73 |
+
"learning_rate": 4.3424317617866004e-07,
|
| 74 |
+
"loss": 0.1898,
|
| 75 |
+
"step": 352
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.06,
|
| 79 |
+
"learning_rate": 4.739454094292804e-07,
|
| 80 |
+
"loss": 0.1665,
|
| 81 |
+
"step": 384
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.06,
|
| 85 |
+
"learning_rate": 5.136476426799007e-07,
|
| 86 |
+
"loss": 0.1358,
|
| 87 |
+
"step": 416
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.07,
|
| 91 |
+
"learning_rate": 5.533498759305211e-07,
|
| 92 |
+
"loss": 0.1713,
|
| 93 |
+
"step": 448
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07,
|
| 97 |
+
"learning_rate": 5.930521091811415e-07,
|
| 98 |
+
"loss": 0.1817,
|
| 99 |
+
"step": 480
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.08,
|
| 103 |
+
"learning_rate": 6.327543424317618e-07,
|
| 104 |
+
"loss": 0.149,
|
| 105 |
+
"step": 512
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.08,
|
| 109 |
+
"learning_rate": 6.724565756823821e-07,
|
| 110 |
+
"loss": 0.165,
|
| 111 |
+
"step": 544
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.09,
|
| 115 |
+
"learning_rate": 7.121588089330024e-07,
|
| 116 |
+
"loss": 0.2282,
|
| 117 |
+
"step": 576
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.09,
|
| 121 |
+
"learning_rate": 7.518610421836227e-07,
|
| 122 |
+
"loss": 0.1666,
|
| 123 |
+
"step": 608
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.1,
|
| 127 |
+
"learning_rate": 7.915632754342431e-07,
|
| 128 |
+
"loss": 0.1229,
|
| 129 |
+
"step": 640
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1,
|
| 133 |
+
"learning_rate": 8.312655086848634e-07,
|
| 134 |
+
"loss": 0.2118,
|
| 135 |
+
"step": 672
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1,
|
| 139 |
+
"learning_rate": 8.709677419354838e-07,
|
| 140 |
+
"loss": 0.1869,
|
| 141 |
+
"step": 704
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.11,
|
| 145 |
+
"learning_rate": 9.106699751861042e-07,
|
| 146 |
+
"loss": 0.2196,
|
| 147 |
+
"step": 736
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.11,
|
| 151 |
+
"learning_rate": 9.503722084367245e-07,
|
| 152 |
+
"loss": 0.0936,
|
| 153 |
+
"step": 768
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.12,
|
| 157 |
+
"learning_rate": 9.90074441687345e-07,
|
| 158 |
+
"loss": 0.1839,
|
| 159 |
+
"step": 800
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.12,
|
| 163 |
+
"learning_rate": 9.980994615140957e-07,
|
| 164 |
+
"loss": 0.1675,
|
| 165 |
+
"step": 832
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.13,
|
| 169 |
+
"learning_rate": 9.955654101995564e-07,
|
| 170 |
+
"loss": 0.1595,
|
| 171 |
+
"step": 864
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13,
|
| 175 |
+
"learning_rate": 9.930313588850174e-07,
|
| 176 |
+
"loss": 0.1556,
|
| 177 |
+
"step": 896
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.14,
|
| 181 |
+
"learning_rate": 9.905764966740576e-07,
|
| 182 |
+
"loss": 0.1711,
|
| 183 |
+
"step": 928
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.14,
|
| 187 |
+
"learning_rate": 9.880424453595185e-07,
|
| 188 |
+
"loss": 0.1777,
|
| 189 |
+
"step": 960
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.15,
|
| 193 |
+
"learning_rate": 9.855083940449792e-07,
|
| 194 |
+
"loss": 0.2031,
|
| 195 |
+
"step": 992
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.15,
|
| 199 |
+
"learning_rate": 9.829743427304402e-07,
|
| 200 |
+
"loss": 0.1529,
|
| 201 |
+
"step": 1024
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.16,
|
| 205 |
+
"learning_rate": 9.804402914159011e-07,
|
| 206 |
+
"loss": 0.1365,
|
| 207 |
+
"step": 1056
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.16,
|
| 211 |
+
"learning_rate": 9.77906240101362e-07,
|
| 212 |
+
"loss": 0.1922,
|
| 213 |
+
"step": 1088
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.17,
|
| 217 |
+
"learning_rate": 9.75372188786823e-07,
|
| 218 |
+
"loss": 0.133,
|
| 219 |
+
"step": 1120
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.17,
|
| 223 |
+
"learning_rate": 9.728381374722837e-07,
|
| 224 |
+
"loss": 0.1692,
|
| 225 |
+
"step": 1152
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.18,
|
| 229 |
+
"learning_rate": 9.703040861577447e-07,
|
| 230 |
+
"loss": 0.1022,
|
| 231 |
+
"step": 1184
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.18,
|
| 235 |
+
"learning_rate": 9.677700348432054e-07,
|
| 236 |
+
"loss": 0.2052,
|
| 237 |
+
"step": 1216
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.19,
|
| 241 |
+
"learning_rate": 9.652359835286664e-07,
|
| 242 |
+
"loss": 0.1546,
|
| 243 |
+
"step": 1248
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.19,
|
| 247 |
+
"learning_rate": 9.627019322141273e-07,
|
| 248 |
+
"loss": 0.149,
|
| 249 |
+
"step": 1280
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.2,
|
| 253 |
+
"learning_rate": 9.601678808995883e-07,
|
| 254 |
+
"loss": 0.1281,
|
| 255 |
+
"step": 1312
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.2,
|
| 259 |
+
"learning_rate": 9.57633829585049e-07,
|
| 260 |
+
"loss": 0.1437,
|
| 261 |
+
"step": 1344
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.2,
|
| 265 |
+
"learning_rate": 9.5509977827051e-07,
|
| 266 |
+
"loss": 0.2097,
|
| 267 |
+
"step": 1376
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.21,
|
| 271 |
+
"learning_rate": 9.525657269559708e-07,
|
| 272 |
+
"loss": 0.1308,
|
| 273 |
+
"step": 1408
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.21,
|
| 277 |
+
"learning_rate": 9.500316756414317e-07,
|
| 278 |
+
"loss": 0.1691,
|
| 279 |
+
"step": 1440
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.22,
|
| 283 |
+
"learning_rate": 9.474976243268927e-07,
|
| 284 |
+
"loss": 0.2319,
|
| 285 |
+
"step": 1472
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.22,
|
| 289 |
+
"learning_rate": 9.449635730123534e-07,
|
| 290 |
+
"loss": 0.2226,
|
| 291 |
+
"step": 1504
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.23,
|
| 295 |
+
"learning_rate": 9.424295216978143e-07,
|
| 296 |
+
"loss": 0.1789,
|
| 297 |
+
"step": 1536
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.23,
|
| 301 |
+
"learning_rate": 9.398954703832752e-07,
|
| 302 |
+
"loss": 0.1932,
|
| 303 |
+
"step": 1568
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.24,
|
| 307 |
+
"learning_rate": 9.373614190687361e-07,
|
| 308 |
+
"loss": 0.1718,
|
| 309 |
+
"step": 1600
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.24,
|
| 313 |
+
"learning_rate": 9.34827367754197e-07,
|
| 314 |
+
"loss": 0.156,
|
| 315 |
+
"step": 1632
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.25,
|
| 319 |
+
"learning_rate": 9.322933164396578e-07,
|
| 320 |
+
"loss": 0.1512,
|
| 321 |
+
"step": 1664
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.25,
|
| 325 |
+
"learning_rate": 9.297592651251187e-07,
|
| 326 |
+
"loss": 0.0968,
|
| 327 |
+
"step": 1696
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.26,
|
| 331 |
+
"learning_rate": 9.272252138105796e-07,
|
| 332 |
+
"loss": 0.0932,
|
| 333 |
+
"step": 1728
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.26,
|
| 337 |
+
"learning_rate": 9.246911624960405e-07,
|
| 338 |
+
"loss": 0.2464,
|
| 339 |
+
"step": 1760
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27,
|
| 343 |
+
"learning_rate": 9.221571111815014e-07,
|
| 344 |
+
"loss": 0.2036,
|
| 345 |
+
"step": 1792
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.27,
|
| 349 |
+
"learning_rate": 9.196230598669623e-07,
|
| 350 |
+
"loss": 0.1245,
|
| 351 |
+
"step": 1824
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.28,
|
| 355 |
+
"learning_rate": 9.170890085524232e-07,
|
| 356 |
+
"loss": 0.1097,
|
| 357 |
+
"step": 1856
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.28,
|
| 361 |
+
"learning_rate": 9.14554957237884e-07,
|
| 362 |
+
"loss": 0.1844,
|
| 363 |
+
"step": 1888
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.29,
|
| 367 |
+
"learning_rate": 9.120209059233449e-07,
|
| 368 |
+
"loss": 0.1114,
|
| 369 |
+
"step": 1920
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.29,
|
| 373 |
+
"learning_rate": 9.094868546088058e-07,
|
| 374 |
+
"loss": 0.1992,
|
| 375 |
+
"step": 1952
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.3,
|
| 379 |
+
"learning_rate": 9.069528032942667e-07,
|
| 380 |
+
"loss": 0.1721,
|
| 381 |
+
"step": 1984
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.3,
|
| 385 |
+
"learning_rate": 9.044187519797275e-07,
|
| 386 |
+
"loss": 0.1473,
|
| 387 |
+
"step": 2016
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.3,
|
| 391 |
+
"learning_rate": 9.018847006651884e-07,
|
| 392 |
+
"loss": 0.1865,
|
| 393 |
+
"step": 2048
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.31,
|
| 397 |
+
"learning_rate": 8.993506493506493e-07,
|
| 398 |
+
"loss": 0.1583,
|
| 399 |
+
"step": 2080
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.31,
|
| 403 |
+
"learning_rate": 8.968165980361102e-07,
|
| 404 |
+
"loss": 0.1866,
|
| 405 |
+
"step": 2112
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.32,
|
| 409 |
+
"learning_rate": 8.942825467215711e-07,
|
| 410 |
+
"loss": 0.1617,
|
| 411 |
+
"step": 2144
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.32,
|
| 415 |
+
"learning_rate": 8.917484954070319e-07,
|
| 416 |
+
"loss": 0.1189,
|
| 417 |
+
"step": 2176
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.33,
|
| 421 |
+
"learning_rate": 8.892144440924928e-07,
|
| 422 |
+
"loss": 0.148,
|
| 423 |
+
"step": 2208
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.33,
|
| 427 |
+
"learning_rate": 8.866803927779537e-07,
|
| 428 |
+
"loss": 0.131,
|
| 429 |
+
"step": 2240
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.34,
|
| 433 |
+
"learning_rate": 8.841463414634146e-07,
|
| 434 |
+
"loss": 0.2261,
|
| 435 |
+
"step": 2272
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.34,
|
| 439 |
+
"learning_rate": 8.816122901488755e-07,
|
| 440 |
+
"loss": 0.1742,
|
| 441 |
+
"step": 2304
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.35,
|
| 445 |
+
"learning_rate": 8.790782388343364e-07,
|
| 446 |
+
"loss": 0.164,
|
| 447 |
+
"step": 2336
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.35,
|
| 451 |
+
"learning_rate": 8.765441875197972e-07,
|
| 452 |
+
"loss": 0.1161,
|
| 453 |
+
"step": 2368
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.36,
|
| 457 |
+
"learning_rate": 8.74010136205258e-07,
|
| 458 |
+
"loss": 0.1636,
|
| 459 |
+
"step": 2400
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.36,
|
| 463 |
+
"learning_rate": 8.71476084890719e-07,
|
| 464 |
+
"loss": 0.2416,
|
| 465 |
+
"step": 2432
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.37,
|
| 469 |
+
"learning_rate": 8.689420335761799e-07,
|
| 470 |
+
"loss": 0.1632,
|
| 471 |
+
"step": 2464
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.37,
|
| 475 |
+
"learning_rate": 8.664079822616408e-07,
|
| 476 |
+
"loss": 0.1477,
|
| 477 |
+
"step": 2496
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.38,
|
| 481 |
+
"learning_rate": 8.638739309471016e-07,
|
| 482 |
+
"loss": 0.2083,
|
| 483 |
+
"step": 2528
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.38,
|
| 487 |
+
"learning_rate": 8.613398796325625e-07,
|
| 488 |
+
"loss": 0.1599,
|
| 489 |
+
"step": 2560
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.39,
|
| 493 |
+
"learning_rate": 8.588058283180234e-07,
|
| 494 |
+
"loss": 0.1817,
|
| 495 |
+
"step": 2592
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.39,
|
| 499 |
+
"learning_rate": 8.562717770034843e-07,
|
| 500 |
+
"loss": 0.1005,
|
| 501 |
+
"step": 2624
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.4,
|
| 505 |
+
"learning_rate": 8.537377256889452e-07,
|
| 506 |
+
"loss": 0.168,
|
| 507 |
+
"step": 2656
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.4,
|
| 511 |
+
"learning_rate": 8.51203674374406e-07,
|
| 512 |
+
"loss": 0.2418,
|
| 513 |
+
"step": 2688
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4,
|
| 517 |
+
"learning_rate": 8.486696230598669e-07,
|
| 518 |
+
"loss": 0.1881,
|
| 519 |
+
"step": 2720
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.41,
|
| 523 |
+
"learning_rate": 8.461355717453278e-07,
|
| 524 |
+
"loss": 0.1829,
|
| 525 |
+
"step": 2752
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.41,
|
| 529 |
+
"learning_rate": 8.436015204307887e-07,
|
| 530 |
+
"loss": 0.1073,
|
| 531 |
+
"step": 2784
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.42,
|
| 535 |
+
"learning_rate": 8.410674691162496e-07,
|
| 536 |
+
"loss": 0.1324,
|
| 537 |
+
"step": 2816
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.42,
|
| 541 |
+
"learning_rate": 8.385334178017105e-07,
|
| 542 |
+
"loss": 0.2077,
|
| 543 |
+
"step": 2848
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.43,
|
| 547 |
+
"learning_rate": 8.359993664871713e-07,
|
| 548 |
+
"loss": 0.2248,
|
| 549 |
+
"step": 2880
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.43,
|
| 553 |
+
"learning_rate": 8.334653151726322e-07,
|
| 554 |
+
"loss": 0.1337,
|
| 555 |
+
"step": 2912
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.44,
|
| 559 |
+
"learning_rate": 8.30931263858093e-07,
|
| 560 |
+
"loss": 0.1906,
|
| 561 |
+
"step": 2944
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.44,
|
| 565 |
+
"learning_rate": 8.28397212543554e-07,
|
| 566 |
+
"loss": 0.1893,
|
| 567 |
+
"step": 2976
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.45,
|
| 571 |
+
"learning_rate": 8.259423503325942e-07,
|
| 572 |
+
"loss": 0.2029,
|
| 573 |
+
"step": 3008
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.45,
|
| 577 |
+
"learning_rate": 8.234082990180551e-07,
|
| 578 |
+
"loss": 0.157,
|
| 579 |
+
"step": 3040
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.46,
|
| 583 |
+
"learning_rate": 8.208742477035159e-07,
|
| 584 |
+
"loss": 0.1433,
|
| 585 |
+
"step": 3072
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.46,
|
| 589 |
+
"learning_rate": 8.183401963889769e-07,
|
| 590 |
+
"loss": 0.1689,
|
| 591 |
+
"step": 3104
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.47,
|
| 595 |
+
"learning_rate": 8.158061450744377e-07,
|
| 596 |
+
"loss": 0.2012,
|
| 597 |
+
"step": 3136
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.47,
|
| 601 |
+
"learning_rate": 8.132720937598986e-07,
|
| 602 |
+
"loss": 0.175,
|
| 603 |
+
"step": 3168
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.48,
|
| 607 |
+
"learning_rate": 8.107380424453595e-07,
|
| 608 |
+
"loss": 0.1961,
|
| 609 |
+
"step": 3200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.48,
|
| 613 |
+
"learning_rate": 8.082039911308203e-07,
|
| 614 |
+
"loss": 0.2547,
|
| 615 |
+
"step": 3232
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.49,
|
| 619 |
+
"learning_rate": 8.056699398162813e-07,
|
| 620 |
+
"loss": 0.1935,
|
| 621 |
+
"step": 3264
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.49,
|
| 625 |
+
"learning_rate": 8.031358885017421e-07,
|
| 626 |
+
"loss": 0.2149,
|
| 627 |
+
"step": 3296
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.5,
|
| 631 |
+
"learning_rate": 8.00601837187203e-07,
|
| 632 |
+
"loss": 0.1809,
|
| 633 |
+
"step": 3328
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5,
|
| 637 |
+
"learning_rate": 7.980677858726639e-07,
|
| 638 |
+
"loss": 0.2072,
|
| 639 |
+
"step": 3360
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5,
|
| 643 |
+
"learning_rate": 7.955337345581247e-07,
|
| 644 |
+
"loss": 0.2116,
|
| 645 |
+
"step": 3392
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.51,
|
| 649 |
+
"learning_rate": 7.929996832435857e-07,
|
| 650 |
+
"loss": 0.1737,
|
| 651 |
+
"step": 3424
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.51,
|
| 655 |
+
"learning_rate": 7.904656319290464e-07,
|
| 656 |
+
"loss": 0.2219,
|
| 657 |
+
"step": 3456
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.52,
|
| 661 |
+
"learning_rate": 7.879315806145074e-07,
|
| 662 |
+
"loss": 0.1849,
|
| 663 |
+
"step": 3488
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.52,
|
| 667 |
+
"learning_rate": 7.853975292999683e-07,
|
| 668 |
+
"loss": 0.1884,
|
| 669 |
+
"step": 3520
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.53,
|
| 673 |
+
"learning_rate": 7.828634779854292e-07,
|
| 674 |
+
"loss": 0.2192,
|
| 675 |
+
"step": 3552
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.53,
|
| 679 |
+
"learning_rate": 7.803294266708901e-07,
|
| 680 |
+
"loss": 0.1958,
|
| 681 |
+
"step": 3584
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.54,
|
| 685 |
+
"learning_rate": 7.777953753563509e-07,
|
| 686 |
+
"loss": 0.1433,
|
| 687 |
+
"step": 3616
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.54,
|
| 691 |
+
"learning_rate": 7.752613240418118e-07,
|
| 692 |
+
"loss": 0.2151,
|
| 693 |
+
"step": 3648
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.55,
|
| 697 |
+
"learning_rate": 7.727272727272727e-07,
|
| 698 |
+
"loss": 0.1675,
|
| 699 |
+
"step": 3680
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.55,
|
| 703 |
+
"learning_rate": 7.701932214127336e-07,
|
| 704 |
+
"loss": 0.1586,
|
| 705 |
+
"step": 3712
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.56,
|
| 709 |
+
"learning_rate": 7.676591700981945e-07,
|
| 710 |
+
"loss": 0.2881,
|
| 711 |
+
"step": 3744
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.56,
|
| 715 |
+
"learning_rate": 7.651251187836553e-07,
|
| 716 |
+
"loss": 0.196,
|
| 717 |
+
"step": 3776
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.57,
|
| 721 |
+
"learning_rate": 7.625910674691162e-07,
|
| 722 |
+
"loss": 0.1285,
|
| 723 |
+
"step": 3808
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.57,
|
| 727 |
+
"learning_rate": 7.60057016154577e-07,
|
| 728 |
+
"loss": 0.2262,
|
| 729 |
+
"step": 3840
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.58,
|
| 733 |
+
"learning_rate": 7.57522964840038e-07,
|
| 734 |
+
"loss": 0.2309,
|
| 735 |
+
"step": 3872
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.58,
|
| 739 |
+
"learning_rate": 7.549889135254989e-07,
|
| 740 |
+
"loss": 0.1533,
|
| 741 |
+
"step": 3904
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.59,
|
| 745 |
+
"learning_rate": 7.524548622109597e-07,
|
| 746 |
+
"loss": 0.1297,
|
| 747 |
+
"step": 3936
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.59,
|
| 751 |
+
"learning_rate": 7.499208108964206e-07,
|
| 752 |
+
"loss": 0.1808,
|
| 753 |
+
"step": 3968
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.6,
|
| 757 |
+
"learning_rate": 7.473867595818814e-07,
|
| 758 |
+
"loss": 0.2401,
|
| 759 |
+
"step": 4000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6,
|
| 763 |
+
"learning_rate": 7.448527082673424e-07,
|
| 764 |
+
"loss": 0.2507,
|
| 765 |
+
"step": 4032
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.61,
|
| 769 |
+
"learning_rate": 7.423186569528033e-07,
|
| 770 |
+
"loss": 0.1562,
|
| 771 |
+
"step": 4064
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 0.61,
|
| 775 |
+
"learning_rate": 7.397846056382642e-07,
|
| 776 |
+
"loss": 0.1912,
|
| 777 |
+
"step": 4096
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.61,
|
| 781 |
+
"learning_rate": 7.373297434273043e-07,
|
| 782 |
+
"loss": 0.1703,
|
| 783 |
+
"step": 4128
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.62,
|
| 787 |
+
"learning_rate": 7.347956921127653e-07,
|
| 788 |
+
"loss": 0.1471,
|
| 789 |
+
"step": 4160
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.62,
|
| 793 |
+
"learning_rate": 7.322616407982262e-07,
|
| 794 |
+
"loss": 0.1539,
|
| 795 |
+
"step": 4192
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.63,
|
| 799 |
+
"learning_rate": 7.297275894836869e-07,
|
| 800 |
+
"loss": 0.1521,
|
| 801 |
+
"step": 4224
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.63,
|
| 805 |
+
"learning_rate": 7.271935381691479e-07,
|
| 806 |
+
"loss": 0.2623,
|
| 807 |
+
"step": 4256
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.64,
|
| 811 |
+
"learning_rate": 7.246594868546087e-07,
|
| 812 |
+
"loss": 0.1753,
|
| 813 |
+
"step": 4288
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.64,
|
| 817 |
+
"learning_rate": 7.221254355400697e-07,
|
| 818 |
+
"loss": 0.1945,
|
| 819 |
+
"step": 4320
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.65,
|
| 823 |
+
"learning_rate": 7.195913842255306e-07,
|
| 824 |
+
"loss": 0.2153,
|
| 825 |
+
"step": 4352
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.65,
|
| 829 |
+
"learning_rate": 7.170573329109915e-07,
|
| 830 |
+
"loss": 0.2841,
|
| 831 |
+
"step": 4384
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.66,
|
| 835 |
+
"learning_rate": 7.145232815964523e-07,
|
| 836 |
+
"loss": 0.1759,
|
| 837 |
+
"step": 4416
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.66,
|
| 841 |
+
"learning_rate": 7.119892302819131e-07,
|
| 842 |
+
"loss": 0.2214,
|
| 843 |
+
"step": 4448
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.67,
|
| 847 |
+
"learning_rate": 7.094551789673741e-07,
|
| 848 |
+
"loss": 0.188,
|
| 849 |
+
"step": 4480
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.67,
|
| 853 |
+
"learning_rate": 7.069211276528349e-07,
|
| 854 |
+
"loss": 0.1579,
|
| 855 |
+
"step": 4512
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.68,
|
| 859 |
+
"learning_rate": 7.043870763382959e-07,
|
| 860 |
+
"loss": 0.2213,
|
| 861 |
+
"step": 4544
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.68,
|
| 865 |
+
"learning_rate": 7.018530250237567e-07,
|
| 866 |
+
"loss": 0.2042,
|
| 867 |
+
"step": 4576
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.69,
|
| 871 |
+
"learning_rate": 6.993189737092175e-07,
|
| 872 |
+
"loss": 0.1852,
|
| 873 |
+
"step": 4608
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.69,
|
| 877 |
+
"learning_rate": 6.967849223946785e-07,
|
| 878 |
+
"loss": 0.1716,
|
| 879 |
+
"step": 4640
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.7,
|
| 883 |
+
"learning_rate": 6.942508710801393e-07,
|
| 884 |
+
"loss": 0.1645,
|
| 885 |
+
"step": 4672
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.7,
|
| 889 |
+
"learning_rate": 6.917168197656003e-07,
|
| 890 |
+
"loss": 0.1986,
|
| 891 |
+
"step": 4704
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.71,
|
| 895 |
+
"learning_rate": 6.89182768451061e-07,
|
| 896 |
+
"loss": 0.2531,
|
| 897 |
+
"step": 4736
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.71,
|
| 901 |
+
"learning_rate": 6.866487171365219e-07,
|
| 902 |
+
"loss": 0.1792,
|
| 903 |
+
"step": 4768
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.71,
|
| 907 |
+
"learning_rate": 6.841146658219829e-07,
|
| 908 |
+
"loss": 0.1843,
|
| 909 |
+
"step": 4800
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.72,
|
| 913 |
+
"learning_rate": 6.815806145074437e-07,
|
| 914 |
+
"loss": 0.2175,
|
| 915 |
+
"step": 4832
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.72,
|
| 919 |
+
"learning_rate": 6.790465631929047e-07,
|
| 920 |
+
"loss": 0.2083,
|
| 921 |
+
"step": 4864
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.73,
|
| 925 |
+
"learning_rate": 6.765125118783655e-07,
|
| 926 |
+
"loss": 0.1729,
|
| 927 |
+
"step": 4896
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.73,
|
| 931 |
+
"learning_rate": 6.739784605638263e-07,
|
| 932 |
+
"loss": 0.1849,
|
| 933 |
+
"step": 4928
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.74,
|
| 937 |
+
"learning_rate": 6.714444092492873e-07,
|
| 938 |
+
"loss": 0.2374,
|
| 939 |
+
"step": 4960
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.74,
|
| 943 |
+
"learning_rate": 6.689103579347481e-07,
|
| 944 |
+
"loss": 0.241,
|
| 945 |
+
"step": 4992
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.75,
|
| 949 |
+
"learning_rate": 6.663763066202091e-07,
|
| 950 |
+
"loss": 0.1853,
|
| 951 |
+
"step": 5024
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 0.75,
|
| 955 |
+
"learning_rate": 6.638422553056699e-07,
|
| 956 |
+
"loss": 0.1957,
|
| 957 |
+
"step": 5056
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 0.76,
|
| 961 |
+
"learning_rate": 6.613082039911308e-07,
|
| 962 |
+
"loss": 0.2052,
|
| 963 |
+
"step": 5088
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.76,
|
| 967 |
+
"learning_rate": 6.587741526765917e-07,
|
| 968 |
+
"loss": 0.2321,
|
| 969 |
+
"step": 5120
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.77,
|
| 973 |
+
"learning_rate": 6.562401013620525e-07,
|
| 974 |
+
"loss": 0.1804,
|
| 975 |
+
"step": 5152
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.77,
|
| 979 |
+
"learning_rate": 6.537060500475135e-07,
|
| 980 |
+
"loss": 0.1842,
|
| 981 |
+
"step": 5184
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.78,
|
| 985 |
+
"learning_rate": 6.511719987329743e-07,
|
| 986 |
+
"loss": 0.2388,
|
| 987 |
+
"step": 5216
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.78,
|
| 991 |
+
"learning_rate": 6.486379474184352e-07,
|
| 992 |
+
"loss": 0.2417,
|
| 993 |
+
"step": 5248
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.79,
|
| 997 |
+
"learning_rate": 6.46103896103896e-07,
|
| 998 |
+
"loss": 0.2224,
|
| 999 |
+
"step": 5280
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.79,
|
| 1003 |
+
"learning_rate": 6.435698447893569e-07,
|
| 1004 |
+
"loss": 0.2029,
|
| 1005 |
+
"step": 5312
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 0.8,
|
| 1009 |
+
"learning_rate": 6.410357934748179e-07,
|
| 1010 |
+
"loss": 0.2807,
|
| 1011 |
+
"step": 5344
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8,
|
| 1015 |
+
"learning_rate": 6.385017421602787e-07,
|
| 1016 |
+
"loss": 0.192,
|
| 1017 |
+
"step": 5376
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.81,
|
| 1021 |
+
"learning_rate": 6.359676908457397e-07,
|
| 1022 |
+
"loss": 0.1848,
|
| 1023 |
+
"step": 5408
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.81,
|
| 1027 |
+
"learning_rate": 6.334336395312004e-07,
|
| 1028 |
+
"loss": 0.2143,
|
| 1029 |
+
"step": 5440
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.81,
|
| 1033 |
+
"learning_rate": 6.308995882166613e-07,
|
| 1034 |
+
"loss": 0.2421,
|
| 1035 |
+
"step": 5472
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.82,
|
| 1039 |
+
"learning_rate": 6.283655369021223e-07,
|
| 1040 |
+
"loss": 0.1724,
|
| 1041 |
+
"step": 5504
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.82,
|
| 1045 |
+
"learning_rate": 6.258314855875831e-07,
|
| 1046 |
+
"loss": 0.1207,
|
| 1047 |
+
"step": 5536
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 0.83,
|
| 1051 |
+
"learning_rate": 6.232974342730441e-07,
|
| 1052 |
+
"loss": 0.2259,
|
| 1053 |
+
"step": 5568
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.83,
|
| 1057 |
+
"learning_rate": 6.207633829585048e-07,
|
| 1058 |
+
"loss": 0.2504,
|
| 1059 |
+
"step": 5600
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.84,
|
| 1063 |
+
"learning_rate": 6.182293316439658e-07,
|
| 1064 |
+
"loss": 0.188,
|
| 1065 |
+
"step": 5632
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.84,
|
| 1069 |
+
"learning_rate": 6.156952803294266e-07,
|
| 1070 |
+
"loss": 0.1893,
|
| 1071 |
+
"step": 5664
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.85,
|
| 1075 |
+
"learning_rate": 6.131612290148875e-07,
|
| 1076 |
+
"loss": 0.1905,
|
| 1077 |
+
"step": 5696
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 0.85,
|
| 1081 |
+
"learning_rate": 6.106271777003485e-07,
|
| 1082 |
+
"loss": 0.2594,
|
| 1083 |
+
"step": 5728
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 0.86,
|
| 1087 |
+
"learning_rate": 6.080931263858092e-07,
|
| 1088 |
+
"loss": 0.3084,
|
| 1089 |
+
"step": 5760
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.86,
|
| 1093 |
+
"learning_rate": 6.055590750712702e-07,
|
| 1094 |
+
"loss": 0.1925,
|
| 1095 |
+
"step": 5792
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.87,
|
| 1099 |
+
"learning_rate": 6.03025023756731e-07,
|
| 1100 |
+
"loss": 0.186,
|
| 1101 |
+
"step": 5824
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.87,
|
| 1105 |
+
"learning_rate": 6.004909724421919e-07,
|
| 1106 |
+
"loss": 0.2302,
|
| 1107 |
+
"step": 5856
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.88,
|
| 1111 |
+
"learning_rate": 5.979569211276529e-07,
|
| 1112 |
+
"loss": 0.1371,
|
| 1113 |
+
"step": 5888
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.88,
|
| 1117 |
+
"learning_rate": 5.954228698131137e-07,
|
| 1118 |
+
"loss": 0.231,
|
| 1119 |
+
"step": 5920
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.89,
|
| 1123 |
+
"learning_rate": 5.928888184985746e-07,
|
| 1124 |
+
"loss": 0.2012,
|
| 1125 |
+
"step": 5952
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.89,
|
| 1129 |
+
"learning_rate": 5.903547671840354e-07,
|
| 1130 |
+
"loss": 0.2006,
|
| 1131 |
+
"step": 5984
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.9,
|
| 1135 |
+
"learning_rate": 5.878207158694963e-07,
|
| 1136 |
+
"loss": 0.215,
|
| 1137 |
+
"step": 6016
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9,
|
| 1141 |
+
"learning_rate": 5.852866645549572e-07,
|
| 1142 |
+
"loss": 0.1471,
|
| 1143 |
+
"step": 6048
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.91,
|
| 1147 |
+
"learning_rate": 5.827526132404181e-07,
|
| 1148 |
+
"loss": 0.2364,
|
| 1149 |
+
"step": 6080
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.91,
|
| 1153 |
+
"learning_rate": 5.80218561925879e-07,
|
| 1154 |
+
"loss": 0.2881,
|
| 1155 |
+
"step": 6112
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.91,
|
| 1159 |
+
"learning_rate": 5.776845106113398e-07,
|
| 1160 |
+
"loss": 0.1536,
|
| 1161 |
+
"step": 6144
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.92,
|
| 1165 |
+
"learning_rate": 5.751504592968008e-07,
|
| 1166 |
+
"loss": 0.2317,
|
| 1167 |
+
"step": 6176
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.92,
|
| 1171 |
+
"learning_rate": 5.726164079822616e-07,
|
| 1172 |
+
"loss": 0.1952,
|
| 1173 |
+
"step": 6208
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.93,
|
| 1177 |
+
"learning_rate": 5.700823566677225e-07,
|
| 1178 |
+
"loss": 0.1602,
|
| 1179 |
+
"step": 6240
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.93,
|
| 1183 |
+
"learning_rate": 5.675483053531834e-07,
|
| 1184 |
+
"loss": 0.212,
|
| 1185 |
+
"step": 6272
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.94,
|
| 1189 |
+
"learning_rate": 5.650142540386442e-07,
|
| 1190 |
+
"loss": 0.2401,
|
| 1191 |
+
"step": 6304
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.94,
|
| 1195 |
+
"learning_rate": 5.624802027241052e-07,
|
| 1196 |
+
"loss": 0.1992,
|
| 1197 |
+
"step": 6336
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.95,
|
| 1201 |
+
"learning_rate": 5.59946151409566e-07,
|
| 1202 |
+
"loss": 0.2616,
|
| 1203 |
+
"step": 6368
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.95,
|
| 1207 |
+
"learning_rate": 5.574121000950269e-07,
|
| 1208 |
+
"loss": 0.146,
|
| 1209 |
+
"step": 6400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.96,
|
| 1213 |
+
"learning_rate": 5.548780487804878e-07,
|
| 1214 |
+
"loss": 0.2081,
|
| 1215 |
+
"step": 6432
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.96,
|
| 1219 |
+
"learning_rate": 5.523439974659486e-07,
|
| 1220 |
+
"loss": 0.207,
|
| 1221 |
+
"step": 6464
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.97,
|
| 1225 |
+
"learning_rate": 5.498099461514096e-07,
|
| 1226 |
+
"loss": 0.2631,
|
| 1227 |
+
"step": 6496
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.97,
|
| 1231 |
+
"learning_rate": 5.472758948368704e-07,
|
| 1232 |
+
"loss": 0.1721,
|
| 1233 |
+
"step": 6528
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.98,
|
| 1237 |
+
"learning_rate": 5.447418435223313e-07,
|
| 1238 |
+
"loss": 0.1908,
|
| 1239 |
+
"step": 6560
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.98,
|
| 1243 |
+
"learning_rate": 5.422869813113715e-07,
|
| 1244 |
+
"loss": 0.2238,
|
| 1245 |
+
"step": 6592
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.99,
|
| 1249 |
+
"learning_rate": 5.397529299968325e-07,
|
| 1250 |
+
"loss": 0.2524,
|
| 1251 |
+
"step": 6624
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.99,
|
| 1255 |
+
"learning_rate": 5.372188786822933e-07,
|
| 1256 |
+
"loss": 0.1968,
|
| 1257 |
+
"step": 6656
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.0,
|
| 1261 |
+
"learning_rate": 5.346848273677542e-07,
|
| 1262 |
+
"loss": 0.2379,
|
| 1263 |
+
"step": 6688
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 1.0,
|
| 1267 |
+
"learning_rate": 5.32150776053215e-07,
|
| 1268 |
+
"loss": 0.246,
|
| 1269 |
+
"step": 6720
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.01,
|
| 1273 |
+
"learning_rate": 5.296167247386759e-07,
|
| 1274 |
+
"loss": 0.104,
|
| 1275 |
+
"step": 6752
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.01,
|
| 1279 |
+
"learning_rate": 5.270826734241369e-07,
|
| 1280 |
+
"loss": 0.102,
|
| 1281 |
+
"step": 6784
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 1.01,
|
| 1285 |
+
"learning_rate": 5.245486221095977e-07,
|
| 1286 |
+
"loss": 0.1432,
|
| 1287 |
+
"step": 6816
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 1.02,
|
| 1291 |
+
"learning_rate": 5.220145707950586e-07,
|
| 1292 |
+
"loss": 0.1262,
|
| 1293 |
+
"step": 6848
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 1.02,
|
| 1297 |
+
"learning_rate": 5.194805194805194e-07,
|
| 1298 |
+
"loss": 0.1299,
|
| 1299 |
+
"step": 6880
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.03,
|
| 1303 |
+
"learning_rate": 5.169464681659803e-07,
|
| 1304 |
+
"loss": 0.1319,
|
| 1305 |
+
"step": 6912
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.03,
|
| 1309 |
+
"learning_rate": 5.144124168514412e-07,
|
| 1310 |
+
"loss": 0.105,
|
| 1311 |
+
"step": 6944
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.04,
|
| 1315 |
+
"learning_rate": 5.118783655369021e-07,
|
| 1316 |
+
"loss": 0.1233,
|
| 1317 |
+
"step": 6976
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.04,
|
| 1321 |
+
"learning_rate": 5.09344314222363e-07,
|
| 1322 |
+
"loss": 0.0922,
|
| 1323 |
+
"step": 7008
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 1.05,
|
| 1327 |
+
"learning_rate": 5.068102629078239e-07,
|
| 1328 |
+
"loss": 0.144,
|
| 1329 |
+
"step": 7040
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.05,
|
| 1333 |
+
"learning_rate": 5.042762115932847e-07,
|
| 1334 |
+
"loss": 0.1828,
|
| 1335 |
+
"step": 7072
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 1.06,
|
| 1339 |
+
"learning_rate": 5.017421602787456e-07,
|
| 1340 |
+
"loss": 0.1097,
|
| 1341 |
+
"step": 7104
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.06,
|
| 1345 |
+
"learning_rate": 4.992081089642065e-07,
|
| 1346 |
+
"loss": 0.1867,
|
| 1347 |
+
"step": 7136
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.07,
|
| 1351 |
+
"learning_rate": 4.966740576496675e-07,
|
| 1352 |
+
"loss": 0.1338,
|
| 1353 |
+
"step": 7168
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.07,
|
| 1357 |
+
"learning_rate": 4.941400063351283e-07,
|
| 1358 |
+
"loss": 0.1535,
|
| 1359 |
+
"step": 7200
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.08,
|
| 1363 |
+
"learning_rate": 4.916059550205891e-07,
|
| 1364 |
+
"loss": 0.1861,
|
| 1365 |
+
"step": 7232
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 1.08,
|
| 1369 |
+
"learning_rate": 4.8907190370605e-07,
|
| 1370 |
+
"loss": 0.1284,
|
| 1371 |
+
"step": 7264
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 1.09,
|
| 1375 |
+
"learning_rate": 4.865378523915109e-07,
|
| 1376 |
+
"loss": 0.1037,
|
| 1377 |
+
"step": 7296
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.09,
|
| 1381 |
+
"learning_rate": 4.840038010769719e-07,
|
| 1382 |
+
"loss": 0.1217,
|
| 1383 |
+
"step": 7328
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.1,
|
| 1387 |
+
"learning_rate": 4.814697497624327e-07,
|
| 1388 |
+
"loss": 0.1469,
|
| 1389 |
+
"step": 7360
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.1,
|
| 1393 |
+
"learning_rate": 4.789356984478935e-07,
|
| 1394 |
+
"loss": 0.1218,
|
| 1395 |
+
"step": 7392
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.11,
|
| 1399 |
+
"learning_rate": 4.764016471333545e-07,
|
| 1400 |
+
"loss": 0.1486,
|
| 1401 |
+
"step": 7424
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.11,
|
| 1405 |
+
"learning_rate": 4.738675958188153e-07,
|
| 1406 |
+
"loss": 0.0796,
|
| 1407 |
+
"step": 7456
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 1.11,
|
| 1411 |
+
"learning_rate": 4.7141273360785554e-07,
|
| 1412 |
+
"loss": 0.1163,
|
| 1413 |
+
"step": 7488
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.12,
|
| 1417 |
+
"learning_rate": 4.688786822933164e-07,
|
| 1418 |
+
"loss": 0.0821,
|
| 1419 |
+
"step": 7520
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 1.12,
|
| 1423 |
+
"learning_rate": 4.663446309787773e-07,
|
| 1424 |
+
"loss": 0.1701,
|
| 1425 |
+
"step": 7552
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 1.13,
|
| 1429 |
+
"learning_rate": 4.638105796642382e-07,
|
| 1430 |
+
"loss": 0.1002,
|
| 1431 |
+
"step": 7584
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.13,
|
| 1435 |
+
"learning_rate": 4.613557174532784e-07,
|
| 1436 |
+
"loss": 0.0914,
|
| 1437 |
+
"step": 7616
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.14,
|
| 1441 |
+
"learning_rate": 4.5882166613873927e-07,
|
| 1442 |
+
"loss": 0.1374,
|
| 1443 |
+
"step": 7648
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.14,
|
| 1447 |
+
"learning_rate": 4.562876148242002e-07,
|
| 1448 |
+
"loss": 0.1142,
|
| 1449 |
+
"step": 7680
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.15,
|
| 1453 |
+
"learning_rate": 4.5375356350966105e-07,
|
| 1454 |
+
"loss": 0.1351,
|
| 1455 |
+
"step": 7712
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.15,
|
| 1459 |
+
"learning_rate": 4.5121951219512194e-07,
|
| 1460 |
+
"loss": 0.1513,
|
| 1461 |
+
"step": 7744
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 1.16,
|
| 1465 |
+
"learning_rate": 4.486854608805828e-07,
|
| 1466 |
+
"loss": 0.0998,
|
| 1467 |
+
"step": 7776
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 1.16,
|
| 1471 |
+
"learning_rate": 4.461514095660437e-07,
|
| 1472 |
+
"loss": 0.1115,
|
| 1473 |
+
"step": 7808
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 1.17,
|
| 1477 |
+
"learning_rate": 4.4361735825150457e-07,
|
| 1478 |
+
"loss": 0.1373,
|
| 1479 |
+
"step": 7840
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.17,
|
| 1483 |
+
"learning_rate": 4.4108330693696546e-07,
|
| 1484 |
+
"loss": 0.1614,
|
| 1485 |
+
"step": 7872
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.18,
|
| 1489 |
+
"learning_rate": 4.3854925562242635e-07,
|
| 1490 |
+
"loss": 0.0826,
|
| 1491 |
+
"step": 7904
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 1.18,
|
| 1495 |
+
"learning_rate": 4.3601520430788724e-07,
|
| 1496 |
+
"loss": 0.1129,
|
| 1497 |
+
"step": 7936
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 1.19,
|
| 1501 |
+
"learning_rate": 4.334811529933481e-07,
|
| 1502 |
+
"loss": 0.0825,
|
| 1503 |
+
"step": 7968
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 1.19,
|
| 1507 |
+
"learning_rate": 4.3094710167880897e-07,
|
| 1508 |
+
"loss": 0.1497,
|
| 1509 |
+
"step": 8000
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 1.2,
|
| 1513 |
+
"learning_rate": 4.284130503642698e-07,
|
| 1514 |
+
"loss": 0.1575,
|
| 1515 |
+
"step": 8032
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 1.2,
|
| 1519 |
+
"learning_rate": 4.2587899904973075e-07,
|
| 1520 |
+
"loss": 0.1377,
|
| 1521 |
+
"step": 8064
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.21,
|
| 1525 |
+
"learning_rate": 4.2334494773519165e-07,
|
| 1526 |
+
"loss": 0.152,
|
| 1527 |
+
"step": 8096
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.21,
|
| 1531 |
+
"learning_rate": 4.208108964206525e-07,
|
| 1532 |
+
"loss": 0.1142,
|
| 1533 |
+
"step": 8128
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 1.21,
|
| 1537 |
+
"learning_rate": 4.182768451061134e-07,
|
| 1538 |
+
"loss": 0.1458,
|
| 1539 |
+
"step": 8160
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 1.22,
|
| 1543 |
+
"learning_rate": 4.1574279379157427e-07,
|
| 1544 |
+
"loss": 0.1216,
|
| 1545 |
+
"step": 8192
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 1.22,
|
| 1549 |
+
"learning_rate": 4.132087424770351e-07,
|
| 1550 |
+
"loss": 0.112,
|
| 1551 |
+
"step": 8224
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 1.23,
|
| 1555 |
+
"learning_rate": 4.1067469116249605e-07,
|
| 1556 |
+
"loss": 0.1517,
|
| 1557 |
+
"step": 8256
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 1.23,
|
| 1561 |
+
"learning_rate": 4.081406398479569e-07,
|
| 1562 |
+
"loss": 0.1046,
|
| 1563 |
+
"step": 8288
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.24,
|
| 1567 |
+
"learning_rate": 4.056065885334178e-07,
|
| 1568 |
+
"loss": 0.1538,
|
| 1569 |
+
"step": 8320
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.24,
|
| 1573 |
+
"learning_rate": 4.030725372188787e-07,
|
| 1574 |
+
"loss": 0.1734,
|
| 1575 |
+
"step": 8352
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 1.25,
|
| 1579 |
+
"learning_rate": 4.005384859043395e-07,
|
| 1580 |
+
"loss": 0.1236,
|
| 1581 |
+
"step": 8384
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 1.25,
|
| 1585 |
+
"learning_rate": 3.980044345898004e-07,
|
| 1586 |
+
"loss": 0.1129,
|
| 1587 |
+
"step": 8416
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 1.26,
|
| 1591 |
+
"learning_rate": 3.9547038327526135e-07,
|
| 1592 |
+
"loss": 0.1193,
|
| 1593 |
+
"step": 8448
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 1.26,
|
| 1597 |
+
"learning_rate": 3.929363319607222e-07,
|
| 1598 |
+
"loss": 0.1124,
|
| 1599 |
+
"step": 8480
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 1.27,
|
| 1603 |
+
"learning_rate": 3.904022806461831e-07,
|
| 1604 |
+
"loss": 0.103,
|
| 1605 |
+
"step": 8512
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 1.27,
|
| 1609 |
+
"learning_rate": 3.878682293316439e-07,
|
| 1610 |
+
"loss": 0.1526,
|
| 1611 |
+
"step": 8544
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.28,
|
| 1615 |
+
"learning_rate": 3.853341780171048e-07,
|
| 1616 |
+
"loss": 0.1747,
|
| 1617 |
+
"step": 8576
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 1.28,
|
| 1621 |
+
"learning_rate": 3.8280012670256575e-07,
|
| 1622 |
+
"loss": 0.0711,
|
| 1623 |
+
"step": 8608
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 1.29,
|
| 1627 |
+
"learning_rate": 3.802660753880266e-07,
|
| 1628 |
+
"loss": 0.131,
|
| 1629 |
+
"step": 8640
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 1.29,
|
| 1633 |
+
"learning_rate": 3.777320240734875e-07,
|
| 1634 |
+
"loss": 0.0695,
|
| 1635 |
+
"step": 8672
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 1.3,
|
| 1639 |
+
"learning_rate": 3.751979727589484e-07,
|
| 1640 |
+
"loss": 0.1176,
|
| 1641 |
+
"step": 8704
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 1.3,
|
| 1645 |
+
"learning_rate": 3.726639214444092e-07,
|
| 1646 |
+
"loss": 0.1141,
|
| 1647 |
+
"step": 8736
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 1.31,
|
| 1651 |
+
"learning_rate": 3.701298701298701e-07,
|
| 1652 |
+
"loss": 0.1437,
|
| 1653 |
+
"step": 8768
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.31,
|
| 1657 |
+
"learning_rate": 3.6759581881533095e-07,
|
| 1658 |
+
"loss": 0.1273,
|
| 1659 |
+
"step": 8800
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 1.31,
|
| 1663 |
+
"learning_rate": 3.650617675007919e-07,
|
| 1664 |
+
"loss": 0.0765,
|
| 1665 |
+
"step": 8832
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 1.32,
|
| 1669 |
+
"learning_rate": 3.625277161862528e-07,
|
| 1670 |
+
"loss": 0.1322,
|
| 1671 |
+
"step": 8864
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 1.32,
|
| 1675 |
+
"learning_rate": 3.599936648717136e-07,
|
| 1676 |
+
"loss": 0.1643,
|
| 1677 |
+
"step": 8896
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 1.33,
|
| 1681 |
+
"learning_rate": 3.574596135571745e-07,
|
| 1682 |
+
"loss": 0.1005,
|
| 1683 |
+
"step": 8928
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 1.33,
|
| 1687 |
+
"learning_rate": 3.549255622426354e-07,
|
| 1688 |
+
"loss": 0.1545,
|
| 1689 |
+
"step": 8960
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 1.34,
|
| 1693 |
+
"learning_rate": 3.523915109280963e-07,
|
| 1694 |
+
"loss": 0.0755,
|
| 1695 |
+
"step": 8992
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.34,
|
| 1699 |
+
"learning_rate": 3.498574596135572e-07,
|
| 1700 |
+
"loss": 0.0917,
|
| 1701 |
+
"step": 9024
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 1.35,
|
| 1705 |
+
"learning_rate": 3.47323408299018e-07,
|
| 1706 |
+
"loss": 0.0893,
|
| 1707 |
+
"step": 9056
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 1.35,
|
| 1711 |
+
"learning_rate": 3.447893569844789e-07,
|
| 1712 |
+
"loss": 0.152,
|
| 1713 |
+
"step": 9088
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 1.36,
|
| 1717 |
+
"learning_rate": 3.422553056699398e-07,
|
| 1718 |
+
"loss": 0.1154,
|
| 1719 |
+
"step": 9120
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 1.36,
|
| 1723 |
+
"learning_rate": 3.3972125435540065e-07,
|
| 1724 |
+
"loss": 0.0817,
|
| 1725 |
+
"step": 9152
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 1.37,
|
| 1729 |
+
"learning_rate": 3.371872030408616e-07,
|
| 1730 |
+
"loss": 0.083,
|
| 1731 |
+
"step": 9184
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 1.37,
|
| 1735 |
+
"learning_rate": 3.346531517263225e-07,
|
| 1736 |
+
"loss": 0.1084,
|
| 1737 |
+
"step": 9216
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.38,
|
| 1741 |
+
"learning_rate": 3.321191004117833e-07,
|
| 1742 |
+
"loss": 0.1514,
|
| 1743 |
+
"step": 9248
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 1.38,
|
| 1747 |
+
"learning_rate": 3.295850490972442e-07,
|
| 1748 |
+
"loss": 0.1195,
|
| 1749 |
+
"step": 9280
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 1.39,
|
| 1753 |
+
"learning_rate": 3.2705099778270505e-07,
|
| 1754 |
+
"loss": 0.0939,
|
| 1755 |
+
"step": 9312
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 1.39,
|
| 1759 |
+
"learning_rate": 3.2451694646816595e-07,
|
| 1760 |
+
"loss": 0.1227,
|
| 1761 |
+
"step": 9344
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 1.4,
|
| 1765 |
+
"learning_rate": 3.219828951536269e-07,
|
| 1766 |
+
"loss": 0.1342,
|
| 1767 |
+
"step": 9376
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 1.4,
|
| 1771 |
+
"learning_rate": 3.1944884383908773e-07,
|
| 1772 |
+
"loss": 0.1214,
|
| 1773 |
+
"step": 9408
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 1.41,
|
| 1777 |
+
"learning_rate": 3.169147925245486e-07,
|
| 1778 |
+
"loss": 0.1624,
|
| 1779 |
+
"step": 9440
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.41,
|
| 1783 |
+
"learning_rate": 3.143807412100095e-07,
|
| 1784 |
+
"loss": 0.1261,
|
| 1785 |
+
"step": 9472
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 1.41,
|
| 1789 |
+
"learning_rate": 3.1184668989547035e-07,
|
| 1790 |
+
"loss": 0.1674,
|
| 1791 |
+
"step": 9504
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 1.42,
|
| 1795 |
+
"learning_rate": 3.0931263858093124e-07,
|
| 1796 |
+
"loss": 0.1613,
|
| 1797 |
+
"step": 9536
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 1.42,
|
| 1801 |
+
"learning_rate": 3.0677858726639213e-07,
|
| 1802 |
+
"loss": 0.1519,
|
| 1803 |
+
"step": 9568
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 1.43,
|
| 1807 |
+
"learning_rate": 3.04244535951853e-07,
|
| 1808 |
+
"loss": 0.1107,
|
| 1809 |
+
"step": 9600
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 1.43,
|
| 1813 |
+
"learning_rate": 3.017104846373139e-07,
|
| 1814 |
+
"loss": 0.0992,
|
| 1815 |
+
"step": 9632
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 1.44,
|
| 1819 |
+
"learning_rate": 2.9917643332277476e-07,
|
| 1820 |
+
"loss": 0.0921,
|
| 1821 |
+
"step": 9664
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.44,
|
| 1825 |
+
"learning_rate": 2.9664238200823565e-07,
|
| 1826 |
+
"loss": 0.1478,
|
| 1827 |
+
"step": 9696
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 1.45,
|
| 1831 |
+
"learning_rate": 2.9410833069369654e-07,
|
| 1832 |
+
"loss": 0.1268,
|
| 1833 |
+
"step": 9728
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 1.45,
|
| 1837 |
+
"learning_rate": 2.9157427937915743e-07,
|
| 1838 |
+
"loss": 0.1086,
|
| 1839 |
+
"step": 9760
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 1.46,
|
| 1843 |
+
"learning_rate": 2.890402280646183e-07,
|
| 1844 |
+
"loss": 0.1913,
|
| 1845 |
+
"step": 9792
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 1.46,
|
| 1849 |
+
"learning_rate": 2.8650617675007916e-07,
|
| 1850 |
+
"loss": 0.1626,
|
| 1851 |
+
"step": 9824
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 1.47,
|
| 1855 |
+
"learning_rate": 2.8397212543554005e-07,
|
| 1856 |
+
"loss": 0.1647,
|
| 1857 |
+
"step": 9856
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 1.47,
|
| 1861 |
+
"learning_rate": 2.8143807412100095e-07,
|
| 1862 |
+
"loss": 0.1245,
|
| 1863 |
+
"step": 9888
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 1.48,
|
| 1867 |
+
"learning_rate": 2.789040228064618e-07,
|
| 1868 |
+
"loss": 0.1326,
|
| 1869 |
+
"step": 9920
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 1.48,
|
| 1873 |
+
"learning_rate": 2.7636997149192273e-07,
|
| 1874 |
+
"loss": 0.1095,
|
| 1875 |
+
"step": 9952
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 1.49,
|
| 1879 |
+
"learning_rate": 2.738359201773836e-07,
|
| 1880 |
+
"loss": 0.1089,
|
| 1881 |
+
"step": 9984
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 1.49,
|
| 1885 |
+
"learning_rate": 2.7130186886284446e-07,
|
| 1886 |
+
"loss": 0.1074,
|
| 1887 |
+
"step": 10016
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 1.5,
|
| 1891 |
+
"learning_rate": 2.6876781754830535e-07,
|
| 1892 |
+
"loss": 0.1084,
|
| 1893 |
+
"step": 10048
|
| 1894 |
+
}
|
| 1895 |
+
],
|
| 1896 |
+
"logging_steps": 32,
|
| 1897 |
+
"max_steps": 13434,
|
| 1898 |
+
"num_input_tokens_seen": 0,
|
| 1899 |
+
"num_train_epochs": 2,
|
| 1900 |
+
"save_steps": 3358,
|
| 1901 |
+
"total_flos": 4.27776497270784e+19,
|
| 1902 |
+
"train_batch_size": 2,
|
| 1903 |
+
"trial_name": null,
|
| 1904 |
+
"trial_params": null
|
| 1905 |
+
}
|
checkpoint-16791/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e4db6f6fcaf2005bb3f583cfbd6e7afdd0dfdcdb7d4db7e107d5c22865fcf47
|
| 3 |
+
size 5048
|
checkpoint-20149/config.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "models/openai/whisper-large-v2/finetune/jacob_filter/checkpoint-6717",
|
| 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-20149/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-20149/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:80523576b409f4707ecea08837512fe2e8a16584f6a339a95f9daa1966598b0a
|
| 3 |
+
size 4992706480
|
checkpoint-20149/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b12b93d32371965d98c848ea6108dd377d01aa81a49d42a6bb585c42f1443db
|
| 3 |
+
size 1180663192
|
checkpoint-20149/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-20149/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:adb2890fea374ef932d48052fb1da275d120c07c4f8b35899d029498eec8409a
|
| 3 |
+
size 3095074288
|
checkpoint-20149/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-20149/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:65e9a4671c9f32d1763620db0c5407681168d62600d835878502ed3977239b4b
|
| 3 |
+
size 14244
|
checkpoint-20149/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b0da0724f15e22687f61a706af43d751e5f3ee0e884a24e5dd70969f919c159
|
| 3 |
+
size 1064
|
checkpoint-20149/trainer_state.json
ADDED
|
@@ -0,0 +1,2535 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9997022480273932,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 13432,
|
| 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": 3.722084367245657e-08,
|
| 14 |
+
"loss": 0.2836,
|
| 15 |
+
"step": 32
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 7.692307692307692e-08,
|
| 20 |
+
"loss": 0.2111,
|
| 21 |
+
"step": 64
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 1.1662531017369727e-07,
|
| 26 |
+
"loss": 0.2145,
|
| 27 |
+
"step": 96
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.02,
|
| 31 |
+
"learning_rate": 1.563275434243176e-07,
|
| 32 |
+
"loss": 0.2375,
|
| 33 |
+
"step": 128
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 1.9602977667493795e-07,
|
| 38 |
+
"loss": 0.1839,
|
| 39 |
+
"step": 160
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03,
|
| 43 |
+
"learning_rate": 2.3573200992555832e-07,
|
| 44 |
+
"loss": 0.2864,
|
| 45 |
+
"step": 192
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03,
|
| 49 |
+
"learning_rate": 2.7543424317617863e-07,
|
| 50 |
+
"loss": 0.1799,
|
| 51 |
+
"step": 224
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.04,
|
| 55 |
+
"learning_rate": 3.1513647642679897e-07,
|
| 56 |
+
"loss": 0.1609,
|
| 57 |
+
"step": 256
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.04,
|
| 61 |
+
"learning_rate": 3.5483870967741936e-07,
|
| 62 |
+
"loss": 0.1434,
|
| 63 |
+
"step": 288
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.05,
|
| 67 |
+
"learning_rate": 3.945409429280397e-07,
|
| 68 |
+
"loss": 0.1699,
|
| 69 |
+
"step": 320
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.05,
|
| 73 |
+
"learning_rate": 4.3424317617866004e-07,
|
| 74 |
+
"loss": 0.1898,
|
| 75 |
+
"step": 352
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.06,
|
| 79 |
+
"learning_rate": 4.739454094292804e-07,
|
| 80 |
+
"loss": 0.1665,
|
| 81 |
+
"step": 384
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.06,
|
| 85 |
+
"learning_rate": 5.136476426799007e-07,
|
| 86 |
+
"loss": 0.1358,
|
| 87 |
+
"step": 416
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.07,
|
| 91 |
+
"learning_rate": 5.533498759305211e-07,
|
| 92 |
+
"loss": 0.1713,
|
| 93 |
+
"step": 448
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07,
|
| 97 |
+
"learning_rate": 5.930521091811415e-07,
|
| 98 |
+
"loss": 0.1817,
|
| 99 |
+
"step": 480
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.08,
|
| 103 |
+
"learning_rate": 6.327543424317618e-07,
|
| 104 |
+
"loss": 0.149,
|
| 105 |
+
"step": 512
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.08,
|
| 109 |
+
"learning_rate": 6.724565756823821e-07,
|
| 110 |
+
"loss": 0.165,
|
| 111 |
+
"step": 544
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.09,
|
| 115 |
+
"learning_rate": 7.121588089330024e-07,
|
| 116 |
+
"loss": 0.2282,
|
| 117 |
+
"step": 576
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.09,
|
| 121 |
+
"learning_rate": 7.518610421836227e-07,
|
| 122 |
+
"loss": 0.1666,
|
| 123 |
+
"step": 608
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.1,
|
| 127 |
+
"learning_rate": 7.915632754342431e-07,
|
| 128 |
+
"loss": 0.1229,
|
| 129 |
+
"step": 640
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1,
|
| 133 |
+
"learning_rate": 8.312655086848634e-07,
|
| 134 |
+
"loss": 0.2118,
|
| 135 |
+
"step": 672
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1,
|
| 139 |
+
"learning_rate": 8.709677419354838e-07,
|
| 140 |
+
"loss": 0.1869,
|
| 141 |
+
"step": 704
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.11,
|
| 145 |
+
"learning_rate": 9.106699751861042e-07,
|
| 146 |
+
"loss": 0.2196,
|
| 147 |
+
"step": 736
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.11,
|
| 151 |
+
"learning_rate": 9.503722084367245e-07,
|
| 152 |
+
"loss": 0.0936,
|
| 153 |
+
"step": 768
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.12,
|
| 157 |
+
"learning_rate": 9.90074441687345e-07,
|
| 158 |
+
"loss": 0.1839,
|
| 159 |
+
"step": 800
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.12,
|
| 163 |
+
"learning_rate": 9.980994615140957e-07,
|
| 164 |
+
"loss": 0.1675,
|
| 165 |
+
"step": 832
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.13,
|
| 169 |
+
"learning_rate": 9.955654101995564e-07,
|
| 170 |
+
"loss": 0.1595,
|
| 171 |
+
"step": 864
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13,
|
| 175 |
+
"learning_rate": 9.930313588850174e-07,
|
| 176 |
+
"loss": 0.1556,
|
| 177 |
+
"step": 896
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.14,
|
| 181 |
+
"learning_rate": 9.905764966740576e-07,
|
| 182 |
+
"loss": 0.1711,
|
| 183 |
+
"step": 928
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.14,
|
| 187 |
+
"learning_rate": 9.880424453595185e-07,
|
| 188 |
+
"loss": 0.1777,
|
| 189 |
+
"step": 960
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.15,
|
| 193 |
+
"learning_rate": 9.855083940449792e-07,
|
| 194 |
+
"loss": 0.2031,
|
| 195 |
+
"step": 992
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.15,
|
| 199 |
+
"learning_rate": 9.829743427304402e-07,
|
| 200 |
+
"loss": 0.1529,
|
| 201 |
+
"step": 1024
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.16,
|
| 205 |
+
"learning_rate": 9.804402914159011e-07,
|
| 206 |
+
"loss": 0.1365,
|
| 207 |
+
"step": 1056
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.16,
|
| 211 |
+
"learning_rate": 9.77906240101362e-07,
|
| 212 |
+
"loss": 0.1922,
|
| 213 |
+
"step": 1088
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.17,
|
| 217 |
+
"learning_rate": 9.75372188786823e-07,
|
| 218 |
+
"loss": 0.133,
|
| 219 |
+
"step": 1120
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.17,
|
| 223 |
+
"learning_rate": 9.728381374722837e-07,
|
| 224 |
+
"loss": 0.1692,
|
| 225 |
+
"step": 1152
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.18,
|
| 229 |
+
"learning_rate": 9.703040861577447e-07,
|
| 230 |
+
"loss": 0.1022,
|
| 231 |
+
"step": 1184
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.18,
|
| 235 |
+
"learning_rate": 9.677700348432054e-07,
|
| 236 |
+
"loss": 0.2052,
|
| 237 |
+
"step": 1216
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.19,
|
| 241 |
+
"learning_rate": 9.652359835286664e-07,
|
| 242 |
+
"loss": 0.1546,
|
| 243 |
+
"step": 1248
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.19,
|
| 247 |
+
"learning_rate": 9.627019322141273e-07,
|
| 248 |
+
"loss": 0.149,
|
| 249 |
+
"step": 1280
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.2,
|
| 253 |
+
"learning_rate": 9.601678808995883e-07,
|
| 254 |
+
"loss": 0.1281,
|
| 255 |
+
"step": 1312
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.2,
|
| 259 |
+
"learning_rate": 9.57633829585049e-07,
|
| 260 |
+
"loss": 0.1437,
|
| 261 |
+
"step": 1344
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.2,
|
| 265 |
+
"learning_rate": 9.5509977827051e-07,
|
| 266 |
+
"loss": 0.2097,
|
| 267 |
+
"step": 1376
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.21,
|
| 271 |
+
"learning_rate": 9.525657269559708e-07,
|
| 272 |
+
"loss": 0.1308,
|
| 273 |
+
"step": 1408
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.21,
|
| 277 |
+
"learning_rate": 9.500316756414317e-07,
|
| 278 |
+
"loss": 0.1691,
|
| 279 |
+
"step": 1440
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.22,
|
| 283 |
+
"learning_rate": 9.474976243268927e-07,
|
| 284 |
+
"loss": 0.2319,
|
| 285 |
+
"step": 1472
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.22,
|
| 289 |
+
"learning_rate": 9.449635730123534e-07,
|
| 290 |
+
"loss": 0.2226,
|
| 291 |
+
"step": 1504
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.23,
|
| 295 |
+
"learning_rate": 9.424295216978143e-07,
|
| 296 |
+
"loss": 0.1789,
|
| 297 |
+
"step": 1536
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.23,
|
| 301 |
+
"learning_rate": 9.398954703832752e-07,
|
| 302 |
+
"loss": 0.1932,
|
| 303 |
+
"step": 1568
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.24,
|
| 307 |
+
"learning_rate": 9.373614190687361e-07,
|
| 308 |
+
"loss": 0.1718,
|
| 309 |
+
"step": 1600
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.24,
|
| 313 |
+
"learning_rate": 9.34827367754197e-07,
|
| 314 |
+
"loss": 0.156,
|
| 315 |
+
"step": 1632
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.25,
|
| 319 |
+
"learning_rate": 9.322933164396578e-07,
|
| 320 |
+
"loss": 0.1512,
|
| 321 |
+
"step": 1664
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.25,
|
| 325 |
+
"learning_rate": 9.297592651251187e-07,
|
| 326 |
+
"loss": 0.0968,
|
| 327 |
+
"step": 1696
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.26,
|
| 331 |
+
"learning_rate": 9.272252138105796e-07,
|
| 332 |
+
"loss": 0.0932,
|
| 333 |
+
"step": 1728
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.26,
|
| 337 |
+
"learning_rate": 9.246911624960405e-07,
|
| 338 |
+
"loss": 0.2464,
|
| 339 |
+
"step": 1760
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27,
|
| 343 |
+
"learning_rate": 9.221571111815014e-07,
|
| 344 |
+
"loss": 0.2036,
|
| 345 |
+
"step": 1792
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.27,
|
| 349 |
+
"learning_rate": 9.196230598669623e-07,
|
| 350 |
+
"loss": 0.1245,
|
| 351 |
+
"step": 1824
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.28,
|
| 355 |
+
"learning_rate": 9.170890085524232e-07,
|
| 356 |
+
"loss": 0.1097,
|
| 357 |
+
"step": 1856
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.28,
|
| 361 |
+
"learning_rate": 9.14554957237884e-07,
|
| 362 |
+
"loss": 0.1844,
|
| 363 |
+
"step": 1888
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.29,
|
| 367 |
+
"learning_rate": 9.120209059233449e-07,
|
| 368 |
+
"loss": 0.1114,
|
| 369 |
+
"step": 1920
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.29,
|
| 373 |
+
"learning_rate": 9.094868546088058e-07,
|
| 374 |
+
"loss": 0.1992,
|
| 375 |
+
"step": 1952
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.3,
|
| 379 |
+
"learning_rate": 9.069528032942667e-07,
|
| 380 |
+
"loss": 0.1721,
|
| 381 |
+
"step": 1984
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.3,
|
| 385 |
+
"learning_rate": 9.044187519797275e-07,
|
| 386 |
+
"loss": 0.1473,
|
| 387 |
+
"step": 2016
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.3,
|
| 391 |
+
"learning_rate": 9.018847006651884e-07,
|
| 392 |
+
"loss": 0.1865,
|
| 393 |
+
"step": 2048
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.31,
|
| 397 |
+
"learning_rate": 8.993506493506493e-07,
|
| 398 |
+
"loss": 0.1583,
|
| 399 |
+
"step": 2080
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.31,
|
| 403 |
+
"learning_rate": 8.968165980361102e-07,
|
| 404 |
+
"loss": 0.1866,
|
| 405 |
+
"step": 2112
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.32,
|
| 409 |
+
"learning_rate": 8.942825467215711e-07,
|
| 410 |
+
"loss": 0.1617,
|
| 411 |
+
"step": 2144
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.32,
|
| 415 |
+
"learning_rate": 8.917484954070319e-07,
|
| 416 |
+
"loss": 0.1189,
|
| 417 |
+
"step": 2176
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.33,
|
| 421 |
+
"learning_rate": 8.892144440924928e-07,
|
| 422 |
+
"loss": 0.148,
|
| 423 |
+
"step": 2208
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.33,
|
| 427 |
+
"learning_rate": 8.866803927779537e-07,
|
| 428 |
+
"loss": 0.131,
|
| 429 |
+
"step": 2240
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.34,
|
| 433 |
+
"learning_rate": 8.841463414634146e-07,
|
| 434 |
+
"loss": 0.2261,
|
| 435 |
+
"step": 2272
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.34,
|
| 439 |
+
"learning_rate": 8.816122901488755e-07,
|
| 440 |
+
"loss": 0.1742,
|
| 441 |
+
"step": 2304
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.35,
|
| 445 |
+
"learning_rate": 8.790782388343364e-07,
|
| 446 |
+
"loss": 0.164,
|
| 447 |
+
"step": 2336
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.35,
|
| 451 |
+
"learning_rate": 8.765441875197972e-07,
|
| 452 |
+
"loss": 0.1161,
|
| 453 |
+
"step": 2368
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.36,
|
| 457 |
+
"learning_rate": 8.74010136205258e-07,
|
| 458 |
+
"loss": 0.1636,
|
| 459 |
+
"step": 2400
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.36,
|
| 463 |
+
"learning_rate": 8.71476084890719e-07,
|
| 464 |
+
"loss": 0.2416,
|
| 465 |
+
"step": 2432
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.37,
|
| 469 |
+
"learning_rate": 8.689420335761799e-07,
|
| 470 |
+
"loss": 0.1632,
|
| 471 |
+
"step": 2464
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.37,
|
| 475 |
+
"learning_rate": 8.664079822616408e-07,
|
| 476 |
+
"loss": 0.1477,
|
| 477 |
+
"step": 2496
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.38,
|
| 481 |
+
"learning_rate": 8.638739309471016e-07,
|
| 482 |
+
"loss": 0.2083,
|
| 483 |
+
"step": 2528
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.38,
|
| 487 |
+
"learning_rate": 8.613398796325625e-07,
|
| 488 |
+
"loss": 0.1599,
|
| 489 |
+
"step": 2560
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.39,
|
| 493 |
+
"learning_rate": 8.588058283180234e-07,
|
| 494 |
+
"loss": 0.1817,
|
| 495 |
+
"step": 2592
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.39,
|
| 499 |
+
"learning_rate": 8.562717770034843e-07,
|
| 500 |
+
"loss": 0.1005,
|
| 501 |
+
"step": 2624
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.4,
|
| 505 |
+
"learning_rate": 8.537377256889452e-07,
|
| 506 |
+
"loss": 0.168,
|
| 507 |
+
"step": 2656
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.4,
|
| 511 |
+
"learning_rate": 8.51203674374406e-07,
|
| 512 |
+
"loss": 0.2418,
|
| 513 |
+
"step": 2688
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4,
|
| 517 |
+
"learning_rate": 8.486696230598669e-07,
|
| 518 |
+
"loss": 0.1881,
|
| 519 |
+
"step": 2720
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.41,
|
| 523 |
+
"learning_rate": 8.461355717453278e-07,
|
| 524 |
+
"loss": 0.1829,
|
| 525 |
+
"step": 2752
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.41,
|
| 529 |
+
"learning_rate": 8.436015204307887e-07,
|
| 530 |
+
"loss": 0.1073,
|
| 531 |
+
"step": 2784
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.42,
|
| 535 |
+
"learning_rate": 8.410674691162496e-07,
|
| 536 |
+
"loss": 0.1324,
|
| 537 |
+
"step": 2816
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.42,
|
| 541 |
+
"learning_rate": 8.385334178017105e-07,
|
| 542 |
+
"loss": 0.2077,
|
| 543 |
+
"step": 2848
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.43,
|
| 547 |
+
"learning_rate": 8.359993664871713e-07,
|
| 548 |
+
"loss": 0.2248,
|
| 549 |
+
"step": 2880
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.43,
|
| 553 |
+
"learning_rate": 8.334653151726322e-07,
|
| 554 |
+
"loss": 0.1337,
|
| 555 |
+
"step": 2912
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.44,
|
| 559 |
+
"learning_rate": 8.30931263858093e-07,
|
| 560 |
+
"loss": 0.1906,
|
| 561 |
+
"step": 2944
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.44,
|
| 565 |
+
"learning_rate": 8.28397212543554e-07,
|
| 566 |
+
"loss": 0.1893,
|
| 567 |
+
"step": 2976
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.45,
|
| 571 |
+
"learning_rate": 8.259423503325942e-07,
|
| 572 |
+
"loss": 0.2029,
|
| 573 |
+
"step": 3008
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.45,
|
| 577 |
+
"learning_rate": 8.234082990180551e-07,
|
| 578 |
+
"loss": 0.157,
|
| 579 |
+
"step": 3040
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.46,
|
| 583 |
+
"learning_rate": 8.208742477035159e-07,
|
| 584 |
+
"loss": 0.1433,
|
| 585 |
+
"step": 3072
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.46,
|
| 589 |
+
"learning_rate": 8.183401963889769e-07,
|
| 590 |
+
"loss": 0.1689,
|
| 591 |
+
"step": 3104
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.47,
|
| 595 |
+
"learning_rate": 8.158061450744377e-07,
|
| 596 |
+
"loss": 0.2012,
|
| 597 |
+
"step": 3136
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.47,
|
| 601 |
+
"learning_rate": 8.132720937598986e-07,
|
| 602 |
+
"loss": 0.175,
|
| 603 |
+
"step": 3168
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.48,
|
| 607 |
+
"learning_rate": 8.107380424453595e-07,
|
| 608 |
+
"loss": 0.1961,
|
| 609 |
+
"step": 3200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.48,
|
| 613 |
+
"learning_rate": 8.082039911308203e-07,
|
| 614 |
+
"loss": 0.2547,
|
| 615 |
+
"step": 3232
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.49,
|
| 619 |
+
"learning_rate": 8.056699398162813e-07,
|
| 620 |
+
"loss": 0.1935,
|
| 621 |
+
"step": 3264
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.49,
|
| 625 |
+
"learning_rate": 8.031358885017421e-07,
|
| 626 |
+
"loss": 0.2149,
|
| 627 |
+
"step": 3296
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.5,
|
| 631 |
+
"learning_rate": 8.00601837187203e-07,
|
| 632 |
+
"loss": 0.1809,
|
| 633 |
+
"step": 3328
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5,
|
| 637 |
+
"learning_rate": 7.980677858726639e-07,
|
| 638 |
+
"loss": 0.2072,
|
| 639 |
+
"step": 3360
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5,
|
| 643 |
+
"learning_rate": 7.955337345581247e-07,
|
| 644 |
+
"loss": 0.2116,
|
| 645 |
+
"step": 3392
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.51,
|
| 649 |
+
"learning_rate": 7.929996832435857e-07,
|
| 650 |
+
"loss": 0.1737,
|
| 651 |
+
"step": 3424
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.51,
|
| 655 |
+
"learning_rate": 7.904656319290464e-07,
|
| 656 |
+
"loss": 0.2219,
|
| 657 |
+
"step": 3456
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.52,
|
| 661 |
+
"learning_rate": 7.879315806145074e-07,
|
| 662 |
+
"loss": 0.1849,
|
| 663 |
+
"step": 3488
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.52,
|
| 667 |
+
"learning_rate": 7.853975292999683e-07,
|
| 668 |
+
"loss": 0.1884,
|
| 669 |
+
"step": 3520
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.53,
|
| 673 |
+
"learning_rate": 7.828634779854292e-07,
|
| 674 |
+
"loss": 0.2192,
|
| 675 |
+
"step": 3552
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.53,
|
| 679 |
+
"learning_rate": 7.803294266708901e-07,
|
| 680 |
+
"loss": 0.1958,
|
| 681 |
+
"step": 3584
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.54,
|
| 685 |
+
"learning_rate": 7.777953753563509e-07,
|
| 686 |
+
"loss": 0.1433,
|
| 687 |
+
"step": 3616
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.54,
|
| 691 |
+
"learning_rate": 7.752613240418118e-07,
|
| 692 |
+
"loss": 0.2151,
|
| 693 |
+
"step": 3648
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.55,
|
| 697 |
+
"learning_rate": 7.727272727272727e-07,
|
| 698 |
+
"loss": 0.1675,
|
| 699 |
+
"step": 3680
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.55,
|
| 703 |
+
"learning_rate": 7.701932214127336e-07,
|
| 704 |
+
"loss": 0.1586,
|
| 705 |
+
"step": 3712
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.56,
|
| 709 |
+
"learning_rate": 7.676591700981945e-07,
|
| 710 |
+
"loss": 0.2881,
|
| 711 |
+
"step": 3744
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.56,
|
| 715 |
+
"learning_rate": 7.651251187836553e-07,
|
| 716 |
+
"loss": 0.196,
|
| 717 |
+
"step": 3776
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.57,
|
| 721 |
+
"learning_rate": 7.625910674691162e-07,
|
| 722 |
+
"loss": 0.1285,
|
| 723 |
+
"step": 3808
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.57,
|
| 727 |
+
"learning_rate": 7.60057016154577e-07,
|
| 728 |
+
"loss": 0.2262,
|
| 729 |
+
"step": 3840
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.58,
|
| 733 |
+
"learning_rate": 7.57522964840038e-07,
|
| 734 |
+
"loss": 0.2309,
|
| 735 |
+
"step": 3872
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.58,
|
| 739 |
+
"learning_rate": 7.549889135254989e-07,
|
| 740 |
+
"loss": 0.1533,
|
| 741 |
+
"step": 3904
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.59,
|
| 745 |
+
"learning_rate": 7.524548622109597e-07,
|
| 746 |
+
"loss": 0.1297,
|
| 747 |
+
"step": 3936
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.59,
|
| 751 |
+
"learning_rate": 7.499208108964206e-07,
|
| 752 |
+
"loss": 0.1808,
|
| 753 |
+
"step": 3968
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.6,
|
| 757 |
+
"learning_rate": 7.473867595818814e-07,
|
| 758 |
+
"loss": 0.2401,
|
| 759 |
+
"step": 4000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6,
|
| 763 |
+
"learning_rate": 7.448527082673424e-07,
|
| 764 |
+
"loss": 0.2507,
|
| 765 |
+
"step": 4032
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.61,
|
| 769 |
+
"learning_rate": 7.423186569528033e-07,
|
| 770 |
+
"loss": 0.1562,
|
| 771 |
+
"step": 4064
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 0.61,
|
| 775 |
+
"learning_rate": 7.397846056382642e-07,
|
| 776 |
+
"loss": 0.1912,
|
| 777 |
+
"step": 4096
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.61,
|
| 781 |
+
"learning_rate": 7.373297434273043e-07,
|
| 782 |
+
"loss": 0.1703,
|
| 783 |
+
"step": 4128
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.62,
|
| 787 |
+
"learning_rate": 7.347956921127653e-07,
|
| 788 |
+
"loss": 0.1471,
|
| 789 |
+
"step": 4160
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.62,
|
| 793 |
+
"learning_rate": 7.322616407982262e-07,
|
| 794 |
+
"loss": 0.1539,
|
| 795 |
+
"step": 4192
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.63,
|
| 799 |
+
"learning_rate": 7.297275894836869e-07,
|
| 800 |
+
"loss": 0.1521,
|
| 801 |
+
"step": 4224
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.63,
|
| 805 |
+
"learning_rate": 7.271935381691479e-07,
|
| 806 |
+
"loss": 0.2623,
|
| 807 |
+
"step": 4256
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.64,
|
| 811 |
+
"learning_rate": 7.246594868546087e-07,
|
| 812 |
+
"loss": 0.1753,
|
| 813 |
+
"step": 4288
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.64,
|
| 817 |
+
"learning_rate": 7.221254355400697e-07,
|
| 818 |
+
"loss": 0.1945,
|
| 819 |
+
"step": 4320
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.65,
|
| 823 |
+
"learning_rate": 7.195913842255306e-07,
|
| 824 |
+
"loss": 0.2153,
|
| 825 |
+
"step": 4352
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.65,
|
| 829 |
+
"learning_rate": 7.170573329109915e-07,
|
| 830 |
+
"loss": 0.2841,
|
| 831 |
+
"step": 4384
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.66,
|
| 835 |
+
"learning_rate": 7.145232815964523e-07,
|
| 836 |
+
"loss": 0.1759,
|
| 837 |
+
"step": 4416
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.66,
|
| 841 |
+
"learning_rate": 7.119892302819131e-07,
|
| 842 |
+
"loss": 0.2214,
|
| 843 |
+
"step": 4448
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.67,
|
| 847 |
+
"learning_rate": 7.094551789673741e-07,
|
| 848 |
+
"loss": 0.188,
|
| 849 |
+
"step": 4480
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.67,
|
| 853 |
+
"learning_rate": 7.069211276528349e-07,
|
| 854 |
+
"loss": 0.1579,
|
| 855 |
+
"step": 4512
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.68,
|
| 859 |
+
"learning_rate": 7.043870763382959e-07,
|
| 860 |
+
"loss": 0.2213,
|
| 861 |
+
"step": 4544
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.68,
|
| 865 |
+
"learning_rate": 7.018530250237567e-07,
|
| 866 |
+
"loss": 0.2042,
|
| 867 |
+
"step": 4576
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.69,
|
| 871 |
+
"learning_rate": 6.993189737092175e-07,
|
| 872 |
+
"loss": 0.1852,
|
| 873 |
+
"step": 4608
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.69,
|
| 877 |
+
"learning_rate": 6.967849223946785e-07,
|
| 878 |
+
"loss": 0.1716,
|
| 879 |
+
"step": 4640
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.7,
|
| 883 |
+
"learning_rate": 6.942508710801393e-07,
|
| 884 |
+
"loss": 0.1645,
|
| 885 |
+
"step": 4672
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.7,
|
| 889 |
+
"learning_rate": 6.917168197656003e-07,
|
| 890 |
+
"loss": 0.1986,
|
| 891 |
+
"step": 4704
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.71,
|
| 895 |
+
"learning_rate": 6.89182768451061e-07,
|
| 896 |
+
"loss": 0.2531,
|
| 897 |
+
"step": 4736
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.71,
|
| 901 |
+
"learning_rate": 6.866487171365219e-07,
|
| 902 |
+
"loss": 0.1792,
|
| 903 |
+
"step": 4768
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.71,
|
| 907 |
+
"learning_rate": 6.841146658219829e-07,
|
| 908 |
+
"loss": 0.1843,
|
| 909 |
+
"step": 4800
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.72,
|
| 913 |
+
"learning_rate": 6.815806145074437e-07,
|
| 914 |
+
"loss": 0.2175,
|
| 915 |
+
"step": 4832
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.72,
|
| 919 |
+
"learning_rate": 6.790465631929047e-07,
|
| 920 |
+
"loss": 0.2083,
|
| 921 |
+
"step": 4864
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.73,
|
| 925 |
+
"learning_rate": 6.765125118783655e-07,
|
| 926 |
+
"loss": 0.1729,
|
| 927 |
+
"step": 4896
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.73,
|
| 931 |
+
"learning_rate": 6.739784605638263e-07,
|
| 932 |
+
"loss": 0.1849,
|
| 933 |
+
"step": 4928
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.74,
|
| 937 |
+
"learning_rate": 6.714444092492873e-07,
|
| 938 |
+
"loss": 0.2374,
|
| 939 |
+
"step": 4960
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.74,
|
| 943 |
+
"learning_rate": 6.689103579347481e-07,
|
| 944 |
+
"loss": 0.241,
|
| 945 |
+
"step": 4992
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.75,
|
| 949 |
+
"learning_rate": 6.663763066202091e-07,
|
| 950 |
+
"loss": 0.1853,
|
| 951 |
+
"step": 5024
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 0.75,
|
| 955 |
+
"learning_rate": 6.638422553056699e-07,
|
| 956 |
+
"loss": 0.1957,
|
| 957 |
+
"step": 5056
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 0.76,
|
| 961 |
+
"learning_rate": 6.613082039911308e-07,
|
| 962 |
+
"loss": 0.2052,
|
| 963 |
+
"step": 5088
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.76,
|
| 967 |
+
"learning_rate": 6.587741526765917e-07,
|
| 968 |
+
"loss": 0.2321,
|
| 969 |
+
"step": 5120
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.77,
|
| 973 |
+
"learning_rate": 6.562401013620525e-07,
|
| 974 |
+
"loss": 0.1804,
|
| 975 |
+
"step": 5152
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.77,
|
| 979 |
+
"learning_rate": 6.537060500475135e-07,
|
| 980 |
+
"loss": 0.1842,
|
| 981 |
+
"step": 5184
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.78,
|
| 985 |
+
"learning_rate": 6.511719987329743e-07,
|
| 986 |
+
"loss": 0.2388,
|
| 987 |
+
"step": 5216
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.78,
|
| 991 |
+
"learning_rate": 6.486379474184352e-07,
|
| 992 |
+
"loss": 0.2417,
|
| 993 |
+
"step": 5248
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.79,
|
| 997 |
+
"learning_rate": 6.46103896103896e-07,
|
| 998 |
+
"loss": 0.2224,
|
| 999 |
+
"step": 5280
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.79,
|
| 1003 |
+
"learning_rate": 6.435698447893569e-07,
|
| 1004 |
+
"loss": 0.2029,
|
| 1005 |
+
"step": 5312
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 0.8,
|
| 1009 |
+
"learning_rate": 6.410357934748179e-07,
|
| 1010 |
+
"loss": 0.2807,
|
| 1011 |
+
"step": 5344
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8,
|
| 1015 |
+
"learning_rate": 6.385017421602787e-07,
|
| 1016 |
+
"loss": 0.192,
|
| 1017 |
+
"step": 5376
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.81,
|
| 1021 |
+
"learning_rate": 6.359676908457397e-07,
|
| 1022 |
+
"loss": 0.1848,
|
| 1023 |
+
"step": 5408
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.81,
|
| 1027 |
+
"learning_rate": 6.334336395312004e-07,
|
| 1028 |
+
"loss": 0.2143,
|
| 1029 |
+
"step": 5440
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.81,
|
| 1033 |
+
"learning_rate": 6.308995882166613e-07,
|
| 1034 |
+
"loss": 0.2421,
|
| 1035 |
+
"step": 5472
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.82,
|
| 1039 |
+
"learning_rate": 6.283655369021223e-07,
|
| 1040 |
+
"loss": 0.1724,
|
| 1041 |
+
"step": 5504
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.82,
|
| 1045 |
+
"learning_rate": 6.258314855875831e-07,
|
| 1046 |
+
"loss": 0.1207,
|
| 1047 |
+
"step": 5536
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 0.83,
|
| 1051 |
+
"learning_rate": 6.232974342730441e-07,
|
| 1052 |
+
"loss": 0.2259,
|
| 1053 |
+
"step": 5568
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.83,
|
| 1057 |
+
"learning_rate": 6.207633829585048e-07,
|
| 1058 |
+
"loss": 0.2504,
|
| 1059 |
+
"step": 5600
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.84,
|
| 1063 |
+
"learning_rate": 6.182293316439658e-07,
|
| 1064 |
+
"loss": 0.188,
|
| 1065 |
+
"step": 5632
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.84,
|
| 1069 |
+
"learning_rate": 6.156952803294266e-07,
|
| 1070 |
+
"loss": 0.1893,
|
| 1071 |
+
"step": 5664
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.85,
|
| 1075 |
+
"learning_rate": 6.131612290148875e-07,
|
| 1076 |
+
"loss": 0.1905,
|
| 1077 |
+
"step": 5696
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 0.85,
|
| 1081 |
+
"learning_rate": 6.106271777003485e-07,
|
| 1082 |
+
"loss": 0.2594,
|
| 1083 |
+
"step": 5728
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 0.86,
|
| 1087 |
+
"learning_rate": 6.080931263858092e-07,
|
| 1088 |
+
"loss": 0.3084,
|
| 1089 |
+
"step": 5760
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.86,
|
| 1093 |
+
"learning_rate": 6.055590750712702e-07,
|
| 1094 |
+
"loss": 0.1925,
|
| 1095 |
+
"step": 5792
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.87,
|
| 1099 |
+
"learning_rate": 6.03025023756731e-07,
|
| 1100 |
+
"loss": 0.186,
|
| 1101 |
+
"step": 5824
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.87,
|
| 1105 |
+
"learning_rate": 6.004909724421919e-07,
|
| 1106 |
+
"loss": 0.2302,
|
| 1107 |
+
"step": 5856
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.88,
|
| 1111 |
+
"learning_rate": 5.979569211276529e-07,
|
| 1112 |
+
"loss": 0.1371,
|
| 1113 |
+
"step": 5888
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.88,
|
| 1117 |
+
"learning_rate": 5.954228698131137e-07,
|
| 1118 |
+
"loss": 0.231,
|
| 1119 |
+
"step": 5920
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.89,
|
| 1123 |
+
"learning_rate": 5.928888184985746e-07,
|
| 1124 |
+
"loss": 0.2012,
|
| 1125 |
+
"step": 5952
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.89,
|
| 1129 |
+
"learning_rate": 5.903547671840354e-07,
|
| 1130 |
+
"loss": 0.2006,
|
| 1131 |
+
"step": 5984
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.9,
|
| 1135 |
+
"learning_rate": 5.878207158694963e-07,
|
| 1136 |
+
"loss": 0.215,
|
| 1137 |
+
"step": 6016
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9,
|
| 1141 |
+
"learning_rate": 5.852866645549572e-07,
|
| 1142 |
+
"loss": 0.1471,
|
| 1143 |
+
"step": 6048
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.91,
|
| 1147 |
+
"learning_rate": 5.827526132404181e-07,
|
| 1148 |
+
"loss": 0.2364,
|
| 1149 |
+
"step": 6080
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.91,
|
| 1153 |
+
"learning_rate": 5.80218561925879e-07,
|
| 1154 |
+
"loss": 0.2881,
|
| 1155 |
+
"step": 6112
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.91,
|
| 1159 |
+
"learning_rate": 5.776845106113398e-07,
|
| 1160 |
+
"loss": 0.1536,
|
| 1161 |
+
"step": 6144
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.92,
|
| 1165 |
+
"learning_rate": 5.751504592968008e-07,
|
| 1166 |
+
"loss": 0.2317,
|
| 1167 |
+
"step": 6176
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.92,
|
| 1171 |
+
"learning_rate": 5.726164079822616e-07,
|
| 1172 |
+
"loss": 0.1952,
|
| 1173 |
+
"step": 6208
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.93,
|
| 1177 |
+
"learning_rate": 5.700823566677225e-07,
|
| 1178 |
+
"loss": 0.1602,
|
| 1179 |
+
"step": 6240
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.93,
|
| 1183 |
+
"learning_rate": 5.675483053531834e-07,
|
| 1184 |
+
"loss": 0.212,
|
| 1185 |
+
"step": 6272
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.94,
|
| 1189 |
+
"learning_rate": 5.650142540386442e-07,
|
| 1190 |
+
"loss": 0.2401,
|
| 1191 |
+
"step": 6304
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.94,
|
| 1195 |
+
"learning_rate": 5.624802027241052e-07,
|
| 1196 |
+
"loss": 0.1992,
|
| 1197 |
+
"step": 6336
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.95,
|
| 1201 |
+
"learning_rate": 5.59946151409566e-07,
|
| 1202 |
+
"loss": 0.2616,
|
| 1203 |
+
"step": 6368
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.95,
|
| 1207 |
+
"learning_rate": 5.574121000950269e-07,
|
| 1208 |
+
"loss": 0.146,
|
| 1209 |
+
"step": 6400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.96,
|
| 1213 |
+
"learning_rate": 5.548780487804878e-07,
|
| 1214 |
+
"loss": 0.2081,
|
| 1215 |
+
"step": 6432
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.96,
|
| 1219 |
+
"learning_rate": 5.523439974659486e-07,
|
| 1220 |
+
"loss": 0.207,
|
| 1221 |
+
"step": 6464
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.97,
|
| 1225 |
+
"learning_rate": 5.498099461514096e-07,
|
| 1226 |
+
"loss": 0.2631,
|
| 1227 |
+
"step": 6496
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.97,
|
| 1231 |
+
"learning_rate": 5.472758948368704e-07,
|
| 1232 |
+
"loss": 0.1721,
|
| 1233 |
+
"step": 6528
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.98,
|
| 1237 |
+
"learning_rate": 5.447418435223313e-07,
|
| 1238 |
+
"loss": 0.1908,
|
| 1239 |
+
"step": 6560
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.98,
|
| 1243 |
+
"learning_rate": 5.422869813113715e-07,
|
| 1244 |
+
"loss": 0.2238,
|
| 1245 |
+
"step": 6592
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.99,
|
| 1249 |
+
"learning_rate": 5.397529299968325e-07,
|
| 1250 |
+
"loss": 0.2524,
|
| 1251 |
+
"step": 6624
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.99,
|
| 1255 |
+
"learning_rate": 5.372188786822933e-07,
|
| 1256 |
+
"loss": 0.1968,
|
| 1257 |
+
"step": 6656
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.0,
|
| 1261 |
+
"learning_rate": 5.346848273677542e-07,
|
| 1262 |
+
"loss": 0.2379,
|
| 1263 |
+
"step": 6688
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 1.0,
|
| 1267 |
+
"learning_rate": 5.32150776053215e-07,
|
| 1268 |
+
"loss": 0.246,
|
| 1269 |
+
"step": 6720
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.01,
|
| 1273 |
+
"learning_rate": 5.296167247386759e-07,
|
| 1274 |
+
"loss": 0.104,
|
| 1275 |
+
"step": 6752
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.01,
|
| 1279 |
+
"learning_rate": 5.270826734241369e-07,
|
| 1280 |
+
"loss": 0.102,
|
| 1281 |
+
"step": 6784
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 1.01,
|
| 1285 |
+
"learning_rate": 5.245486221095977e-07,
|
| 1286 |
+
"loss": 0.1432,
|
| 1287 |
+
"step": 6816
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 1.02,
|
| 1291 |
+
"learning_rate": 5.220145707950586e-07,
|
| 1292 |
+
"loss": 0.1262,
|
| 1293 |
+
"step": 6848
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 1.02,
|
| 1297 |
+
"learning_rate": 5.194805194805194e-07,
|
| 1298 |
+
"loss": 0.1299,
|
| 1299 |
+
"step": 6880
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.03,
|
| 1303 |
+
"learning_rate": 5.169464681659803e-07,
|
| 1304 |
+
"loss": 0.1319,
|
| 1305 |
+
"step": 6912
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.03,
|
| 1309 |
+
"learning_rate": 5.144124168514412e-07,
|
| 1310 |
+
"loss": 0.105,
|
| 1311 |
+
"step": 6944
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.04,
|
| 1315 |
+
"learning_rate": 5.118783655369021e-07,
|
| 1316 |
+
"loss": 0.1233,
|
| 1317 |
+
"step": 6976
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.04,
|
| 1321 |
+
"learning_rate": 5.09344314222363e-07,
|
| 1322 |
+
"loss": 0.0922,
|
| 1323 |
+
"step": 7008
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 1.05,
|
| 1327 |
+
"learning_rate": 5.068102629078239e-07,
|
| 1328 |
+
"loss": 0.144,
|
| 1329 |
+
"step": 7040
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.05,
|
| 1333 |
+
"learning_rate": 5.042762115932847e-07,
|
| 1334 |
+
"loss": 0.1828,
|
| 1335 |
+
"step": 7072
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 1.06,
|
| 1339 |
+
"learning_rate": 5.017421602787456e-07,
|
| 1340 |
+
"loss": 0.1097,
|
| 1341 |
+
"step": 7104
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.06,
|
| 1345 |
+
"learning_rate": 4.992081089642065e-07,
|
| 1346 |
+
"loss": 0.1867,
|
| 1347 |
+
"step": 7136
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.07,
|
| 1351 |
+
"learning_rate": 4.966740576496675e-07,
|
| 1352 |
+
"loss": 0.1338,
|
| 1353 |
+
"step": 7168
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.07,
|
| 1357 |
+
"learning_rate": 4.941400063351283e-07,
|
| 1358 |
+
"loss": 0.1535,
|
| 1359 |
+
"step": 7200
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.08,
|
| 1363 |
+
"learning_rate": 4.916059550205891e-07,
|
| 1364 |
+
"loss": 0.1861,
|
| 1365 |
+
"step": 7232
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 1.08,
|
| 1369 |
+
"learning_rate": 4.8907190370605e-07,
|
| 1370 |
+
"loss": 0.1284,
|
| 1371 |
+
"step": 7264
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 1.09,
|
| 1375 |
+
"learning_rate": 4.865378523915109e-07,
|
| 1376 |
+
"loss": 0.1037,
|
| 1377 |
+
"step": 7296
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.09,
|
| 1381 |
+
"learning_rate": 4.840038010769719e-07,
|
| 1382 |
+
"loss": 0.1217,
|
| 1383 |
+
"step": 7328
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.1,
|
| 1387 |
+
"learning_rate": 4.814697497624327e-07,
|
| 1388 |
+
"loss": 0.1469,
|
| 1389 |
+
"step": 7360
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.1,
|
| 1393 |
+
"learning_rate": 4.789356984478935e-07,
|
| 1394 |
+
"loss": 0.1218,
|
| 1395 |
+
"step": 7392
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.11,
|
| 1399 |
+
"learning_rate": 4.764016471333545e-07,
|
| 1400 |
+
"loss": 0.1486,
|
| 1401 |
+
"step": 7424
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.11,
|
| 1405 |
+
"learning_rate": 4.738675958188153e-07,
|
| 1406 |
+
"loss": 0.0796,
|
| 1407 |
+
"step": 7456
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 1.11,
|
| 1411 |
+
"learning_rate": 4.7141273360785554e-07,
|
| 1412 |
+
"loss": 0.1163,
|
| 1413 |
+
"step": 7488
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.12,
|
| 1417 |
+
"learning_rate": 4.688786822933164e-07,
|
| 1418 |
+
"loss": 0.0821,
|
| 1419 |
+
"step": 7520
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 1.12,
|
| 1423 |
+
"learning_rate": 4.663446309787773e-07,
|
| 1424 |
+
"loss": 0.1701,
|
| 1425 |
+
"step": 7552
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 1.13,
|
| 1429 |
+
"learning_rate": 4.638105796642382e-07,
|
| 1430 |
+
"loss": 0.1002,
|
| 1431 |
+
"step": 7584
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.13,
|
| 1435 |
+
"learning_rate": 4.613557174532784e-07,
|
| 1436 |
+
"loss": 0.0914,
|
| 1437 |
+
"step": 7616
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.14,
|
| 1441 |
+
"learning_rate": 4.5882166613873927e-07,
|
| 1442 |
+
"loss": 0.1374,
|
| 1443 |
+
"step": 7648
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.14,
|
| 1447 |
+
"learning_rate": 4.562876148242002e-07,
|
| 1448 |
+
"loss": 0.1142,
|
| 1449 |
+
"step": 7680
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.15,
|
| 1453 |
+
"learning_rate": 4.5375356350966105e-07,
|
| 1454 |
+
"loss": 0.1351,
|
| 1455 |
+
"step": 7712
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.15,
|
| 1459 |
+
"learning_rate": 4.5121951219512194e-07,
|
| 1460 |
+
"loss": 0.1513,
|
| 1461 |
+
"step": 7744
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 1.16,
|
| 1465 |
+
"learning_rate": 4.486854608805828e-07,
|
| 1466 |
+
"loss": 0.0998,
|
| 1467 |
+
"step": 7776
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 1.16,
|
| 1471 |
+
"learning_rate": 4.461514095660437e-07,
|
| 1472 |
+
"loss": 0.1115,
|
| 1473 |
+
"step": 7808
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 1.17,
|
| 1477 |
+
"learning_rate": 4.4361735825150457e-07,
|
| 1478 |
+
"loss": 0.1373,
|
| 1479 |
+
"step": 7840
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.17,
|
| 1483 |
+
"learning_rate": 4.4108330693696546e-07,
|
| 1484 |
+
"loss": 0.1614,
|
| 1485 |
+
"step": 7872
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.18,
|
| 1489 |
+
"learning_rate": 4.3854925562242635e-07,
|
| 1490 |
+
"loss": 0.0826,
|
| 1491 |
+
"step": 7904
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 1.18,
|
| 1495 |
+
"learning_rate": 4.3601520430788724e-07,
|
| 1496 |
+
"loss": 0.1129,
|
| 1497 |
+
"step": 7936
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 1.19,
|
| 1501 |
+
"learning_rate": 4.334811529933481e-07,
|
| 1502 |
+
"loss": 0.0825,
|
| 1503 |
+
"step": 7968
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 1.19,
|
| 1507 |
+
"learning_rate": 4.3094710167880897e-07,
|
| 1508 |
+
"loss": 0.1497,
|
| 1509 |
+
"step": 8000
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 1.2,
|
| 1513 |
+
"learning_rate": 4.284130503642698e-07,
|
| 1514 |
+
"loss": 0.1575,
|
| 1515 |
+
"step": 8032
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 1.2,
|
| 1519 |
+
"learning_rate": 4.2587899904973075e-07,
|
| 1520 |
+
"loss": 0.1377,
|
| 1521 |
+
"step": 8064
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.21,
|
| 1525 |
+
"learning_rate": 4.2334494773519165e-07,
|
| 1526 |
+
"loss": 0.152,
|
| 1527 |
+
"step": 8096
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.21,
|
| 1531 |
+
"learning_rate": 4.208108964206525e-07,
|
| 1532 |
+
"loss": 0.1142,
|
| 1533 |
+
"step": 8128
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 1.21,
|
| 1537 |
+
"learning_rate": 4.182768451061134e-07,
|
| 1538 |
+
"loss": 0.1458,
|
| 1539 |
+
"step": 8160
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 1.22,
|
| 1543 |
+
"learning_rate": 4.1574279379157427e-07,
|
| 1544 |
+
"loss": 0.1216,
|
| 1545 |
+
"step": 8192
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 1.22,
|
| 1549 |
+
"learning_rate": 4.132087424770351e-07,
|
| 1550 |
+
"loss": 0.112,
|
| 1551 |
+
"step": 8224
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 1.23,
|
| 1555 |
+
"learning_rate": 4.1067469116249605e-07,
|
| 1556 |
+
"loss": 0.1517,
|
| 1557 |
+
"step": 8256
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 1.23,
|
| 1561 |
+
"learning_rate": 4.081406398479569e-07,
|
| 1562 |
+
"loss": 0.1046,
|
| 1563 |
+
"step": 8288
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.24,
|
| 1567 |
+
"learning_rate": 4.056065885334178e-07,
|
| 1568 |
+
"loss": 0.1538,
|
| 1569 |
+
"step": 8320
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.24,
|
| 1573 |
+
"learning_rate": 4.030725372188787e-07,
|
| 1574 |
+
"loss": 0.1734,
|
| 1575 |
+
"step": 8352
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 1.25,
|
| 1579 |
+
"learning_rate": 4.005384859043395e-07,
|
| 1580 |
+
"loss": 0.1236,
|
| 1581 |
+
"step": 8384
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 1.25,
|
| 1585 |
+
"learning_rate": 3.980044345898004e-07,
|
| 1586 |
+
"loss": 0.1129,
|
| 1587 |
+
"step": 8416
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 1.26,
|
| 1591 |
+
"learning_rate": 3.9547038327526135e-07,
|
| 1592 |
+
"loss": 0.1193,
|
| 1593 |
+
"step": 8448
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 1.26,
|
| 1597 |
+
"learning_rate": 3.929363319607222e-07,
|
| 1598 |
+
"loss": 0.1124,
|
| 1599 |
+
"step": 8480
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 1.27,
|
| 1603 |
+
"learning_rate": 3.904022806461831e-07,
|
| 1604 |
+
"loss": 0.103,
|
| 1605 |
+
"step": 8512
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 1.27,
|
| 1609 |
+
"learning_rate": 3.878682293316439e-07,
|
| 1610 |
+
"loss": 0.1526,
|
| 1611 |
+
"step": 8544
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.28,
|
| 1615 |
+
"learning_rate": 3.853341780171048e-07,
|
| 1616 |
+
"loss": 0.1747,
|
| 1617 |
+
"step": 8576
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 1.28,
|
| 1621 |
+
"learning_rate": 3.8280012670256575e-07,
|
| 1622 |
+
"loss": 0.0711,
|
| 1623 |
+
"step": 8608
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 1.29,
|
| 1627 |
+
"learning_rate": 3.802660753880266e-07,
|
| 1628 |
+
"loss": 0.131,
|
| 1629 |
+
"step": 8640
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 1.29,
|
| 1633 |
+
"learning_rate": 3.777320240734875e-07,
|
| 1634 |
+
"loss": 0.0695,
|
| 1635 |
+
"step": 8672
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 1.3,
|
| 1639 |
+
"learning_rate": 3.751979727589484e-07,
|
| 1640 |
+
"loss": 0.1176,
|
| 1641 |
+
"step": 8704
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 1.3,
|
| 1645 |
+
"learning_rate": 3.726639214444092e-07,
|
| 1646 |
+
"loss": 0.1141,
|
| 1647 |
+
"step": 8736
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 1.31,
|
| 1651 |
+
"learning_rate": 3.701298701298701e-07,
|
| 1652 |
+
"loss": 0.1437,
|
| 1653 |
+
"step": 8768
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.31,
|
| 1657 |
+
"learning_rate": 3.6759581881533095e-07,
|
| 1658 |
+
"loss": 0.1273,
|
| 1659 |
+
"step": 8800
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 1.31,
|
| 1663 |
+
"learning_rate": 3.650617675007919e-07,
|
| 1664 |
+
"loss": 0.0765,
|
| 1665 |
+
"step": 8832
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 1.32,
|
| 1669 |
+
"learning_rate": 3.625277161862528e-07,
|
| 1670 |
+
"loss": 0.1322,
|
| 1671 |
+
"step": 8864
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 1.32,
|
| 1675 |
+
"learning_rate": 3.599936648717136e-07,
|
| 1676 |
+
"loss": 0.1643,
|
| 1677 |
+
"step": 8896
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 1.33,
|
| 1681 |
+
"learning_rate": 3.574596135571745e-07,
|
| 1682 |
+
"loss": 0.1005,
|
| 1683 |
+
"step": 8928
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 1.33,
|
| 1687 |
+
"learning_rate": 3.549255622426354e-07,
|
| 1688 |
+
"loss": 0.1545,
|
| 1689 |
+
"step": 8960
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 1.34,
|
| 1693 |
+
"learning_rate": 3.523915109280963e-07,
|
| 1694 |
+
"loss": 0.0755,
|
| 1695 |
+
"step": 8992
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.34,
|
| 1699 |
+
"learning_rate": 3.498574596135572e-07,
|
| 1700 |
+
"loss": 0.0917,
|
| 1701 |
+
"step": 9024
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 1.35,
|
| 1705 |
+
"learning_rate": 3.47323408299018e-07,
|
| 1706 |
+
"loss": 0.0893,
|
| 1707 |
+
"step": 9056
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 1.35,
|
| 1711 |
+
"learning_rate": 3.447893569844789e-07,
|
| 1712 |
+
"loss": 0.152,
|
| 1713 |
+
"step": 9088
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 1.36,
|
| 1717 |
+
"learning_rate": 3.422553056699398e-07,
|
| 1718 |
+
"loss": 0.1154,
|
| 1719 |
+
"step": 9120
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 1.36,
|
| 1723 |
+
"learning_rate": 3.3972125435540065e-07,
|
| 1724 |
+
"loss": 0.0817,
|
| 1725 |
+
"step": 9152
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 1.37,
|
| 1729 |
+
"learning_rate": 3.371872030408616e-07,
|
| 1730 |
+
"loss": 0.083,
|
| 1731 |
+
"step": 9184
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 1.37,
|
| 1735 |
+
"learning_rate": 3.346531517263225e-07,
|
| 1736 |
+
"loss": 0.1084,
|
| 1737 |
+
"step": 9216
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.38,
|
| 1741 |
+
"learning_rate": 3.321191004117833e-07,
|
| 1742 |
+
"loss": 0.1514,
|
| 1743 |
+
"step": 9248
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 1.38,
|
| 1747 |
+
"learning_rate": 3.295850490972442e-07,
|
| 1748 |
+
"loss": 0.1195,
|
| 1749 |
+
"step": 9280
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 1.39,
|
| 1753 |
+
"learning_rate": 3.2705099778270505e-07,
|
| 1754 |
+
"loss": 0.0939,
|
| 1755 |
+
"step": 9312
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 1.39,
|
| 1759 |
+
"learning_rate": 3.2451694646816595e-07,
|
| 1760 |
+
"loss": 0.1227,
|
| 1761 |
+
"step": 9344
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 1.4,
|
| 1765 |
+
"learning_rate": 3.219828951536269e-07,
|
| 1766 |
+
"loss": 0.1342,
|
| 1767 |
+
"step": 9376
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 1.4,
|
| 1771 |
+
"learning_rate": 3.1944884383908773e-07,
|
| 1772 |
+
"loss": 0.1214,
|
| 1773 |
+
"step": 9408
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 1.41,
|
| 1777 |
+
"learning_rate": 3.169147925245486e-07,
|
| 1778 |
+
"loss": 0.1624,
|
| 1779 |
+
"step": 9440
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.41,
|
| 1783 |
+
"learning_rate": 3.143807412100095e-07,
|
| 1784 |
+
"loss": 0.1261,
|
| 1785 |
+
"step": 9472
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 1.41,
|
| 1789 |
+
"learning_rate": 3.1184668989547035e-07,
|
| 1790 |
+
"loss": 0.1674,
|
| 1791 |
+
"step": 9504
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 1.42,
|
| 1795 |
+
"learning_rate": 3.0931263858093124e-07,
|
| 1796 |
+
"loss": 0.1613,
|
| 1797 |
+
"step": 9536
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 1.42,
|
| 1801 |
+
"learning_rate": 3.0677858726639213e-07,
|
| 1802 |
+
"loss": 0.1519,
|
| 1803 |
+
"step": 9568
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 1.43,
|
| 1807 |
+
"learning_rate": 3.04244535951853e-07,
|
| 1808 |
+
"loss": 0.1107,
|
| 1809 |
+
"step": 9600
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 1.43,
|
| 1813 |
+
"learning_rate": 3.017104846373139e-07,
|
| 1814 |
+
"loss": 0.0992,
|
| 1815 |
+
"step": 9632
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 1.44,
|
| 1819 |
+
"learning_rate": 2.9917643332277476e-07,
|
| 1820 |
+
"loss": 0.0921,
|
| 1821 |
+
"step": 9664
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.44,
|
| 1825 |
+
"learning_rate": 2.9664238200823565e-07,
|
| 1826 |
+
"loss": 0.1478,
|
| 1827 |
+
"step": 9696
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 1.45,
|
| 1831 |
+
"learning_rate": 2.9410833069369654e-07,
|
| 1832 |
+
"loss": 0.1268,
|
| 1833 |
+
"step": 9728
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 1.45,
|
| 1837 |
+
"learning_rate": 2.9157427937915743e-07,
|
| 1838 |
+
"loss": 0.1086,
|
| 1839 |
+
"step": 9760
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 1.46,
|
| 1843 |
+
"learning_rate": 2.890402280646183e-07,
|
| 1844 |
+
"loss": 0.1913,
|
| 1845 |
+
"step": 9792
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 1.46,
|
| 1849 |
+
"learning_rate": 2.8650617675007916e-07,
|
| 1850 |
+
"loss": 0.1626,
|
| 1851 |
+
"step": 9824
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 1.47,
|
| 1855 |
+
"learning_rate": 2.8397212543554005e-07,
|
| 1856 |
+
"loss": 0.1647,
|
| 1857 |
+
"step": 9856
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 1.47,
|
| 1861 |
+
"learning_rate": 2.8143807412100095e-07,
|
| 1862 |
+
"loss": 0.1245,
|
| 1863 |
+
"step": 9888
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 1.48,
|
| 1867 |
+
"learning_rate": 2.789040228064618e-07,
|
| 1868 |
+
"loss": 0.1326,
|
| 1869 |
+
"step": 9920
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 1.48,
|
| 1873 |
+
"learning_rate": 2.7636997149192273e-07,
|
| 1874 |
+
"loss": 0.1095,
|
| 1875 |
+
"step": 9952
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 1.49,
|
| 1879 |
+
"learning_rate": 2.738359201773836e-07,
|
| 1880 |
+
"loss": 0.1089,
|
| 1881 |
+
"step": 9984
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 1.49,
|
| 1885 |
+
"learning_rate": 2.7130186886284446e-07,
|
| 1886 |
+
"loss": 0.1074,
|
| 1887 |
+
"step": 10016
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 1.5,
|
| 1891 |
+
"learning_rate": 2.6876781754830535e-07,
|
| 1892 |
+
"loss": 0.1084,
|
| 1893 |
+
"step": 10048
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 1.5,
|
| 1897 |
+
"learning_rate": 2.662337662337662e-07,
|
| 1898 |
+
"loss": 0.162,
|
| 1899 |
+
"step": 10080
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 1.51,
|
| 1903 |
+
"learning_rate": 2.636997149192271e-07,
|
| 1904 |
+
"loss": 0.1419,
|
| 1905 |
+
"step": 10112
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 1.51,
|
| 1909 |
+
"learning_rate": 2.61165663604688e-07,
|
| 1910 |
+
"loss": 0.1125,
|
| 1911 |
+
"step": 10144
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 1.51,
|
| 1915 |
+
"learning_rate": 2.5863161229014886e-07,
|
| 1916 |
+
"loss": 0.0937,
|
| 1917 |
+
"step": 10176
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 1.52,
|
| 1921 |
+
"learning_rate": 2.5609756097560976e-07,
|
| 1922 |
+
"loss": 0.1234,
|
| 1923 |
+
"step": 10208
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 1.52,
|
| 1927 |
+
"learning_rate": 2.5356350966107065e-07,
|
| 1928 |
+
"loss": 0.138,
|
| 1929 |
+
"step": 10240
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 1.53,
|
| 1933 |
+
"learning_rate": 2.510294583465315e-07,
|
| 1934 |
+
"loss": 0.1753,
|
| 1935 |
+
"step": 10272
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 1.53,
|
| 1939 |
+
"learning_rate": 2.484954070319924e-07,
|
| 1940 |
+
"loss": 0.1185,
|
| 1941 |
+
"step": 10304
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 1.54,
|
| 1945 |
+
"learning_rate": 2.4596135571745327e-07,
|
| 1946 |
+
"loss": 0.0995,
|
| 1947 |
+
"step": 10336
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 1.54,
|
| 1951 |
+
"learning_rate": 2.4342730440291416e-07,
|
| 1952 |
+
"loss": 0.0922,
|
| 1953 |
+
"step": 10368
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 1.55,
|
| 1957 |
+
"learning_rate": 2.40893253088375e-07,
|
| 1958 |
+
"loss": 0.1387,
|
| 1959 |
+
"step": 10400
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 1.55,
|
| 1963 |
+
"learning_rate": 2.3835920177383592e-07,
|
| 1964 |
+
"loss": 0.1478,
|
| 1965 |
+
"step": 10432
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 1.56,
|
| 1969 |
+
"learning_rate": 2.3582515045929678e-07,
|
| 1970 |
+
"loss": 0.112,
|
| 1971 |
+
"step": 10464
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 1.56,
|
| 1975 |
+
"learning_rate": 2.3329109914475768e-07,
|
| 1976 |
+
"loss": 0.1357,
|
| 1977 |
+
"step": 10496
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 1.57,
|
| 1981 |
+
"learning_rate": 2.3075704783021854e-07,
|
| 1982 |
+
"loss": 0.1086,
|
| 1983 |
+
"step": 10528
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 1.57,
|
| 1987 |
+
"learning_rate": 2.2822299651567943e-07,
|
| 1988 |
+
"loss": 0.1271,
|
| 1989 |
+
"step": 10560
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.58,
|
| 1993 |
+
"learning_rate": 2.2568894520114032e-07,
|
| 1994 |
+
"loss": 0.1474,
|
| 1995 |
+
"step": 10592
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 1.58,
|
| 1999 |
+
"learning_rate": 2.231548938866012e-07,
|
| 2000 |
+
"loss": 0.1282,
|
| 2001 |
+
"step": 10624
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 1.59,
|
| 2005 |
+
"learning_rate": 2.2062084257206205e-07,
|
| 2006 |
+
"loss": 0.1723,
|
| 2007 |
+
"step": 10656
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 1.59,
|
| 2011 |
+
"learning_rate": 2.1808679125752297e-07,
|
| 2012 |
+
"loss": 0.161,
|
| 2013 |
+
"step": 10688
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 1.6,
|
| 2017 |
+
"learning_rate": 2.156319290465632e-07,
|
| 2018 |
+
"loss": 0.1467,
|
| 2019 |
+
"step": 10720
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 1.6,
|
| 2023 |
+
"learning_rate": 2.1309787773202406e-07,
|
| 2024 |
+
"loss": 0.0872,
|
| 2025 |
+
"step": 10752
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.61,
|
| 2029 |
+
"learning_rate": 2.1056382641748495e-07,
|
| 2030 |
+
"loss": 0.1393,
|
| 2031 |
+
"step": 10784
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.61,
|
| 2035 |
+
"learning_rate": 2.0802977510294584e-07,
|
| 2036 |
+
"loss": 0.1885,
|
| 2037 |
+
"step": 10816
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 1.62,
|
| 2041 |
+
"learning_rate": 2.054957237884067e-07,
|
| 2042 |
+
"loss": 0.0775,
|
| 2043 |
+
"step": 10848
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 1.62,
|
| 2047 |
+
"learning_rate": 2.029616724738676e-07,
|
| 2048 |
+
"loss": 0.1347,
|
| 2049 |
+
"step": 10880
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 1.62,
|
| 2053 |
+
"learning_rate": 2.0042762115932846e-07,
|
| 2054 |
+
"loss": 0.086,
|
| 2055 |
+
"step": 10912
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 1.63,
|
| 2059 |
+
"learning_rate": 1.9789356984478935e-07,
|
| 2060 |
+
"loss": 0.1257,
|
| 2061 |
+
"step": 10944
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 1.63,
|
| 2065 |
+
"learning_rate": 1.9535951853025024e-07,
|
| 2066 |
+
"loss": 0.169,
|
| 2067 |
+
"step": 10976
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 1.64,
|
| 2071 |
+
"learning_rate": 1.928254672157111e-07,
|
| 2072 |
+
"loss": 0.1343,
|
| 2073 |
+
"step": 11008
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 1.64,
|
| 2077 |
+
"learning_rate": 1.9029141590117197e-07,
|
| 2078 |
+
"loss": 0.097,
|
| 2079 |
+
"step": 11040
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 1.65,
|
| 2083 |
+
"learning_rate": 1.877573645866329e-07,
|
| 2084 |
+
"loss": 0.1061,
|
| 2085 |
+
"step": 11072
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 1.65,
|
| 2089 |
+
"learning_rate": 1.8522331327209376e-07,
|
| 2090 |
+
"loss": 0.1503,
|
| 2091 |
+
"step": 11104
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 1.66,
|
| 2095 |
+
"learning_rate": 1.8268926195755462e-07,
|
| 2096 |
+
"loss": 0.1025,
|
| 2097 |
+
"step": 11136
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 1.66,
|
| 2101 |
+
"learning_rate": 1.8015521064301551e-07,
|
| 2102 |
+
"loss": 0.0772,
|
| 2103 |
+
"step": 11168
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 1.67,
|
| 2107 |
+
"learning_rate": 1.776211593284764e-07,
|
| 2108 |
+
"loss": 0.1272,
|
| 2109 |
+
"step": 11200
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 1.67,
|
| 2113 |
+
"learning_rate": 1.7508710801393727e-07,
|
| 2114 |
+
"loss": 0.1134,
|
| 2115 |
+
"step": 11232
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 1.68,
|
| 2119 |
+
"learning_rate": 1.7255305669939816e-07,
|
| 2120 |
+
"loss": 0.1774,
|
| 2121 |
+
"step": 11264
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 1.68,
|
| 2125 |
+
"learning_rate": 1.7001900538485903e-07,
|
| 2126 |
+
"loss": 0.1119,
|
| 2127 |
+
"step": 11296
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 1.69,
|
| 2131 |
+
"learning_rate": 1.6748495407031992e-07,
|
| 2132 |
+
"loss": 0.1709,
|
| 2133 |
+
"step": 11328
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 1.69,
|
| 2137 |
+
"learning_rate": 1.649509027557808e-07,
|
| 2138 |
+
"loss": 0.1346,
|
| 2139 |
+
"step": 11360
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 1.7,
|
| 2143 |
+
"learning_rate": 1.6241685144124168e-07,
|
| 2144 |
+
"loss": 0.1499,
|
| 2145 |
+
"step": 11392
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 1.7,
|
| 2149 |
+
"learning_rate": 1.5988280012670254e-07,
|
| 2150 |
+
"loss": 0.1406,
|
| 2151 |
+
"step": 11424
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 1.71,
|
| 2155 |
+
"learning_rate": 1.5734874881216346e-07,
|
| 2156 |
+
"loss": 0.1105,
|
| 2157 |
+
"step": 11456
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.71,
|
| 2161 |
+
"learning_rate": 1.5481469749762433e-07,
|
| 2162 |
+
"loss": 0.1298,
|
| 2163 |
+
"step": 11488
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 1.72,
|
| 2167 |
+
"learning_rate": 1.522806461830852e-07,
|
| 2168 |
+
"loss": 0.1309,
|
| 2169 |
+
"step": 11520
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 1.72,
|
| 2173 |
+
"learning_rate": 1.4974659486854608e-07,
|
| 2174 |
+
"loss": 0.1059,
|
| 2175 |
+
"step": 11552
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 1.72,
|
| 2179 |
+
"learning_rate": 1.4721254355400697e-07,
|
| 2180 |
+
"loss": 0.1232,
|
| 2181 |
+
"step": 11584
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 1.73,
|
| 2185 |
+
"learning_rate": 1.4467849223946784e-07,
|
| 2186 |
+
"loss": 0.1357,
|
| 2187 |
+
"step": 11616
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 1.73,
|
| 2191 |
+
"learning_rate": 1.4214444092492873e-07,
|
| 2192 |
+
"loss": 0.0904,
|
| 2193 |
+
"step": 11648
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 1.74,
|
| 2197 |
+
"learning_rate": 1.396103896103896e-07,
|
| 2198 |
+
"loss": 0.1196,
|
| 2199 |
+
"step": 11680
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.74,
|
| 2203 |
+
"learning_rate": 1.370763382958505e-07,
|
| 2204 |
+
"loss": 0.1525,
|
| 2205 |
+
"step": 11712
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 1.75,
|
| 2209 |
+
"learning_rate": 1.3454228698131138e-07,
|
| 2210 |
+
"loss": 0.1775,
|
| 2211 |
+
"step": 11744
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 1.75,
|
| 2215 |
+
"learning_rate": 1.3200823566677224e-07,
|
| 2216 |
+
"loss": 0.1018,
|
| 2217 |
+
"step": 11776
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 1.76,
|
| 2221 |
+
"learning_rate": 1.294741843522331e-07,
|
| 2222 |
+
"loss": 0.0977,
|
| 2223 |
+
"step": 11808
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 1.76,
|
| 2227 |
+
"learning_rate": 1.2694013303769403e-07,
|
| 2228 |
+
"loss": 0.118,
|
| 2229 |
+
"step": 11840
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 1.77,
|
| 2233 |
+
"learning_rate": 1.244060817231549e-07,
|
| 2234 |
+
"loss": 0.0621,
|
| 2235 |
+
"step": 11872
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 1.77,
|
| 2239 |
+
"learning_rate": 1.2187203040861576e-07,
|
| 2240 |
+
"loss": 0.1363,
|
| 2241 |
+
"step": 11904
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.78,
|
| 2245 |
+
"learning_rate": 1.1933797909407665e-07,
|
| 2246 |
+
"loss": 0.1073,
|
| 2247 |
+
"step": 11936
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 1.78,
|
| 2251 |
+
"learning_rate": 1.1680392777953753e-07,
|
| 2252 |
+
"loss": 0.1046,
|
| 2253 |
+
"step": 11968
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 1.79,
|
| 2257 |
+
"learning_rate": 1.1426987646499842e-07,
|
| 2258 |
+
"loss": 0.0813,
|
| 2259 |
+
"step": 12000
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 1.79,
|
| 2263 |
+
"learning_rate": 1.1173582515045929e-07,
|
| 2264 |
+
"loss": 0.1154,
|
| 2265 |
+
"step": 12032
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 1.8,
|
| 2269 |
+
"learning_rate": 1.0920177383592018e-07,
|
| 2270 |
+
"loss": 0.1207,
|
| 2271 |
+
"step": 12064
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 1.8,
|
| 2275 |
+
"learning_rate": 1.0666772252138106e-07,
|
| 2276 |
+
"loss": 0.1177,
|
| 2277 |
+
"step": 12096
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 1.81,
|
| 2281 |
+
"learning_rate": 1.0413367120684193e-07,
|
| 2282 |
+
"loss": 0.1019,
|
| 2283 |
+
"step": 12128
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.81,
|
| 2287 |
+
"learning_rate": 1.0159961989230281e-07,
|
| 2288 |
+
"loss": 0.0776,
|
| 2289 |
+
"step": 12160
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 1.82,
|
| 2293 |
+
"learning_rate": 9.90655685777637e-08,
|
| 2294 |
+
"loss": 0.176,
|
| 2295 |
+
"step": 12192
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 1.82,
|
| 2299 |
+
"learning_rate": 9.653151726322457e-08,
|
| 2300 |
+
"loss": 0.1396,
|
| 2301 |
+
"step": 12224
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 1.82,
|
| 2305 |
+
"learning_rate": 9.399746594868546e-08,
|
| 2306 |
+
"loss": 0.1733,
|
| 2307 |
+
"step": 12256
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 1.83,
|
| 2311 |
+
"learning_rate": 9.146341463414634e-08,
|
| 2312 |
+
"loss": 0.1555,
|
| 2313 |
+
"step": 12288
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.83,
|
| 2317 |
+
"learning_rate": 8.892936331960722e-08,
|
| 2318 |
+
"loss": 0.1402,
|
| 2319 |
+
"step": 12320
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 1.84,
|
| 2323 |
+
"learning_rate": 8.63953120050681e-08,
|
| 2324 |
+
"loss": 0.1134,
|
| 2325 |
+
"step": 12352
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.84,
|
| 2329 |
+
"learning_rate": 8.386126069052899e-08,
|
| 2330 |
+
"loss": 0.1745,
|
| 2331 |
+
"step": 12384
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 1.85,
|
| 2335 |
+
"learning_rate": 8.132720937598985e-08,
|
| 2336 |
+
"loss": 0.1001,
|
| 2337 |
+
"step": 12416
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 1.85,
|
| 2341 |
+
"learning_rate": 7.879315806145075e-08,
|
| 2342 |
+
"loss": 0.0996,
|
| 2343 |
+
"step": 12448
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.86,
|
| 2347 |
+
"learning_rate": 7.625910674691162e-08,
|
| 2348 |
+
"loss": 0.0759,
|
| 2349 |
+
"step": 12480
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 1.86,
|
| 2353 |
+
"learning_rate": 7.37250554323725e-08,
|
| 2354 |
+
"loss": 0.0885,
|
| 2355 |
+
"step": 12512
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 1.87,
|
| 2359 |
+
"learning_rate": 7.119100411783338e-08,
|
| 2360 |
+
"loss": 0.1775,
|
| 2361 |
+
"step": 12544
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 1.87,
|
| 2365 |
+
"learning_rate": 6.865695280329427e-08,
|
| 2366 |
+
"loss": 0.0895,
|
| 2367 |
+
"step": 12576
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.88,
|
| 2371 |
+
"learning_rate": 6.612290148875514e-08,
|
| 2372 |
+
"loss": 0.1475,
|
| 2373 |
+
"step": 12608
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 1.88,
|
| 2377 |
+
"learning_rate": 6.358885017421603e-08,
|
| 2378 |
+
"loss": 0.1242,
|
| 2379 |
+
"step": 12640
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 1.89,
|
| 2383 |
+
"learning_rate": 6.105479885967691e-08,
|
| 2384 |
+
"loss": 0.149,
|
| 2385 |
+
"step": 12672
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.89,
|
| 2389 |
+
"learning_rate": 5.8520747545137786e-08,
|
| 2390 |
+
"loss": 0.0646,
|
| 2391 |
+
"step": 12704
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 1.9,
|
| 2395 |
+
"learning_rate": 5.5986696230598664e-08,
|
| 2396 |
+
"loss": 0.1511,
|
| 2397 |
+
"step": 12736
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 1.9,
|
| 2401 |
+
"learning_rate": 5.345264491605954e-08,
|
| 2402 |
+
"loss": 0.1567,
|
| 2403 |
+
"step": 12768
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 1.91,
|
| 2407 |
+
"learning_rate": 5.091859360152043e-08,
|
| 2408 |
+
"loss": 0.1383,
|
| 2409 |
+
"step": 12800
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.91,
|
| 2413 |
+
"learning_rate": 4.8384542286981306e-08,
|
| 2414 |
+
"loss": 0.153,
|
| 2415 |
+
"step": 12832
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 1.92,
|
| 2419 |
+
"learning_rate": 4.5850490972442185e-08,
|
| 2420 |
+
"loss": 0.15,
|
| 2421 |
+
"step": 12864
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 1.92,
|
| 2425 |
+
"learning_rate": 4.331643965790307e-08,
|
| 2426 |
+
"loss": 0.1292,
|
| 2427 |
+
"step": 12896
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.92,
|
| 2431 |
+
"learning_rate": 4.078238834336395e-08,
|
| 2432 |
+
"loss": 0.1298,
|
| 2433 |
+
"step": 12928
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 1.93,
|
| 2437 |
+
"learning_rate": 3.824833702882483e-08,
|
| 2438 |
+
"loss": 0.1317,
|
| 2439 |
+
"step": 12960
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 1.93,
|
| 2443 |
+
"learning_rate": 3.571428571428571e-08,
|
| 2444 |
+
"loss": 0.0836,
|
| 2445 |
+
"step": 12992
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 1.94,
|
| 2449 |
+
"learning_rate": 3.318023439974659e-08,
|
| 2450 |
+
"loss": 0.0985,
|
| 2451 |
+
"step": 13024
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 1.94,
|
| 2455 |
+
"learning_rate": 3.0646183085207475e-08,
|
| 2456 |
+
"loss": 0.0995,
|
| 2457 |
+
"step": 13056
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 1.95,
|
| 2461 |
+
"learning_rate": 2.8112131770668357e-08,
|
| 2462 |
+
"loss": 0.0664,
|
| 2463 |
+
"step": 13088
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 1.95,
|
| 2467 |
+
"learning_rate": 2.5657269559708584e-08,
|
| 2468 |
+
"loss": 0.1424,
|
| 2469 |
+
"step": 13120
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 1.96,
|
| 2473 |
+
"learning_rate": 2.3123218245169466e-08,
|
| 2474 |
+
"loss": 0.1044,
|
| 2475 |
+
"step": 13152
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 1.96,
|
| 2479 |
+
"learning_rate": 2.0589166930630344e-08,
|
| 2480 |
+
"loss": 0.1377,
|
| 2481 |
+
"step": 13184
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 1.97,
|
| 2485 |
+
"learning_rate": 1.8055115616091226e-08,
|
| 2486 |
+
"loss": 0.1469,
|
| 2487 |
+
"step": 13216
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 1.97,
|
| 2491 |
+
"learning_rate": 1.5521064301552108e-08,
|
| 2492 |
+
"loss": 0.0877,
|
| 2493 |
+
"step": 13248
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 1.98,
|
| 2497 |
+
"learning_rate": 1.2987012987012988e-08,
|
| 2498 |
+
"loss": 0.139,
|
| 2499 |
+
"step": 13280
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 1.98,
|
| 2503 |
+
"learning_rate": 1.0452961672473868e-08,
|
| 2504 |
+
"loss": 0.1598,
|
| 2505 |
+
"step": 13312
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 1.99,
|
| 2509 |
+
"learning_rate": 7.918910357934748e-09,
|
| 2510 |
+
"loss": 0.1874,
|
| 2511 |
+
"step": 13344
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 1.99,
|
| 2515 |
+
"learning_rate": 5.384859043395628e-09,
|
| 2516 |
+
"loss": 0.1875,
|
| 2517 |
+
"step": 13376
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 2.0,
|
| 2521 |
+
"learning_rate": 2.8508077288565093e-09,
|
| 2522 |
+
"loss": 0.1206,
|
| 2523 |
+
"step": 13408
|
| 2524 |
+
}
|
| 2525 |
+
],
|
| 2526 |
+
"logging_steps": 32,
|
| 2527 |
+
"max_steps": 13434,
|
| 2528 |
+
"num_input_tokens_seen": 0,
|
| 2529 |
+
"num_train_epochs": 2,
|
| 2530 |
+
"save_steps": 3358,
|
| 2531 |
+
"total_flos": 5.70368663027712e+19,
|
| 2532 |
+
"train_batch_size": 2,
|
| 2533 |
+
"trial_name": null,
|
| 2534 |
+
"trial_params": null
|
| 2535 |
+
}
|
checkpoint-20149/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e4db6f6fcaf2005bb3f583cfbd6e7afdd0dfdcdb7d4db7e107d5c22865fcf47
|
| 3 |
+
size 5048
|
checkpoint-6717/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-6717/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-6717/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3cbea7bf3e0a471c26c40067b5e0a880f36a63f3e9024e97afdec04184851c33
|
| 3 |
+
size 4992706480
|
checkpoint-6717/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49150d638e0ced917107f0af865392158918de17427afb8e34d0daa72ed7fbeb
|
| 3 |
+
size 1180663192
|
checkpoint-6717/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-6717/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ced3b7553c984f99429fdd37b1eb83bcda1bed0400a6c81c76635140fe4a5653
|
| 3 |
+
size 3095074288
|