Hugging Face's logo

ArtFair
/
whisper-large-v2-finetune-exp3-Jacob-1_0-mask

Safetensors
whisper
Model card Files Community
hsethu commited on
Commit
2abeb7e
·
verified ·
1 Parent(s): 16a35c2

Upload folder using huggingface_hub

Browse files
Files changed (40) hide show
  1. checkpoint-22486/config.json +66 -0
  2. checkpoint-22486/generation_config.json +317 -0
  3. checkpoint-22486/model-00001-of-00002.safetensors +3 -0
  4. checkpoint-22486/model-00002-of-00002.safetensors +3 -0
  5. checkpoint-22486/model.safetensors.index.json +0 -0
  6. checkpoint-22486/optimizer.pt +3 -0
  7. checkpoint-22486/preprocessor_config.json +14 -0
  8. checkpoint-22486/rng_state.pth +3 -0
  9. checkpoint-22486/scheduler.pt +3 -0
  10. checkpoint-22486/trainer_state.json +2127 -0
  11. checkpoint-22486/training_args.bin +3 -0
  12. checkpoint-44972/config.json +66 -0
  13. checkpoint-44972/generation_config.json +317 -0
  14. checkpoint-44972/model-00001-of-00002.safetensors +3 -0
  15. checkpoint-44972/model-00002-of-00002.safetensors +3 -0
  16. checkpoint-44972/model.safetensors.index.json +0 -0
  17. checkpoint-44972/optimizer.pt +3 -0
  18. checkpoint-44972/preprocessor_config.json +14 -0
  19. checkpoint-44972/rng_state.pth +3 -0
  20. checkpoint-44972/scheduler.pt +3 -0
  21. checkpoint-44972/trainer_state.json +4233 -0
  22. checkpoint-44972/training_args.bin +3 -0
  23. checkpoint-67458/config.json +66 -0
  24. checkpoint-67458/generation_config.json +317 -0
  25. checkpoint-67458/model-00001-of-00002.safetensors +3 -0
  26. checkpoint-67458/model-00002-of-00002.safetensors +3 -0
  27. checkpoint-67458/model.safetensors.index.json +0 -0
  28. checkpoint-67458/optimizer.pt +3 -0
  29. checkpoint-67458/preprocessor_config.json +14 -0
  30. checkpoint-67458/rng_state.pth +3 -0
  31. checkpoint-67458/scheduler.pt +3 -0
  32. checkpoint-67458/trainer_state.json +0 -0
  33. checkpoint-67458/training_args.bin +3 -0
  34. config.json +66 -0
  35. generation_config.json +317 -0
  36. model-00001-of-00002.safetensors +3 -0
  37. model-00002-of-00002.safetensors +3 -0
  38. model.safetensors.index.json +0 -0
  39. preprocessor_config.json +14 -0
  40. training_args.bin +3 -0
checkpoint-22486/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-22486/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-22486/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e35040b6e8e18c5c6844bd592e3a59d4c62fc804ece9d6d1a2ae92636537d481
3
+ size 4992706480
checkpoint-22486/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54b12159bbcd58cfd461019d1192865ad52c93d254154a1a338b383b574265ca
3
+ size 1180663192
checkpoint-22486/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-22486/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:780f93f2600fc79ac44d1a379c4f08de761fced81ad3d426789eba61fa236a89
3
+ size 3095074288
checkpoint-22486/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-22486/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:655f5250c055fcc41a90cb93533bcf9c5086e4ad547fe41f3891222514fde2a6
3
+ size 14244
checkpoint-22486/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6281b82ea535ea86d94b25530b5a7902a6a0d92802f8d7a70ddec0d585b4164
3
+ size 1064
checkpoint-22486/trainer_state.json ADDED
@@ -0,0 +1,2127 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 22486,
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": 9.991105576803343e-07,
14
+ "loss": 3.5693,
15
+ "step": 64
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 9.981618192060246e-07,
20
+ "loss": 1.3039,
21
+ "step": 128
22
+ },
23
+ {
24
+ "epoch": 0.01,
25
+ "learning_rate": 9.972130807317145e-07,
26
+ "loss": 0.6178,
27
+ "step": 192
28
+ },
29
+ {
30
+ "epoch": 0.01,
31
+ "learning_rate": 9.962643422574046e-07,
32
+ "loss": 0.2217,
33
+ "step": 256
34
+ },
35
+ {
36
+ "epoch": 0.01,
37
+ "learning_rate": 9.953156037830946e-07,
38
+ "loss": 0.135,
39
+ "step": 320
40
+ },
41
+ {
42
+ "epoch": 0.02,
43
+ "learning_rate": 9.943668653087847e-07,
44
+ "loss": 0.2547,
45
+ "step": 384
46
+ },
47
+ {
48
+ "epoch": 0.02,
49
+ "learning_rate": 9.934181268344748e-07,
50
+ "loss": 0.2357,
51
+ "step": 448
52
+ },
53
+ {
54
+ "epoch": 0.02,
55
+ "learning_rate": 9.92513860476148e-07,
56
+ "loss": 0.4032,
57
+ "step": 512
58
+ },
59
+ {
60
+ "epoch": 0.03,
61
+ "learning_rate": 9.915651220018381e-07,
62
+ "loss": 0.2513,
63
+ "step": 576
64
+ },
65
+ {
66
+ "epoch": 0.03,
67
+ "learning_rate": 9.906163835275282e-07,
68
+ "loss": 0.2058,
69
+ "step": 640
70
+ },
71
+ {
72
+ "epoch": 0.03,
73
+ "learning_rate": 9.89667645053218e-07,
74
+ "loss": 0.2276,
75
+ "step": 704
76
+ },
77
+ {
78
+ "epoch": 0.03,
79
+ "learning_rate": 9.887189065789084e-07,
80
+ "loss": 0.1798,
81
+ "step": 768
82
+ },
83
+ {
84
+ "epoch": 0.04,
85
+ "learning_rate": 9.877701681045983e-07,
86
+ "loss": 0.2646,
87
+ "step": 832
88
+ },
89
+ {
90
+ "epoch": 0.04,
91
+ "learning_rate": 9.868214296302884e-07,
92
+ "loss": 0.2434,
93
+ "step": 896
94
+ },
95
+ {
96
+ "epoch": 0.04,
97
+ "learning_rate": 9.858726911559785e-07,
98
+ "loss": 0.3211,
99
+ "step": 960
100
+ },
101
+ {
102
+ "epoch": 0.05,
103
+ "learning_rate": 9.849239526816686e-07,
104
+ "loss": 0.1601,
105
+ "step": 1024
106
+ },
107
+ {
108
+ "epoch": 0.05,
109
+ "learning_rate": 9.839752142073586e-07,
110
+ "loss": 0.2012,
111
+ "step": 1088
112
+ },
113
+ {
114
+ "epoch": 0.05,
115
+ "learning_rate": 9.830264757330487e-07,
116
+ "loss": 0.1635,
117
+ "step": 1152
118
+ },
119
+ {
120
+ "epoch": 0.05,
121
+ "learning_rate": 9.820777372587388e-07,
122
+ "loss": 0.2808,
123
+ "step": 1216
124
+ },
125
+ {
126
+ "epoch": 0.06,
127
+ "learning_rate": 9.811289987844287e-07,
128
+ "loss": 0.1622,
129
+ "step": 1280
130
+ },
131
+ {
132
+ "epoch": 0.06,
133
+ "learning_rate": 9.801802603101188e-07,
134
+ "loss": 0.23,
135
+ "step": 1344
136
+ },
137
+ {
138
+ "epoch": 0.06,
139
+ "learning_rate": 9.792315218358089e-07,
140
+ "loss": 0.1615,
141
+ "step": 1408
142
+ },
143
+ {
144
+ "epoch": 0.07,
145
+ "learning_rate": 9.78282783361499e-07,
146
+ "loss": 0.1424,
147
+ "step": 1472
148
+ },
149
+ {
150
+ "epoch": 0.07,
151
+ "learning_rate": 9.77334044887189e-07,
152
+ "loss": 0.2007,
153
+ "step": 1536
154
+ },
155
+ {
156
+ "epoch": 0.07,
157
+ "learning_rate": 9.763853064128792e-07,
158
+ "loss": 0.3649,
159
+ "step": 1600
160
+ },
161
+ {
162
+ "epoch": 0.07,
163
+ "learning_rate": 9.75436567938569e-07,
164
+ "loss": 0.1149,
165
+ "step": 1664
166
+ },
167
+ {
168
+ "epoch": 0.08,
169
+ "learning_rate": 9.744878294642591e-07,
170
+ "loss": 0.1882,
171
+ "step": 1728
172
+ },
173
+ {
174
+ "epoch": 0.08,
175
+ "learning_rate": 9.735390909899492e-07,
176
+ "loss": 0.2598,
177
+ "step": 1792
178
+ },
179
+ {
180
+ "epoch": 0.08,
181
+ "learning_rate": 9.725903525156393e-07,
182
+ "loss": 0.2198,
183
+ "step": 1856
184
+ },
185
+ {
186
+ "epoch": 0.09,
187
+ "learning_rate": 9.716416140413294e-07,
188
+ "loss": 0.1786,
189
+ "step": 1920
190
+ },
191
+ {
192
+ "epoch": 0.09,
193
+ "learning_rate": 9.706928755670195e-07,
194
+ "loss": 0.2713,
195
+ "step": 1984
196
+ },
197
+ {
198
+ "epoch": 0.09,
199
+ "learning_rate": 9.697441370927096e-07,
200
+ "loss": 0.2076,
201
+ "step": 2048
202
+ },
203
+ {
204
+ "epoch": 0.09,
205
+ "learning_rate": 9.687953986183995e-07,
206
+ "loss": 0.2343,
207
+ "step": 2112
208
+ },
209
+ {
210
+ "epoch": 0.1,
211
+ "learning_rate": 9.678466601440898e-07,
212
+ "loss": 0.3182,
213
+ "step": 2176
214
+ },
215
+ {
216
+ "epoch": 0.1,
217
+ "learning_rate": 9.668979216697796e-07,
218
+ "loss": 0.2102,
219
+ "step": 2240
220
+ },
221
+ {
222
+ "epoch": 0.1,
223
+ "learning_rate": 9.659491831954697e-07,
224
+ "loss": 0.2687,
225
+ "step": 2304
226
+ },
227
+ {
228
+ "epoch": 0.11,
229
+ "learning_rate": 9.650004447211598e-07,
230
+ "loss": 0.2186,
231
+ "step": 2368
232
+ },
233
+ {
234
+ "epoch": 0.11,
235
+ "learning_rate": 9.6405170624685e-07,
236
+ "loss": 0.1295,
237
+ "step": 2432
238
+ },
239
+ {
240
+ "epoch": 0.11,
241
+ "learning_rate": 9.631029677725398e-07,
242
+ "loss": 0.1542,
243
+ "step": 2496
244
+ },
245
+ {
246
+ "epoch": 0.11,
247
+ "learning_rate": 9.621542292982299e-07,
248
+ "loss": 0.1365,
249
+ "step": 2560
250
+ },
251
+ {
252
+ "epoch": 0.12,
253
+ "learning_rate": 9.6120549082392e-07,
254
+ "loss": 0.2106,
255
+ "step": 2624
256
+ },
257
+ {
258
+ "epoch": 0.12,
259
+ "learning_rate": 9.6025675234961e-07,
260
+ "loss": 0.2753,
261
+ "step": 2688
262
+ },
263
+ {
264
+ "epoch": 0.12,
265
+ "learning_rate": 9.593080138753002e-07,
266
+ "loss": 0.2162,
267
+ "step": 2752
268
+ },
269
+ {
270
+ "epoch": 0.13,
271
+ "learning_rate": 9.583592754009902e-07,
272
+ "loss": 0.2186,
273
+ "step": 2816
274
+ },
275
+ {
276
+ "epoch": 0.13,
277
+ "learning_rate": 9.574105369266803e-07,
278
+ "loss": 0.0991,
279
+ "step": 2880
280
+ },
281
+ {
282
+ "epoch": 0.13,
283
+ "learning_rate": 9.564617984523702e-07,
284
+ "loss": 0.1616,
285
+ "step": 2944
286
+ },
287
+ {
288
+ "epoch": 0.13,
289
+ "learning_rate": 9.555130599780605e-07,
290
+ "loss": 0.2029,
291
+ "step": 3008
292
+ },
293
+ {
294
+ "epoch": 0.14,
295
+ "learning_rate": 9.545643215037504e-07,
296
+ "loss": 0.1659,
297
+ "step": 3072
298
+ },
299
+ {
300
+ "epoch": 0.14,
301
+ "learning_rate": 9.536155830294405e-07,
302
+ "loss": 0.1887,
303
+ "step": 3136
304
+ },
305
+ {
306
+ "epoch": 0.14,
307
+ "learning_rate": 9.526668445551306e-07,
308
+ "loss": 0.2626,
309
+ "step": 3200
310
+ },
311
+ {
312
+ "epoch": 0.15,
313
+ "learning_rate": 9.517181060808206e-07,
314
+ "loss": 0.1325,
315
+ "step": 3264
316
+ },
317
+ {
318
+ "epoch": 0.15,
319
+ "learning_rate": 9.507693676065108e-07,
320
+ "loss": 0.2125,
321
+ "step": 3328
322
+ },
323
+ {
324
+ "epoch": 0.15,
325
+ "learning_rate": 9.498206291322007e-07,
326
+ "loss": 0.1926,
327
+ "step": 3392
328
+ },
329
+ {
330
+ "epoch": 0.15,
331
+ "learning_rate": 9.488718906578907e-07,
332
+ "loss": 0.2237,
333
+ "step": 3456
334
+ },
335
+ {
336
+ "epoch": 0.16,
337
+ "learning_rate": 9.479231521835809e-07,
338
+ "loss": 0.1283,
339
+ "step": 3520
340
+ },
341
+ {
342
+ "epoch": 0.16,
343
+ "learning_rate": 9.469744137092709e-07,
344
+ "loss": 0.2052,
345
+ "step": 3584
346
+ },
347
+ {
348
+ "epoch": 0.16,
349
+ "learning_rate": 9.46025675234961e-07,
350
+ "loss": 0.1983,
351
+ "step": 3648
352
+ },
353
+ {
354
+ "epoch": 0.17,
355
+ "learning_rate": 9.45076936760651e-07,
356
+ "loss": 0.1976,
357
+ "step": 3712
358
+ },
359
+ {
360
+ "epoch": 0.17,
361
+ "learning_rate": 9.441281982863411e-07,
362
+ "loss": 0.1999,
363
+ "step": 3776
364
+ },
365
+ {
366
+ "epoch": 0.17,
367
+ "learning_rate": 9.431794598120312e-07,
368
+ "loss": 0.2231,
369
+ "step": 3840
370
+ },
371
+ {
372
+ "epoch": 0.17,
373
+ "learning_rate": 9.422307213377212e-07,
374
+ "loss": 0.2057,
375
+ "step": 3904
376
+ },
377
+ {
378
+ "epoch": 0.18,
379
+ "learning_rate": 9.412819828634113e-07,
380
+ "loss": 0.2376,
381
+ "step": 3968
382
+ },
383
+ {
384
+ "epoch": 0.18,
385
+ "learning_rate": 9.403332443891013e-07,
386
+ "loss": 0.1325,
387
+ "step": 4032
388
+ },
389
+ {
390
+ "epoch": 0.18,
391
+ "learning_rate": 9.393845059147913e-07,
392
+ "loss": 0.2595,
393
+ "step": 4096
394
+ },
395
+ {
396
+ "epoch": 0.19,
397
+ "learning_rate": 9.384357674404815e-07,
398
+ "loss": 0.1817,
399
+ "step": 4160
400
+ },
401
+ {
402
+ "epoch": 0.19,
403
+ "learning_rate": 9.374870289661715e-07,
404
+ "loss": 0.2224,
405
+ "step": 4224
406
+ },
407
+ {
408
+ "epoch": 0.19,
409
+ "learning_rate": 9.365382904918616e-07,
410
+ "loss": 0.1444,
411
+ "step": 4288
412
+ },
413
+ {
414
+ "epoch": 0.19,
415
+ "learning_rate": 9.355895520175517e-07,
416
+ "loss": 0.2695,
417
+ "step": 4352
418
+ },
419
+ {
420
+ "epoch": 0.2,
421
+ "learning_rate": 9.346408135432417e-07,
422
+ "loss": 0.1859,
423
+ "step": 4416
424
+ },
425
+ {
426
+ "epoch": 0.2,
427
+ "learning_rate": 9.336920750689318e-07,
428
+ "loss": 0.1551,
429
+ "step": 4480
430
+ },
431
+ {
432
+ "epoch": 0.2,
433
+ "learning_rate": 9.327433365946217e-07,
434
+ "loss": 0.2093,
435
+ "step": 4544
436
+ },
437
+ {
438
+ "epoch": 0.2,
439
+ "learning_rate": 9.317945981203118e-07,
440
+ "loss": 0.1922,
441
+ "step": 4608
442
+ },
443
+ {
444
+ "epoch": 0.21,
445
+ "learning_rate": 9.308458596460019e-07,
446
+ "loss": 0.2056,
447
+ "step": 4672
448
+ },
449
+ {
450
+ "epoch": 0.21,
451
+ "learning_rate": 9.298971211716919e-07,
452
+ "loss": 0.2087,
453
+ "step": 4736
454
+ },
455
+ {
456
+ "epoch": 0.21,
457
+ "learning_rate": 9.289483826973821e-07,
458
+ "loss": 0.1552,
459
+ "step": 4800
460
+ },
461
+ {
462
+ "epoch": 0.22,
463
+ "learning_rate": 9.279996442230721e-07,
464
+ "loss": 0.1619,
465
+ "step": 4864
466
+ },
467
+ {
468
+ "epoch": 0.22,
469
+ "learning_rate": 9.270509057487621e-07,
470
+ "loss": 0.2634,
471
+ "step": 4928
472
+ },
473
+ {
474
+ "epoch": 0.22,
475
+ "learning_rate": 9.261021672744523e-07,
476
+ "loss": 0.1459,
477
+ "step": 4992
478
+ },
479
+ {
480
+ "epoch": 0.22,
481
+ "learning_rate": 9.251534288001423e-07,
482
+ "loss": 0.225,
483
+ "step": 5056
484
+ },
485
+ {
486
+ "epoch": 0.23,
487
+ "learning_rate": 9.242046903258323e-07,
488
+ "loss": 0.1878,
489
+ "step": 5120
490
+ },
491
+ {
492
+ "epoch": 0.23,
493
+ "learning_rate": 9.232559518515224e-07,
494
+ "loss": 0.1286,
495
+ "step": 5184
496
+ },
497
+ {
498
+ "epoch": 0.23,
499
+ "learning_rate": 9.223072133772124e-07,
500
+ "loss": 0.1059,
501
+ "step": 5248
502
+ },
503
+ {
504
+ "epoch": 0.24,
505
+ "learning_rate": 9.213584749029025e-07,
506
+ "loss": 0.1541,
507
+ "step": 5312
508
+ },
509
+ {
510
+ "epoch": 0.24,
511
+ "learning_rate": 9.204097364285925e-07,
512
+ "loss": 0.2427,
513
+ "step": 5376
514
+ },
515
+ {
516
+ "epoch": 0.24,
517
+ "learning_rate": 9.194609979542827e-07,
518
+ "loss": 0.1368,
519
+ "step": 5440
520
+ },
521
+ {
522
+ "epoch": 0.24,
523
+ "learning_rate": 9.185122594799727e-07,
524
+ "loss": 0.2477,
525
+ "step": 5504
526
+ },
527
+ {
528
+ "epoch": 0.25,
529
+ "learning_rate": 9.175635210056627e-07,
530
+ "loss": 0.1385,
531
+ "step": 5568
532
+ },
533
+ {
534
+ "epoch": 0.25,
535
+ "learning_rate": 9.166147825313529e-07,
536
+ "loss": 0.2537,
537
+ "step": 5632
538
+ },
539
+ {
540
+ "epoch": 0.25,
541
+ "learning_rate": 9.156660440570428e-07,
542
+ "loss": 0.2374,
543
+ "step": 5696
544
+ },
545
+ {
546
+ "epoch": 0.26,
547
+ "learning_rate": 9.147173055827329e-07,
548
+ "loss": 0.1942,
549
+ "step": 5760
550
+ },
551
+ {
552
+ "epoch": 0.26,
553
+ "learning_rate": 9.13768567108423e-07,
554
+ "loss": 0.1416,
555
+ "step": 5824
556
+ },
557
+ {
558
+ "epoch": 0.26,
559
+ "learning_rate": 9.12819828634113e-07,
560
+ "loss": 0.2205,
561
+ "step": 5888
562
+ },
563
+ {
564
+ "epoch": 0.26,
565
+ "learning_rate": 9.118710901598031e-07,
566
+ "loss": 0.1265,
567
+ "step": 5952
568
+ },
569
+ {
570
+ "epoch": 0.27,
571
+ "learning_rate": 9.109223516854932e-07,
572
+ "loss": 0.1972,
573
+ "step": 6016
574
+ },
575
+ {
576
+ "epoch": 0.27,
577
+ "learning_rate": 9.099736132111833e-07,
578
+ "loss": 0.2119,
579
+ "step": 6080
580
+ },
581
+ {
582
+ "epoch": 0.27,
583
+ "learning_rate": 9.090248747368733e-07,
584
+ "loss": 0.1284,
585
+ "step": 6144
586
+ },
587
+ {
588
+ "epoch": 0.28,
589
+ "learning_rate": 9.080761362625632e-07,
590
+ "loss": 0.1665,
591
+ "step": 6208
592
+ },
593
+ {
594
+ "epoch": 0.28,
595
+ "learning_rate": 9.071273977882534e-07,
596
+ "loss": 0.1545,
597
+ "step": 6272
598
+ },
599
+ {
600
+ "epoch": 0.28,
601
+ "learning_rate": 9.061786593139434e-07,
602
+ "loss": 0.1505,
603
+ "step": 6336
604
+ },
605
+ {
606
+ "epoch": 0.28,
607
+ "learning_rate": 9.052299208396334e-07,
608
+ "loss": 0.1871,
609
+ "step": 6400
610
+ },
611
+ {
612
+ "epoch": 0.29,
613
+ "learning_rate": 9.042811823653236e-07,
614
+ "loss": 0.2302,
615
+ "step": 6464
616
+ },
617
+ {
618
+ "epoch": 0.29,
619
+ "learning_rate": 9.033324438910136e-07,
620
+ "loss": 0.1395,
621
+ "step": 6528
622
+ },
623
+ {
624
+ "epoch": 0.29,
625
+ "learning_rate": 9.023837054167037e-07,
626
+ "loss": 0.2195,
627
+ "step": 6592
628
+ },
629
+ {
630
+ "epoch": 0.3,
631
+ "learning_rate": 9.014349669423938e-07,
632
+ "loss": 0.2419,
633
+ "step": 6656
634
+ },
635
+ {
636
+ "epoch": 0.3,
637
+ "learning_rate": 9.004862284680838e-07,
638
+ "loss": 0.2173,
639
+ "step": 6720
640
+ },
641
+ {
642
+ "epoch": 0.3,
643
+ "learning_rate": 8.995374899937739e-07,
644
+ "loss": 0.2522,
645
+ "step": 6784
646
+ },
647
+ {
648
+ "epoch": 0.3,
649
+ "learning_rate": 8.985887515194639e-07,
650
+ "loss": 0.2317,
651
+ "step": 6848
652
+ },
653
+ {
654
+ "epoch": 0.31,
655
+ "learning_rate": 8.97640013045154e-07,
656
+ "loss": 0.1959,
657
+ "step": 6912
658
+ },
659
+ {
660
+ "epoch": 0.31,
661
+ "learning_rate": 8.96691274570844e-07,
662
+ "loss": 0.1889,
663
+ "step": 6976
664
+ },
665
+ {
666
+ "epoch": 0.31,
667
+ "learning_rate": 8.95742536096534e-07,
668
+ "loss": 0.2131,
669
+ "step": 7040
670
+ },
671
+ {
672
+ "epoch": 0.32,
673
+ "learning_rate": 8.947937976222242e-07,
674
+ "loss": 0.1321,
675
+ "step": 7104
676
+ },
677
+ {
678
+ "epoch": 0.32,
679
+ "learning_rate": 8.938450591479142e-07,
680
+ "loss": 0.1679,
681
+ "step": 7168
682
+ },
683
+ {
684
+ "epoch": 0.32,
685
+ "learning_rate": 8.928963206736043e-07,
686
+ "loss": 0.1826,
687
+ "step": 7232
688
+ },
689
+ {
690
+ "epoch": 0.32,
691
+ "learning_rate": 8.919475821992944e-07,
692
+ "loss": 0.2144,
693
+ "step": 7296
694
+ },
695
+ {
696
+ "epoch": 0.33,
697
+ "learning_rate": 8.909988437249843e-07,
698
+ "loss": 0.1614,
699
+ "step": 7360
700
+ },
701
+ {
702
+ "epoch": 0.33,
703
+ "learning_rate": 8.900501052506744e-07,
704
+ "loss": 0.241,
705
+ "step": 7424
706
+ },
707
+ {
708
+ "epoch": 0.33,
709
+ "learning_rate": 8.891013667763645e-07,
710
+ "loss": 0.1825,
711
+ "step": 7488
712
+ },
713
+ {
714
+ "epoch": 0.34,
715
+ "learning_rate": 8.881526283020546e-07,
716
+ "loss": 0.17,
717
+ "step": 7552
718
+ },
719
+ {
720
+ "epoch": 0.34,
721
+ "learning_rate": 8.872038898277446e-07,
722
+ "loss": 0.1562,
723
+ "step": 7616
724
+ },
725
+ {
726
+ "epoch": 0.34,
727
+ "learning_rate": 8.862551513534347e-07,
728
+ "loss": 0.2264,
729
+ "step": 7680
730
+ },
731
+ {
732
+ "epoch": 0.34,
733
+ "learning_rate": 8.853064128791248e-07,
734
+ "loss": 0.1325,
735
+ "step": 7744
736
+ },
737
+ {
738
+ "epoch": 0.35,
739
+ "learning_rate": 8.843576744048148e-07,
740
+ "loss": 0.1601,
741
+ "step": 7808
742
+ },
743
+ {
744
+ "epoch": 0.35,
745
+ "learning_rate": 8.83408935930505e-07,
746
+ "loss": 0.1784,
747
+ "step": 7872
748
+ },
749
+ {
750
+ "epoch": 0.35,
751
+ "learning_rate": 8.82460197456195e-07,
752
+ "loss": 0.1447,
753
+ "step": 7936
754
+ },
755
+ {
756
+ "epoch": 0.36,
757
+ "learning_rate": 8.815114589818849e-07,
758
+ "loss": 0.166,
759
+ "step": 8000
760
+ },
761
+ {
762
+ "epoch": 0.36,
763
+ "learning_rate": 8.80562720507575e-07,
764
+ "loss": 0.1554,
765
+ "step": 8064
766
+ },
767
+ {
768
+ "epoch": 0.36,
769
+ "learning_rate": 8.796139820332651e-07,
770
+ "loss": 0.1097,
771
+ "step": 8128
772
+ },
773
+ {
774
+ "epoch": 0.36,
775
+ "learning_rate": 8.786652435589552e-07,
776
+ "loss": 0.1322,
777
+ "step": 8192
778
+ },
779
+ {
780
+ "epoch": 0.37,
781
+ "learning_rate": 8.777165050846452e-07,
782
+ "loss": 0.262,
783
+ "step": 8256
784
+ },
785
+ {
786
+ "epoch": 0.37,
787
+ "learning_rate": 8.767677666103353e-07,
788
+ "loss": 0.1755,
789
+ "step": 8320
790
+ },
791
+ {
792
+ "epoch": 0.37,
793
+ "learning_rate": 8.758190281360254e-07,
794
+ "loss": 0.1646,
795
+ "step": 8384
796
+ },
797
+ {
798
+ "epoch": 0.38,
799
+ "learning_rate": 8.748702896617154e-07,
800
+ "loss": 0.1481,
801
+ "step": 8448
802
+ },
803
+ {
804
+ "epoch": 0.38,
805
+ "learning_rate": 8.739215511874055e-07,
806
+ "loss": 0.0985,
807
+ "step": 8512
808
+ },
809
+ {
810
+ "epoch": 0.38,
811
+ "learning_rate": 8.729728127130955e-07,
812
+ "loss": 0.1401,
813
+ "step": 8576
814
+ },
815
+ {
816
+ "epoch": 0.38,
817
+ "learning_rate": 8.720240742387855e-07,
818
+ "loss": 0.2057,
819
+ "step": 8640
820
+ },
821
+ {
822
+ "epoch": 0.39,
823
+ "learning_rate": 8.710753357644757e-07,
824
+ "loss": 0.1677,
825
+ "step": 8704
826
+ },
827
+ {
828
+ "epoch": 0.39,
829
+ "learning_rate": 8.701265972901657e-07,
830
+ "loss": 0.1398,
831
+ "step": 8768
832
+ },
833
+ {
834
+ "epoch": 0.39,
835
+ "learning_rate": 8.691778588158557e-07,
836
+ "loss": 0.1665,
837
+ "step": 8832
838
+ },
839
+ {
840
+ "epoch": 0.4,
841
+ "learning_rate": 8.682439443802069e-07,
842
+ "loss": 0.1737,
843
+ "step": 8896
844
+ },
845
+ {
846
+ "epoch": 0.4,
847
+ "learning_rate": 8.67295205905897e-07,
848
+ "loss": 0.1642,
849
+ "step": 8960
850
+ },
851
+ {
852
+ "epoch": 0.4,
853
+ "learning_rate": 8.66346467431587e-07,
854
+ "loss": 0.1487,
855
+ "step": 9024
856
+ },
857
+ {
858
+ "epoch": 0.4,
859
+ "learning_rate": 8.653977289572771e-07,
860
+ "loss": 0.186,
861
+ "step": 9088
862
+ },
863
+ {
864
+ "epoch": 0.41,
865
+ "learning_rate": 8.644489904829672e-07,
866
+ "loss": 0.1902,
867
+ "step": 9152
868
+ },
869
+ {
870
+ "epoch": 0.41,
871
+ "learning_rate": 8.635002520086571e-07,
872
+ "loss": 0.1977,
873
+ "step": 9216
874
+ },
875
+ {
876
+ "epoch": 0.41,
877
+ "learning_rate": 8.625515135343473e-07,
878
+ "loss": 0.1853,
879
+ "step": 9280
880
+ },
881
+ {
882
+ "epoch": 0.42,
883
+ "learning_rate": 8.616027750600373e-07,
884
+ "loss": 0.1156,
885
+ "step": 9344
886
+ },
887
+ {
888
+ "epoch": 0.42,
889
+ "learning_rate": 8.606540365857273e-07,
890
+ "loss": 0.179,
891
+ "step": 9408
892
+ },
893
+ {
894
+ "epoch": 0.42,
895
+ "learning_rate": 8.597052981114175e-07,
896
+ "loss": 0.1978,
897
+ "step": 9472
898
+ },
899
+ {
900
+ "epoch": 0.42,
901
+ "learning_rate": 8.587565596371075e-07,
902
+ "loss": 0.1735,
903
+ "step": 9536
904
+ },
905
+ {
906
+ "epoch": 0.43,
907
+ "learning_rate": 8.578078211627976e-07,
908
+ "loss": 0.1579,
909
+ "step": 9600
910
+ },
911
+ {
912
+ "epoch": 0.43,
913
+ "learning_rate": 8.568590826884876e-07,
914
+ "loss": 0.1444,
915
+ "step": 9664
916
+ },
917
+ {
918
+ "epoch": 0.43,
919
+ "learning_rate": 8.559103442141777e-07,
920
+ "loss": 0.1664,
921
+ "step": 9728
922
+ },
923
+ {
924
+ "epoch": 0.44,
925
+ "learning_rate": 8.549616057398677e-07,
926
+ "loss": 0.1715,
927
+ "step": 9792
928
+ },
929
+ {
930
+ "epoch": 0.44,
931
+ "learning_rate": 8.540128672655577e-07,
932
+ "loss": 0.1189,
933
+ "step": 9856
934
+ },
935
+ {
936
+ "epoch": 0.44,
937
+ "learning_rate": 8.530641287912479e-07,
938
+ "loss": 0.1255,
939
+ "step": 9920
940
+ },
941
+ {
942
+ "epoch": 0.44,
943
+ "learning_rate": 8.521153903169379e-07,
944
+ "loss": 0.2314,
945
+ "step": 9984
946
+ },
947
+ {
948
+ "epoch": 0.45,
949
+ "learning_rate": 8.511814758812891e-07,
950
+ "loss": 0.18,
951
+ "step": 10048
952
+ },
953
+ {
954
+ "epoch": 0.45,
955
+ "learning_rate": 8.502327374069791e-07,
956
+ "loss": 0.1948,
957
+ "step": 10112
958
+ },
959
+ {
960
+ "epoch": 0.45,
961
+ "learning_rate": 8.492839989326692e-07,
962
+ "loss": 0.1763,
963
+ "step": 10176
964
+ },
965
+ {
966
+ "epoch": 0.46,
967
+ "learning_rate": 8.483352604583593e-07,
968
+ "loss": 0.1984,
969
+ "step": 10240
970
+ },
971
+ {
972
+ "epoch": 0.46,
973
+ "learning_rate": 8.473865219840493e-07,
974
+ "loss": 0.169,
975
+ "step": 10304
976
+ },
977
+ {
978
+ "epoch": 0.46,
979
+ "learning_rate": 8.464377835097394e-07,
980
+ "loss": 0.1268,
981
+ "step": 10368
982
+ },
983
+ {
984
+ "epoch": 0.46,
985
+ "learning_rate": 8.454890450354294e-07,
986
+ "loss": 0.2261,
987
+ "step": 10432
988
+ },
989
+ {
990
+ "epoch": 0.47,
991
+ "learning_rate": 8.445403065611195e-07,
992
+ "loss": 0.1751,
993
+ "step": 10496
994
+ },
995
+ {
996
+ "epoch": 0.47,
997
+ "learning_rate": 8.435915680868095e-07,
998
+ "loss": 0.1758,
999
+ "step": 10560
1000
+ },
1001
+ {
1002
+ "epoch": 0.47,
1003
+ "learning_rate": 8.426428296124995e-07,
1004
+ "loss": 0.1565,
1005
+ "step": 10624
1006
+ },
1007
+ {
1008
+ "epoch": 0.48,
1009
+ "learning_rate": 8.416940911381897e-07,
1010
+ "loss": 0.0861,
1011
+ "step": 10688
1012
+ },
1013
+ {
1014
+ "epoch": 0.48,
1015
+ "learning_rate": 8.407453526638797e-07,
1016
+ "loss": 0.1382,
1017
+ "step": 10752
1018
+ },
1019
+ {
1020
+ "epoch": 0.48,
1021
+ "learning_rate": 8.397966141895698e-07,
1022
+ "loss": 0.1268,
1023
+ "step": 10816
1024
+ },
1025
+ {
1026
+ "epoch": 0.48,
1027
+ "learning_rate": 8.388478757152599e-07,
1028
+ "loss": 0.1418,
1029
+ "step": 10880
1030
+ },
1031
+ {
1032
+ "epoch": 0.49,
1033
+ "learning_rate": 8.378991372409498e-07,
1034
+ "loss": 0.2417,
1035
+ "step": 10944
1036
+ },
1037
+ {
1038
+ "epoch": 0.49,
1039
+ "learning_rate": 8.369503987666399e-07,
1040
+ "loss": 0.1285,
1041
+ "step": 11008
1042
+ },
1043
+ {
1044
+ "epoch": 0.49,
1045
+ "learning_rate": 8.3600166029233e-07,
1046
+ "loss": 0.1719,
1047
+ "step": 11072
1048
+ },
1049
+ {
1050
+ "epoch": 0.5,
1051
+ "learning_rate": 8.350529218180201e-07,
1052
+ "loss": 0.1432,
1053
+ "step": 11136
1054
+ },
1055
+ {
1056
+ "epoch": 0.5,
1057
+ "learning_rate": 8.341041833437101e-07,
1058
+ "loss": 0.2443,
1059
+ "step": 11200
1060
+ },
1061
+ {
1062
+ "epoch": 0.5,
1063
+ "learning_rate": 8.331554448694002e-07,
1064
+ "loss": 0.1348,
1065
+ "step": 11264
1066
+ },
1067
+ {
1068
+ "epoch": 0.5,
1069
+ "learning_rate": 8.322067063950903e-07,
1070
+ "loss": 0.1251,
1071
+ "step": 11328
1072
+ },
1073
+ {
1074
+ "epoch": 0.51,
1075
+ "learning_rate": 8.312579679207803e-07,
1076
+ "loss": 0.156,
1077
+ "step": 11392
1078
+ },
1079
+ {
1080
+ "epoch": 0.51,
1081
+ "learning_rate": 8.303092294464703e-07,
1082
+ "loss": 0.3104,
1083
+ "step": 11456
1084
+ },
1085
+ {
1086
+ "epoch": 0.51,
1087
+ "learning_rate": 8.293604909721605e-07,
1088
+ "loss": 0.1834,
1089
+ "step": 11520
1090
+ },
1091
+ {
1092
+ "epoch": 0.52,
1093
+ "learning_rate": 8.284117524978504e-07,
1094
+ "loss": 0.1312,
1095
+ "step": 11584
1096
+ },
1097
+ {
1098
+ "epoch": 0.52,
1099
+ "learning_rate": 8.274630140235405e-07,
1100
+ "loss": 0.1026,
1101
+ "step": 11648
1102
+ },
1103
+ {
1104
+ "epoch": 0.52,
1105
+ "learning_rate": 8.265142755492306e-07,
1106
+ "loss": 0.1805,
1107
+ "step": 11712
1108
+ },
1109
+ {
1110
+ "epoch": 0.52,
1111
+ "learning_rate": 8.255655370749206e-07,
1112
+ "loss": 0.2334,
1113
+ "step": 11776
1114
+ },
1115
+ {
1116
+ "epoch": 0.53,
1117
+ "learning_rate": 8.246167986006107e-07,
1118
+ "loss": 0.1606,
1119
+ "step": 11840
1120
+ },
1121
+ {
1122
+ "epoch": 0.53,
1123
+ "learning_rate": 8.236680601263008e-07,
1124
+ "loss": 0.1009,
1125
+ "step": 11904
1126
+ },
1127
+ {
1128
+ "epoch": 0.53,
1129
+ "learning_rate": 8.227193216519909e-07,
1130
+ "loss": 0.1337,
1131
+ "step": 11968
1132
+ },
1133
+ {
1134
+ "epoch": 0.54,
1135
+ "learning_rate": 8.217705831776809e-07,
1136
+ "loss": 0.2247,
1137
+ "step": 12032
1138
+ },
1139
+ {
1140
+ "epoch": 0.54,
1141
+ "learning_rate": 8.20821844703371e-07,
1142
+ "loss": 0.163,
1143
+ "step": 12096
1144
+ },
1145
+ {
1146
+ "epoch": 0.54,
1147
+ "learning_rate": 8.19873106229061e-07,
1148
+ "loss": 0.1729,
1149
+ "step": 12160
1150
+ },
1151
+ {
1152
+ "epoch": 0.54,
1153
+ "learning_rate": 8.18924367754751e-07,
1154
+ "loss": 0.2133,
1155
+ "step": 12224
1156
+ },
1157
+ {
1158
+ "epoch": 0.55,
1159
+ "learning_rate": 8.179756292804412e-07,
1160
+ "loss": 0.2887,
1161
+ "step": 12288
1162
+ },
1163
+ {
1164
+ "epoch": 0.55,
1165
+ "learning_rate": 8.170268908061312e-07,
1166
+ "loss": 0.128,
1167
+ "step": 12352
1168
+ },
1169
+ {
1170
+ "epoch": 0.55,
1171
+ "learning_rate": 8.160781523318212e-07,
1172
+ "loss": 0.2019,
1173
+ "step": 12416
1174
+ },
1175
+ {
1176
+ "epoch": 0.56,
1177
+ "learning_rate": 8.151294138575113e-07,
1178
+ "loss": 0.128,
1179
+ "step": 12480
1180
+ },
1181
+ {
1182
+ "epoch": 0.56,
1183
+ "learning_rate": 8.141806753832014e-07,
1184
+ "loss": 0.1888,
1185
+ "step": 12544
1186
+ },
1187
+ {
1188
+ "epoch": 0.56,
1189
+ "learning_rate": 8.132319369088915e-07,
1190
+ "loss": 0.2176,
1191
+ "step": 12608
1192
+ },
1193
+ {
1194
+ "epoch": 0.56,
1195
+ "learning_rate": 8.122831984345814e-07,
1196
+ "loss": 0.1555,
1197
+ "step": 12672
1198
+ },
1199
+ {
1200
+ "epoch": 0.57,
1201
+ "learning_rate": 8.113344599602715e-07,
1202
+ "loss": 0.1948,
1203
+ "step": 12736
1204
+ },
1205
+ {
1206
+ "epoch": 0.57,
1207
+ "learning_rate": 8.103857214859616e-07,
1208
+ "loss": 0.233,
1209
+ "step": 12800
1210
+ },
1211
+ {
1212
+ "epoch": 0.57,
1213
+ "learning_rate": 8.094369830116516e-07,
1214
+ "loss": 0.1574,
1215
+ "step": 12864
1216
+ },
1217
+ {
1218
+ "epoch": 0.57,
1219
+ "learning_rate": 8.084882445373418e-07,
1220
+ "loss": 0.1377,
1221
+ "step": 12928
1222
+ },
1223
+ {
1224
+ "epoch": 0.58,
1225
+ "learning_rate": 8.075395060630318e-07,
1226
+ "loss": 0.1563,
1227
+ "step": 12992
1228
+ },
1229
+ {
1230
+ "epoch": 0.58,
1231
+ "learning_rate": 8.065907675887218e-07,
1232
+ "loss": 0.1345,
1233
+ "step": 13056
1234
+ },
1235
+ {
1236
+ "epoch": 0.58,
1237
+ "learning_rate": 8.05642029114412e-07,
1238
+ "loss": 0.1489,
1239
+ "step": 13120
1240
+ },
1241
+ {
1242
+ "epoch": 0.59,
1243
+ "learning_rate": 8.047081146787631e-07,
1244
+ "loss": 0.2524,
1245
+ "step": 13184
1246
+ },
1247
+ {
1248
+ "epoch": 0.59,
1249
+ "learning_rate": 8.037742002431142e-07,
1250
+ "loss": 0.1534,
1251
+ "step": 13248
1252
+ },
1253
+ {
1254
+ "epoch": 0.59,
1255
+ "learning_rate": 8.028254617688043e-07,
1256
+ "loss": 0.136,
1257
+ "step": 13312
1258
+ },
1259
+ {
1260
+ "epoch": 0.59,
1261
+ "learning_rate": 8.018767232944942e-07,
1262
+ "loss": 0.151,
1263
+ "step": 13376
1264
+ },
1265
+ {
1266
+ "epoch": 0.6,
1267
+ "learning_rate": 8.009279848201844e-07,
1268
+ "loss": 0.1423,
1269
+ "step": 13440
1270
+ },
1271
+ {
1272
+ "epoch": 0.6,
1273
+ "learning_rate": 7.999792463458744e-07,
1274
+ "loss": 0.2284,
1275
+ "step": 13504
1276
+ },
1277
+ {
1278
+ "epoch": 0.6,
1279
+ "learning_rate": 7.990305078715644e-07,
1280
+ "loss": 0.1515,
1281
+ "step": 13568
1282
+ },
1283
+ {
1284
+ "epoch": 0.61,
1285
+ "learning_rate": 7.980817693972546e-07,
1286
+ "loss": 0.1431,
1287
+ "step": 13632
1288
+ },
1289
+ {
1290
+ "epoch": 0.61,
1291
+ "learning_rate": 7.971330309229446e-07,
1292
+ "loss": 0.1759,
1293
+ "step": 13696
1294
+ },
1295
+ {
1296
+ "epoch": 0.61,
1297
+ "learning_rate": 7.961842924486347e-07,
1298
+ "loss": 0.2942,
1299
+ "step": 13760
1300
+ },
1301
+ {
1302
+ "epoch": 0.61,
1303
+ "learning_rate": 7.952355539743248e-07,
1304
+ "loss": 0.1382,
1305
+ "step": 13824
1306
+ },
1307
+ {
1308
+ "epoch": 0.62,
1309
+ "learning_rate": 7.942868155000148e-07,
1310
+ "loss": 0.181,
1311
+ "step": 13888
1312
+ },
1313
+ {
1314
+ "epoch": 0.62,
1315
+ "learning_rate": 7.933380770257049e-07,
1316
+ "loss": 0.2471,
1317
+ "step": 13952
1318
+ },
1319
+ {
1320
+ "epoch": 0.62,
1321
+ "learning_rate": 7.923893385513949e-07,
1322
+ "loss": 0.1487,
1323
+ "step": 14016
1324
+ },
1325
+ {
1326
+ "epoch": 0.63,
1327
+ "learning_rate": 7.91440600077085e-07,
1328
+ "loss": 0.1653,
1329
+ "step": 14080
1330
+ },
1331
+ {
1332
+ "epoch": 0.63,
1333
+ "learning_rate": 7.90491861602775e-07,
1334
+ "loss": 0.193,
1335
+ "step": 14144
1336
+ },
1337
+ {
1338
+ "epoch": 0.63,
1339
+ "learning_rate": 7.89543123128465e-07,
1340
+ "loss": 0.115,
1341
+ "step": 14208
1342
+ },
1343
+ {
1344
+ "epoch": 0.63,
1345
+ "learning_rate": 7.885943846541552e-07,
1346
+ "loss": 0.1413,
1347
+ "step": 14272
1348
+ },
1349
+ {
1350
+ "epoch": 0.64,
1351
+ "learning_rate": 7.876456461798452e-07,
1352
+ "loss": 0.1508,
1353
+ "step": 14336
1354
+ },
1355
+ {
1356
+ "epoch": 0.64,
1357
+ "learning_rate": 7.866969077055352e-07,
1358
+ "loss": 0.1752,
1359
+ "step": 14400
1360
+ },
1361
+ {
1362
+ "epoch": 0.64,
1363
+ "learning_rate": 7.857481692312254e-07,
1364
+ "loss": 0.2432,
1365
+ "step": 14464
1366
+ },
1367
+ {
1368
+ "epoch": 0.65,
1369
+ "learning_rate": 7.847994307569153e-07,
1370
+ "loss": 0.1978,
1371
+ "step": 14528
1372
+ },
1373
+ {
1374
+ "epoch": 0.65,
1375
+ "learning_rate": 7.838506922826054e-07,
1376
+ "loss": 0.1445,
1377
+ "step": 14592
1378
+ },
1379
+ {
1380
+ "epoch": 0.65,
1381
+ "learning_rate": 7.829019538082955e-07,
1382
+ "loss": 0.1484,
1383
+ "step": 14656
1384
+ },
1385
+ {
1386
+ "epoch": 0.65,
1387
+ "learning_rate": 7.819532153339855e-07,
1388
+ "loss": 0.1887,
1389
+ "step": 14720
1390
+ },
1391
+ {
1392
+ "epoch": 0.66,
1393
+ "learning_rate": 7.810044768596756e-07,
1394
+ "loss": 0.216,
1395
+ "step": 14784
1396
+ },
1397
+ {
1398
+ "epoch": 0.66,
1399
+ "learning_rate": 7.800557383853657e-07,
1400
+ "loss": 0.1803,
1401
+ "step": 14848
1402
+ },
1403
+ {
1404
+ "epoch": 0.66,
1405
+ "learning_rate": 7.791069999110558e-07,
1406
+ "loss": 0.1332,
1407
+ "step": 14912
1408
+ },
1409
+ {
1410
+ "epoch": 0.67,
1411
+ "learning_rate": 7.781582614367458e-07,
1412
+ "loss": 0.2439,
1413
+ "step": 14976
1414
+ },
1415
+ {
1416
+ "epoch": 0.67,
1417
+ "learning_rate": 7.772095229624358e-07,
1418
+ "loss": 0.1689,
1419
+ "step": 15040
1420
+ },
1421
+ {
1422
+ "epoch": 0.67,
1423
+ "learning_rate": 7.76260784488126e-07,
1424
+ "loss": 0.1823,
1425
+ "step": 15104
1426
+ },
1427
+ {
1428
+ "epoch": 0.67,
1429
+ "learning_rate": 7.753120460138159e-07,
1430
+ "loss": 0.1905,
1431
+ "step": 15168
1432
+ },
1433
+ {
1434
+ "epoch": 0.68,
1435
+ "learning_rate": 7.74363307539506e-07,
1436
+ "loss": 0.2558,
1437
+ "step": 15232
1438
+ },
1439
+ {
1440
+ "epoch": 0.68,
1441
+ "learning_rate": 7.734145690651961e-07,
1442
+ "loss": 0.1531,
1443
+ "step": 15296
1444
+ },
1445
+ {
1446
+ "epoch": 0.68,
1447
+ "learning_rate": 7.724658305908861e-07,
1448
+ "loss": 0.1849,
1449
+ "step": 15360
1450
+ },
1451
+ {
1452
+ "epoch": 0.69,
1453
+ "learning_rate": 7.715170921165762e-07,
1454
+ "loss": 0.1317,
1455
+ "step": 15424
1456
+ },
1457
+ {
1458
+ "epoch": 0.69,
1459
+ "learning_rate": 7.705683536422663e-07,
1460
+ "loss": 0.1096,
1461
+ "step": 15488
1462
+ },
1463
+ {
1464
+ "epoch": 0.69,
1465
+ "learning_rate": 7.696196151679564e-07,
1466
+ "loss": 0.2193,
1467
+ "step": 15552
1468
+ },
1469
+ {
1470
+ "epoch": 0.69,
1471
+ "learning_rate": 7.686708766936464e-07,
1472
+ "loss": 0.1658,
1473
+ "step": 15616
1474
+ },
1475
+ {
1476
+ "epoch": 0.7,
1477
+ "learning_rate": 7.677221382193365e-07,
1478
+ "loss": 0.1553,
1479
+ "step": 15680
1480
+ },
1481
+ {
1482
+ "epoch": 0.7,
1483
+ "learning_rate": 7.667733997450265e-07,
1484
+ "loss": 0.1772,
1485
+ "step": 15744
1486
+ },
1487
+ {
1488
+ "epoch": 0.7,
1489
+ "learning_rate": 7.658246612707165e-07,
1490
+ "loss": 0.2147,
1491
+ "step": 15808
1492
+ },
1493
+ {
1494
+ "epoch": 0.71,
1495
+ "learning_rate": 7.648759227964067e-07,
1496
+ "loss": 0.1096,
1497
+ "step": 15872
1498
+ },
1499
+ {
1500
+ "epoch": 0.71,
1501
+ "learning_rate": 7.639271843220967e-07,
1502
+ "loss": 0.1613,
1503
+ "step": 15936
1504
+ },
1505
+ {
1506
+ "epoch": 0.71,
1507
+ "learning_rate": 7.629784458477867e-07,
1508
+ "loss": 0.1488,
1509
+ "step": 16000
1510
+ },
1511
+ {
1512
+ "epoch": 0.71,
1513
+ "learning_rate": 7.620297073734768e-07,
1514
+ "loss": 0.2256,
1515
+ "step": 16064
1516
+ },
1517
+ {
1518
+ "epoch": 0.72,
1519
+ "learning_rate": 7.610809688991669e-07,
1520
+ "loss": 0.2512,
1521
+ "step": 16128
1522
+ },
1523
+ {
1524
+ "epoch": 0.72,
1525
+ "learning_rate": 7.60132230424857e-07,
1526
+ "loss": 0.1264,
1527
+ "step": 16192
1528
+ },
1529
+ {
1530
+ "epoch": 0.72,
1531
+ "learning_rate": 7.59183491950547e-07,
1532
+ "loss": 0.1162,
1533
+ "step": 16256
1534
+ },
1535
+ {
1536
+ "epoch": 0.73,
1537
+ "learning_rate": 7.58234753476237e-07,
1538
+ "loss": 0.1401,
1539
+ "step": 16320
1540
+ },
1541
+ {
1542
+ "epoch": 0.73,
1543
+ "learning_rate": 7.572860150019271e-07,
1544
+ "loss": 0.1336,
1545
+ "step": 16384
1546
+ },
1547
+ {
1548
+ "epoch": 0.73,
1549
+ "learning_rate": 7.563372765276171e-07,
1550
+ "loss": 0.1234,
1551
+ "step": 16448
1552
+ },
1553
+ {
1554
+ "epoch": 0.73,
1555
+ "learning_rate": 7.553885380533072e-07,
1556
+ "loss": 0.1195,
1557
+ "step": 16512
1558
+ },
1559
+ {
1560
+ "epoch": 0.74,
1561
+ "learning_rate": 7.544397995789973e-07,
1562
+ "loss": 0.2435,
1563
+ "step": 16576
1564
+ },
1565
+ {
1566
+ "epoch": 0.74,
1567
+ "learning_rate": 7.534910611046873e-07,
1568
+ "loss": 0.1109,
1569
+ "step": 16640
1570
+ },
1571
+ {
1572
+ "epoch": 0.74,
1573
+ "learning_rate": 7.525423226303775e-07,
1574
+ "loss": 0.2088,
1575
+ "step": 16704
1576
+ },
1577
+ {
1578
+ "epoch": 0.75,
1579
+ "learning_rate": 7.515935841560675e-07,
1580
+ "loss": 0.141,
1581
+ "step": 16768
1582
+ },
1583
+ {
1584
+ "epoch": 0.75,
1585
+ "learning_rate": 7.506448456817574e-07,
1586
+ "loss": 0.1428,
1587
+ "step": 16832
1588
+ },
1589
+ {
1590
+ "epoch": 0.75,
1591
+ "learning_rate": 7.496961072074475e-07,
1592
+ "loss": 0.1505,
1593
+ "step": 16896
1594
+ },
1595
+ {
1596
+ "epoch": 0.75,
1597
+ "learning_rate": 7.487473687331376e-07,
1598
+ "loss": 0.2152,
1599
+ "step": 16960
1600
+ },
1601
+ {
1602
+ "epoch": 0.76,
1603
+ "learning_rate": 7.477986302588277e-07,
1604
+ "loss": 0.2008,
1605
+ "step": 17024
1606
+ },
1607
+ {
1608
+ "epoch": 0.76,
1609
+ "learning_rate": 7.468498917845177e-07,
1610
+ "loss": 0.1872,
1611
+ "step": 17088
1612
+ },
1613
+ {
1614
+ "epoch": 0.76,
1615
+ "learning_rate": 7.459011533102078e-07,
1616
+ "loss": 0.1313,
1617
+ "step": 17152
1618
+ },
1619
+ {
1620
+ "epoch": 0.77,
1621
+ "learning_rate": 7.449524148358979e-07,
1622
+ "loss": 0.1099,
1623
+ "step": 17216
1624
+ },
1625
+ {
1626
+ "epoch": 0.77,
1627
+ "learning_rate": 7.440036763615879e-07,
1628
+ "loss": 0.138,
1629
+ "step": 17280
1630
+ },
1631
+ {
1632
+ "epoch": 0.77,
1633
+ "learning_rate": 7.430549378872781e-07,
1634
+ "loss": 0.1871,
1635
+ "step": 17344
1636
+ },
1637
+ {
1638
+ "epoch": 0.77,
1639
+ "learning_rate": 7.42106199412968e-07,
1640
+ "loss": 0.18,
1641
+ "step": 17408
1642
+ },
1643
+ {
1644
+ "epoch": 0.78,
1645
+ "learning_rate": 7.41157460938658e-07,
1646
+ "loss": 0.1337,
1647
+ "step": 17472
1648
+ },
1649
+ {
1650
+ "epoch": 0.78,
1651
+ "learning_rate": 7.402087224643482e-07,
1652
+ "loss": 0.1222,
1653
+ "step": 17536
1654
+ },
1655
+ {
1656
+ "epoch": 0.78,
1657
+ "learning_rate": 7.392599839900382e-07,
1658
+ "loss": 0.1434,
1659
+ "step": 17600
1660
+ },
1661
+ {
1662
+ "epoch": 0.79,
1663
+ "learning_rate": 7.383112455157283e-07,
1664
+ "loss": 0.1538,
1665
+ "step": 17664
1666
+ },
1667
+ {
1668
+ "epoch": 0.79,
1669
+ "learning_rate": 7.373625070414183e-07,
1670
+ "loss": 0.1908,
1671
+ "step": 17728
1672
+ },
1673
+ {
1674
+ "epoch": 0.79,
1675
+ "learning_rate": 7.364137685671084e-07,
1676
+ "loss": 0.1244,
1677
+ "step": 17792
1678
+ },
1679
+ {
1680
+ "epoch": 0.79,
1681
+ "learning_rate": 7.354650300927985e-07,
1682
+ "loss": 0.1593,
1683
+ "step": 17856
1684
+ },
1685
+ {
1686
+ "epoch": 0.8,
1687
+ "learning_rate": 7.345162916184885e-07,
1688
+ "loss": 0.1588,
1689
+ "step": 17920
1690
+ },
1691
+ {
1692
+ "epoch": 0.8,
1693
+ "learning_rate": 7.335675531441787e-07,
1694
+ "loss": 0.1639,
1695
+ "step": 17984
1696
+ },
1697
+ {
1698
+ "epoch": 0.8,
1699
+ "learning_rate": 7.326188146698686e-07,
1700
+ "loss": 0.1431,
1701
+ "step": 18048
1702
+ },
1703
+ {
1704
+ "epoch": 0.81,
1705
+ "learning_rate": 7.316700761955586e-07,
1706
+ "loss": 0.2002,
1707
+ "step": 18112
1708
+ },
1709
+ {
1710
+ "epoch": 0.81,
1711
+ "learning_rate": 7.307213377212488e-07,
1712
+ "loss": 0.1761,
1713
+ "step": 18176
1714
+ },
1715
+ {
1716
+ "epoch": 0.81,
1717
+ "learning_rate": 7.297725992469388e-07,
1718
+ "loss": 0.1597,
1719
+ "step": 18240
1720
+ },
1721
+ {
1722
+ "epoch": 0.81,
1723
+ "learning_rate": 7.288238607726288e-07,
1724
+ "loss": 0.1952,
1725
+ "step": 18304
1726
+ },
1727
+ {
1728
+ "epoch": 0.82,
1729
+ "learning_rate": 7.27875122298319e-07,
1730
+ "loss": 0.1843,
1731
+ "step": 18368
1732
+ },
1733
+ {
1734
+ "epoch": 0.82,
1735
+ "learning_rate": 7.26926383824009e-07,
1736
+ "loss": 0.1032,
1737
+ "step": 18432
1738
+ },
1739
+ {
1740
+ "epoch": 0.82,
1741
+ "learning_rate": 7.259776453496991e-07,
1742
+ "loss": 0.1952,
1743
+ "step": 18496
1744
+ },
1745
+ {
1746
+ "epoch": 0.83,
1747
+ "learning_rate": 7.25028906875389e-07,
1748
+ "loss": 0.193,
1749
+ "step": 18560
1750
+ },
1751
+ {
1752
+ "epoch": 0.83,
1753
+ "learning_rate": 7.240801684010791e-07,
1754
+ "loss": 0.137,
1755
+ "step": 18624
1756
+ },
1757
+ {
1758
+ "epoch": 0.83,
1759
+ "learning_rate": 7.231314299267692e-07,
1760
+ "loss": 0.1992,
1761
+ "step": 18688
1762
+ },
1763
+ {
1764
+ "epoch": 0.83,
1765
+ "learning_rate": 7.221826914524592e-07,
1766
+ "loss": 0.138,
1767
+ "step": 18752
1768
+ },
1769
+ {
1770
+ "epoch": 0.84,
1771
+ "learning_rate": 7.212339529781494e-07,
1772
+ "loss": 0.2263,
1773
+ "step": 18816
1774
+ },
1775
+ {
1776
+ "epoch": 0.84,
1777
+ "learning_rate": 7.202852145038394e-07,
1778
+ "loss": 0.2101,
1779
+ "step": 18880
1780
+ },
1781
+ {
1782
+ "epoch": 0.84,
1783
+ "learning_rate": 7.193364760295294e-07,
1784
+ "loss": 0.1731,
1785
+ "step": 18944
1786
+ },
1787
+ {
1788
+ "epoch": 0.85,
1789
+ "learning_rate": 7.183877375552196e-07,
1790
+ "loss": 0.1523,
1791
+ "step": 19008
1792
+ },
1793
+ {
1794
+ "epoch": 0.85,
1795
+ "learning_rate": 7.174389990809096e-07,
1796
+ "loss": 0.1671,
1797
+ "step": 19072
1798
+ },
1799
+ {
1800
+ "epoch": 0.85,
1801
+ "learning_rate": 7.164902606065996e-07,
1802
+ "loss": 0.1549,
1803
+ "step": 19136
1804
+ },
1805
+ {
1806
+ "epoch": 0.85,
1807
+ "learning_rate": 7.155415221322897e-07,
1808
+ "loss": 0.1346,
1809
+ "step": 19200
1810
+ },
1811
+ {
1812
+ "epoch": 0.86,
1813
+ "learning_rate": 7.145927836579797e-07,
1814
+ "loss": 0.2403,
1815
+ "step": 19264
1816
+ },
1817
+ {
1818
+ "epoch": 0.86,
1819
+ "learning_rate": 7.136588692223308e-07,
1820
+ "loss": 0.1909,
1821
+ "step": 19328
1822
+ },
1823
+ {
1824
+ "epoch": 0.86,
1825
+ "learning_rate": 7.12710130748021e-07,
1826
+ "loss": 0.1801,
1827
+ "step": 19392
1828
+ },
1829
+ {
1830
+ "epoch": 0.87,
1831
+ "learning_rate": 7.11761392273711e-07,
1832
+ "loss": 0.1196,
1833
+ "step": 19456
1834
+ },
1835
+ {
1836
+ "epoch": 0.87,
1837
+ "learning_rate": 7.10812653799401e-07,
1838
+ "loss": 0.0749,
1839
+ "step": 19520
1840
+ },
1841
+ {
1842
+ "epoch": 0.87,
1843
+ "learning_rate": 7.098639153250912e-07,
1844
+ "loss": 0.1386,
1845
+ "step": 19584
1846
+ },
1847
+ {
1848
+ "epoch": 0.87,
1849
+ "learning_rate": 7.089151768507812e-07,
1850
+ "loss": 0.219,
1851
+ "step": 19648
1852
+ },
1853
+ {
1854
+ "epoch": 0.88,
1855
+ "learning_rate": 7.079664383764713e-07,
1856
+ "loss": 0.1572,
1857
+ "step": 19712
1858
+ },
1859
+ {
1860
+ "epoch": 0.88,
1861
+ "learning_rate": 7.070176999021614e-07,
1862
+ "loss": 0.191,
1863
+ "step": 19776
1864
+ },
1865
+ {
1866
+ "epoch": 0.88,
1867
+ "learning_rate": 7.060689614278513e-07,
1868
+ "loss": 0.1882,
1869
+ "step": 19840
1870
+ },
1871
+ {
1872
+ "epoch": 0.89,
1873
+ "learning_rate": 7.051202229535414e-07,
1874
+ "loss": 0.1654,
1875
+ "step": 19904
1876
+ },
1877
+ {
1878
+ "epoch": 0.89,
1879
+ "learning_rate": 7.041714844792315e-07,
1880
+ "loss": 0.1397,
1881
+ "step": 19968
1882
+ },
1883
+ {
1884
+ "epoch": 0.89,
1885
+ "learning_rate": 7.032227460049216e-07,
1886
+ "loss": 0.1948,
1887
+ "step": 20032
1888
+ },
1889
+ {
1890
+ "epoch": 0.89,
1891
+ "learning_rate": 7.022740075306116e-07,
1892
+ "loss": 0.2171,
1893
+ "step": 20096
1894
+ },
1895
+ {
1896
+ "epoch": 0.9,
1897
+ "learning_rate": 7.013252690563016e-07,
1898
+ "loss": 0.2474,
1899
+ "step": 20160
1900
+ },
1901
+ {
1902
+ "epoch": 0.9,
1903
+ "learning_rate": 7.003765305819918e-07,
1904
+ "loss": 0.2014,
1905
+ "step": 20224
1906
+ },
1907
+ {
1908
+ "epoch": 0.9,
1909
+ "learning_rate": 6.994277921076818e-07,
1910
+ "loss": 0.1256,
1911
+ "step": 20288
1912
+ },
1913
+ {
1914
+ "epoch": 0.91,
1915
+ "learning_rate": 6.984790536333718e-07,
1916
+ "loss": 0.1634,
1917
+ "step": 20352
1918
+ },
1919
+ {
1920
+ "epoch": 0.91,
1921
+ "learning_rate": 6.975303151590619e-07,
1922
+ "loss": 0.1672,
1923
+ "step": 20416
1924
+ },
1925
+ {
1926
+ "epoch": 0.91,
1927
+ "learning_rate": 6.965815766847519e-07,
1928
+ "loss": 0.1773,
1929
+ "step": 20480
1930
+ },
1931
+ {
1932
+ "epoch": 0.91,
1933
+ "learning_rate": 6.95632838210442e-07,
1934
+ "loss": 0.1157,
1935
+ "step": 20544
1936
+ },
1937
+ {
1938
+ "epoch": 0.92,
1939
+ "learning_rate": 6.946840997361321e-07,
1940
+ "loss": 0.2241,
1941
+ "step": 20608
1942
+ },
1943
+ {
1944
+ "epoch": 0.92,
1945
+ "learning_rate": 6.937353612618222e-07,
1946
+ "loss": 0.1108,
1947
+ "step": 20672
1948
+ },
1949
+ {
1950
+ "epoch": 0.92,
1951
+ "learning_rate": 6.927866227875122e-07,
1952
+ "loss": 0.1821,
1953
+ "step": 20736
1954
+ },
1955
+ {
1956
+ "epoch": 0.93,
1957
+ "learning_rate": 6.918378843132023e-07,
1958
+ "loss": 0.1459,
1959
+ "step": 20800
1960
+ },
1961
+ {
1962
+ "epoch": 0.93,
1963
+ "learning_rate": 6.908891458388924e-07,
1964
+ "loss": 0.2022,
1965
+ "step": 20864
1966
+ },
1967
+ {
1968
+ "epoch": 0.93,
1969
+ "learning_rate": 6.899404073645823e-07,
1970
+ "loss": 0.1864,
1971
+ "step": 20928
1972
+ },
1973
+ {
1974
+ "epoch": 0.93,
1975
+ "learning_rate": 6.889916688902723e-07,
1976
+ "loss": 0.1436,
1977
+ "step": 20992
1978
+ },
1979
+ {
1980
+ "epoch": 0.94,
1981
+ "learning_rate": 6.880429304159625e-07,
1982
+ "loss": 0.1771,
1983
+ "step": 21056
1984
+ },
1985
+ {
1986
+ "epoch": 0.94,
1987
+ "learning_rate": 6.870941919416525e-07,
1988
+ "loss": 0.1782,
1989
+ "step": 21120
1990
+ },
1991
+ {
1992
+ "epoch": 0.94,
1993
+ "learning_rate": 6.861454534673426e-07,
1994
+ "loss": 0.1754,
1995
+ "step": 21184
1996
+ },
1997
+ {
1998
+ "epoch": 0.94,
1999
+ "learning_rate": 6.851967149930327e-07,
2000
+ "loss": 0.1483,
2001
+ "step": 21248
2002
+ },
2003
+ {
2004
+ "epoch": 0.95,
2005
+ "learning_rate": 6.842479765187227e-07,
2006
+ "loss": 0.1373,
2007
+ "step": 21312
2008
+ },
2009
+ {
2010
+ "epoch": 0.95,
2011
+ "learning_rate": 6.833140620830739e-07,
2012
+ "loss": 0.219,
2013
+ "step": 21376
2014
+ },
2015
+ {
2016
+ "epoch": 0.95,
2017
+ "learning_rate": 6.82365323608764e-07,
2018
+ "loss": 0.1474,
2019
+ "step": 21440
2020
+ },
2021
+ {
2022
+ "epoch": 0.96,
2023
+ "learning_rate": 6.81416585134454e-07,
2024
+ "loss": 0.1713,
2025
+ "step": 21504
2026
+ },
2027
+ {
2028
+ "epoch": 0.96,
2029
+ "learning_rate": 6.80467846660144e-07,
2030
+ "loss": 0.1034,
2031
+ "step": 21568
2032
+ },
2033
+ {
2034
+ "epoch": 0.96,
2035
+ "learning_rate": 6.795191081858341e-07,
2036
+ "loss": 0.185,
2037
+ "step": 21632
2038
+ },
2039
+ {
2040
+ "epoch": 0.96,
2041
+ "learning_rate": 6.785703697115241e-07,
2042
+ "loss": 0.284,
2043
+ "step": 21696
2044
+ },
2045
+ {
2046
+ "epoch": 0.97,
2047
+ "learning_rate": 6.776216312372143e-07,
2048
+ "loss": 0.1953,
2049
+ "step": 21760
2050
+ },
2051
+ {
2052
+ "epoch": 0.97,
2053
+ "learning_rate": 6.766728927629043e-07,
2054
+ "loss": 0.168,
2055
+ "step": 21824
2056
+ },
2057
+ {
2058
+ "epoch": 0.97,
2059
+ "learning_rate": 6.757241542885943e-07,
2060
+ "loss": 0.1852,
2061
+ "step": 21888
2062
+ },
2063
+ {
2064
+ "epoch": 0.98,
2065
+ "learning_rate": 6.747754158142845e-07,
2066
+ "loss": 0.1358,
2067
+ "step": 21952
2068
+ },
2069
+ {
2070
+ "epoch": 0.98,
2071
+ "learning_rate": 6.738266773399745e-07,
2072
+ "loss": 0.1885,
2073
+ "step": 22016
2074
+ },
2075
+ {
2076
+ "epoch": 0.98,
2077
+ "learning_rate": 6.728779388656646e-07,
2078
+ "loss": 0.22,
2079
+ "step": 22080
2080
+ },
2081
+ {
2082
+ "epoch": 0.98,
2083
+ "learning_rate": 6.719292003913545e-07,
2084
+ "loss": 0.214,
2085
+ "step": 22144
2086
+ },
2087
+ {
2088
+ "epoch": 0.99,
2089
+ "learning_rate": 6.709804619170446e-07,
2090
+ "loss": 0.1198,
2091
+ "step": 22208
2092
+ },
2093
+ {
2094
+ "epoch": 0.99,
2095
+ "learning_rate": 6.700317234427347e-07,
2096
+ "loss": 0.1458,
2097
+ "step": 22272
2098
+ },
2099
+ {
2100
+ "epoch": 0.99,
2101
+ "learning_rate": 6.690829849684247e-07,
2102
+ "loss": 0.2405,
2103
+ "step": 22336
2104
+ },
2105
+ {
2106
+ "epoch": 1.0,
2107
+ "learning_rate": 6.681342464941149e-07,
2108
+ "loss": 0.1162,
2109
+ "step": 22400
2110
+ },
2111
+ {
2112
+ "epoch": 1.0,
2113
+ "learning_rate": 6.671855080198049e-07,
2114
+ "loss": 0.1508,
2115
+ "step": 22464
2116
+ }
2117
+ ],
2118
+ "logging_steps": 64,
2119
+ "max_steps": 67458,
2120
+ "num_input_tokens_seen": 0,
2121
+ "num_train_epochs": 3,
2122
+ "save_steps": 22486,
2123
+ "total_flos": 4.77416235737088e+19,
2124
+ "train_batch_size": 1,
2125
+ "trial_name": null,
2126
+ "trial_params": null
2127
+ }
checkpoint-22486/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08851645278532aec84b541ed50bff863f0a3babdffb268c3cf70f72a1796795
3
+ size 5048
checkpoint-44972/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-44972/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-44972/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95bb2a5c0ff9fcc9188a594fdf14a3526edd270a195898ef3ad8285e4a796659
3
+ size 4992706480
checkpoint-44972/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:793633ac02334f9d053c4810b804fa578d227cd26fcb603c2079e8046638d54c
3
+ size 1180663192
checkpoint-44972/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-44972/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48925f797b8d64cdea1c8cbcbce572d0ae353054cc6ebadc534fc185f6243289
3
+ size 3095074288
checkpoint-44972/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-44972/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ff58d3a8c580e707cb2056fe8f0f2e992e02fb0cd93c2f47bffb6c0b711439f4
3
+ size 14244
checkpoint-44972/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f790d5a7c30d8e158794933fdaa3a3c147b10fba3eede14a6a83a32f26717af
3
+ size 1064
checkpoint-44972/trainer_state.json ADDED
@@ -0,0 +1,4233 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
+ "eval_steps": 500,
6
+ "global_step": 44972,
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": 9.991105576803343e-07,
14
+ "loss": 3.5693,
15
+ "step": 64
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 9.981618192060246e-07,
20
+ "loss": 1.3039,
21
+ "step": 128
22
+ },
23
+ {
24
+ "epoch": 0.01,
25
+ "learning_rate": 9.972130807317145e-07,
26
+ "loss": 0.6178,
27
+ "step": 192
28
+ },
29
+ {
30
+ "epoch": 0.01,
31
+ "learning_rate": 9.962643422574046e-07,
32
+ "loss": 0.2217,
33
+ "step": 256
34
+ },
35
+ {
36
+ "epoch": 0.01,
37
+ "learning_rate": 9.953156037830946e-07,
38
+ "loss": 0.135,
39
+ "step": 320
40
+ },
41
+ {
42
+ "epoch": 0.02,
43
+ "learning_rate": 9.943668653087847e-07,
44
+ "loss": 0.2547,
45
+ "step": 384
46
+ },
47
+ {
48
+ "epoch": 0.02,
49
+ "learning_rate": 9.934181268344748e-07,
50
+ "loss": 0.2357,
51
+ "step": 448
52
+ },
53
+ {
54
+ "epoch": 0.02,
55
+ "learning_rate": 9.92513860476148e-07,
56
+ "loss": 0.4032,
57
+ "step": 512
58
+ },
59
+ {
60
+ "epoch": 0.03,
61
+ "learning_rate": 9.915651220018381e-07,
62
+ "loss": 0.2513,
63
+ "step": 576
64
+ },
65
+ {
66
+ "epoch": 0.03,
67
+ "learning_rate": 9.906163835275282e-07,
68
+ "loss": 0.2058,
69
+ "step": 640
70
+ },
71
+ {
72
+ "epoch": 0.03,
73
+ "learning_rate": 9.89667645053218e-07,
74
+ "loss": 0.2276,
75
+ "step": 704
76
+ },
77
+ {
78
+ "epoch": 0.03,
79
+ "learning_rate": 9.887189065789084e-07,
80
+ "loss": 0.1798,
81
+ "step": 768
82
+ },
83
+ {
84
+ "epoch": 0.04,
85
+ "learning_rate": 9.877701681045983e-07,
86
+ "loss": 0.2646,
87
+ "step": 832
88
+ },
89
+ {
90
+ "epoch": 0.04,
91
+ "learning_rate": 9.868214296302884e-07,
92
+ "loss": 0.2434,
93
+ "step": 896
94
+ },
95
+ {
96
+ "epoch": 0.04,
97
+ "learning_rate": 9.858726911559785e-07,
98
+ "loss": 0.3211,
99
+ "step": 960
100
+ },
101
+ {
102
+ "epoch": 0.05,
103
+ "learning_rate": 9.849239526816686e-07,
104
+ "loss": 0.1601,
105
+ "step": 1024
106
+ },
107
+ {
108
+ "epoch": 0.05,
109
+ "learning_rate": 9.839752142073586e-07,
110
+ "loss": 0.2012,
111
+ "step": 1088
112
+ },
113
+ {
114
+ "epoch": 0.05,
115
+ "learning_rate": 9.830264757330487e-07,
116
+ "loss": 0.1635,
117
+ "step": 1152
118
+ },
119
+ {
120
+ "epoch": 0.05,
121
+ "learning_rate": 9.820777372587388e-07,
122
+ "loss": 0.2808,
123
+ "step": 1216
124
+ },
125
+ {
126
+ "epoch": 0.06,
127
+ "learning_rate": 9.811289987844287e-07,
128
+ "loss": 0.1622,
129
+ "step": 1280
130
+ },
131
+ {
132
+ "epoch": 0.06,
133
+ "learning_rate": 9.801802603101188e-07,
134
+ "loss": 0.23,
135
+ "step": 1344
136
+ },
137
+ {
138
+ "epoch": 0.06,
139
+ "learning_rate": 9.792315218358089e-07,
140
+ "loss": 0.1615,
141
+ "step": 1408
142
+ },
143
+ {
144
+ "epoch": 0.07,
145
+ "learning_rate": 9.78282783361499e-07,
146
+ "loss": 0.1424,
147
+ "step": 1472
148
+ },
149
+ {
150
+ "epoch": 0.07,
151
+ "learning_rate": 9.77334044887189e-07,
152
+ "loss": 0.2007,
153
+ "step": 1536
154
+ },
155
+ {
156
+ "epoch": 0.07,
157
+ "learning_rate": 9.763853064128792e-07,
158
+ "loss": 0.3649,
159
+ "step": 1600
160
+ },
161
+ {
162
+ "epoch": 0.07,
163
+ "learning_rate": 9.75436567938569e-07,
164
+ "loss": 0.1149,
165
+ "step": 1664
166
+ },
167
+ {
168
+ "epoch": 0.08,
169
+ "learning_rate": 9.744878294642591e-07,
170
+ "loss": 0.1882,
171
+ "step": 1728
172
+ },
173
+ {
174
+ "epoch": 0.08,
175
+ "learning_rate": 9.735390909899492e-07,
176
+ "loss": 0.2598,
177
+ "step": 1792
178
+ },
179
+ {
180
+ "epoch": 0.08,
181
+ "learning_rate": 9.725903525156393e-07,
182
+ "loss": 0.2198,
183
+ "step": 1856
184
+ },
185
+ {
186
+ "epoch": 0.09,
187
+ "learning_rate": 9.716416140413294e-07,
188
+ "loss": 0.1786,
189
+ "step": 1920
190
+ },
191
+ {
192
+ "epoch": 0.09,
193
+ "learning_rate": 9.706928755670195e-07,
194
+ "loss": 0.2713,
195
+ "step": 1984
196
+ },
197
+ {
198
+ "epoch": 0.09,
199
+ "learning_rate": 9.697441370927096e-07,
200
+ "loss": 0.2076,
201
+ "step": 2048
202
+ },
203
+ {
204
+ "epoch": 0.09,
205
+ "learning_rate": 9.687953986183995e-07,
206
+ "loss": 0.2343,
207
+ "step": 2112
208
+ },
209
+ {
210
+ "epoch": 0.1,
211
+ "learning_rate": 9.678466601440898e-07,
212
+ "loss": 0.3182,
213
+ "step": 2176
214
+ },
215
+ {
216
+ "epoch": 0.1,
217
+ "learning_rate": 9.668979216697796e-07,
218
+ "loss": 0.2102,
219
+ "step": 2240
220
+ },
221
+ {
222
+ "epoch": 0.1,
223
+ "learning_rate": 9.659491831954697e-07,
224
+ "loss": 0.2687,
225
+ "step": 2304
226
+ },
227
+ {
228
+ "epoch": 0.11,
229
+ "learning_rate": 9.650004447211598e-07,
230
+ "loss": 0.2186,
231
+ "step": 2368
232
+ },
233
+ {
234
+ "epoch": 0.11,
235
+ "learning_rate": 9.6405170624685e-07,
236
+ "loss": 0.1295,
237
+ "step": 2432
238
+ },
239
+ {
240
+ "epoch": 0.11,
241
+ "learning_rate": 9.631029677725398e-07,
242
+ "loss": 0.1542,
243
+ "step": 2496
244
+ },
245
+ {
246
+ "epoch": 0.11,
247
+ "learning_rate": 9.621542292982299e-07,
248
+ "loss": 0.1365,
249
+ "step": 2560
250
+ },
251
+ {
252
+ "epoch": 0.12,
253
+ "learning_rate": 9.6120549082392e-07,
254
+ "loss": 0.2106,
255
+ "step": 2624
256
+ },
257
+ {
258
+ "epoch": 0.12,
259
+ "learning_rate": 9.6025675234961e-07,
260
+ "loss": 0.2753,
261
+ "step": 2688
262
+ },
263
+ {
264
+ "epoch": 0.12,
265
+ "learning_rate": 9.593080138753002e-07,
266
+ "loss": 0.2162,
267
+ "step": 2752
268
+ },
269
+ {
270
+ "epoch": 0.13,
271
+ "learning_rate": 9.583592754009902e-07,
272
+ "loss": 0.2186,
273
+ "step": 2816
274
+ },
275
+ {
276
+ "epoch": 0.13,
277
+ "learning_rate": 9.574105369266803e-07,
278
+ "loss": 0.0991,
279
+ "step": 2880
280
+ },
281
+ {
282
+ "epoch": 0.13,
283
+ "learning_rate": 9.564617984523702e-07,
284
+ "loss": 0.1616,
285
+ "step": 2944
286
+ },
287
+ {
288
+ "epoch": 0.13,
289
+ "learning_rate": 9.555130599780605e-07,
290
+ "loss": 0.2029,
291
+ "step": 3008
292
+ },
293
+ {
294
+ "epoch": 0.14,
295
+ "learning_rate": 9.545643215037504e-07,
296
+ "loss": 0.1659,
297
+ "step": 3072
298
+ },
299
+ {
300
+ "epoch": 0.14,
301
+ "learning_rate": 9.536155830294405e-07,
302
+ "loss": 0.1887,
303
+ "step": 3136
304
+ },
305
+ {
306
+ "epoch": 0.14,
307
+ "learning_rate": 9.526668445551306e-07,
308
+ "loss": 0.2626,
309
+ "step": 3200
310
+ },
311
+ {
312
+ "epoch": 0.15,
313
+ "learning_rate": 9.517181060808206e-07,
314
+ "loss": 0.1325,
315
+ "step": 3264
316
+ },
317
+ {
318
+ "epoch": 0.15,
319
+ "learning_rate": 9.507693676065108e-07,
320
+ "loss": 0.2125,
321
+ "step": 3328
322
+ },
323
+ {
324
+ "epoch": 0.15,
325
+ "learning_rate": 9.498206291322007e-07,
326
+ "loss": 0.1926,
327
+ "step": 3392
328
+ },
329
+ {
330
+ "epoch": 0.15,
331
+ "learning_rate": 9.488718906578907e-07,
332
+ "loss": 0.2237,
333
+ "step": 3456
334
+ },
335
+ {
336
+ "epoch": 0.16,
337
+ "learning_rate": 9.479231521835809e-07,
338
+ "loss": 0.1283,
339
+ "step": 3520
340
+ },
341
+ {
342
+ "epoch": 0.16,
343
+ "learning_rate": 9.469744137092709e-07,
344
+ "loss": 0.2052,
345
+ "step": 3584
346
+ },
347
+ {
348
+ "epoch": 0.16,
349
+ "learning_rate": 9.46025675234961e-07,
350
+ "loss": 0.1983,
351
+ "step": 3648
352
+ },
353
+ {
354
+ "epoch": 0.17,
355
+ "learning_rate": 9.45076936760651e-07,
356
+ "loss": 0.1976,
357
+ "step": 3712
358
+ },
359
+ {
360
+ "epoch": 0.17,
361
+ "learning_rate": 9.441281982863411e-07,
362
+ "loss": 0.1999,
363
+ "step": 3776
364
+ },
365
+ {
366
+ "epoch": 0.17,
367
+ "learning_rate": 9.431794598120312e-07,
368
+ "loss": 0.2231,
369
+ "step": 3840
370
+ },
371
+ {
372
+ "epoch": 0.17,
373
+ "learning_rate": 9.422307213377212e-07,
374
+ "loss": 0.2057,
375
+ "step": 3904
376
+ },
377
+ {
378
+ "epoch": 0.18,
379
+ "learning_rate": 9.412819828634113e-07,
380
+ "loss": 0.2376,
381
+ "step": 3968
382
+ },
383
+ {
384
+ "epoch": 0.18,
385
+ "learning_rate": 9.403332443891013e-07,
386
+ "loss": 0.1325,
387
+ "step": 4032
388
+ },
389
+ {
390
+ "epoch": 0.18,
391
+ "learning_rate": 9.393845059147913e-07,
392
+ "loss": 0.2595,
393
+ "step": 4096
394
+ },
395
+ {
396
+ "epoch": 0.19,
397
+ "learning_rate": 9.384357674404815e-07,
398
+ "loss": 0.1817,
399
+ "step": 4160
400
+ },
401
+ {
402
+ "epoch": 0.19,
403
+ "learning_rate": 9.374870289661715e-07,
404
+ "loss": 0.2224,
405
+ "step": 4224
406
+ },
407
+ {
408
+ "epoch": 0.19,
409
+ "learning_rate": 9.365382904918616e-07,
410
+ "loss": 0.1444,
411
+ "step": 4288
412
+ },
413
+ {
414
+ "epoch": 0.19,
415
+ "learning_rate": 9.355895520175517e-07,
416
+ "loss": 0.2695,
417
+ "step": 4352
418
+ },
419
+ {
420
+ "epoch": 0.2,
421
+ "learning_rate": 9.346408135432417e-07,
422
+ "loss": 0.1859,
423
+ "step": 4416
424
+ },
425
+ {
426
+ "epoch": 0.2,
427
+ "learning_rate": 9.336920750689318e-07,
428
+ "loss": 0.1551,
429
+ "step": 4480
430
+ },
431
+ {
432
+ "epoch": 0.2,
433
+ "learning_rate": 9.327433365946217e-07,
434
+ "loss": 0.2093,
435
+ "step": 4544
436
+ },
437
+ {
438
+ "epoch": 0.2,
439
+ "learning_rate": 9.317945981203118e-07,
440
+ "loss": 0.1922,
441
+ "step": 4608
442
+ },
443
+ {
444
+ "epoch": 0.21,
445
+ "learning_rate": 9.308458596460019e-07,
446
+ "loss": 0.2056,
447
+ "step": 4672
448
+ },
449
+ {
450
+ "epoch": 0.21,
451
+ "learning_rate": 9.298971211716919e-07,
452
+ "loss": 0.2087,
453
+ "step": 4736
454
+ },
455
+ {
456
+ "epoch": 0.21,
457
+ "learning_rate": 9.289483826973821e-07,
458
+ "loss": 0.1552,
459
+ "step": 4800
460
+ },
461
+ {
462
+ "epoch": 0.22,
463
+ "learning_rate": 9.279996442230721e-07,
464
+ "loss": 0.1619,
465
+ "step": 4864
466
+ },
467
+ {
468
+ "epoch": 0.22,
469
+ "learning_rate": 9.270509057487621e-07,
470
+ "loss": 0.2634,
471
+ "step": 4928
472
+ },
473
+ {
474
+ "epoch": 0.22,
475
+ "learning_rate": 9.261021672744523e-07,
476
+ "loss": 0.1459,
477
+ "step": 4992
478
+ },
479
+ {
480
+ "epoch": 0.22,
481
+ "learning_rate": 9.251534288001423e-07,
482
+ "loss": 0.225,
483
+ "step": 5056
484
+ },
485
+ {
486
+ "epoch": 0.23,
487
+ "learning_rate": 9.242046903258323e-07,
488
+ "loss": 0.1878,
489
+ "step": 5120
490
+ },
491
+ {
492
+ "epoch": 0.23,
493
+ "learning_rate": 9.232559518515224e-07,
494
+ "loss": 0.1286,
495
+ "step": 5184
496
+ },
497
+ {
498
+ "epoch": 0.23,
499
+ "learning_rate": 9.223072133772124e-07,
500
+ "loss": 0.1059,
501
+ "step": 5248
502
+ },
503
+ {
504
+ "epoch": 0.24,
505
+ "learning_rate": 9.213584749029025e-07,
506
+ "loss": 0.1541,
507
+ "step": 5312
508
+ },
509
+ {
510
+ "epoch": 0.24,
511
+ "learning_rate": 9.204097364285925e-07,
512
+ "loss": 0.2427,
513
+ "step": 5376
514
+ },
515
+ {
516
+ "epoch": 0.24,
517
+ "learning_rate": 9.194609979542827e-07,
518
+ "loss": 0.1368,
519
+ "step": 5440
520
+ },
521
+ {
522
+ "epoch": 0.24,
523
+ "learning_rate": 9.185122594799727e-07,
524
+ "loss": 0.2477,
525
+ "step": 5504
526
+ },
527
+ {
528
+ "epoch": 0.25,
529
+ "learning_rate": 9.175635210056627e-07,
530
+ "loss": 0.1385,
531
+ "step": 5568
532
+ },
533
+ {
534
+ "epoch": 0.25,
535
+ "learning_rate": 9.166147825313529e-07,
536
+ "loss": 0.2537,
537
+ "step": 5632
538
+ },
539
+ {
540
+ "epoch": 0.25,
541
+ "learning_rate": 9.156660440570428e-07,
542
+ "loss": 0.2374,
543
+ "step": 5696
544
+ },
545
+ {
546
+ "epoch": 0.26,
547
+ "learning_rate": 9.147173055827329e-07,
548
+ "loss": 0.1942,
549
+ "step": 5760
550
+ },
551
+ {
552
+ "epoch": 0.26,
553
+ "learning_rate": 9.13768567108423e-07,
554
+ "loss": 0.1416,
555
+ "step": 5824
556
+ },
557
+ {
558
+ "epoch": 0.26,
559
+ "learning_rate": 9.12819828634113e-07,
560
+ "loss": 0.2205,
561
+ "step": 5888
562
+ },
563
+ {
564
+ "epoch": 0.26,
565
+ "learning_rate": 9.118710901598031e-07,
566
+ "loss": 0.1265,
567
+ "step": 5952
568
+ },
569
+ {
570
+ "epoch": 0.27,
571
+ "learning_rate": 9.109223516854932e-07,
572
+ "loss": 0.1972,
573
+ "step": 6016
574
+ },
575
+ {
576
+ "epoch": 0.27,
577
+ "learning_rate": 9.099736132111833e-07,
578
+ "loss": 0.2119,
579
+ "step": 6080
580
+ },
581
+ {
582
+ "epoch": 0.27,
583
+ "learning_rate": 9.090248747368733e-07,
584
+ "loss": 0.1284,
585
+ "step": 6144
586
+ },
587
+ {
588
+ "epoch": 0.28,
589
+ "learning_rate": 9.080761362625632e-07,
590
+ "loss": 0.1665,
591
+ "step": 6208
592
+ },
593
+ {
594
+ "epoch": 0.28,
595
+ "learning_rate": 9.071273977882534e-07,
596
+ "loss": 0.1545,
597
+ "step": 6272
598
+ },
599
+ {
600
+ "epoch": 0.28,
601
+ "learning_rate": 9.061786593139434e-07,
602
+ "loss": 0.1505,
603
+ "step": 6336
604
+ },
605
+ {
606
+ "epoch": 0.28,
607
+ "learning_rate": 9.052299208396334e-07,
608
+ "loss": 0.1871,
609
+ "step": 6400
610
+ },
611
+ {
612
+ "epoch": 0.29,
613
+ "learning_rate": 9.042811823653236e-07,
614
+ "loss": 0.2302,
615
+ "step": 6464
616
+ },
617
+ {
618
+ "epoch": 0.29,
619
+ "learning_rate": 9.033324438910136e-07,
620
+ "loss": 0.1395,
621
+ "step": 6528
622
+ },
623
+ {
624
+ "epoch": 0.29,
625
+ "learning_rate": 9.023837054167037e-07,
626
+ "loss": 0.2195,
627
+ "step": 6592
628
+ },
629
+ {
630
+ "epoch": 0.3,
631
+ "learning_rate": 9.014349669423938e-07,
632
+ "loss": 0.2419,
633
+ "step": 6656
634
+ },
635
+ {
636
+ "epoch": 0.3,
637
+ "learning_rate": 9.004862284680838e-07,
638
+ "loss": 0.2173,
639
+ "step": 6720
640
+ },
641
+ {
642
+ "epoch": 0.3,
643
+ "learning_rate": 8.995374899937739e-07,
644
+ "loss": 0.2522,
645
+ "step": 6784
646
+ },
647
+ {
648
+ "epoch": 0.3,
649
+ "learning_rate": 8.985887515194639e-07,
650
+ "loss": 0.2317,
651
+ "step": 6848
652
+ },
653
+ {
654
+ "epoch": 0.31,
655
+ "learning_rate": 8.97640013045154e-07,
656
+ "loss": 0.1959,
657
+ "step": 6912
658
+ },
659
+ {
660
+ "epoch": 0.31,
661
+ "learning_rate": 8.96691274570844e-07,
662
+ "loss": 0.1889,
663
+ "step": 6976
664
+ },
665
+ {
666
+ "epoch": 0.31,
667
+ "learning_rate": 8.95742536096534e-07,
668
+ "loss": 0.2131,
669
+ "step": 7040
670
+ },
671
+ {
672
+ "epoch": 0.32,
673
+ "learning_rate": 8.947937976222242e-07,
674
+ "loss": 0.1321,
675
+ "step": 7104
676
+ },
677
+ {
678
+ "epoch": 0.32,
679
+ "learning_rate": 8.938450591479142e-07,
680
+ "loss": 0.1679,
681
+ "step": 7168
682
+ },
683
+ {
684
+ "epoch": 0.32,
685
+ "learning_rate": 8.928963206736043e-07,
686
+ "loss": 0.1826,
687
+ "step": 7232
688
+ },
689
+ {
690
+ "epoch": 0.32,
691
+ "learning_rate": 8.919475821992944e-07,
692
+ "loss": 0.2144,
693
+ "step": 7296
694
+ },
695
+ {
696
+ "epoch": 0.33,
697
+ "learning_rate": 8.909988437249843e-07,
698
+ "loss": 0.1614,
699
+ "step": 7360
700
+ },
701
+ {
702
+ "epoch": 0.33,
703
+ "learning_rate": 8.900501052506744e-07,
704
+ "loss": 0.241,
705
+ "step": 7424
706
+ },
707
+ {
708
+ "epoch": 0.33,
709
+ "learning_rate": 8.891013667763645e-07,
710
+ "loss": 0.1825,
711
+ "step": 7488
712
+ },
713
+ {
714
+ "epoch": 0.34,
715
+ "learning_rate": 8.881526283020546e-07,
716
+ "loss": 0.17,
717
+ "step": 7552
718
+ },
719
+ {
720
+ "epoch": 0.34,
721
+ "learning_rate": 8.872038898277446e-07,
722
+ "loss": 0.1562,
723
+ "step": 7616
724
+ },
725
+ {
726
+ "epoch": 0.34,
727
+ "learning_rate": 8.862551513534347e-07,
728
+ "loss": 0.2264,
729
+ "step": 7680
730
+ },
731
+ {
732
+ "epoch": 0.34,
733
+ "learning_rate": 8.853064128791248e-07,
734
+ "loss": 0.1325,
735
+ "step": 7744
736
+ },
737
+ {
738
+ "epoch": 0.35,
739
+ "learning_rate": 8.843576744048148e-07,
740
+ "loss": 0.1601,
741
+ "step": 7808
742
+ },
743
+ {
744
+ "epoch": 0.35,
745
+ "learning_rate": 8.83408935930505e-07,
746
+ "loss": 0.1784,
747
+ "step": 7872
748
+ },
749
+ {
750
+ "epoch": 0.35,
751
+ "learning_rate": 8.82460197456195e-07,
752
+ "loss": 0.1447,
753
+ "step": 7936
754
+ },
755
+ {
756
+ "epoch": 0.36,
757
+ "learning_rate": 8.815114589818849e-07,
758
+ "loss": 0.166,
759
+ "step": 8000
760
+ },
761
+ {
762
+ "epoch": 0.36,
763
+ "learning_rate": 8.80562720507575e-07,
764
+ "loss": 0.1554,
765
+ "step": 8064
766
+ },
767
+ {
768
+ "epoch": 0.36,
769
+ "learning_rate": 8.796139820332651e-07,
770
+ "loss": 0.1097,
771
+ "step": 8128
772
+ },
773
+ {
774
+ "epoch": 0.36,
775
+ "learning_rate": 8.786652435589552e-07,
776
+ "loss": 0.1322,
777
+ "step": 8192
778
+ },
779
+ {
780
+ "epoch": 0.37,
781
+ "learning_rate": 8.777165050846452e-07,
782
+ "loss": 0.262,
783
+ "step": 8256
784
+ },
785
+ {
786
+ "epoch": 0.37,
787
+ "learning_rate": 8.767677666103353e-07,
788
+ "loss": 0.1755,
789
+ "step": 8320
790
+ },
791
+ {
792
+ "epoch": 0.37,
793
+ "learning_rate": 8.758190281360254e-07,
794
+ "loss": 0.1646,
795
+ "step": 8384
796
+ },
797
+ {
798
+ "epoch": 0.38,
799
+ "learning_rate": 8.748702896617154e-07,
800
+ "loss": 0.1481,
801
+ "step": 8448
802
+ },
803
+ {
804
+ "epoch": 0.38,
805
+ "learning_rate": 8.739215511874055e-07,
806
+ "loss": 0.0985,
807
+ "step": 8512
808
+ },
809
+ {
810
+ "epoch": 0.38,
811
+ "learning_rate": 8.729728127130955e-07,
812
+ "loss": 0.1401,
813
+ "step": 8576
814
+ },
815
+ {
816
+ "epoch": 0.38,
817
+ "learning_rate": 8.720240742387855e-07,
818
+ "loss": 0.2057,
819
+ "step": 8640
820
+ },
821
+ {
822
+ "epoch": 0.39,
823
+ "learning_rate": 8.710753357644757e-07,
824
+ "loss": 0.1677,
825
+ "step": 8704
826
+ },
827
+ {
828
+ "epoch": 0.39,
829
+ "learning_rate": 8.701265972901657e-07,
830
+ "loss": 0.1398,
831
+ "step": 8768
832
+ },
833
+ {
834
+ "epoch": 0.39,
835
+ "learning_rate": 8.691778588158557e-07,
836
+ "loss": 0.1665,
837
+ "step": 8832
838
+ },
839
+ {
840
+ "epoch": 0.4,
841
+ "learning_rate": 8.682439443802069e-07,
842
+ "loss": 0.1737,
843
+ "step": 8896
844
+ },
845
+ {
846
+ "epoch": 0.4,
847
+ "learning_rate": 8.67295205905897e-07,
848
+ "loss": 0.1642,
849
+ "step": 8960
850
+ },
851
+ {
852
+ "epoch": 0.4,
853
+ "learning_rate": 8.66346467431587e-07,
854
+ "loss": 0.1487,
855
+ "step": 9024
856
+ },
857
+ {
858
+ "epoch": 0.4,
859
+ "learning_rate": 8.653977289572771e-07,
860
+ "loss": 0.186,
861
+ "step": 9088
862
+ },
863
+ {
864
+ "epoch": 0.41,
865
+ "learning_rate": 8.644489904829672e-07,
866
+ "loss": 0.1902,
867
+ "step": 9152
868
+ },
869
+ {
870
+ "epoch": 0.41,
871
+ "learning_rate": 8.635002520086571e-07,
872
+ "loss": 0.1977,
873
+ "step": 9216
874
+ },
875
+ {
876
+ "epoch": 0.41,
877
+ "learning_rate": 8.625515135343473e-07,
878
+ "loss": 0.1853,
879
+ "step": 9280
880
+ },
881
+ {
882
+ "epoch": 0.42,
883
+ "learning_rate": 8.616027750600373e-07,
884
+ "loss": 0.1156,
885
+ "step": 9344
886
+ },
887
+ {
888
+ "epoch": 0.42,
889
+ "learning_rate": 8.606540365857273e-07,
890
+ "loss": 0.179,
891
+ "step": 9408
892
+ },
893
+ {
894
+ "epoch": 0.42,
895
+ "learning_rate": 8.597052981114175e-07,
896
+ "loss": 0.1978,
897
+ "step": 9472
898
+ },
899
+ {
900
+ "epoch": 0.42,
901
+ "learning_rate": 8.587565596371075e-07,
902
+ "loss": 0.1735,
903
+ "step": 9536
904
+ },
905
+ {
906
+ "epoch": 0.43,
907
+ "learning_rate": 8.578078211627976e-07,
908
+ "loss": 0.1579,
909
+ "step": 9600
910
+ },
911
+ {
912
+ "epoch": 0.43,
913
+ "learning_rate": 8.568590826884876e-07,
914
+ "loss": 0.1444,
915
+ "step": 9664
916
+ },
917
+ {
918
+ "epoch": 0.43,
919
+ "learning_rate": 8.559103442141777e-07,
920
+ "loss": 0.1664,
921
+ "step": 9728
922
+ },
923
+ {
924
+ "epoch": 0.44,
925
+ "learning_rate": 8.549616057398677e-07,
926
+ "loss": 0.1715,
927
+ "step": 9792
928
+ },
929
+ {
930
+ "epoch": 0.44,
931
+ "learning_rate": 8.540128672655577e-07,
932
+ "loss": 0.1189,
933
+ "step": 9856
934
+ },
935
+ {
936
+ "epoch": 0.44,
937
+ "learning_rate": 8.530641287912479e-07,
938
+ "loss": 0.1255,
939
+ "step": 9920
940
+ },
941
+ {
942
+ "epoch": 0.44,
943
+ "learning_rate": 8.521153903169379e-07,
944
+ "loss": 0.2314,
945
+ "step": 9984
946
+ },
947
+ {
948
+ "epoch": 0.45,
949
+ "learning_rate": 8.511814758812891e-07,
950
+ "loss": 0.18,
951
+ "step": 10048
952
+ },
953
+ {
954
+ "epoch": 0.45,
955
+ "learning_rate": 8.502327374069791e-07,
956
+ "loss": 0.1948,
957
+ "step": 10112
958
+ },
959
+ {
960
+ "epoch": 0.45,
961
+ "learning_rate": 8.492839989326692e-07,
962
+ "loss": 0.1763,
963
+ "step": 10176
964
+ },
965
+ {
966
+ "epoch": 0.46,
967
+ "learning_rate": 8.483352604583593e-07,
968
+ "loss": 0.1984,
969
+ "step": 10240
970
+ },
971
+ {
972
+ "epoch": 0.46,
973
+ "learning_rate": 8.473865219840493e-07,
974
+ "loss": 0.169,
975
+ "step": 10304
976
+ },
977
+ {
978
+ "epoch": 0.46,
979
+ "learning_rate": 8.464377835097394e-07,
980
+ "loss": 0.1268,
981
+ "step": 10368
982
+ },
983
+ {
984
+ "epoch": 0.46,
985
+ "learning_rate": 8.454890450354294e-07,
986
+ "loss": 0.2261,
987
+ "step": 10432
988
+ },
989
+ {
990
+ "epoch": 0.47,
991
+ "learning_rate": 8.445403065611195e-07,
992
+ "loss": 0.1751,
993
+ "step": 10496
994
+ },
995
+ {
996
+ "epoch": 0.47,
997
+ "learning_rate": 8.435915680868095e-07,
998
+ "loss": 0.1758,
999
+ "step": 10560
1000
+ },
1001
+ {
1002
+ "epoch": 0.47,
1003
+ "learning_rate": 8.426428296124995e-07,
1004
+ "loss": 0.1565,
1005
+ "step": 10624
1006
+ },
1007
+ {
1008
+ "epoch": 0.48,
1009
+ "learning_rate": 8.416940911381897e-07,
1010
+ "loss": 0.0861,
1011
+ "step": 10688
1012
+ },
1013
+ {
1014
+ "epoch": 0.48,
1015
+ "learning_rate": 8.407453526638797e-07,
1016
+ "loss": 0.1382,
1017
+ "step": 10752
1018
+ },
1019
+ {
1020
+ "epoch": 0.48,
1021
+ "learning_rate": 8.397966141895698e-07,
1022
+ "loss": 0.1268,
1023
+ "step": 10816
1024
+ },
1025
+ {
1026
+ "epoch": 0.48,
1027
+ "learning_rate": 8.388478757152599e-07,
1028
+ "loss": 0.1418,
1029
+ "step": 10880
1030
+ },
1031
+ {
1032
+ "epoch": 0.49,
1033
+ "learning_rate": 8.378991372409498e-07,
1034
+ "loss": 0.2417,
1035
+ "step": 10944
1036
+ },
1037
+ {
1038
+ "epoch": 0.49,
1039
+ "learning_rate": 8.369503987666399e-07,
1040
+ "loss": 0.1285,
1041
+ "step": 11008
1042
+ },
1043
+ {
1044
+ "epoch": 0.49,
1045
+ "learning_rate": 8.3600166029233e-07,
1046
+ "loss": 0.1719,
1047
+ "step": 11072
1048
+ },
1049
+ {
1050
+ "epoch": 0.5,
1051
+ "learning_rate": 8.350529218180201e-07,
1052
+ "loss": 0.1432,
1053
+ "step": 11136
1054
+ },
1055
+ {
1056
+ "epoch": 0.5,
1057
+ "learning_rate": 8.341041833437101e-07,
1058
+ "loss": 0.2443,
1059
+ "step": 11200
1060
+ },
1061
+ {
1062
+ "epoch": 0.5,
1063
+ "learning_rate": 8.331554448694002e-07,
1064
+ "loss": 0.1348,
1065
+ "step": 11264
1066
+ },
1067
+ {
1068
+ "epoch": 0.5,
1069
+ "learning_rate": 8.322067063950903e-07,
1070
+ "loss": 0.1251,
1071
+ "step": 11328
1072
+ },
1073
+ {
1074
+ "epoch": 0.51,
1075
+ "learning_rate": 8.312579679207803e-07,
1076
+ "loss": 0.156,
1077
+ "step": 11392
1078
+ },
1079
+ {
1080
+ "epoch": 0.51,
1081
+ "learning_rate": 8.303092294464703e-07,
1082
+ "loss": 0.3104,
1083
+ "step": 11456
1084
+ },
1085
+ {
1086
+ "epoch": 0.51,
1087
+ "learning_rate": 8.293604909721605e-07,
1088
+ "loss": 0.1834,
1089
+ "step": 11520
1090
+ },
1091
+ {
1092
+ "epoch": 0.52,
1093
+ "learning_rate": 8.284117524978504e-07,
1094
+ "loss": 0.1312,
1095
+ "step": 11584
1096
+ },
1097
+ {
1098
+ "epoch": 0.52,
1099
+ "learning_rate": 8.274630140235405e-07,
1100
+ "loss": 0.1026,
1101
+ "step": 11648
1102
+ },
1103
+ {
1104
+ "epoch": 0.52,
1105
+ "learning_rate": 8.265142755492306e-07,
1106
+ "loss": 0.1805,
1107
+ "step": 11712
1108
+ },
1109
+ {
1110
+ "epoch": 0.52,
1111
+ "learning_rate": 8.255655370749206e-07,
1112
+ "loss": 0.2334,
1113
+ "step": 11776
1114
+ },
1115
+ {
1116
+ "epoch": 0.53,
1117
+ "learning_rate": 8.246167986006107e-07,
1118
+ "loss": 0.1606,
1119
+ "step": 11840
1120
+ },
1121
+ {
1122
+ "epoch": 0.53,
1123
+ "learning_rate": 8.236680601263008e-07,
1124
+ "loss": 0.1009,
1125
+ "step": 11904
1126
+ },
1127
+ {
1128
+ "epoch": 0.53,
1129
+ "learning_rate": 8.227193216519909e-07,
1130
+ "loss": 0.1337,
1131
+ "step": 11968
1132
+ },
1133
+ {
1134
+ "epoch": 0.54,
1135
+ "learning_rate": 8.217705831776809e-07,
1136
+ "loss": 0.2247,
1137
+ "step": 12032
1138
+ },
1139
+ {
1140
+ "epoch": 0.54,
1141
+ "learning_rate": 8.20821844703371e-07,
1142
+ "loss": 0.163,
1143
+ "step": 12096
1144
+ },
1145
+ {
1146
+ "epoch": 0.54,
1147
+ "learning_rate": 8.19873106229061e-07,
1148
+ "loss": 0.1729,
1149
+ "step": 12160
1150
+ },
1151
+ {
1152
+ "epoch": 0.54,
1153
+ "learning_rate": 8.18924367754751e-07,
1154
+ "loss": 0.2133,
1155
+ "step": 12224
1156
+ },
1157
+ {
1158
+ "epoch": 0.55,
1159
+ "learning_rate": 8.179756292804412e-07,
1160
+ "loss": 0.2887,
1161
+ "step": 12288
1162
+ },
1163
+ {
1164
+ "epoch": 0.55,
1165
+ "learning_rate": 8.170268908061312e-07,
1166
+ "loss": 0.128,
1167
+ "step": 12352
1168
+ },
1169
+ {
1170
+ "epoch": 0.55,
1171
+ "learning_rate": 8.160781523318212e-07,
1172
+ "loss": 0.2019,
1173
+ "step": 12416
1174
+ },
1175
+ {
1176
+ "epoch": 0.56,
1177
+ "learning_rate": 8.151294138575113e-07,
1178
+ "loss": 0.128,
1179
+ "step": 12480
1180
+ },
1181
+ {
1182
+ "epoch": 0.56,
1183
+ "learning_rate": 8.141806753832014e-07,
1184
+ "loss": 0.1888,
1185
+ "step": 12544
1186
+ },
1187
+ {
1188
+ "epoch": 0.56,
1189
+ "learning_rate": 8.132319369088915e-07,
1190
+ "loss": 0.2176,
1191
+ "step": 12608
1192
+ },
1193
+ {
1194
+ "epoch": 0.56,
1195
+ "learning_rate": 8.122831984345814e-07,
1196
+ "loss": 0.1555,
1197
+ "step": 12672
1198
+ },
1199
+ {
1200
+ "epoch": 0.57,
1201
+ "learning_rate": 8.113344599602715e-07,
1202
+ "loss": 0.1948,
1203
+ "step": 12736
1204
+ },
1205
+ {
1206
+ "epoch": 0.57,
1207
+ "learning_rate": 8.103857214859616e-07,
1208
+ "loss": 0.233,
1209
+ "step": 12800
1210
+ },
1211
+ {
1212
+ "epoch": 0.57,
1213
+ "learning_rate": 8.094369830116516e-07,
1214
+ "loss": 0.1574,
1215
+ "step": 12864
1216
+ },
1217
+ {
1218
+ "epoch": 0.57,
1219
+ "learning_rate": 8.084882445373418e-07,
1220
+ "loss": 0.1377,
1221
+ "step": 12928
1222
+ },
1223
+ {
1224
+ "epoch": 0.58,
1225
+ "learning_rate": 8.075395060630318e-07,
1226
+ "loss": 0.1563,
1227
+ "step": 12992
1228
+ },
1229
+ {
1230
+ "epoch": 0.58,
1231
+ "learning_rate": 8.065907675887218e-07,
1232
+ "loss": 0.1345,
1233
+ "step": 13056
1234
+ },
1235
+ {
1236
+ "epoch": 0.58,
1237
+ "learning_rate": 8.05642029114412e-07,
1238
+ "loss": 0.1489,
1239
+ "step": 13120
1240
+ },
1241
+ {
1242
+ "epoch": 0.59,
1243
+ "learning_rate": 8.047081146787631e-07,
1244
+ "loss": 0.2524,
1245
+ "step": 13184
1246
+ },
1247
+ {
1248
+ "epoch": 0.59,
1249
+ "learning_rate": 8.037742002431142e-07,
1250
+ "loss": 0.1534,
1251
+ "step": 13248
1252
+ },
1253
+ {
1254
+ "epoch": 0.59,
1255
+ "learning_rate": 8.028254617688043e-07,
1256
+ "loss": 0.136,
1257
+ "step": 13312
1258
+ },
1259
+ {
1260
+ "epoch": 0.59,
1261
+ "learning_rate": 8.018767232944942e-07,
1262
+ "loss": 0.151,
1263
+ "step": 13376
1264
+ },
1265
+ {
1266
+ "epoch": 0.6,
1267
+ "learning_rate": 8.009279848201844e-07,
1268
+ "loss": 0.1423,
1269
+ "step": 13440
1270
+ },
1271
+ {
1272
+ "epoch": 0.6,
1273
+ "learning_rate": 7.999792463458744e-07,
1274
+ "loss": 0.2284,
1275
+ "step": 13504
1276
+ },
1277
+ {
1278
+ "epoch": 0.6,
1279
+ "learning_rate": 7.990305078715644e-07,
1280
+ "loss": 0.1515,
1281
+ "step": 13568
1282
+ },
1283
+ {
1284
+ "epoch": 0.61,
1285
+ "learning_rate": 7.980817693972546e-07,
1286
+ "loss": 0.1431,
1287
+ "step": 13632
1288
+ },
1289
+ {
1290
+ "epoch": 0.61,
1291
+ "learning_rate": 7.971330309229446e-07,
1292
+ "loss": 0.1759,
1293
+ "step": 13696
1294
+ },
1295
+ {
1296
+ "epoch": 0.61,
1297
+ "learning_rate": 7.961842924486347e-07,
1298
+ "loss": 0.2942,
1299
+ "step": 13760
1300
+ },
1301
+ {
1302
+ "epoch": 0.61,
1303
+ "learning_rate": 7.952355539743248e-07,
1304
+ "loss": 0.1382,
1305
+ "step": 13824
1306
+ },
1307
+ {
1308
+ "epoch": 0.62,
1309
+ "learning_rate": 7.942868155000148e-07,
1310
+ "loss": 0.181,
1311
+ "step": 13888
1312
+ },
1313
+ {
1314
+ "epoch": 0.62,
1315
+ "learning_rate": 7.933380770257049e-07,
1316
+ "loss": 0.2471,
1317
+ "step": 13952
1318
+ },
1319
+ {
1320
+ "epoch": 0.62,
1321
+ "learning_rate": 7.923893385513949e-07,
1322
+ "loss": 0.1487,
1323
+ "step": 14016
1324
+ },
1325
+ {
1326
+ "epoch": 0.63,
1327
+ "learning_rate": 7.91440600077085e-07,
1328
+ "loss": 0.1653,
1329
+ "step": 14080
1330
+ },
1331
+ {
1332
+ "epoch": 0.63,
1333
+ "learning_rate": 7.90491861602775e-07,
1334
+ "loss": 0.193,
1335
+ "step": 14144
1336
+ },
1337
+ {
1338
+ "epoch": 0.63,
1339
+ "learning_rate": 7.89543123128465e-07,
1340
+ "loss": 0.115,
1341
+ "step": 14208
1342
+ },
1343
+ {
1344
+ "epoch": 0.63,
1345
+ "learning_rate": 7.885943846541552e-07,
1346
+ "loss": 0.1413,
1347
+ "step": 14272
1348
+ },
1349
+ {
1350
+ "epoch": 0.64,
1351
+ "learning_rate": 7.876456461798452e-07,
1352
+ "loss": 0.1508,
1353
+ "step": 14336
1354
+ },
1355
+ {
1356
+ "epoch": 0.64,
1357
+ "learning_rate": 7.866969077055352e-07,
1358
+ "loss": 0.1752,
1359
+ "step": 14400
1360
+ },
1361
+ {
1362
+ "epoch": 0.64,
1363
+ "learning_rate": 7.857481692312254e-07,
1364
+ "loss": 0.2432,
1365
+ "step": 14464
1366
+ },
1367
+ {
1368
+ "epoch": 0.65,
1369
+ "learning_rate": 7.847994307569153e-07,
1370
+ "loss": 0.1978,
1371
+ "step": 14528
1372
+ },
1373
+ {
1374
+ "epoch": 0.65,
1375
+ "learning_rate": 7.838506922826054e-07,
1376
+ "loss": 0.1445,
1377
+ "step": 14592
1378
+ },
1379
+ {
1380
+ "epoch": 0.65,
1381
+ "learning_rate": 7.829019538082955e-07,
1382
+ "loss": 0.1484,
1383
+ "step": 14656
1384
+ },
1385
+ {
1386
+ "epoch": 0.65,
1387
+ "learning_rate": 7.819532153339855e-07,
1388
+ "loss": 0.1887,
1389
+ "step": 14720
1390
+ },
1391
+ {
1392
+ "epoch": 0.66,
1393
+ "learning_rate": 7.810044768596756e-07,
1394
+ "loss": 0.216,
1395
+ "step": 14784
1396
+ },
1397
+ {
1398
+ "epoch": 0.66,
1399
+ "learning_rate": 7.800557383853657e-07,
1400
+ "loss": 0.1803,
1401
+ "step": 14848
1402
+ },
1403
+ {
1404
+ "epoch": 0.66,
1405
+ "learning_rate": 7.791069999110558e-07,
1406
+ "loss": 0.1332,
1407
+ "step": 14912
1408
+ },
1409
+ {
1410
+ "epoch": 0.67,
1411
+ "learning_rate": 7.781582614367458e-07,
1412
+ "loss": 0.2439,
1413
+ "step": 14976
1414
+ },
1415
+ {
1416
+ "epoch": 0.67,
1417
+ "learning_rate": 7.772095229624358e-07,
1418
+ "loss": 0.1689,
1419
+ "step": 15040
1420
+ },
1421
+ {
1422
+ "epoch": 0.67,
1423
+ "learning_rate": 7.76260784488126e-07,
1424
+ "loss": 0.1823,
1425
+ "step": 15104
1426
+ },
1427
+ {
1428
+ "epoch": 0.67,
1429
+ "learning_rate": 7.753120460138159e-07,
1430
+ "loss": 0.1905,
1431
+ "step": 15168
1432
+ },
1433
+ {
1434
+ "epoch": 0.68,
1435
+ "learning_rate": 7.74363307539506e-07,
1436
+ "loss": 0.2558,
1437
+ "step": 15232
1438
+ },
1439
+ {
1440
+ "epoch": 0.68,
1441
+ "learning_rate": 7.734145690651961e-07,
1442
+ "loss": 0.1531,
1443
+ "step": 15296
1444
+ },
1445
+ {
1446
+ "epoch": 0.68,
1447
+ "learning_rate": 7.724658305908861e-07,
1448
+ "loss": 0.1849,
1449
+ "step": 15360
1450
+ },
1451
+ {
1452
+ "epoch": 0.69,
1453
+ "learning_rate": 7.715170921165762e-07,
1454
+ "loss": 0.1317,
1455
+ "step": 15424
1456
+ },
1457
+ {
1458
+ "epoch": 0.69,
1459
+ "learning_rate": 7.705683536422663e-07,
1460
+ "loss": 0.1096,
1461
+ "step": 15488
1462
+ },
1463
+ {
1464
+ "epoch": 0.69,
1465
+ "learning_rate": 7.696196151679564e-07,
1466
+ "loss": 0.2193,
1467
+ "step": 15552
1468
+ },
1469
+ {
1470
+ "epoch": 0.69,
1471
+ "learning_rate": 7.686708766936464e-07,
1472
+ "loss": 0.1658,
1473
+ "step": 15616
1474
+ },
1475
+ {
1476
+ "epoch": 0.7,
1477
+ "learning_rate": 7.677221382193365e-07,
1478
+ "loss": 0.1553,
1479
+ "step": 15680
1480
+ },
1481
+ {
1482
+ "epoch": 0.7,
1483
+ "learning_rate": 7.667733997450265e-07,
1484
+ "loss": 0.1772,
1485
+ "step": 15744
1486
+ },
1487
+ {
1488
+ "epoch": 0.7,
1489
+ "learning_rate": 7.658246612707165e-07,
1490
+ "loss": 0.2147,
1491
+ "step": 15808
1492
+ },
1493
+ {
1494
+ "epoch": 0.71,
1495
+ "learning_rate": 7.648759227964067e-07,
1496
+ "loss": 0.1096,
1497
+ "step": 15872
1498
+ },
1499
+ {
1500
+ "epoch": 0.71,
1501
+ "learning_rate": 7.639271843220967e-07,
1502
+ "loss": 0.1613,
1503
+ "step": 15936
1504
+ },
1505
+ {
1506
+ "epoch": 0.71,
1507
+ "learning_rate": 7.629784458477867e-07,
1508
+ "loss": 0.1488,
1509
+ "step": 16000
1510
+ },
1511
+ {
1512
+ "epoch": 0.71,
1513
+ "learning_rate": 7.620297073734768e-07,
1514
+ "loss": 0.2256,
1515
+ "step": 16064
1516
+ },
1517
+ {
1518
+ "epoch": 0.72,
1519
+ "learning_rate": 7.610809688991669e-07,
1520
+ "loss": 0.2512,
1521
+ "step": 16128
1522
+ },
1523
+ {
1524
+ "epoch": 0.72,
1525
+ "learning_rate": 7.60132230424857e-07,
1526
+ "loss": 0.1264,
1527
+ "step": 16192
1528
+ },
1529
+ {
1530
+ "epoch": 0.72,
1531
+ "learning_rate": 7.59183491950547e-07,
1532
+ "loss": 0.1162,
1533
+ "step": 16256
1534
+ },
1535
+ {
1536
+ "epoch": 0.73,
1537
+ "learning_rate": 7.58234753476237e-07,
1538
+ "loss": 0.1401,
1539
+ "step": 16320
1540
+ },
1541
+ {
1542
+ "epoch": 0.73,
1543
+ "learning_rate": 7.572860150019271e-07,
1544
+ "loss": 0.1336,
1545
+ "step": 16384
1546
+ },
1547
+ {
1548
+ "epoch": 0.73,
1549
+ "learning_rate": 7.563372765276171e-07,
1550
+ "loss": 0.1234,
1551
+ "step": 16448
1552
+ },
1553
+ {
1554
+ "epoch": 0.73,
1555
+ "learning_rate": 7.553885380533072e-07,
1556
+ "loss": 0.1195,
1557
+ "step": 16512
1558
+ },
1559
+ {
1560
+ "epoch": 0.74,
1561
+ "learning_rate": 7.544397995789973e-07,
1562
+ "loss": 0.2435,
1563
+ "step": 16576
1564
+ },
1565
+ {
1566
+ "epoch": 0.74,
1567
+ "learning_rate": 7.534910611046873e-07,
1568
+ "loss": 0.1109,
1569
+ "step": 16640
1570
+ },
1571
+ {
1572
+ "epoch": 0.74,
1573
+ "learning_rate": 7.525423226303775e-07,
1574
+ "loss": 0.2088,
1575
+ "step": 16704
1576
+ },
1577
+ {
1578
+ "epoch": 0.75,
1579
+ "learning_rate": 7.515935841560675e-07,
1580
+ "loss": 0.141,
1581
+ "step": 16768
1582
+ },
1583
+ {
1584
+ "epoch": 0.75,
1585
+ "learning_rate": 7.506448456817574e-07,
1586
+ "loss": 0.1428,
1587
+ "step": 16832
1588
+ },
1589
+ {
1590
+ "epoch": 0.75,
1591
+ "learning_rate": 7.496961072074475e-07,
1592
+ "loss": 0.1505,
1593
+ "step": 16896
1594
+ },
1595
+ {
1596
+ "epoch": 0.75,
1597
+ "learning_rate": 7.487473687331376e-07,
1598
+ "loss": 0.2152,
1599
+ "step": 16960
1600
+ },
1601
+ {
1602
+ "epoch": 0.76,
1603
+ "learning_rate": 7.477986302588277e-07,
1604
+ "loss": 0.2008,
1605
+ "step": 17024
1606
+ },
1607
+ {
1608
+ "epoch": 0.76,
1609
+ "learning_rate": 7.468498917845177e-07,
1610
+ "loss": 0.1872,
1611
+ "step": 17088
1612
+ },
1613
+ {
1614
+ "epoch": 0.76,
1615
+ "learning_rate": 7.459011533102078e-07,
1616
+ "loss": 0.1313,
1617
+ "step": 17152
1618
+ },
1619
+ {
1620
+ "epoch": 0.77,
1621
+ "learning_rate": 7.449524148358979e-07,
1622
+ "loss": 0.1099,
1623
+ "step": 17216
1624
+ },
1625
+ {
1626
+ "epoch": 0.77,
1627
+ "learning_rate": 7.440036763615879e-07,
1628
+ "loss": 0.138,
1629
+ "step": 17280
1630
+ },
1631
+ {
1632
+ "epoch": 0.77,
1633
+ "learning_rate": 7.430549378872781e-07,
1634
+ "loss": 0.1871,
1635
+ "step": 17344
1636
+ },
1637
+ {
1638
+ "epoch": 0.77,
1639
+ "learning_rate": 7.42106199412968e-07,
1640
+ "loss": 0.18,
1641
+ "step": 17408
1642
+ },
1643
+ {
1644
+ "epoch": 0.78,
1645
+ "learning_rate": 7.41157460938658e-07,
1646
+ "loss": 0.1337,
1647
+ "step": 17472
1648
+ },
1649
+ {
1650
+ "epoch": 0.78,
1651
+ "learning_rate": 7.402087224643482e-07,
1652
+ "loss": 0.1222,
1653
+ "step": 17536
1654
+ },
1655
+ {
1656
+ "epoch": 0.78,
1657
+ "learning_rate": 7.392599839900382e-07,
1658
+ "loss": 0.1434,
1659
+ "step": 17600
1660
+ },
1661
+ {
1662
+ "epoch": 0.79,
1663
+ "learning_rate": 7.383112455157283e-07,
1664
+ "loss": 0.1538,
1665
+ "step": 17664
1666
+ },
1667
+ {
1668
+ "epoch": 0.79,
1669
+ "learning_rate": 7.373625070414183e-07,
1670
+ "loss": 0.1908,
1671
+ "step": 17728
1672
+ },
1673
+ {
1674
+ "epoch": 0.79,
1675
+ "learning_rate": 7.364137685671084e-07,
1676
+ "loss": 0.1244,
1677
+ "step": 17792
1678
+ },
1679
+ {
1680
+ "epoch": 0.79,
1681
+ "learning_rate": 7.354650300927985e-07,
1682
+ "loss": 0.1593,
1683
+ "step": 17856
1684
+ },
1685
+ {
1686
+ "epoch": 0.8,
1687
+ "learning_rate": 7.345162916184885e-07,
1688
+ "loss": 0.1588,
1689
+ "step": 17920
1690
+ },
1691
+ {
1692
+ "epoch": 0.8,
1693
+ "learning_rate": 7.335675531441787e-07,
1694
+ "loss": 0.1639,
1695
+ "step": 17984
1696
+ },
1697
+ {
1698
+ "epoch": 0.8,
1699
+ "learning_rate": 7.326188146698686e-07,
1700
+ "loss": 0.1431,
1701
+ "step": 18048
1702
+ },
1703
+ {
1704
+ "epoch": 0.81,
1705
+ "learning_rate": 7.316700761955586e-07,
1706
+ "loss": 0.2002,
1707
+ "step": 18112
1708
+ },
1709
+ {
1710
+ "epoch": 0.81,
1711
+ "learning_rate": 7.307213377212488e-07,
1712
+ "loss": 0.1761,
1713
+ "step": 18176
1714
+ },
1715
+ {
1716
+ "epoch": 0.81,
1717
+ "learning_rate": 7.297725992469388e-07,
1718
+ "loss": 0.1597,
1719
+ "step": 18240
1720
+ },
1721
+ {
1722
+ "epoch": 0.81,
1723
+ "learning_rate": 7.288238607726288e-07,
1724
+ "loss": 0.1952,
1725
+ "step": 18304
1726
+ },
1727
+ {
1728
+ "epoch": 0.82,
1729
+ "learning_rate": 7.27875122298319e-07,
1730
+ "loss": 0.1843,
1731
+ "step": 18368
1732
+ },
1733
+ {
1734
+ "epoch": 0.82,
1735
+ "learning_rate": 7.26926383824009e-07,
1736
+ "loss": 0.1032,
1737
+ "step": 18432
1738
+ },
1739
+ {
1740
+ "epoch": 0.82,
1741
+ "learning_rate": 7.259776453496991e-07,
1742
+ "loss": 0.1952,
1743
+ "step": 18496
1744
+ },
1745
+ {
1746
+ "epoch": 0.83,
1747
+ "learning_rate": 7.25028906875389e-07,
1748
+ "loss": 0.193,
1749
+ "step": 18560
1750
+ },
1751
+ {
1752
+ "epoch": 0.83,
1753
+ "learning_rate": 7.240801684010791e-07,
1754
+ "loss": 0.137,
1755
+ "step": 18624
1756
+ },
1757
+ {
1758
+ "epoch": 0.83,
1759
+ "learning_rate": 7.231314299267692e-07,
1760
+ "loss": 0.1992,
1761
+ "step": 18688
1762
+ },
1763
+ {
1764
+ "epoch": 0.83,
1765
+ "learning_rate": 7.221826914524592e-07,
1766
+ "loss": 0.138,
1767
+ "step": 18752
1768
+ },
1769
+ {
1770
+ "epoch": 0.84,
1771
+ "learning_rate": 7.212339529781494e-07,
1772
+ "loss": 0.2263,
1773
+ "step": 18816
1774
+ },
1775
+ {
1776
+ "epoch": 0.84,
1777
+ "learning_rate": 7.202852145038394e-07,
1778
+ "loss": 0.2101,
1779
+ "step": 18880
1780
+ },
1781
+ {
1782
+ "epoch": 0.84,
1783
+ "learning_rate": 7.193364760295294e-07,
1784
+ "loss": 0.1731,
1785
+ "step": 18944
1786
+ },
1787
+ {
1788
+ "epoch": 0.85,
1789
+ "learning_rate": 7.183877375552196e-07,
1790
+ "loss": 0.1523,
1791
+ "step": 19008
1792
+ },
1793
+ {
1794
+ "epoch": 0.85,
1795
+ "learning_rate": 7.174389990809096e-07,
1796
+ "loss": 0.1671,
1797
+ "step": 19072
1798
+ },
1799
+ {
1800
+ "epoch": 0.85,
1801
+ "learning_rate": 7.164902606065996e-07,
1802
+ "loss": 0.1549,
1803
+ "step": 19136
1804
+ },
1805
+ {
1806
+ "epoch": 0.85,
1807
+ "learning_rate": 7.155415221322897e-07,
1808
+ "loss": 0.1346,
1809
+ "step": 19200
1810
+ },
1811
+ {
1812
+ "epoch": 0.86,
1813
+ "learning_rate": 7.145927836579797e-07,
1814
+ "loss": 0.2403,
1815
+ "step": 19264
1816
+ },
1817
+ {
1818
+ "epoch": 0.86,
1819
+ "learning_rate": 7.136588692223308e-07,
1820
+ "loss": 0.1909,
1821
+ "step": 19328
1822
+ },
1823
+ {
1824
+ "epoch": 0.86,
1825
+ "learning_rate": 7.12710130748021e-07,
1826
+ "loss": 0.1801,
1827
+ "step": 19392
1828
+ },
1829
+ {
1830
+ "epoch": 0.87,
1831
+ "learning_rate": 7.11761392273711e-07,
1832
+ "loss": 0.1196,
1833
+ "step": 19456
1834
+ },
1835
+ {
1836
+ "epoch": 0.87,
1837
+ "learning_rate": 7.10812653799401e-07,
1838
+ "loss": 0.0749,
1839
+ "step": 19520
1840
+ },
1841
+ {
1842
+ "epoch": 0.87,
1843
+ "learning_rate": 7.098639153250912e-07,
1844
+ "loss": 0.1386,
1845
+ "step": 19584
1846
+ },
1847
+ {
1848
+ "epoch": 0.87,
1849
+ "learning_rate": 7.089151768507812e-07,
1850
+ "loss": 0.219,
1851
+ "step": 19648
1852
+ },
1853
+ {
1854
+ "epoch": 0.88,
1855
+ "learning_rate": 7.079664383764713e-07,
1856
+ "loss": 0.1572,
1857
+ "step": 19712
1858
+ },
1859
+ {
1860
+ "epoch": 0.88,
1861
+ "learning_rate": 7.070176999021614e-07,
1862
+ "loss": 0.191,
1863
+ "step": 19776
1864
+ },
1865
+ {
1866
+ "epoch": 0.88,
1867
+ "learning_rate": 7.060689614278513e-07,
1868
+ "loss": 0.1882,
1869
+ "step": 19840
1870
+ },
1871
+ {
1872
+ "epoch": 0.89,
1873
+ "learning_rate": 7.051202229535414e-07,
1874
+ "loss": 0.1654,
1875
+ "step": 19904
1876
+ },
1877
+ {
1878
+ "epoch": 0.89,
1879
+ "learning_rate": 7.041714844792315e-07,
1880
+ "loss": 0.1397,
1881
+ "step": 19968
1882
+ },
1883
+ {
1884
+ "epoch": 0.89,
1885
+ "learning_rate": 7.032227460049216e-07,
1886
+ "loss": 0.1948,
1887
+ "step": 20032
1888
+ },
1889
+ {
1890
+ "epoch": 0.89,
1891
+ "learning_rate": 7.022740075306116e-07,
1892
+ "loss": 0.2171,
1893
+ "step": 20096
1894
+ },
1895
+ {
1896
+ "epoch": 0.9,
1897
+ "learning_rate": 7.013252690563016e-07,
1898
+ "loss": 0.2474,
1899
+ "step": 20160
1900
+ },
1901
+ {
1902
+ "epoch": 0.9,
1903
+ "learning_rate": 7.003765305819918e-07,
1904
+ "loss": 0.2014,
1905
+ "step": 20224
1906
+ },
1907
+ {
1908
+ "epoch": 0.9,
1909
+ "learning_rate": 6.994277921076818e-07,
1910
+ "loss": 0.1256,
1911
+ "step": 20288
1912
+ },
1913
+ {
1914
+ "epoch": 0.91,
1915
+ "learning_rate": 6.984790536333718e-07,
1916
+ "loss": 0.1634,
1917
+ "step": 20352
1918
+ },
1919
+ {
1920
+ "epoch": 0.91,
1921
+ "learning_rate": 6.975303151590619e-07,
1922
+ "loss": 0.1672,
1923
+ "step": 20416
1924
+ },
1925
+ {
1926
+ "epoch": 0.91,
1927
+ "learning_rate": 6.965815766847519e-07,
1928
+ "loss": 0.1773,
1929
+ "step": 20480
1930
+ },
1931
+ {
1932
+ "epoch": 0.91,
1933
+ "learning_rate": 6.95632838210442e-07,
1934
+ "loss": 0.1157,
1935
+ "step": 20544
1936
+ },
1937
+ {
1938
+ "epoch": 0.92,
1939
+ "learning_rate": 6.946840997361321e-07,
1940
+ "loss": 0.2241,
1941
+ "step": 20608
1942
+ },
1943
+ {
1944
+ "epoch": 0.92,
1945
+ "learning_rate": 6.937353612618222e-07,
1946
+ "loss": 0.1108,
1947
+ "step": 20672
1948
+ },
1949
+ {
1950
+ "epoch": 0.92,
1951
+ "learning_rate": 6.927866227875122e-07,
1952
+ "loss": 0.1821,
1953
+ "step": 20736
1954
+ },
1955
+ {
1956
+ "epoch": 0.93,
1957
+ "learning_rate": 6.918378843132023e-07,
1958
+ "loss": 0.1459,
1959
+ "step": 20800
1960
+ },
1961
+ {
1962
+ "epoch": 0.93,
1963
+ "learning_rate": 6.908891458388924e-07,
1964
+ "loss": 0.2022,
1965
+ "step": 20864
1966
+ },
1967
+ {
1968
+ "epoch": 0.93,
1969
+ "learning_rate": 6.899404073645823e-07,
1970
+ "loss": 0.1864,
1971
+ "step": 20928
1972
+ },
1973
+ {
1974
+ "epoch": 0.93,
1975
+ "learning_rate": 6.889916688902723e-07,
1976
+ "loss": 0.1436,
1977
+ "step": 20992
1978
+ },
1979
+ {
1980
+ "epoch": 0.94,
1981
+ "learning_rate": 6.880429304159625e-07,
1982
+ "loss": 0.1771,
1983
+ "step": 21056
1984
+ },
1985
+ {
1986
+ "epoch": 0.94,
1987
+ "learning_rate": 6.870941919416525e-07,
1988
+ "loss": 0.1782,
1989
+ "step": 21120
1990
+ },
1991
+ {
1992
+ "epoch": 0.94,
1993
+ "learning_rate": 6.861454534673426e-07,
1994
+ "loss": 0.1754,
1995
+ "step": 21184
1996
+ },
1997
+ {
1998
+ "epoch": 0.94,
1999
+ "learning_rate": 6.851967149930327e-07,
2000
+ "loss": 0.1483,
2001
+ "step": 21248
2002
+ },
2003
+ {
2004
+ "epoch": 0.95,
2005
+ "learning_rate": 6.842479765187227e-07,
2006
+ "loss": 0.1373,
2007
+ "step": 21312
2008
+ },
2009
+ {
2010
+ "epoch": 0.95,
2011
+ "learning_rate": 6.833140620830739e-07,
2012
+ "loss": 0.219,
2013
+ "step": 21376
2014
+ },
2015
+ {
2016
+ "epoch": 0.95,
2017
+ "learning_rate": 6.82365323608764e-07,
2018
+ "loss": 0.1474,
2019
+ "step": 21440
2020
+ },
2021
+ {
2022
+ "epoch": 0.96,
2023
+ "learning_rate": 6.81416585134454e-07,
2024
+ "loss": 0.1713,
2025
+ "step": 21504
2026
+ },
2027
+ {
2028
+ "epoch": 0.96,
2029
+ "learning_rate": 6.80467846660144e-07,
2030
+ "loss": 0.1034,
2031
+ "step": 21568
2032
+ },
2033
+ {
2034
+ "epoch": 0.96,
2035
+ "learning_rate": 6.795191081858341e-07,
2036
+ "loss": 0.185,
2037
+ "step": 21632
2038
+ },
2039
+ {
2040
+ "epoch": 0.96,
2041
+ "learning_rate": 6.785703697115241e-07,
2042
+ "loss": 0.284,
2043
+ "step": 21696
2044
+ },
2045
+ {
2046
+ "epoch": 0.97,
2047
+ "learning_rate": 6.776216312372143e-07,
2048
+ "loss": 0.1953,
2049
+ "step": 21760
2050
+ },
2051
+ {
2052
+ "epoch": 0.97,
2053
+ "learning_rate": 6.766728927629043e-07,
2054
+ "loss": 0.168,
2055
+ "step": 21824
2056
+ },
2057
+ {
2058
+ "epoch": 0.97,
2059
+ "learning_rate": 6.757241542885943e-07,
2060
+ "loss": 0.1852,
2061
+ "step": 21888
2062
+ },
2063
+ {
2064
+ "epoch": 0.98,
2065
+ "learning_rate": 6.747754158142845e-07,
2066
+ "loss": 0.1358,
2067
+ "step": 21952
2068
+ },
2069
+ {
2070
+ "epoch": 0.98,
2071
+ "learning_rate": 6.738266773399745e-07,
2072
+ "loss": 0.1885,
2073
+ "step": 22016
2074
+ },
2075
+ {
2076
+ "epoch": 0.98,
2077
+ "learning_rate": 6.728779388656646e-07,
2078
+ "loss": 0.22,
2079
+ "step": 22080
2080
+ },
2081
+ {
2082
+ "epoch": 0.98,
2083
+ "learning_rate": 6.719292003913545e-07,
2084
+ "loss": 0.214,
2085
+ "step": 22144
2086
+ },
2087
+ {
2088
+ "epoch": 0.99,
2089
+ "learning_rate": 6.709804619170446e-07,
2090
+ "loss": 0.1198,
2091
+ "step": 22208
2092
+ },
2093
+ {
2094
+ "epoch": 0.99,
2095
+ "learning_rate": 6.700317234427347e-07,
2096
+ "loss": 0.1458,
2097
+ "step": 22272
2098
+ },
2099
+ {
2100
+ "epoch": 0.99,
2101
+ "learning_rate": 6.690829849684247e-07,
2102
+ "loss": 0.2405,
2103
+ "step": 22336
2104
+ },
2105
+ {
2106
+ "epoch": 1.0,
2107
+ "learning_rate": 6.681342464941149e-07,
2108
+ "loss": 0.1162,
2109
+ "step": 22400
2110
+ },
2111
+ {
2112
+ "epoch": 1.0,
2113
+ "learning_rate": 6.671855080198049e-07,
2114
+ "loss": 0.1508,
2115
+ "step": 22464
2116
+ },
2117
+ {
2118
+ "epoch": 1.0,
2119
+ "learning_rate": 6.662367695454949e-07,
2120
+ "loss": 0.1805,
2121
+ "step": 22528
2122
+ },
2123
+ {
2124
+ "epoch": 1.0,
2125
+ "learning_rate": 6.652880310711851e-07,
2126
+ "loss": 0.0954,
2127
+ "step": 22592
2128
+ },
2129
+ {
2130
+ "epoch": 1.01,
2131
+ "learning_rate": 6.643392925968751e-07,
2132
+ "loss": 0.0756,
2133
+ "step": 22656
2134
+ },
2135
+ {
2136
+ "epoch": 1.01,
2137
+ "learning_rate": 6.633905541225651e-07,
2138
+ "loss": 0.1514,
2139
+ "step": 22720
2140
+ },
2141
+ {
2142
+ "epoch": 1.01,
2143
+ "learning_rate": 6.624418156482551e-07,
2144
+ "loss": 0.1408,
2145
+ "step": 22784
2146
+ },
2147
+ {
2148
+ "epoch": 1.02,
2149
+ "learning_rate": 6.614930771739452e-07,
2150
+ "loss": 0.1979,
2151
+ "step": 22848
2152
+ },
2153
+ {
2154
+ "epoch": 1.02,
2155
+ "learning_rate": 6.605443386996353e-07,
2156
+ "loss": 0.1277,
2157
+ "step": 22912
2158
+ },
2159
+ {
2160
+ "epoch": 1.02,
2161
+ "learning_rate": 6.595956002253253e-07,
2162
+ "loss": 0.0607,
2163
+ "step": 22976
2164
+ },
2165
+ {
2166
+ "epoch": 1.02,
2167
+ "learning_rate": 6.586468617510155e-07,
2168
+ "loss": 0.126,
2169
+ "step": 23040
2170
+ },
2171
+ {
2172
+ "epoch": 1.03,
2173
+ "learning_rate": 6.576981232767055e-07,
2174
+ "loss": 0.116,
2175
+ "step": 23104
2176
+ },
2177
+ {
2178
+ "epoch": 1.03,
2179
+ "learning_rate": 6.567493848023955e-07,
2180
+ "loss": 0.1233,
2181
+ "step": 23168
2182
+ },
2183
+ {
2184
+ "epoch": 1.03,
2185
+ "learning_rate": 6.558006463280857e-07,
2186
+ "loss": 0.0871,
2187
+ "step": 23232
2188
+ },
2189
+ {
2190
+ "epoch": 1.04,
2191
+ "learning_rate": 6.548519078537756e-07,
2192
+ "loss": 0.0974,
2193
+ "step": 23296
2194
+ },
2195
+ {
2196
+ "epoch": 1.04,
2197
+ "learning_rate": 6.539031693794656e-07,
2198
+ "loss": 0.1102,
2199
+ "step": 23360
2200
+ },
2201
+ {
2202
+ "epoch": 1.04,
2203
+ "learning_rate": 6.529544309051558e-07,
2204
+ "loss": 0.0905,
2205
+ "step": 23424
2206
+ },
2207
+ {
2208
+ "epoch": 1.04,
2209
+ "learning_rate": 6.520056924308458e-07,
2210
+ "loss": 0.0783,
2211
+ "step": 23488
2212
+ },
2213
+ {
2214
+ "epoch": 1.05,
2215
+ "learning_rate": 6.510569539565359e-07,
2216
+ "loss": 0.0835,
2217
+ "step": 23552
2218
+ },
2219
+ {
2220
+ "epoch": 1.05,
2221
+ "learning_rate": 6.501082154822259e-07,
2222
+ "loss": 0.1459,
2223
+ "step": 23616
2224
+ },
2225
+ {
2226
+ "epoch": 1.05,
2227
+ "learning_rate": 6.49159477007916e-07,
2228
+ "loss": 0.1042,
2229
+ "step": 23680
2230
+ },
2231
+ {
2232
+ "epoch": 1.06,
2233
+ "learning_rate": 6.482107385336061e-07,
2234
+ "loss": 0.1063,
2235
+ "step": 23744
2236
+ },
2237
+ {
2238
+ "epoch": 1.06,
2239
+ "learning_rate": 6.472620000592961e-07,
2240
+ "loss": 0.1107,
2241
+ "step": 23808
2242
+ },
2243
+ {
2244
+ "epoch": 1.06,
2245
+ "learning_rate": 6.463132615849863e-07,
2246
+ "loss": 0.1557,
2247
+ "step": 23872
2248
+ },
2249
+ {
2250
+ "epoch": 1.06,
2251
+ "learning_rate": 6.453645231106762e-07,
2252
+ "loss": 0.1015,
2253
+ "step": 23936
2254
+ },
2255
+ {
2256
+ "epoch": 1.07,
2257
+ "learning_rate": 6.444157846363662e-07,
2258
+ "loss": 0.1518,
2259
+ "step": 24000
2260
+ },
2261
+ {
2262
+ "epoch": 1.07,
2263
+ "learning_rate": 6.434670461620564e-07,
2264
+ "loss": 0.1283,
2265
+ "step": 24064
2266
+ },
2267
+ {
2268
+ "epoch": 1.07,
2269
+ "learning_rate": 6.425183076877464e-07,
2270
+ "loss": 0.1181,
2271
+ "step": 24128
2272
+ },
2273
+ {
2274
+ "epoch": 1.08,
2275
+ "learning_rate": 6.415695692134365e-07,
2276
+ "loss": 0.1414,
2277
+ "step": 24192
2278
+ },
2279
+ {
2280
+ "epoch": 1.08,
2281
+ "learning_rate": 6.406208307391266e-07,
2282
+ "loss": 0.0739,
2283
+ "step": 24256
2284
+ },
2285
+ {
2286
+ "epoch": 1.08,
2287
+ "learning_rate": 6.396869163034777e-07,
2288
+ "loss": 0.1461,
2289
+ "step": 24320
2290
+ },
2291
+ {
2292
+ "epoch": 1.08,
2293
+ "learning_rate": 6.387381778291678e-07,
2294
+ "loss": 0.1104,
2295
+ "step": 24384
2296
+ },
2297
+ {
2298
+ "epoch": 1.09,
2299
+ "learning_rate": 6.377894393548579e-07,
2300
+ "loss": 0.0982,
2301
+ "step": 24448
2302
+ },
2303
+ {
2304
+ "epoch": 1.09,
2305
+ "learning_rate": 6.368407008805478e-07,
2306
+ "loss": 0.1016,
2307
+ "step": 24512
2308
+ },
2309
+ {
2310
+ "epoch": 1.09,
2311
+ "learning_rate": 6.358919624062378e-07,
2312
+ "loss": 0.1343,
2313
+ "step": 24576
2314
+ },
2315
+ {
2316
+ "epoch": 1.1,
2317
+ "learning_rate": 6.34943223931928e-07,
2318
+ "loss": 0.0936,
2319
+ "step": 24640
2320
+ },
2321
+ {
2322
+ "epoch": 1.1,
2323
+ "learning_rate": 6.33994485457618e-07,
2324
+ "loss": 0.1245,
2325
+ "step": 24704
2326
+ },
2327
+ {
2328
+ "epoch": 1.1,
2329
+ "learning_rate": 6.330457469833081e-07,
2330
+ "loss": 0.1656,
2331
+ "step": 24768
2332
+ },
2333
+ {
2334
+ "epoch": 1.1,
2335
+ "learning_rate": 6.320970085089982e-07,
2336
+ "loss": 0.1573,
2337
+ "step": 24832
2338
+ },
2339
+ {
2340
+ "epoch": 1.11,
2341
+ "learning_rate": 6.311482700346882e-07,
2342
+ "loss": 0.1143,
2343
+ "step": 24896
2344
+ },
2345
+ {
2346
+ "epoch": 1.11,
2347
+ "learning_rate": 6.301995315603783e-07,
2348
+ "loss": 0.1048,
2349
+ "step": 24960
2350
+ },
2351
+ {
2352
+ "epoch": 1.11,
2353
+ "learning_rate": 6.292507930860684e-07,
2354
+ "loss": 0.0797,
2355
+ "step": 25024
2356
+ },
2357
+ {
2358
+ "epoch": 1.12,
2359
+ "learning_rate": 6.283020546117584e-07,
2360
+ "loss": 0.0676,
2361
+ "step": 25088
2362
+ },
2363
+ {
2364
+ "epoch": 1.12,
2365
+ "learning_rate": 6.273533161374484e-07,
2366
+ "loss": 0.1028,
2367
+ "step": 25152
2368
+ },
2369
+ {
2370
+ "epoch": 1.12,
2371
+ "learning_rate": 6.264045776631385e-07,
2372
+ "loss": 0.1148,
2373
+ "step": 25216
2374
+ },
2375
+ {
2376
+ "epoch": 1.12,
2377
+ "learning_rate": 6.254558391888286e-07,
2378
+ "loss": 0.1185,
2379
+ "step": 25280
2380
+ },
2381
+ {
2382
+ "epoch": 1.13,
2383
+ "learning_rate": 6.245071007145186e-07,
2384
+ "loss": 0.1288,
2385
+ "step": 25344
2386
+ },
2387
+ {
2388
+ "epoch": 1.13,
2389
+ "learning_rate": 6.235583622402088e-07,
2390
+ "loss": 0.1474,
2391
+ "step": 25408
2392
+ },
2393
+ {
2394
+ "epoch": 1.13,
2395
+ "learning_rate": 6.226096237658988e-07,
2396
+ "loss": 0.1075,
2397
+ "step": 25472
2398
+ },
2399
+ {
2400
+ "epoch": 1.14,
2401
+ "learning_rate": 6.216608852915888e-07,
2402
+ "loss": 0.1531,
2403
+ "step": 25536
2404
+ },
2405
+ {
2406
+ "epoch": 1.14,
2407
+ "learning_rate": 6.207121468172789e-07,
2408
+ "loss": 0.1072,
2409
+ "step": 25600
2410
+ },
2411
+ {
2412
+ "epoch": 1.14,
2413
+ "learning_rate": 6.19763408342969e-07,
2414
+ "loss": 0.0739,
2415
+ "step": 25664
2416
+ },
2417
+ {
2418
+ "epoch": 1.14,
2419
+ "learning_rate": 6.188146698686589e-07,
2420
+ "loss": 0.0674,
2421
+ "step": 25728
2422
+ },
2423
+ {
2424
+ "epoch": 1.15,
2425
+ "learning_rate": 6.17865931394349e-07,
2426
+ "loss": 0.0712,
2427
+ "step": 25792
2428
+ },
2429
+ {
2430
+ "epoch": 1.15,
2431
+ "learning_rate": 6.169171929200391e-07,
2432
+ "loss": 0.116,
2433
+ "step": 25856
2434
+ },
2435
+ {
2436
+ "epoch": 1.15,
2437
+ "learning_rate": 6.159684544457292e-07,
2438
+ "loss": 0.1242,
2439
+ "step": 25920
2440
+ },
2441
+ {
2442
+ "epoch": 1.16,
2443
+ "learning_rate": 6.150197159714192e-07,
2444
+ "loss": 0.0647,
2445
+ "step": 25984
2446
+ },
2447
+ {
2448
+ "epoch": 1.16,
2449
+ "learning_rate": 6.140709774971093e-07,
2450
+ "loss": 0.1281,
2451
+ "step": 26048
2452
+ },
2453
+ {
2454
+ "epoch": 1.16,
2455
+ "learning_rate": 6.131222390227994e-07,
2456
+ "loss": 0.1245,
2457
+ "step": 26112
2458
+ },
2459
+ {
2460
+ "epoch": 1.16,
2461
+ "learning_rate": 6.121735005484894e-07,
2462
+ "loss": 0.1712,
2463
+ "step": 26176
2464
+ },
2465
+ {
2466
+ "epoch": 1.17,
2467
+ "learning_rate": 6.112247620741796e-07,
2468
+ "loss": 0.1322,
2469
+ "step": 26240
2470
+ },
2471
+ {
2472
+ "epoch": 1.17,
2473
+ "learning_rate": 6.102760235998695e-07,
2474
+ "loss": 0.1391,
2475
+ "step": 26304
2476
+ },
2477
+ {
2478
+ "epoch": 1.17,
2479
+ "learning_rate": 6.093272851255595e-07,
2480
+ "loss": 0.0661,
2481
+ "step": 26368
2482
+ },
2483
+ {
2484
+ "epoch": 1.18,
2485
+ "learning_rate": 6.083785466512496e-07,
2486
+ "loss": 0.153,
2487
+ "step": 26432
2488
+ },
2489
+ {
2490
+ "epoch": 1.18,
2491
+ "learning_rate": 6.074298081769397e-07,
2492
+ "loss": 0.0904,
2493
+ "step": 26496
2494
+ },
2495
+ {
2496
+ "epoch": 1.18,
2497
+ "learning_rate": 6.064810697026298e-07,
2498
+ "loss": 0.1427,
2499
+ "step": 26560
2500
+ },
2501
+ {
2502
+ "epoch": 1.18,
2503
+ "learning_rate": 6.055323312283198e-07,
2504
+ "loss": 0.0734,
2505
+ "step": 26624
2506
+ },
2507
+ {
2508
+ "epoch": 1.19,
2509
+ "learning_rate": 6.045835927540099e-07,
2510
+ "loss": 0.0615,
2511
+ "step": 26688
2512
+ },
2513
+ {
2514
+ "epoch": 1.19,
2515
+ "learning_rate": 6.036348542797e-07,
2516
+ "loss": 0.1454,
2517
+ "step": 26752
2518
+ },
2519
+ {
2520
+ "epoch": 1.19,
2521
+ "learning_rate": 6.027009398440512e-07,
2522
+ "loss": 0.1196,
2523
+ "step": 26816
2524
+ },
2525
+ {
2526
+ "epoch": 1.2,
2527
+ "learning_rate": 6.017522013697411e-07,
2528
+ "loss": 0.1487,
2529
+ "step": 26880
2530
+ },
2531
+ {
2532
+ "epoch": 1.2,
2533
+ "learning_rate": 6.008182869340923e-07,
2534
+ "loss": 0.1299,
2535
+ "step": 26944
2536
+ },
2537
+ {
2538
+ "epoch": 1.2,
2539
+ "learning_rate": 5.998695484597823e-07,
2540
+ "loss": 0.111,
2541
+ "step": 27008
2542
+ },
2543
+ {
2544
+ "epoch": 1.2,
2545
+ "learning_rate": 5.989208099854724e-07,
2546
+ "loss": 0.0529,
2547
+ "step": 27072
2548
+ },
2549
+ {
2550
+ "epoch": 1.21,
2551
+ "learning_rate": 5.979720715111625e-07,
2552
+ "loss": 0.1411,
2553
+ "step": 27136
2554
+ },
2555
+ {
2556
+ "epoch": 1.21,
2557
+ "learning_rate": 5.970233330368525e-07,
2558
+ "loss": 0.1335,
2559
+ "step": 27200
2560
+ },
2561
+ {
2562
+ "epoch": 1.21,
2563
+ "learning_rate": 5.960745945625426e-07,
2564
+ "loss": 0.1229,
2565
+ "step": 27264
2566
+ },
2567
+ {
2568
+ "epoch": 1.22,
2569
+ "learning_rate": 5.951258560882326e-07,
2570
+ "loss": 0.0501,
2571
+ "step": 27328
2572
+ },
2573
+ {
2574
+ "epoch": 1.22,
2575
+ "learning_rate": 5.941771176139228e-07,
2576
+ "loss": 0.0866,
2577
+ "step": 27392
2578
+ },
2579
+ {
2580
+ "epoch": 1.22,
2581
+ "learning_rate": 5.932283791396128e-07,
2582
+ "loss": 0.0643,
2583
+ "step": 27456
2584
+ },
2585
+ {
2586
+ "epoch": 1.22,
2587
+ "learning_rate": 5.922796406653027e-07,
2588
+ "loss": 0.0793,
2589
+ "step": 27520
2590
+ },
2591
+ {
2592
+ "epoch": 1.23,
2593
+ "learning_rate": 5.913309021909929e-07,
2594
+ "loss": 0.0627,
2595
+ "step": 27584
2596
+ },
2597
+ {
2598
+ "epoch": 1.23,
2599
+ "learning_rate": 5.903821637166829e-07,
2600
+ "loss": 0.1354,
2601
+ "step": 27648
2602
+ },
2603
+ {
2604
+ "epoch": 1.23,
2605
+ "learning_rate": 5.89433425242373e-07,
2606
+ "loss": 0.1341,
2607
+ "step": 27712
2608
+ },
2609
+ {
2610
+ "epoch": 1.24,
2611
+ "learning_rate": 5.884846867680631e-07,
2612
+ "loss": 0.1205,
2613
+ "step": 27776
2614
+ },
2615
+ {
2616
+ "epoch": 1.24,
2617
+ "learning_rate": 5.875359482937531e-07,
2618
+ "loss": 0.1247,
2619
+ "step": 27840
2620
+ },
2621
+ {
2622
+ "epoch": 1.24,
2623
+ "learning_rate": 5.865872098194432e-07,
2624
+ "loss": 0.1071,
2625
+ "step": 27904
2626
+ },
2627
+ {
2628
+ "epoch": 1.24,
2629
+ "learning_rate": 5.856384713451333e-07,
2630
+ "loss": 0.125,
2631
+ "step": 27968
2632
+ },
2633
+ {
2634
+ "epoch": 1.25,
2635
+ "learning_rate": 5.846897328708234e-07,
2636
+ "loss": 0.0815,
2637
+ "step": 28032
2638
+ },
2639
+ {
2640
+ "epoch": 1.25,
2641
+ "learning_rate": 5.837409943965133e-07,
2642
+ "loss": 0.0916,
2643
+ "step": 28096
2644
+ },
2645
+ {
2646
+ "epoch": 1.25,
2647
+ "learning_rate": 5.827922559222033e-07,
2648
+ "loss": 0.1076,
2649
+ "step": 28160
2650
+ },
2651
+ {
2652
+ "epoch": 1.26,
2653
+ "learning_rate": 5.818435174478935e-07,
2654
+ "loss": 0.1217,
2655
+ "step": 28224
2656
+ },
2657
+ {
2658
+ "epoch": 1.26,
2659
+ "learning_rate": 5.808947789735835e-07,
2660
+ "loss": 0.0918,
2661
+ "step": 28288
2662
+ },
2663
+ {
2664
+ "epoch": 1.26,
2665
+ "learning_rate": 5.799460404992736e-07,
2666
+ "loss": 0.1532,
2667
+ "step": 28352
2668
+ },
2669
+ {
2670
+ "epoch": 1.26,
2671
+ "learning_rate": 5.789973020249637e-07,
2672
+ "loss": 0.0839,
2673
+ "step": 28416
2674
+ },
2675
+ {
2676
+ "epoch": 1.27,
2677
+ "learning_rate": 5.780485635506537e-07,
2678
+ "loss": 0.1425,
2679
+ "step": 28480
2680
+ },
2681
+ {
2682
+ "epoch": 1.27,
2683
+ "learning_rate": 5.770998250763438e-07,
2684
+ "loss": 0.0679,
2685
+ "step": 28544
2686
+ },
2687
+ {
2688
+ "epoch": 1.27,
2689
+ "learning_rate": 5.761510866020339e-07,
2690
+ "loss": 0.0667,
2691
+ "step": 28608
2692
+ },
2693
+ {
2694
+ "epoch": 1.28,
2695
+ "learning_rate": 5.75202348127724e-07,
2696
+ "loss": 0.1245,
2697
+ "step": 28672
2698
+ },
2699
+ {
2700
+ "epoch": 1.28,
2701
+ "learning_rate": 5.742536096534139e-07,
2702
+ "loss": 0.1255,
2703
+ "step": 28736
2704
+ },
2705
+ {
2706
+ "epoch": 1.28,
2707
+ "learning_rate": 5.73304871179104e-07,
2708
+ "loss": 0.0819,
2709
+ "step": 28800
2710
+ },
2711
+ {
2712
+ "epoch": 1.28,
2713
+ "learning_rate": 5.723561327047941e-07,
2714
+ "loss": 0.1035,
2715
+ "step": 28864
2716
+ },
2717
+ {
2718
+ "epoch": 1.29,
2719
+ "learning_rate": 5.714073942304841e-07,
2720
+ "loss": 0.1238,
2721
+ "step": 28928
2722
+ },
2723
+ {
2724
+ "epoch": 1.29,
2725
+ "learning_rate": 5.704586557561741e-07,
2726
+ "loss": 0.1312,
2727
+ "step": 28992
2728
+ },
2729
+ {
2730
+ "epoch": 1.29,
2731
+ "learning_rate": 5.695099172818643e-07,
2732
+ "loss": 0.0645,
2733
+ "step": 29056
2734
+ },
2735
+ {
2736
+ "epoch": 1.3,
2737
+ "learning_rate": 5.685611788075543e-07,
2738
+ "loss": 0.1143,
2739
+ "step": 29120
2740
+ },
2741
+ {
2742
+ "epoch": 1.3,
2743
+ "learning_rate": 5.676124403332444e-07,
2744
+ "loss": 0.0848,
2745
+ "step": 29184
2746
+ },
2747
+ {
2748
+ "epoch": 1.3,
2749
+ "learning_rate": 5.666637018589344e-07,
2750
+ "loss": 0.1051,
2751
+ "step": 29248
2752
+ },
2753
+ {
2754
+ "epoch": 1.3,
2755
+ "learning_rate": 5.657149633846244e-07,
2756
+ "loss": 0.0981,
2757
+ "step": 29312
2758
+ },
2759
+ {
2760
+ "epoch": 1.31,
2761
+ "learning_rate": 5.647662249103145e-07,
2762
+ "loss": 0.0894,
2763
+ "step": 29376
2764
+ },
2765
+ {
2766
+ "epoch": 1.31,
2767
+ "learning_rate": 5.638174864360046e-07,
2768
+ "loss": 0.1164,
2769
+ "step": 29440
2770
+ },
2771
+ {
2772
+ "epoch": 1.31,
2773
+ "learning_rate": 5.628687479616947e-07,
2774
+ "loss": 0.0816,
2775
+ "step": 29504
2776
+ },
2777
+ {
2778
+ "epoch": 1.31,
2779
+ "learning_rate": 5.619200094873847e-07,
2780
+ "loss": 0.1426,
2781
+ "step": 29568
2782
+ },
2783
+ {
2784
+ "epoch": 1.32,
2785
+ "learning_rate": 5.609712710130748e-07,
2786
+ "loss": 0.1748,
2787
+ "step": 29632
2788
+ },
2789
+ {
2790
+ "epoch": 1.32,
2791
+ "learning_rate": 5.600225325387649e-07,
2792
+ "loss": 0.0979,
2793
+ "step": 29696
2794
+ },
2795
+ {
2796
+ "epoch": 1.32,
2797
+ "learning_rate": 5.590737940644549e-07,
2798
+ "loss": 0.1499,
2799
+ "step": 29760
2800
+ },
2801
+ {
2802
+ "epoch": 1.33,
2803
+ "learning_rate": 5.58125055590145e-07,
2804
+ "loss": 0.0845,
2805
+ "step": 29824
2806
+ },
2807
+ {
2808
+ "epoch": 1.33,
2809
+ "learning_rate": 5.57176317115835e-07,
2810
+ "loss": 0.1298,
2811
+ "step": 29888
2812
+ },
2813
+ {
2814
+ "epoch": 1.33,
2815
+ "learning_rate": 5.56227578641525e-07,
2816
+ "loss": 0.1241,
2817
+ "step": 29952
2818
+ },
2819
+ {
2820
+ "epoch": 1.33,
2821
+ "learning_rate": 5.552788401672151e-07,
2822
+ "loss": 0.1174,
2823
+ "step": 30016
2824
+ },
2825
+ {
2826
+ "epoch": 1.34,
2827
+ "learning_rate": 5.543301016929052e-07,
2828
+ "loss": 0.1209,
2829
+ "step": 30080
2830
+ },
2831
+ {
2832
+ "epoch": 1.34,
2833
+ "learning_rate": 5.533813632185953e-07,
2834
+ "loss": 0.1071,
2835
+ "step": 30144
2836
+ },
2837
+ {
2838
+ "epoch": 1.34,
2839
+ "learning_rate": 5.524326247442853e-07,
2840
+ "loss": 0.0615,
2841
+ "step": 30208
2842
+ },
2843
+ {
2844
+ "epoch": 1.35,
2845
+ "learning_rate": 5.514838862699754e-07,
2846
+ "loss": 0.0797,
2847
+ "step": 30272
2848
+ },
2849
+ {
2850
+ "epoch": 1.35,
2851
+ "learning_rate": 5.505351477956655e-07,
2852
+ "loss": 0.1182,
2853
+ "step": 30336
2854
+ },
2855
+ {
2856
+ "epoch": 1.35,
2857
+ "learning_rate": 5.495864093213554e-07,
2858
+ "loss": 0.1027,
2859
+ "step": 30400
2860
+ },
2861
+ {
2862
+ "epoch": 1.35,
2863
+ "learning_rate": 5.486376708470456e-07,
2864
+ "loss": 0.0997,
2865
+ "step": 30464
2866
+ },
2867
+ {
2868
+ "epoch": 1.36,
2869
+ "learning_rate": 5.476889323727356e-07,
2870
+ "loss": 0.1624,
2871
+ "step": 30528
2872
+ },
2873
+ {
2874
+ "epoch": 1.36,
2875
+ "learning_rate": 5.467401938984256e-07,
2876
+ "loss": 0.0811,
2877
+ "step": 30592
2878
+ },
2879
+ {
2880
+ "epoch": 1.36,
2881
+ "learning_rate": 5.457914554241158e-07,
2882
+ "loss": 0.0667,
2883
+ "step": 30656
2884
+ },
2885
+ {
2886
+ "epoch": 1.37,
2887
+ "learning_rate": 5.448427169498058e-07,
2888
+ "loss": 0.1017,
2889
+ "step": 30720
2890
+ },
2891
+ {
2892
+ "epoch": 1.37,
2893
+ "learning_rate": 5.438939784754958e-07,
2894
+ "loss": 0.0942,
2895
+ "step": 30784
2896
+ },
2897
+ {
2898
+ "epoch": 1.37,
2899
+ "learning_rate": 5.429452400011859e-07,
2900
+ "loss": 0.1741,
2901
+ "step": 30848
2902
+ },
2903
+ {
2904
+ "epoch": 1.37,
2905
+ "learning_rate": 5.41996501526876e-07,
2906
+ "loss": 0.1527,
2907
+ "step": 30912
2908
+ },
2909
+ {
2910
+ "epoch": 1.38,
2911
+ "learning_rate": 5.41047763052566e-07,
2912
+ "loss": 0.0979,
2913
+ "step": 30976
2914
+ },
2915
+ {
2916
+ "epoch": 1.38,
2917
+ "learning_rate": 5.40099024578256e-07,
2918
+ "loss": 0.0569,
2919
+ "step": 31040
2920
+ },
2921
+ {
2922
+ "epoch": 1.38,
2923
+ "learning_rate": 5.391502861039461e-07,
2924
+ "loss": 0.1476,
2925
+ "step": 31104
2926
+ },
2927
+ {
2928
+ "epoch": 1.39,
2929
+ "learning_rate": 5.382015476296362e-07,
2930
+ "loss": 0.097,
2931
+ "step": 31168
2932
+ },
2933
+ {
2934
+ "epoch": 1.39,
2935
+ "learning_rate": 5.372676331939874e-07,
2936
+ "loss": 0.0647,
2937
+ "step": 31232
2938
+ },
2939
+ {
2940
+ "epoch": 1.39,
2941
+ "learning_rate": 5.363188947196774e-07,
2942
+ "loss": 0.103,
2943
+ "step": 31296
2944
+ },
2945
+ {
2946
+ "epoch": 1.39,
2947
+ "learning_rate": 5.353701562453674e-07,
2948
+ "loss": 0.0903,
2949
+ "step": 31360
2950
+ },
2951
+ {
2952
+ "epoch": 1.4,
2953
+ "learning_rate": 5.344214177710576e-07,
2954
+ "loss": 0.1414,
2955
+ "step": 31424
2956
+ },
2957
+ {
2958
+ "epoch": 1.4,
2959
+ "learning_rate": 5.334726792967476e-07,
2960
+ "loss": 0.0945,
2961
+ "step": 31488
2962
+ },
2963
+ {
2964
+ "epoch": 1.4,
2965
+ "learning_rate": 5.325239408224377e-07,
2966
+ "loss": 0.1426,
2967
+ "step": 31552
2968
+ },
2969
+ {
2970
+ "epoch": 1.41,
2971
+ "learning_rate": 5.315752023481276e-07,
2972
+ "loss": 0.1063,
2973
+ "step": 31616
2974
+ },
2975
+ {
2976
+ "epoch": 1.41,
2977
+ "learning_rate": 5.306264638738177e-07,
2978
+ "loss": 0.0921,
2979
+ "step": 31680
2980
+ },
2981
+ {
2982
+ "epoch": 1.41,
2983
+ "learning_rate": 5.296777253995078e-07,
2984
+ "loss": 0.0904,
2985
+ "step": 31744
2986
+ },
2987
+ {
2988
+ "epoch": 1.41,
2989
+ "learning_rate": 5.287289869251978e-07,
2990
+ "loss": 0.1278,
2991
+ "step": 31808
2992
+ },
2993
+ {
2994
+ "epoch": 1.42,
2995
+ "learning_rate": 5.27780248450888e-07,
2996
+ "loss": 0.1034,
2997
+ "step": 31872
2998
+ },
2999
+ {
3000
+ "epoch": 1.42,
3001
+ "learning_rate": 5.26831509976578e-07,
3002
+ "loss": 0.0908,
3003
+ "step": 31936
3004
+ },
3005
+ {
3006
+ "epoch": 1.42,
3007
+ "learning_rate": 5.25882771502268e-07,
3008
+ "loss": 0.1339,
3009
+ "step": 32000
3010
+ },
3011
+ {
3012
+ "epoch": 1.43,
3013
+ "learning_rate": 5.249340330279582e-07,
3014
+ "loss": 0.0777,
3015
+ "step": 32064
3016
+ },
3017
+ {
3018
+ "epoch": 1.43,
3019
+ "learning_rate": 5.239852945536482e-07,
3020
+ "loss": 0.1353,
3021
+ "step": 32128
3022
+ },
3023
+ {
3024
+ "epoch": 1.43,
3025
+ "learning_rate": 5.230365560793382e-07,
3026
+ "loss": 0.1369,
3027
+ "step": 32192
3028
+ },
3029
+ {
3030
+ "epoch": 1.43,
3031
+ "learning_rate": 5.220878176050283e-07,
3032
+ "loss": 0.087,
3033
+ "step": 32256
3034
+ },
3035
+ {
3036
+ "epoch": 1.44,
3037
+ "learning_rate": 5.211390791307183e-07,
3038
+ "loss": 0.129,
3039
+ "step": 32320
3040
+ },
3041
+ {
3042
+ "epoch": 1.44,
3043
+ "learning_rate": 5.201903406564084e-07,
3044
+ "loss": 0.1012,
3045
+ "step": 32384
3046
+ },
3047
+ {
3048
+ "epoch": 1.44,
3049
+ "learning_rate": 5.192416021820984e-07,
3050
+ "loss": 0.1114,
3051
+ "step": 32448
3052
+ },
3053
+ {
3054
+ "epoch": 1.45,
3055
+ "learning_rate": 5.182928637077886e-07,
3056
+ "loss": 0.0917,
3057
+ "step": 32512
3058
+ },
3059
+ {
3060
+ "epoch": 1.45,
3061
+ "learning_rate": 5.173441252334786e-07,
3062
+ "loss": 0.0852,
3063
+ "step": 32576
3064
+ },
3065
+ {
3066
+ "epoch": 1.45,
3067
+ "learning_rate": 5.163953867591686e-07,
3068
+ "loss": 0.0937,
3069
+ "step": 32640
3070
+ },
3071
+ {
3072
+ "epoch": 1.45,
3073
+ "learning_rate": 5.154466482848588e-07,
3074
+ "loss": 0.1014,
3075
+ "step": 32704
3076
+ },
3077
+ {
3078
+ "epoch": 1.46,
3079
+ "learning_rate": 5.144979098105487e-07,
3080
+ "loss": 0.0621,
3081
+ "step": 32768
3082
+ },
3083
+ {
3084
+ "epoch": 1.46,
3085
+ "learning_rate": 5.135491713362388e-07,
3086
+ "loss": 0.0927,
3087
+ "step": 32832
3088
+ },
3089
+ {
3090
+ "epoch": 1.46,
3091
+ "learning_rate": 5.126004328619289e-07,
3092
+ "loss": 0.1182,
3093
+ "step": 32896
3094
+ },
3095
+ {
3096
+ "epoch": 1.47,
3097
+ "learning_rate": 5.116516943876189e-07,
3098
+ "loss": 0.1206,
3099
+ "step": 32960
3100
+ },
3101
+ {
3102
+ "epoch": 1.47,
3103
+ "learning_rate": 5.10702955913309e-07,
3104
+ "loss": 0.1605,
3105
+ "step": 33024
3106
+ },
3107
+ {
3108
+ "epoch": 1.47,
3109
+ "learning_rate": 5.097542174389991e-07,
3110
+ "loss": 0.1525,
3111
+ "step": 33088
3112
+ },
3113
+ {
3114
+ "epoch": 1.47,
3115
+ "learning_rate": 5.088054789646892e-07,
3116
+ "loss": 0.0679,
3117
+ "step": 33152
3118
+ },
3119
+ {
3120
+ "epoch": 1.48,
3121
+ "learning_rate": 5.078567404903792e-07,
3122
+ "loss": 0.1432,
3123
+ "step": 33216
3124
+ },
3125
+ {
3126
+ "epoch": 1.48,
3127
+ "learning_rate": 5.069080020160692e-07,
3128
+ "loss": 0.1117,
3129
+ "step": 33280
3130
+ },
3131
+ {
3132
+ "epoch": 1.48,
3133
+ "learning_rate": 5.059592635417593e-07,
3134
+ "loss": 0.0866,
3135
+ "step": 33344
3136
+ },
3137
+ {
3138
+ "epoch": 1.49,
3139
+ "learning_rate": 5.050105250674493e-07,
3140
+ "loss": 0.0727,
3141
+ "step": 33408
3142
+ },
3143
+ {
3144
+ "epoch": 1.49,
3145
+ "learning_rate": 5.040617865931393e-07,
3146
+ "loss": 0.0671,
3147
+ "step": 33472
3148
+ },
3149
+ {
3150
+ "epoch": 1.49,
3151
+ "learning_rate": 5.031130481188295e-07,
3152
+ "loss": 0.1083,
3153
+ "step": 33536
3154
+ },
3155
+ {
3156
+ "epoch": 1.49,
3157
+ "learning_rate": 5.021791336831806e-07,
3158
+ "loss": 0.1442,
3159
+ "step": 33600
3160
+ },
3161
+ {
3162
+ "epoch": 1.5,
3163
+ "learning_rate": 5.012303952088707e-07,
3164
+ "loss": 0.1382,
3165
+ "step": 33664
3166
+ },
3167
+ {
3168
+ "epoch": 1.5,
3169
+ "learning_rate": 5.002816567345607e-07,
3170
+ "loss": 0.1495,
3171
+ "step": 33728
3172
+ },
3173
+ {
3174
+ "epoch": 1.5,
3175
+ "learning_rate": 4.993329182602508e-07,
3176
+ "loss": 0.1246,
3177
+ "step": 33792
3178
+ },
3179
+ {
3180
+ "epoch": 1.51,
3181
+ "learning_rate": 4.983841797859409e-07,
3182
+ "loss": 0.0701,
3183
+ "step": 33856
3184
+ },
3185
+ {
3186
+ "epoch": 1.51,
3187
+ "learning_rate": 4.974354413116309e-07,
3188
+ "loss": 0.1176,
3189
+ "step": 33920
3190
+ },
3191
+ {
3192
+ "epoch": 1.51,
3193
+ "learning_rate": 4.964867028373209e-07,
3194
+ "loss": 0.1294,
3195
+ "step": 33984
3196
+ },
3197
+ {
3198
+ "epoch": 1.51,
3199
+ "learning_rate": 4.95537964363011e-07,
3200
+ "loss": 0.1144,
3201
+ "step": 34048
3202
+ },
3203
+ {
3204
+ "epoch": 1.52,
3205
+ "learning_rate": 4.945892258887011e-07,
3206
+ "loss": 0.1356,
3207
+ "step": 34112
3208
+ },
3209
+ {
3210
+ "epoch": 1.52,
3211
+ "learning_rate": 4.936404874143911e-07,
3212
+ "loss": 0.0847,
3213
+ "step": 34176
3214
+ },
3215
+ {
3216
+ "epoch": 1.52,
3217
+ "learning_rate": 4.926917489400812e-07,
3218
+ "loss": 0.1499,
3219
+ "step": 34240
3220
+ },
3221
+ {
3222
+ "epoch": 1.53,
3223
+ "learning_rate": 4.917430104657713e-07,
3224
+ "loss": 0.1345,
3225
+ "step": 34304
3226
+ },
3227
+ {
3228
+ "epoch": 1.53,
3229
+ "learning_rate": 4.907942719914614e-07,
3230
+ "loss": 0.0594,
3231
+ "step": 34368
3232
+ },
3233
+ {
3234
+ "epoch": 1.53,
3235
+ "learning_rate": 4.898455335171514e-07,
3236
+ "loss": 0.1239,
3237
+ "step": 34432
3238
+ },
3239
+ {
3240
+ "epoch": 1.53,
3241
+ "learning_rate": 4.888967950428415e-07,
3242
+ "loss": 0.0936,
3243
+ "step": 34496
3244
+ },
3245
+ {
3246
+ "epoch": 1.54,
3247
+ "learning_rate": 4.879480565685315e-07,
3248
+ "loss": 0.0963,
3249
+ "step": 34560
3250
+ },
3251
+ {
3252
+ "epoch": 1.54,
3253
+ "learning_rate": 4.869993180942215e-07,
3254
+ "loss": 0.1015,
3255
+ "step": 34624
3256
+ },
3257
+ {
3258
+ "epoch": 1.54,
3259
+ "learning_rate": 4.860505796199116e-07,
3260
+ "loss": 0.0758,
3261
+ "step": 34688
3262
+ },
3263
+ {
3264
+ "epoch": 1.55,
3265
+ "learning_rate": 4.851018411456017e-07,
3266
+ "loss": 0.0954,
3267
+ "step": 34752
3268
+ },
3269
+ {
3270
+ "epoch": 1.55,
3271
+ "learning_rate": 4.841531026712917e-07,
3272
+ "loss": 0.1693,
3273
+ "step": 34816
3274
+ },
3275
+ {
3276
+ "epoch": 1.55,
3277
+ "learning_rate": 4.832043641969818e-07,
3278
+ "loss": 0.1428,
3279
+ "step": 34880
3280
+ },
3281
+ {
3282
+ "epoch": 1.55,
3283
+ "learning_rate": 4.822556257226719e-07,
3284
+ "loss": 0.1161,
3285
+ "step": 34944
3286
+ },
3287
+ {
3288
+ "epoch": 1.56,
3289
+ "learning_rate": 4.81306887248362e-07,
3290
+ "loss": 0.0599,
3291
+ "step": 35008
3292
+ },
3293
+ {
3294
+ "epoch": 1.56,
3295
+ "learning_rate": 4.803581487740521e-07,
3296
+ "loss": 0.1093,
3297
+ "step": 35072
3298
+ },
3299
+ {
3300
+ "epoch": 1.56,
3301
+ "learning_rate": 4.79409410299742e-07,
3302
+ "loss": 0.0763,
3303
+ "step": 35136
3304
+ },
3305
+ {
3306
+ "epoch": 1.57,
3307
+ "learning_rate": 4.784606718254321e-07,
3308
+ "loss": 0.119,
3309
+ "step": 35200
3310
+ },
3311
+ {
3312
+ "epoch": 1.57,
3313
+ "learning_rate": 4.775119333511221e-07,
3314
+ "loss": 0.1467,
3315
+ "step": 35264
3316
+ },
3317
+ {
3318
+ "epoch": 1.57,
3319
+ "learning_rate": 4.765631948768122e-07,
3320
+ "loss": 0.1311,
3321
+ "step": 35328
3322
+ },
3323
+ {
3324
+ "epoch": 1.57,
3325
+ "learning_rate": 4.7561445640250225e-07,
3326
+ "loss": 0.1628,
3327
+ "step": 35392
3328
+ },
3329
+ {
3330
+ "epoch": 1.58,
3331
+ "learning_rate": 4.7466571792819234e-07,
3332
+ "loss": 0.0847,
3333
+ "step": 35456
3334
+ },
3335
+ {
3336
+ "epoch": 1.58,
3337
+ "learning_rate": 4.737169794538824e-07,
3338
+ "loss": 0.0888,
3339
+ "step": 35520
3340
+ },
3341
+ {
3342
+ "epoch": 1.58,
3343
+ "learning_rate": 4.7276824097957247e-07,
3344
+ "loss": 0.1145,
3345
+ "step": 35584
3346
+ },
3347
+ {
3348
+ "epoch": 1.59,
3349
+ "learning_rate": 4.7181950250526256e-07,
3350
+ "loss": 0.1296,
3351
+ "step": 35648
3352
+ },
3353
+ {
3354
+ "epoch": 1.59,
3355
+ "learning_rate": 4.7087076403095254e-07,
3356
+ "loss": 0.0984,
3357
+ "step": 35712
3358
+ },
3359
+ {
3360
+ "epoch": 1.59,
3361
+ "learning_rate": 4.6992202555664263e-07,
3362
+ "loss": 0.1509,
3363
+ "step": 35776
3364
+ },
3365
+ {
3366
+ "epoch": 1.59,
3367
+ "learning_rate": 4.6897328708233267e-07,
3368
+ "loss": 0.1288,
3369
+ "step": 35840
3370
+ },
3371
+ {
3372
+ "epoch": 1.6,
3373
+ "learning_rate": 4.6802454860802276e-07,
3374
+ "loss": 0.1383,
3375
+ "step": 35904
3376
+ },
3377
+ {
3378
+ "epoch": 1.6,
3379
+ "learning_rate": 4.6707581013371285e-07,
3380
+ "loss": 0.1004,
3381
+ "step": 35968
3382
+ },
3383
+ {
3384
+ "epoch": 1.6,
3385
+ "learning_rate": 4.6614189569806395e-07,
3386
+ "loss": 0.1098,
3387
+ "step": 36032
3388
+ },
3389
+ {
3390
+ "epoch": 1.61,
3391
+ "learning_rate": 4.6519315722375404e-07,
3392
+ "loss": 0.1092,
3393
+ "step": 36096
3394
+ },
3395
+ {
3396
+ "epoch": 1.61,
3397
+ "learning_rate": 4.642444187494441e-07,
3398
+ "loss": 0.1028,
3399
+ "step": 36160
3400
+ },
3401
+ {
3402
+ "epoch": 1.61,
3403
+ "learning_rate": 4.6329568027513417e-07,
3404
+ "loss": 0.1196,
3405
+ "step": 36224
3406
+ },
3407
+ {
3408
+ "epoch": 1.61,
3409
+ "learning_rate": 4.6234694180082415e-07,
3410
+ "loss": 0.1209,
3411
+ "step": 36288
3412
+ },
3413
+ {
3414
+ "epoch": 1.62,
3415
+ "learning_rate": 4.6139820332651424e-07,
3416
+ "loss": 0.0756,
3417
+ "step": 36352
3418
+ },
3419
+ {
3420
+ "epoch": 1.62,
3421
+ "learning_rate": 4.6044946485220433e-07,
3422
+ "loss": 0.0949,
3423
+ "step": 36416
3424
+ },
3425
+ {
3426
+ "epoch": 1.62,
3427
+ "learning_rate": 4.5950072637789437e-07,
3428
+ "loss": 0.0852,
3429
+ "step": 36480
3430
+ },
3431
+ {
3432
+ "epoch": 1.63,
3433
+ "learning_rate": 4.585668119422455e-07,
3434
+ "loss": 0.1649,
3435
+ "step": 36544
3436
+ },
3437
+ {
3438
+ "epoch": 1.63,
3439
+ "learning_rate": 4.5761807346793556e-07,
3440
+ "loss": 0.1239,
3441
+ "step": 36608
3442
+ },
3443
+ {
3444
+ "epoch": 1.63,
3445
+ "learning_rate": 4.5666933499362565e-07,
3446
+ "loss": 0.0986,
3447
+ "step": 36672
3448
+ },
3449
+ {
3450
+ "epoch": 1.63,
3451
+ "learning_rate": 4.5572059651931574e-07,
3452
+ "loss": 0.1478,
3453
+ "step": 36736
3454
+ },
3455
+ {
3456
+ "epoch": 1.64,
3457
+ "learning_rate": 4.547718580450058e-07,
3458
+ "loss": 0.1032,
3459
+ "step": 36800
3460
+ },
3461
+ {
3462
+ "epoch": 1.64,
3463
+ "learning_rate": 4.538231195706958e-07,
3464
+ "loss": 0.1364,
3465
+ "step": 36864
3466
+ },
3467
+ {
3468
+ "epoch": 1.64,
3469
+ "learning_rate": 4.5287438109638585e-07,
3470
+ "loss": 0.0667,
3471
+ "step": 36928
3472
+ },
3473
+ {
3474
+ "epoch": 1.65,
3475
+ "learning_rate": 4.5192564262207594e-07,
3476
+ "loss": 0.0836,
3477
+ "step": 36992
3478
+ },
3479
+ {
3480
+ "epoch": 1.65,
3481
+ "learning_rate": 4.5097690414776603e-07,
3482
+ "loss": 0.1446,
3483
+ "step": 37056
3484
+ },
3485
+ {
3486
+ "epoch": 1.65,
3487
+ "learning_rate": 4.50028165673456e-07,
3488
+ "loss": 0.0684,
3489
+ "step": 37120
3490
+ },
3491
+ {
3492
+ "epoch": 1.65,
3493
+ "learning_rate": 4.490794271991461e-07,
3494
+ "loss": 0.1048,
3495
+ "step": 37184
3496
+ },
3497
+ {
3498
+ "epoch": 1.66,
3499
+ "learning_rate": 4.4813068872483615e-07,
3500
+ "loss": 0.0935,
3501
+ "step": 37248
3502
+ },
3503
+ {
3504
+ "epoch": 1.66,
3505
+ "learning_rate": 4.4718195025052624e-07,
3506
+ "loss": 0.1176,
3507
+ "step": 37312
3508
+ },
3509
+ {
3510
+ "epoch": 1.66,
3511
+ "learning_rate": 4.4623321177621633e-07,
3512
+ "loss": 0.095,
3513
+ "step": 37376
3514
+ },
3515
+ {
3516
+ "epoch": 1.67,
3517
+ "learning_rate": 4.452844733019063e-07,
3518
+ "loss": 0.139,
3519
+ "step": 37440
3520
+ },
3521
+ {
3522
+ "epoch": 1.67,
3523
+ "learning_rate": 4.443357348275964e-07,
3524
+ "loss": 0.0939,
3525
+ "step": 37504
3526
+ },
3527
+ {
3528
+ "epoch": 1.67,
3529
+ "learning_rate": 4.433869963532865e-07,
3530
+ "loss": 0.0793,
3531
+ "step": 37568
3532
+ },
3533
+ {
3534
+ "epoch": 1.67,
3535
+ "learning_rate": 4.4243825787897653e-07,
3536
+ "loss": 0.1213,
3537
+ "step": 37632
3538
+ },
3539
+ {
3540
+ "epoch": 1.68,
3541
+ "learning_rate": 4.414895194046666e-07,
3542
+ "loss": 0.1258,
3543
+ "step": 37696
3544
+ },
3545
+ {
3546
+ "epoch": 1.68,
3547
+ "learning_rate": 4.405407809303566e-07,
3548
+ "loss": 0.0902,
3549
+ "step": 37760
3550
+ },
3551
+ {
3552
+ "epoch": 1.68,
3553
+ "learning_rate": 4.395920424560467e-07,
3554
+ "loss": 0.1153,
3555
+ "step": 37824
3556
+ },
3557
+ {
3558
+ "epoch": 1.68,
3559
+ "learning_rate": 4.386433039817368e-07,
3560
+ "loss": 0.1569,
3561
+ "step": 37888
3562
+ },
3563
+ {
3564
+ "epoch": 1.69,
3565
+ "learning_rate": 4.376945655074268e-07,
3566
+ "loss": 0.0657,
3567
+ "step": 37952
3568
+ },
3569
+ {
3570
+ "epoch": 1.69,
3571
+ "learning_rate": 4.367458270331169e-07,
3572
+ "loss": 0.1206,
3573
+ "step": 38016
3574
+ },
3575
+ {
3576
+ "epoch": 1.69,
3577
+ "learning_rate": 4.357970885588069e-07,
3578
+ "loss": 0.1313,
3579
+ "step": 38080
3580
+ },
3581
+ {
3582
+ "epoch": 1.7,
3583
+ "learning_rate": 4.34848350084497e-07,
3584
+ "loss": 0.1358,
3585
+ "step": 38144
3586
+ },
3587
+ {
3588
+ "epoch": 1.7,
3589
+ "learning_rate": 4.338996116101871e-07,
3590
+ "loss": 0.0711,
3591
+ "step": 38208
3592
+ },
3593
+ {
3594
+ "epoch": 1.7,
3595
+ "learning_rate": 4.329508731358771e-07,
3596
+ "loss": 0.1359,
3597
+ "step": 38272
3598
+ },
3599
+ {
3600
+ "epoch": 1.7,
3601
+ "learning_rate": 4.3200213466156716e-07,
3602
+ "loss": 0.087,
3603
+ "step": 38336
3604
+ },
3605
+ {
3606
+ "epoch": 1.71,
3607
+ "learning_rate": 4.3105339618725725e-07,
3608
+ "loss": 0.0831,
3609
+ "step": 38400
3610
+ },
3611
+ {
3612
+ "epoch": 1.71,
3613
+ "learning_rate": 4.301046577129473e-07,
3614
+ "loss": 0.065,
3615
+ "step": 38464
3616
+ },
3617
+ {
3618
+ "epoch": 1.71,
3619
+ "learning_rate": 4.291559192386374e-07,
3620
+ "loss": 0.1132,
3621
+ "step": 38528
3622
+ },
3623
+ {
3624
+ "epoch": 1.72,
3625
+ "learning_rate": 4.282071807643274e-07,
3626
+ "loss": 0.1098,
3627
+ "step": 38592
3628
+ },
3629
+ {
3630
+ "epoch": 1.72,
3631
+ "learning_rate": 4.2725844229001745e-07,
3632
+ "loss": 0.0918,
3633
+ "step": 38656
3634
+ },
3635
+ {
3636
+ "epoch": 1.72,
3637
+ "learning_rate": 4.2630970381570754e-07,
3638
+ "loss": 0.1517,
3639
+ "step": 38720
3640
+ },
3641
+ {
3642
+ "epoch": 1.72,
3643
+ "learning_rate": 4.253609653413976e-07,
3644
+ "loss": 0.0869,
3645
+ "step": 38784
3646
+ },
3647
+ {
3648
+ "epoch": 1.73,
3649
+ "learning_rate": 4.2441222686708767e-07,
3650
+ "loss": 0.1281,
3651
+ "step": 38848
3652
+ },
3653
+ {
3654
+ "epoch": 1.73,
3655
+ "learning_rate": 4.234634883927777e-07,
3656
+ "loss": 0.0536,
3657
+ "step": 38912
3658
+ },
3659
+ {
3660
+ "epoch": 1.73,
3661
+ "learning_rate": 4.2251474991846775e-07,
3662
+ "loss": 0.1776,
3663
+ "step": 38976
3664
+ },
3665
+ {
3666
+ "epoch": 1.74,
3667
+ "learning_rate": 4.2156601144415784e-07,
3668
+ "loss": 0.15,
3669
+ "step": 39040
3670
+ },
3671
+ {
3672
+ "epoch": 1.74,
3673
+ "learning_rate": 4.206172729698479e-07,
3674
+ "loss": 0.0522,
3675
+ "step": 39104
3676
+ },
3677
+ {
3678
+ "epoch": 1.74,
3679
+ "learning_rate": 4.1966853449553797e-07,
3680
+ "loss": 0.0863,
3681
+ "step": 39168
3682
+ },
3683
+ {
3684
+ "epoch": 1.74,
3685
+ "learning_rate": 4.18719796021228e-07,
3686
+ "loss": 0.0848,
3687
+ "step": 39232
3688
+ },
3689
+ {
3690
+ "epoch": 1.75,
3691
+ "learning_rate": 4.1777105754691804e-07,
3692
+ "loss": 0.1043,
3693
+ "step": 39296
3694
+ },
3695
+ {
3696
+ "epoch": 1.75,
3697
+ "learning_rate": 4.1682231907260813e-07,
3698
+ "loss": 0.0975,
3699
+ "step": 39360
3700
+ },
3701
+ {
3702
+ "epoch": 1.75,
3703
+ "learning_rate": 4.1587358059829817e-07,
3704
+ "loss": 0.155,
3705
+ "step": 39424
3706
+ },
3707
+ {
3708
+ "epoch": 1.76,
3709
+ "learning_rate": 4.1492484212398826e-07,
3710
+ "loss": 0.0518,
3711
+ "step": 39488
3712
+ },
3713
+ {
3714
+ "epoch": 1.76,
3715
+ "learning_rate": 4.139761036496783e-07,
3716
+ "loss": 0.1139,
3717
+ "step": 39552
3718
+ },
3719
+ {
3720
+ "epoch": 1.76,
3721
+ "learning_rate": 4.1304218921402945e-07,
3722
+ "loss": 0.1094,
3723
+ "step": 39616
3724
+ },
3725
+ {
3726
+ "epoch": 1.76,
3727
+ "learning_rate": 4.1209345073971954e-07,
3728
+ "loss": 0.0979,
3729
+ "step": 39680
3730
+ },
3731
+ {
3732
+ "epoch": 1.77,
3733
+ "learning_rate": 4.111447122654096e-07,
3734
+ "loss": 0.0968,
3735
+ "step": 39744
3736
+ },
3737
+ {
3738
+ "epoch": 1.77,
3739
+ "learning_rate": 4.101959737910996e-07,
3740
+ "loss": 0.0828,
3741
+ "step": 39808
3742
+ },
3743
+ {
3744
+ "epoch": 1.77,
3745
+ "learning_rate": 4.0924723531678965e-07,
3746
+ "loss": 0.1267,
3747
+ "step": 39872
3748
+ },
3749
+ {
3750
+ "epoch": 1.78,
3751
+ "learning_rate": 4.0829849684247974e-07,
3752
+ "loss": 0.1223,
3753
+ "step": 39936
3754
+ },
3755
+ {
3756
+ "epoch": 1.78,
3757
+ "learning_rate": 4.0734975836816983e-07,
3758
+ "loss": 0.0536,
3759
+ "step": 40000
3760
+ },
3761
+ {
3762
+ "epoch": 1.78,
3763
+ "learning_rate": 4.0640101989385987e-07,
3764
+ "loss": 0.0832,
3765
+ "step": 40064
3766
+ },
3767
+ {
3768
+ "epoch": 1.78,
3769
+ "learning_rate": 4.054522814195499e-07,
3770
+ "loss": 0.1293,
3771
+ "step": 40128
3772
+ },
3773
+ {
3774
+ "epoch": 1.79,
3775
+ "learning_rate": 4.0450354294524e-07,
3776
+ "loss": 0.0894,
3777
+ "step": 40192
3778
+ },
3779
+ {
3780
+ "epoch": 1.79,
3781
+ "learning_rate": 4.0355480447093004e-07,
3782
+ "loss": 0.0937,
3783
+ "step": 40256
3784
+ },
3785
+ {
3786
+ "epoch": 1.79,
3787
+ "learning_rate": 4.0260606599662013e-07,
3788
+ "loss": 0.0785,
3789
+ "step": 40320
3790
+ },
3791
+ {
3792
+ "epoch": 1.8,
3793
+ "learning_rate": 4.0165732752231016e-07,
3794
+ "loss": 0.1403,
3795
+ "step": 40384
3796
+ },
3797
+ {
3798
+ "epoch": 1.8,
3799
+ "learning_rate": 4.007085890480002e-07,
3800
+ "loss": 0.0647,
3801
+ "step": 40448
3802
+ },
3803
+ {
3804
+ "epoch": 1.8,
3805
+ "learning_rate": 3.997598505736903e-07,
3806
+ "loss": 0.108,
3807
+ "step": 40512
3808
+ },
3809
+ {
3810
+ "epoch": 1.8,
3811
+ "learning_rate": 3.9881111209938033e-07,
3812
+ "loss": 0.1352,
3813
+ "step": 40576
3814
+ },
3815
+ {
3816
+ "epoch": 1.81,
3817
+ "learning_rate": 3.978623736250704e-07,
3818
+ "loss": 0.1345,
3819
+ "step": 40640
3820
+ },
3821
+ {
3822
+ "epoch": 1.81,
3823
+ "learning_rate": 3.969136351507604e-07,
3824
+ "loss": 0.0638,
3825
+ "step": 40704
3826
+ },
3827
+ {
3828
+ "epoch": 1.81,
3829
+ "learning_rate": 3.959648966764505e-07,
3830
+ "loss": 0.1575,
3831
+ "step": 40768
3832
+ },
3833
+ {
3834
+ "epoch": 1.82,
3835
+ "learning_rate": 3.950161582021406e-07,
3836
+ "loss": 0.0485,
3837
+ "step": 40832
3838
+ },
3839
+ {
3840
+ "epoch": 1.82,
3841
+ "learning_rate": 3.940674197278306e-07,
3842
+ "loss": 0.1627,
3843
+ "step": 40896
3844
+ },
3845
+ {
3846
+ "epoch": 1.82,
3847
+ "learning_rate": 3.931186812535207e-07,
3848
+ "loss": 0.096,
3849
+ "step": 40960
3850
+ },
3851
+ {
3852
+ "epoch": 1.82,
3853
+ "learning_rate": 3.9216994277921075e-07,
3854
+ "loss": 0.1305,
3855
+ "step": 41024
3856
+ },
3857
+ {
3858
+ "epoch": 1.83,
3859
+ "learning_rate": 3.912212043049008e-07,
3860
+ "loss": 0.0839,
3861
+ "step": 41088
3862
+ },
3863
+ {
3864
+ "epoch": 1.83,
3865
+ "learning_rate": 3.902724658305909e-07,
3866
+ "loss": 0.1202,
3867
+ "step": 41152
3868
+ },
3869
+ {
3870
+ "epoch": 1.83,
3871
+ "learning_rate": 3.893237273562809e-07,
3872
+ "loss": 0.1068,
3873
+ "step": 41216
3874
+ },
3875
+ {
3876
+ "epoch": 1.84,
3877
+ "learning_rate": 3.88374988881971e-07,
3878
+ "loss": 0.1165,
3879
+ "step": 41280
3880
+ },
3881
+ {
3882
+ "epoch": 1.84,
3883
+ "learning_rate": 3.8742625040766105e-07,
3884
+ "loss": 0.0656,
3885
+ "step": 41344
3886
+ },
3887
+ {
3888
+ "epoch": 1.84,
3889
+ "learning_rate": 3.864775119333511e-07,
3890
+ "loss": 0.0999,
3891
+ "step": 41408
3892
+ },
3893
+ {
3894
+ "epoch": 1.84,
3895
+ "learning_rate": 3.855287734590412e-07,
3896
+ "loss": 0.1246,
3897
+ "step": 41472
3898
+ },
3899
+ {
3900
+ "epoch": 1.85,
3901
+ "learning_rate": 3.845800349847312e-07,
3902
+ "loss": 0.1357,
3903
+ "step": 41536
3904
+ },
3905
+ {
3906
+ "epoch": 1.85,
3907
+ "learning_rate": 3.836312965104213e-07,
3908
+ "loss": 0.1208,
3909
+ "step": 41600
3910
+ },
3911
+ {
3912
+ "epoch": 1.85,
3913
+ "learning_rate": 3.8268255803611134e-07,
3914
+ "loss": 0.1149,
3915
+ "step": 41664
3916
+ },
3917
+ {
3918
+ "epoch": 1.86,
3919
+ "learning_rate": 3.817338195618014e-07,
3920
+ "loss": 0.0814,
3921
+ "step": 41728
3922
+ },
3923
+ {
3924
+ "epoch": 1.86,
3925
+ "learning_rate": 3.8078508108749147e-07,
3926
+ "loss": 0.1499,
3927
+ "step": 41792
3928
+ },
3929
+ {
3930
+ "epoch": 1.86,
3931
+ "learning_rate": 3.7983634261318156e-07,
3932
+ "loss": 0.088,
3933
+ "step": 41856
3934
+ },
3935
+ {
3936
+ "epoch": 1.86,
3937
+ "learning_rate": 3.7888760413887155e-07,
3938
+ "loss": 0.0644,
3939
+ "step": 41920
3940
+ },
3941
+ {
3942
+ "epoch": 1.87,
3943
+ "learning_rate": 3.7793886566456164e-07,
3944
+ "loss": 0.0924,
3945
+ "step": 41984
3946
+ },
3947
+ {
3948
+ "epoch": 1.87,
3949
+ "learning_rate": 3.769901271902517e-07,
3950
+ "loss": 0.1307,
3951
+ "step": 42048
3952
+ },
3953
+ {
3954
+ "epoch": 1.87,
3955
+ "learning_rate": 3.7604138871594176e-07,
3956
+ "loss": 0.0862,
3957
+ "step": 42112
3958
+ },
3959
+ {
3960
+ "epoch": 1.88,
3961
+ "learning_rate": 3.7509265024163185e-07,
3962
+ "loss": 0.1854,
3963
+ "step": 42176
3964
+ },
3965
+ {
3966
+ "epoch": 1.88,
3967
+ "learning_rate": 3.7414391176732184e-07,
3968
+ "loss": 0.0671,
3969
+ "step": 42240
3970
+ },
3971
+ {
3972
+ "epoch": 1.88,
3973
+ "learning_rate": 3.7319517329301193e-07,
3974
+ "loss": 0.0915,
3975
+ "step": 42304
3976
+ },
3977
+ {
3978
+ "epoch": 1.88,
3979
+ "learning_rate": 3.7224643481870197e-07,
3980
+ "loss": 0.0638,
3981
+ "step": 42368
3982
+ },
3983
+ {
3984
+ "epoch": 1.89,
3985
+ "learning_rate": 3.7129769634439206e-07,
3986
+ "loss": 0.0935,
3987
+ "step": 42432
3988
+ },
3989
+ {
3990
+ "epoch": 1.89,
3991
+ "learning_rate": 3.7034895787008215e-07,
3992
+ "loss": 0.1093,
3993
+ "step": 42496
3994
+ },
3995
+ {
3996
+ "epoch": 1.89,
3997
+ "learning_rate": 3.6940021939577213e-07,
3998
+ "loss": 0.0964,
3999
+ "step": 42560
4000
+ },
4001
+ {
4002
+ "epoch": 1.9,
4003
+ "learning_rate": 3.684514809214622e-07,
4004
+ "loss": 0.1378,
4005
+ "step": 42624
4006
+ },
4007
+ {
4008
+ "epoch": 1.9,
4009
+ "learning_rate": 3.675027424471523e-07,
4010
+ "loss": 0.157,
4011
+ "step": 42688
4012
+ },
4013
+ {
4014
+ "epoch": 1.9,
4015
+ "learning_rate": 3.6655400397284235e-07,
4016
+ "loss": 0.1068,
4017
+ "step": 42752
4018
+ },
4019
+ {
4020
+ "epoch": 1.9,
4021
+ "learning_rate": 3.656052654985324e-07,
4022
+ "loss": 0.1646,
4023
+ "step": 42816
4024
+ },
4025
+ {
4026
+ "epoch": 1.91,
4027
+ "learning_rate": 3.6465652702422243e-07,
4028
+ "loss": 0.0418,
4029
+ "step": 42880
4030
+ },
4031
+ {
4032
+ "epoch": 1.91,
4033
+ "learning_rate": 3.637077885499125e-07,
4034
+ "loss": 0.1197,
4035
+ "step": 42944
4036
+ },
4037
+ {
4038
+ "epoch": 1.91,
4039
+ "learning_rate": 3.627590500756026e-07,
4040
+ "loss": 0.1105,
4041
+ "step": 43008
4042
+ },
4043
+ {
4044
+ "epoch": 1.92,
4045
+ "learning_rate": 3.6181031160129265e-07,
4046
+ "loss": 0.0999,
4047
+ "step": 43072
4048
+ },
4049
+ {
4050
+ "epoch": 1.92,
4051
+ "learning_rate": 3.608615731269827e-07,
4052
+ "loss": 0.1232,
4053
+ "step": 43136
4054
+ },
4055
+ {
4056
+ "epoch": 1.92,
4057
+ "learning_rate": 3.599128346526727e-07,
4058
+ "loss": 0.0982,
4059
+ "step": 43200
4060
+ },
4061
+ {
4062
+ "epoch": 1.92,
4063
+ "learning_rate": 3.589640961783628e-07,
4064
+ "loss": 0.0958,
4065
+ "step": 43264
4066
+ },
4067
+ {
4068
+ "epoch": 1.93,
4069
+ "learning_rate": 3.580153577040529e-07,
4070
+ "loss": 0.1199,
4071
+ "step": 43328
4072
+ },
4073
+ {
4074
+ "epoch": 1.93,
4075
+ "learning_rate": 3.5706661922974294e-07,
4076
+ "loss": 0.1282,
4077
+ "step": 43392
4078
+ },
4079
+ {
4080
+ "epoch": 1.93,
4081
+ "learning_rate": 3.56117880755433e-07,
4082
+ "loss": 0.1198,
4083
+ "step": 43456
4084
+ },
4085
+ {
4086
+ "epoch": 1.94,
4087
+ "learning_rate": 3.5516914228112307e-07,
4088
+ "loss": 0.078,
4089
+ "step": 43520
4090
+ },
4091
+ {
4092
+ "epoch": 1.94,
4093
+ "learning_rate": 3.542204038068131e-07,
4094
+ "loss": 0.1085,
4095
+ "step": 43584
4096
+ },
4097
+ {
4098
+ "epoch": 1.94,
4099
+ "learning_rate": 3.532716653325032e-07,
4100
+ "loss": 0.0971,
4101
+ "step": 43648
4102
+ },
4103
+ {
4104
+ "epoch": 1.94,
4105
+ "learning_rate": 3.523229268581932e-07,
4106
+ "loss": 0.1054,
4107
+ "step": 43712
4108
+ },
4109
+ {
4110
+ "epoch": 1.95,
4111
+ "learning_rate": 3.5137418838388327e-07,
4112
+ "loss": 0.0871,
4113
+ "step": 43776
4114
+ },
4115
+ {
4116
+ "epoch": 1.95,
4117
+ "learning_rate": 3.5042544990957336e-07,
4118
+ "loss": 0.0859,
4119
+ "step": 43840
4120
+ },
4121
+ {
4122
+ "epoch": 1.95,
4123
+ "learning_rate": 3.494767114352634e-07,
4124
+ "loss": 0.1574,
4125
+ "step": 43904
4126
+ },
4127
+ {
4128
+ "epoch": 1.96,
4129
+ "learning_rate": 3.485279729609535e-07,
4130
+ "loss": 0.151,
4131
+ "step": 43968
4132
+ },
4133
+ {
4134
+ "epoch": 1.96,
4135
+ "learning_rate": 3.475792344866435e-07,
4136
+ "loss": 0.0996,
4137
+ "step": 44032
4138
+ },
4139
+ {
4140
+ "epoch": 1.96,
4141
+ "learning_rate": 3.4663049601233357e-07,
4142
+ "loss": 0.1001,
4143
+ "step": 44096
4144
+ },
4145
+ {
4146
+ "epoch": 1.96,
4147
+ "learning_rate": 3.4568175753802366e-07,
4148
+ "loss": 0.0738,
4149
+ "step": 44160
4150
+ },
4151
+ {
4152
+ "epoch": 1.97,
4153
+ "learning_rate": 3.447330190637137e-07,
4154
+ "loss": 0.0682,
4155
+ "step": 44224
4156
+ },
4157
+ {
4158
+ "epoch": 1.97,
4159
+ "learning_rate": 3.437842805894038e-07,
4160
+ "loss": 0.0821,
4161
+ "step": 44288
4162
+ },
4163
+ {
4164
+ "epoch": 1.97,
4165
+ "learning_rate": 3.428355421150938e-07,
4166
+ "loss": 0.0613,
4167
+ "step": 44352
4168
+ },
4169
+ {
4170
+ "epoch": 1.98,
4171
+ "learning_rate": 3.4188680364078386e-07,
4172
+ "loss": 0.137,
4173
+ "step": 44416
4174
+ },
4175
+ {
4176
+ "epoch": 1.98,
4177
+ "learning_rate": 3.4093806516647395e-07,
4178
+ "loss": 0.0937,
4179
+ "step": 44480
4180
+ },
4181
+ {
4182
+ "epoch": 1.98,
4183
+ "learning_rate": 3.39989326692164e-07,
4184
+ "loss": 0.1133,
4185
+ "step": 44544
4186
+ },
4187
+ {
4188
+ "epoch": 1.98,
4189
+ "learning_rate": 3.390405882178541e-07,
4190
+ "loss": 0.1079,
4191
+ "step": 44608
4192
+ },
4193
+ {
4194
+ "epoch": 1.99,
4195
+ "learning_rate": 3.380918497435441e-07,
4196
+ "loss": 0.089,
4197
+ "step": 44672
4198
+ },
4199
+ {
4200
+ "epoch": 1.99,
4201
+ "learning_rate": 3.3714311126923416e-07,
4202
+ "loss": 0.1321,
4203
+ "step": 44736
4204
+ },
4205
+ {
4206
+ "epoch": 1.99,
4207
+ "learning_rate": 3.3619437279492425e-07,
4208
+ "loss": 0.1646,
4209
+ "step": 44800
4210
+ },
4211
+ {
4212
+ "epoch": 2.0,
4213
+ "learning_rate": 3.3524563432061434e-07,
4214
+ "loss": 0.1659,
4215
+ "step": 44864
4216
+ },
4217
+ {
4218
+ "epoch": 2.0,
4219
+ "learning_rate": 3.342968958463043e-07,
4220
+ "loss": 0.084,
4221
+ "step": 44928
4222
+ }
4223
+ ],
4224
+ "logging_steps": 64,
4225
+ "max_steps": 67458,
4226
+ "num_input_tokens_seen": 0,
4227
+ "num_train_epochs": 3,
4228
+ "save_steps": 22486,
4229
+ "total_flos": 9.54832471474176e+19,
4230
+ "train_batch_size": 1,
4231
+ "trial_name": null,
4232
+ "trial_params": null
4233
+ }
checkpoint-44972/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08851645278532aec84b541ed50bff863f0a3babdffb268c3cf70f72a1796795
3
+ size 5048
checkpoint-67458/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-67458/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-67458/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc1c06984de4d009028a10bdd81691a8db2ab89d8a6ec6ea88aff4d34fffb60e
3
+ size 4992706480
checkpoint-67458/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11537408ced3f03759bf55d00a2bbe548e4cb039552400cfa32b889c1b12b865
3
+ size 1180663192
checkpoint-67458/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-67458/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f2756587ea8af78b1148f9174c711411acfa050f2478bca8f52589950ea1fe0
3
+ size 3095076784
checkpoint-67458/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-67458/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f8cfd5367affdbeb9b9192acba07b4dcdbff54676166acf3d02e0e85344337b
3
+ size 14244
checkpoint-67458/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e93c76bd70669666f692355d9f1972d57fe320263e9e603a69fce3da4790e35c
3
+ size 1064
checkpoint-67458/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-67458/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08851645278532aec84b541ed50bff863f0a3babdffb268c3cf70f72a1796795
3
+ size 5048
config.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-large-v2",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 1280,
17
+ "decoder_attention_heads": 20,
18
+ "decoder_ffn_dim": 5120,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 32,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 20,
24
+ "encoder_ffn_dim": 5120,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 32,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": [
29
+ [
30
+ 1,
31
+ 50259
32
+ ],
33
+ [
34
+ 2,
35
+ 50359
36
+ ],
37
+ [
38
+ 3,
39
+ 50363
40
+ ]
41
+ ],
42
+ "init_std": 0.02,
43
+ "is_encoder_decoder": true,
44
+ "language": "English",
45
+ "mask_feature_length": 10,
46
+ "mask_feature_min_masks": 0,
47
+ "mask_feature_prob": 0.0,
48
+ "mask_time_length": 10,
49
+ "mask_time_min_masks": 2,
50
+ "mask_time_prob": 0.05,
51
+ "max_length": 448,
52
+ "max_source_positions": 1500,
53
+ "max_target_positions": 448,
54
+ "median_filter_width": 7,
55
+ "model_type": "whisper",
56
+ "num_hidden_layers": 32,
57
+ "num_mel_bins": 80,
58
+ "pad_token_id": 50257,
59
+ "scale_embedding": false,
60
+ "suppress_tokens": [],
61
+ "torch_dtype": "float32",
62
+ "transformers_version": "4.37.2",
63
+ "use_cache": true,
64
+ "use_weighted_layer_sum": false,
65
+ "vocab_size": 51865
66
+ }
generation_config.json ADDED
@@ -0,0 +1,317 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 10,
5
+ 12
6
+ ],
7
+ [
8
+ 13,
9
+ 17
10
+ ],
11
+ [
12
+ 16,
13
+ 11
14
+ ],
15
+ [
16
+ 16,
17
+ 12
18
+ ],
19
+ [
20
+ 16,
21
+ 13
22
+ ],
23
+ [
24
+ 17,
25
+ 15
26
+ ],
27
+ [
28
+ 17,
29
+ 16
30
+ ],
31
+ [
32
+ 18,
33
+ 4
34
+ ],
35
+ [
36
+ 18,
37
+ 11
38
+ ],
39
+ [
40
+ 18,
41
+ 19
42
+ ],
43
+ [
44
+ 19,
45
+ 11
46
+ ],
47
+ [
48
+ 21,
49
+ 2
50
+ ],
51
+ [
52
+ 21,
53
+ 3
54
+ ],
55
+ [
56
+ 22,
57
+ 3
58
+ ],
59
+ [
60
+ 22,
61
+ 9
62
+ ],
63
+ [
64
+ 22,
65
+ 12
66
+ ],
67
+ [
68
+ 23,
69
+ 5
70
+ ],
71
+ [
72
+ 23,
73
+ 7
74
+ ],
75
+ [
76
+ 23,
77
+ 13
78
+ ],
79
+ [
80
+ 25,
81
+ 5
82
+ ],
83
+ [
84
+ 26,
85
+ 1
86
+ ],
87
+ [
88
+ 26,
89
+ 12
90
+ ],
91
+ [
92
+ 27,
93
+ 15
94
+ ]
95
+ ],
96
+ "begin_suppress_tokens": [
97
+ 220,
98
+ 50257
99
+ ],
100
+ "bos_token_id": 50257,
101
+ "decoder_start_token_id": 50258,
102
+ "eos_token_id": 50257,
103
+ "forced_decoder_ids": [
104
+ [
105
+ 1,
106
+ null
107
+ ],
108
+ [
109
+ 2,
110
+ 50359
111
+ ]
112
+ ],
113
+ "is_multilingual": true,
114
+ "lang_to_id": {
115
+ "<|af|>": 50327,
116
+ "<|am|>": 50334,
117
+ "<|ar|>": 50272,
118
+ "<|as|>": 50350,
119
+ "<|az|>": 50304,
120
+ "<|ba|>": 50355,
121
+ "<|be|>": 50330,
122
+ "<|bg|>": 50292,
123
+ "<|bn|>": 50302,
124
+ "<|bo|>": 50347,
125
+ "<|br|>": 50309,
126
+ "<|bs|>": 50315,
127
+ "<|ca|>": 50270,
128
+ "<|cs|>": 50283,
129
+ "<|cy|>": 50297,
130
+ "<|da|>": 50285,
131
+ "<|de|>": 50261,
132
+ "<|el|>": 50281,
133
+ "<|en|>": 50259,
134
+ "<|es|>": 50262,
135
+ "<|et|>": 50307,
136
+ "<|eu|>": 50310,
137
+ "<|fa|>": 50300,
138
+ "<|fi|>": 50277,
139
+ "<|fo|>": 50338,
140
+ "<|fr|>": 50265,
141
+ "<|gl|>": 50319,
142
+ "<|gu|>": 50333,
143
+ "<|haw|>": 50352,
144
+ "<|ha|>": 50354,
145
+ "<|he|>": 50279,
146
+ "<|hi|>": 50276,
147
+ "<|hr|>": 50291,
148
+ "<|ht|>": 50339,
149
+ "<|hu|>": 50286,
150
+ "<|hy|>": 50312,
151
+ "<|id|>": 50275,
152
+ "<|is|>": 50311,
153
+ "<|it|>": 50274,
154
+ "<|ja|>": 50266,
155
+ "<|jw|>": 50356,
156
+ "<|ka|>": 50329,
157
+ "<|kk|>": 50316,
158
+ "<|km|>": 50323,
159
+ "<|kn|>": 50306,
160
+ "<|ko|>": 50264,
161
+ "<|la|>": 50294,
162
+ "<|lb|>": 50345,
163
+ "<|ln|>": 50353,
164
+ "<|lo|>": 50336,
165
+ "<|lt|>": 50293,
166
+ "<|lv|>": 50301,
167
+ "<|mg|>": 50349,
168
+ "<|mi|>": 50295,
169
+ "<|mk|>": 50308,
170
+ "<|ml|>": 50296,
171
+ "<|mn|>": 50314,
172
+ "<|mr|>": 50320,
173
+ "<|ms|>": 50282,
174
+ "<|mt|>": 50343,
175
+ "<|my|>": 50346,
176
+ "<|ne|>": 50313,
177
+ "<|nl|>": 50271,
178
+ "<|nn|>": 50342,
179
+ "<|no|>": 50288,
180
+ "<|oc|>": 50328,
181
+ "<|pa|>": 50321,
182
+ "<|pl|>": 50269,
183
+ "<|ps|>": 50340,
184
+ "<|pt|>": 50267,
185
+ "<|ro|>": 50284,
186
+ "<|ru|>": 50263,
187
+ "<|sa|>": 50344,
188
+ "<|sd|>": 50332,
189
+ "<|si|>": 50322,
190
+ "<|sk|>": 50298,
191
+ "<|sl|>": 50305,
192
+ "<|sn|>": 50324,
193
+ "<|so|>": 50326,
194
+ "<|sq|>": 50317,
195
+ "<|sr|>": 50303,
196
+ "<|su|>": 50357,
197
+ "<|sv|>": 50273,
198
+ "<|sw|>": 50318,
199
+ "<|ta|>": 50287,
200
+ "<|te|>": 50299,
201
+ "<|tg|>": 50331,
202
+ "<|th|>": 50289,
203
+ "<|tk|>": 50341,
204
+ "<|tl|>": 50348,
205
+ "<|tr|>": 50268,
206
+ "<|tt|>": 50351,
207
+ "<|uk|>": 50280,
208
+ "<|ur|>": 50290,
209
+ "<|uz|>": 50337,
210
+ "<|vi|>": 50278,
211
+ "<|yi|>": 50335,
212
+ "<|yo|>": 50325,
213
+ "<|zh|>": 50260
214
+ },
215
+ "max_initial_timestamp_index": 50,
216
+ "max_length": 448,
217
+ "no_timestamps_token_id": 50363,
218
+ "pad_token_id": 50257,
219
+ "prev_sot_token_id": 50361,
220
+ "return_timestamps": false,
221
+ "suppress_tokens": [
222
+ 1,
223
+ 2,
224
+ 7,
225
+ 8,
226
+ 9,
227
+ 10,
228
+ 14,
229
+ 25,
230
+ 26,
231
+ 27,
232
+ 28,
233
+ 29,
234
+ 31,
235
+ 58,
236
+ 59,
237
+ 60,
238
+ 61,
239
+ 62,
240
+ 63,
241
+ 90,
242
+ 91,
243
+ 92,
244
+ 93,
245
+ 359,
246
+ 503,
247
+ 522,
248
+ 542,
249
+ 873,
250
+ 893,
251
+ 902,
252
+ 918,
253
+ 922,
254
+ 931,
255
+ 1350,
256
+ 1853,
257
+ 1982,
258
+ 2460,
259
+ 2627,
260
+ 3246,
261
+ 3253,
262
+ 3268,
263
+ 3536,
264
+ 3846,
265
+ 3961,
266
+ 4183,
267
+ 4667,
268
+ 6585,
269
+ 6647,
270
+ 7273,
271
+ 9061,
272
+ 9383,
273
+ 10428,
274
+ 10929,
275
+ 11938,
276
+ 12033,
277
+ 12331,
278
+ 12562,
279
+ 13793,
280
+ 14157,
281
+ 14635,
282
+ 15265,
283
+ 15618,
284
+ 16553,
285
+ 16604,
286
+ 18362,
287
+ 18956,
288
+ 20075,
289
+ 21675,
290
+ 22520,
291
+ 26130,
292
+ 26161,
293
+ 26435,
294
+ 28279,
295
+ 29464,
296
+ 31650,
297
+ 32302,
298
+ 32470,
299
+ 36865,
300
+ 42863,
301
+ 47425,
302
+ 49870,
303
+ 50254,
304
+ 50258,
305
+ 50358,
306
+ 50359,
307
+ 50360,
308
+ 50361,
309
+ 50362
310
+ ],
311
+ "task": "transcribe",
312
+ "task_to_id": {
313
+ "transcribe": 50359,
314
+ "translate": 50358
315
+ },
316
+ "transformers_version": "4.37.2"
317
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc1c06984de4d009028a10bdd81691a8db2ab89d8a6ec6ea88aff4d34fffb60e
3
+ size 4992706480
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11537408ced3f03759bf55d00a2bbe548e4cb039552400cfa32b889c1b12b865
3
+ size 1180663192
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08851645278532aec84b541ed50bff863f0a3babdffb268c3cf70f72a1796795
3
+ size 5048