Hugging Face's logo

lw2333
/
hinghwa-with-ipa-tok

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
whisper
Inference Endpoints
Model card Files Metrics Community
lw2333 commited on
Commit
563379f
•
1 Parent(s): c9f4a2d

first commit

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. checkpoint-1000/config.json +52 -0
  2. checkpoint-1000/generation_config.json +263 -0
  3. checkpoint-1000/optimizer.pt +3 -0
  4. checkpoint-1000/preprocessor_config.json +14 -0
  5. checkpoint-1000/pytorch_model.bin +3 -0
  6. checkpoint-1000/rng_state.pth +3 -0
  7. checkpoint-1000/scheduler.pt +3 -0
  8. checkpoint-1000/trainer_state.json +265 -0
  9. checkpoint-1000/training_args.bin +3 -0
  10. checkpoint-2000/config.json +52 -0
  11. checkpoint-2000/generation_config.json +263 -0
  12. checkpoint-2000/optimizer.pt +3 -0
  13. checkpoint-2000/preprocessor_config.json +14 -0
  14. checkpoint-2000/pytorch_model.bin +3 -0
  15. checkpoint-2000/rng_state.pth +3 -0
  16. checkpoint-2000/scheduler.pt +3 -0
  17. checkpoint-2000/trainer_state.json +514 -0
  18. checkpoint-2000/training_args.bin +3 -0
  19. checkpoint-3000/config.json +52 -0
  20. checkpoint-3000/generation_config.json +263 -0
  21. checkpoint-3000/optimizer.pt +3 -0
  22. checkpoint-3000/preprocessor_config.json +14 -0
  23. checkpoint-3000/pytorch_model.bin +3 -0
  24. checkpoint-3000/rng_state.pth +3 -0
  25. checkpoint-3000/scheduler.pt +3 -0
  26. checkpoint-3000/trainer_state.json +763 -0
  27. checkpoint-3000/training_args.bin +3 -0
  28. checkpoint-4000/config.json +52 -0
  29. checkpoint-4000/generation_config.json +263 -0
  30. checkpoint-4000/optimizer.pt +3 -0
  31. checkpoint-4000/preprocessor_config.json +14 -0
  32. checkpoint-4000/pytorch_model.bin +3 -0
  33. checkpoint-4000/rng_state.pth +3 -0
  34. checkpoint-4000/scheduler.pt +3 -0
  35. checkpoint-4000/trainer_state.json +1012 -0
  36. checkpoint-4000/training_args.bin +3 -0
  37. checkpoint-5000/config.json +52 -0
  38. checkpoint-5000/generation_config.json +263 -0
  39. checkpoint-5000/optimizer.pt +3 -0
  40. checkpoint-5000/preprocessor_config.json +14 -0
  41. checkpoint-5000/pytorch_model.bin +3 -0
  42. checkpoint-5000/rng_state.pth +3 -0
  43. checkpoint-5000/scheduler.pt +3 -0
  44. checkpoint-5000/trainer_state.json +1261 -0
  45. checkpoint-5000/training_args.bin +3 -0
  46. checkpoint-6000/config.json +52 -0
  47. checkpoint-6000/generation_config.json +263 -0
  48. checkpoint-6000/optimizer.pt +3 -0
  49. checkpoint-6000/preprocessor_config.json +14 -0
  50. checkpoint-6000/pytorch_model.bin +3 -0
checkpoint-1000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
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": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.31.0",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,263 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "max_initial_timestamp_index": 1,
164
+ "max_length": 448,
165
+ "no_timestamps_token_id": 50363,
166
+ "pad_token_id": 50257,
167
+ "return_timestamps": false,
168
+ "suppress_tokens": [
169
+ 1,
170
+ 2,
171
+ 7,
172
+ 8,
173
+ 9,
174
+ 10,
175
+ 14,
176
+ 25,
177
+ 26,
178
+ 27,
179
+ 28,
180
+ 29,
181
+ 31,
182
+ 58,
183
+ 59,
184
+ 60,
185
+ 61,
186
+ 62,
187
+ 63,
188
+ 90,
189
+ 91,
190
+ 92,
191
+ 93,
192
+ 359,
193
+ 503,
194
+ 522,
195
+ 542,
196
+ 873,
197
+ 893,
198
+ 902,
199
+ 918,
200
+ 922,
201
+ 931,
202
+ 1350,
203
+ 1853,
204
+ 1982,
205
+ 2460,
206
+ 2627,
207
+ 3246,
208
+ 3253,
209
+ 3268,
210
+ 3536,
211
+ 3846,
212
+ 3961,
213
+ 4183,
214
+ 4667,
215
+ 6585,
216
+ 6647,
217
+ 7273,
218
+ 9061,
219
+ 9383,
220
+ 10428,
221
+ 10929,
222
+ 11938,
223
+ 12033,
224
+ 12331,
225
+ 12562,
226
+ 13793,
227
+ 14157,
228
+ 14635,
229
+ 15265,
230
+ 15618,
231
+ 16553,
232
+ 16604,
233
+ 18362,
234
+ 18956,
235
+ 20075,
236
+ 21675,
237
+ 22520,
238
+ 26130,
239
+ 26161,
240
+ 26435,
241
+ 28279,
242
+ 29464,
243
+ 31650,
244
+ 32302,
245
+ 32470,
246
+ 36865,
247
+ 42863,
248
+ 47425,
249
+ 49870,
250
+ 50254,
251
+ 50258,
252
+ 50358,
253
+ 50359,
254
+ 50360,
255
+ 50361,
256
+ 50362
257
+ ],
258
+ "task_to_id": {
259
+ "transcribe": 50359,
260
+ "translate": 50358
261
+ },
262
+ "transformers_version": "4.31.0"
263
+ }
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cb607022ed6f7e1f03731d0dd3f5e027fa8728c3d6cd82bdb5121dcd6cd7604
3
+ size 1934161157
checkpoint-1000/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-1000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b5c6a3adf034ae1a496075d77c5ceed88602fd608f8a8b47573a83c2589e710
3
+ size 967102729
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:543cf3abb35dfcf7e085bfee9f5238988edd0801049824c3ab6aa7457c7230af
3
+ size 14575
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84d8d4a47c4440e4292f72bacf04b03bc7fbf8c721c0e8b2d4fe3f3f171d575a
3
+ size 627
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,265 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 5215.41095890411,
3
+ "best_model_checkpoint": "./20230912-output\\checkpoint-1000",
4
+ "epoch": 8.390141583639224,
5
+ "global_step": 1000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.21,
12
+ "learning_rate": 3.8e-07,
13
+ "loss": 18.9312,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.42,
18
+ "learning_rate": 8.8e-07,
19
+ "loss": 17.0566,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.63,
24
+ "learning_rate": 1.3800000000000001e-06,
25
+ "loss": 13.4189,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.84,
30
+ "learning_rate": 1.8800000000000002e-06,
31
+ "loss": 10.4803,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 1.05,
36
+ "learning_rate": 2.38e-06,
37
+ "loss": 9.3304,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 1.26,
42
+ "learning_rate": 2.88e-06,
43
+ "loss": 8.3203,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 1.47,
48
+ "learning_rate": 3.3600000000000004e-06,
49
+ "loss": 8.9822,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 1.68,
54
+ "learning_rate": 3.86e-06,
55
+ "loss": 10.7581,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 1.89,
60
+ "learning_rate": 4.360000000000001e-06,
61
+ "loss": 13.2164,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 2.1,
66
+ "learning_rate": 4.86e-06,
67
+ "loss": 12.7052,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 2.31,
72
+ "learning_rate": 5.36e-06,
73
+ "loss": 10.3904,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 2.52,
78
+ "learning_rate": 5.86e-06,
79
+ "loss": 8.2617,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 2.73,
84
+ "learning_rate": 6.360000000000001e-06,
85
+ "loss": 6.8613,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 2.94,
90
+ "learning_rate": 6.860000000000001e-06,
91
+ "loss": 6.1093,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 3.15,
96
+ "learning_rate": 7.360000000000001e-06,
97
+ "loss": 5.8769,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 3.36,
102
+ "learning_rate": 7.860000000000001e-06,
103
+ "loss": 5.6272,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 3.57,
108
+ "learning_rate": 8.36e-06,
109
+ "loss": 5.674,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 3.78,
114
+ "learning_rate": 8.860000000000002e-06,
115
+ "loss": 5.5667,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 3.99,
120
+ "learning_rate": 9.360000000000002e-06,
121
+ "loss": 5.4705,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 4.2,
126
+ "learning_rate": 9.86e-06,
127
+ "loss": 5.1703,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 4.4,
132
+ "learning_rate": 9.967272727272728e-06,
133
+ "loss": 5.1044,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 4.61,
138
+ "learning_rate": 9.921818181818183e-06,
139
+ "loss": 4.9244,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 4.82,
144
+ "learning_rate": 9.876363636363637e-06,
145
+ "loss": 4.8602,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 5.03,
150
+ "learning_rate": 9.830909090909092e-06,
151
+ "loss": 4.6437,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 5.24,
156
+ "learning_rate": 9.785454545454546e-06,
157
+ "loss": 4.0014,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 5.45,
162
+ "learning_rate": 9.74e-06,
163
+ "loss": 3.8653,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 5.66,
168
+ "learning_rate": 9.694545454545455e-06,
169
+ "loss": 3.8299,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 5.87,
174
+ "learning_rate": 9.64909090909091e-06,
175
+ "loss": 3.6928,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 6.08,
180
+ "learning_rate": 9.603636363636365e-06,
181
+ "loss": 3.1672,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 6.29,
186
+ "learning_rate": 9.558181818181818e-06,
187
+ "loss": 2.5309,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 6.5,
192
+ "learning_rate": 9.512727272727274e-06,
193
+ "loss": 2.4801,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 6.71,
198
+ "learning_rate": 9.467272727272727e-06,
199
+ "loss": 2.2236,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 6.92,
204
+ "learning_rate": 9.421818181818183e-06,
205
+ "loss": 2.3091,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 7.13,
210
+ "learning_rate": 9.376363636363637e-06,
211
+ "loss": 1.9126,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 7.34,
216
+ "learning_rate": 9.330909090909092e-06,
217
+ "loss": 1.4071,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 7.55,
222
+ "learning_rate": 9.285454545454546e-06,
223
+ "loss": 1.5367,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 7.76,
228
+ "learning_rate": 9.240000000000001e-06,
229
+ "loss": 1.3686,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 7.97,
234
+ "learning_rate": 9.194545454545455e-06,
235
+ "loss": 1.3177,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 8.18,
240
+ "learning_rate": 9.14909090909091e-06,
241
+ "loss": 0.9167,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 8.39,
246
+ "learning_rate": 9.103636363636364e-06,
247
+ "loss": 0.9502,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 8.39,
252
+ "eval_loss": 3.2547426223754883,
253
+ "eval_runtime": 507.4588,
254
+ "eval_samples_per_second": 1.255,
255
+ "eval_steps_per_second": 0.158,
256
+ "eval_wer": 5215.41095890411,
257
+ "step": 1000
258
+ }
259
+ ],
260
+ "max_steps": 6000,
261
+ "num_train_epochs": 51,
262
+ "total_flos": 4.62371528392704e+18,
263
+ "trial_name": null,
264
+ "trial_params": null
265
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cbeda0929f5bade1f2682f85820a8eb42b209173de61e856a1e94b350f2c7d1
3
+ size 4091
checkpoint-2000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
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": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.31.0",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-2000/generation_config.json ADDED
@@ -0,0 +1,263 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "max_initial_timestamp_index": 1,
164
+ "max_length": 448,
165
+ "no_timestamps_token_id": 50363,
166
+ "pad_token_id": 50257,
167
+ "return_timestamps": false,
168
+ "suppress_tokens": [
169
+ 1,
170
+ 2,
171
+ 7,
172
+ 8,
173
+ 9,
174
+ 10,
175
+ 14,
176
+ 25,
177
+ 26,
178
+ 27,
179
+ 28,
180
+ 29,
181
+ 31,
182
+ 58,
183
+ 59,
184
+ 60,
185
+ 61,
186
+ 62,
187
+ 63,
188
+ 90,
189
+ 91,
190
+ 92,
191
+ 93,
192
+ 359,
193
+ 503,
194
+ 522,
195
+ 542,
196
+ 873,
197
+ 893,
198
+ 902,
199
+ 918,
200
+ 922,
201
+ 931,
202
+ 1350,
203
+ 1853,
204
+ 1982,
205
+ 2460,
206
+ 2627,
207
+ 3246,
208
+ 3253,
209
+ 3268,
210
+ 3536,
211
+ 3846,
212
+ 3961,
213
+ 4183,
214
+ 4667,
215
+ 6585,
216
+ 6647,
217
+ 7273,
218
+ 9061,
219
+ 9383,
220
+ 10428,
221
+ 10929,
222
+ 11938,
223
+ 12033,
224
+ 12331,
225
+ 12562,
226
+ 13793,
227
+ 14157,
228
+ 14635,
229
+ 15265,
230
+ 15618,
231
+ 16553,
232
+ 16604,
233
+ 18362,
234
+ 18956,
235
+ 20075,
236
+ 21675,
237
+ 22520,
238
+ 26130,
239
+ 26161,
240
+ 26435,
241
+ 28279,
242
+ 29464,
243
+ 31650,
244
+ 32302,
245
+ 32470,
246
+ 36865,
247
+ 42863,
248
+ 47425,
249
+ 49870,
250
+ 50254,
251
+ 50258,
252
+ 50358,
253
+ 50359,
254
+ 50360,
255
+ 50361,
256
+ 50362
257
+ ],
258
+ "task_to_id": {
259
+ "transcribe": 50359,
260
+ "translate": 50358
261
+ },
262
+ "transformers_version": "4.31.0"
263
+ }
checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2cbbd7fcccd547d5b84855f322d2c1c010e0d684038428548c22e4e0b3a7709f
3
+ size 1934161157
checkpoint-2000/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-2000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0337cc6b297e394420535da5f952ed3c6704cd5eb53a31ed0d7aaf5aa12b1188
3
+ size 967102729
checkpoint-2000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc6b92840554bcef4ed4331b6d0b4b41eb6f43eccb0e7ba62cb89b7c408357dc
3
+ size 14575
checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b594c2ba696182b1ac2e0bd4ab3823913909439464dda6da5f57afa5d5ac4ab1
3
+ size 627
checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,514 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 5006.62100456621,
3
+ "best_model_checkpoint": "./20230912-output\\checkpoint-2000",
4
+ "epoch": 16.780283167278448,
5
+ "global_step": 2000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.21,
12
+ "learning_rate": 3.8e-07,
13
+ "loss": 18.9312,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.42,
18
+ "learning_rate": 8.8e-07,
19
+ "loss": 17.0566,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.63,
24
+ "learning_rate": 1.3800000000000001e-06,
25
+ "loss": 13.4189,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.84,
30
+ "learning_rate": 1.8800000000000002e-06,
31
+ "loss": 10.4803,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 1.05,
36
+ "learning_rate": 2.38e-06,
37
+ "loss": 9.3304,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 1.26,
42
+ "learning_rate": 2.88e-06,
43
+ "loss": 8.3203,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 1.47,
48
+ "learning_rate": 3.3600000000000004e-06,
49
+ "loss": 8.9822,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 1.68,
54
+ "learning_rate": 3.86e-06,
55
+ "loss": 10.7581,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 1.89,
60
+ "learning_rate": 4.360000000000001e-06,
61
+ "loss": 13.2164,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 2.1,
66
+ "learning_rate": 4.86e-06,
67
+ "loss": 12.7052,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 2.31,
72
+ "learning_rate": 5.36e-06,
73
+ "loss": 10.3904,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 2.52,
78
+ "learning_rate": 5.86e-06,
79
+ "loss": 8.2617,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 2.73,
84
+ "learning_rate": 6.360000000000001e-06,
85
+ "loss": 6.8613,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 2.94,
90
+ "learning_rate": 6.860000000000001e-06,
91
+ "loss": 6.1093,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 3.15,
96
+ "learning_rate": 7.360000000000001e-06,
97
+ "loss": 5.8769,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 3.36,
102
+ "learning_rate": 7.860000000000001e-06,
103
+ "loss": 5.6272,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 3.57,
108
+ "learning_rate": 8.36e-06,
109
+ "loss": 5.674,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 3.78,
114
+ "learning_rate": 8.860000000000002e-06,
115
+ "loss": 5.5667,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 3.99,
120
+ "learning_rate": 9.360000000000002e-06,
121
+ "loss": 5.4705,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 4.2,
126
+ "learning_rate": 9.86e-06,
127
+ "loss": 5.1703,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 4.4,
132
+ "learning_rate": 9.967272727272728e-06,
133
+ "loss": 5.1044,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 4.61,
138
+ "learning_rate": 9.921818181818183e-06,
139
+ "loss": 4.9244,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 4.82,
144
+ "learning_rate": 9.876363636363637e-06,
145
+ "loss": 4.8602,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 5.03,
150
+ "learning_rate": 9.830909090909092e-06,
151
+ "loss": 4.6437,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 5.24,
156
+ "learning_rate": 9.785454545454546e-06,
157
+ "loss": 4.0014,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 5.45,
162
+ "learning_rate": 9.74e-06,
163
+ "loss": 3.8653,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 5.66,
168
+ "learning_rate": 9.694545454545455e-06,
169
+ "loss": 3.8299,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 5.87,
174
+ "learning_rate": 9.64909090909091e-06,
175
+ "loss": 3.6928,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 6.08,
180
+ "learning_rate": 9.603636363636365e-06,
181
+ "loss": 3.1672,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 6.29,
186
+ "learning_rate": 9.558181818181818e-06,
187
+ "loss": 2.5309,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 6.5,
192
+ "learning_rate": 9.512727272727274e-06,
193
+ "loss": 2.4801,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 6.71,
198
+ "learning_rate": 9.467272727272727e-06,
199
+ "loss": 2.2236,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 6.92,
204
+ "learning_rate": 9.421818181818183e-06,
205
+ "loss": 2.3091,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 7.13,
210
+ "learning_rate": 9.376363636363637e-06,
211
+ "loss": 1.9126,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 7.34,
216
+ "learning_rate": 9.330909090909092e-06,
217
+ "loss": 1.4071,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 7.55,
222
+ "learning_rate": 9.285454545454546e-06,
223
+ "loss": 1.5367,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 7.76,
228
+ "learning_rate": 9.240000000000001e-06,
229
+ "loss": 1.3686,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 7.97,
234
+ "learning_rate": 9.194545454545455e-06,
235
+ "loss": 1.3177,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 8.18,
240
+ "learning_rate": 9.14909090909091e-06,
241
+ "loss": 0.9167,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 8.39,
246
+ "learning_rate": 9.103636363636364e-06,
247
+ "loss": 0.9502,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 8.39,
252
+ "eval_loss": 3.2547426223754883,
253
+ "eval_runtime": 507.4588,
254
+ "eval_samples_per_second": 1.255,
255
+ "eval_steps_per_second": 0.158,
256
+ "eval_wer": 5215.41095890411,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 8.6,
261
+ "learning_rate": 9.05818181818182e-06,
262
+ "loss": 0.774,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 8.81,
267
+ "learning_rate": 9.012727272727273e-06,
268
+ "loss": 0.8792,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 9.02,
273
+ "learning_rate": 8.967272727272727e-06,
274
+ "loss": 0.8182,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 9.23,
279
+ "learning_rate": 8.921818181818183e-06,
280
+ "loss": 0.5773,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 9.44,
285
+ "learning_rate": 8.876363636363638e-06,
286
+ "loss": 0.5618,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 9.65,
291
+ "learning_rate": 8.830909090909092e-06,
292
+ "loss": 0.5499,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 9.86,
297
+ "learning_rate": 8.785454545454546e-06,
298
+ "loss": 0.5145,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 10.07,
303
+ "learning_rate": 8.740000000000001e-06,
304
+ "loss": 0.4436,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 10.28,
309
+ "learning_rate": 8.694545454545455e-06,
310
+ "loss": 0.3293,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 10.49,
315
+ "learning_rate": 8.64909090909091e-06,
316
+ "loss": 0.3603,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 10.7,
321
+ "learning_rate": 8.603636363636364e-06,
322
+ "loss": 0.3796,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 10.91,
327
+ "learning_rate": 8.55818181818182e-06,
328
+ "loss": 0.347,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 11.12,
333
+ "learning_rate": 8.512727272727273e-06,
334
+ "loss": 0.263,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 11.33,
339
+ "learning_rate": 8.467272727272729e-06,
340
+ "loss": 0.272,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 11.54,
345
+ "learning_rate": 8.421818181818182e-06,
346
+ "loss": 0.2657,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 11.75,
351
+ "learning_rate": 8.376363636363636e-06,
352
+ "loss": 0.2299,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 11.96,
357
+ "learning_rate": 8.330909090909092e-06,
358
+ "loss": 0.2563,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 12.17,
363
+ "learning_rate": 8.285454545454547e-06,
364
+ "loss": 0.1685,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 12.38,
369
+ "learning_rate": 8.24e-06,
370
+ "loss": 0.2457,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 12.59,
375
+ "learning_rate": 8.194545454545454e-06,
376
+ "loss": 0.1859,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 12.79,
381
+ "learning_rate": 8.14909090909091e-06,
382
+ "loss": 0.1883,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 13.0,
387
+ "learning_rate": 8.103636363636365e-06,
388
+ "loss": 0.1484,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 13.21,
393
+ "learning_rate": 8.058181818181819e-06,
394
+ "loss": 0.1322,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 13.42,
399
+ "learning_rate": 8.012727272727273e-06,
400
+ "loss": 0.1099,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 13.63,
405
+ "learning_rate": 7.967272727272728e-06,
406
+ "loss": 0.1455,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 13.84,
411
+ "learning_rate": 7.921818181818182e-06,
412
+ "loss": 0.1534,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 14.05,
417
+ "learning_rate": 7.876363636363637e-06,
418
+ "loss": 0.1459,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 14.26,
423
+ "learning_rate": 7.830909090909091e-06,
424
+ "loss": 0.1286,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 14.47,
429
+ "learning_rate": 7.785454545454547e-06,
430
+ "loss": 0.0906,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 14.68,
435
+ "learning_rate": 7.74e-06,
436
+ "loss": 0.1186,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 14.89,
441
+ "learning_rate": 7.694545454545456e-06,
442
+ "loss": 0.0855,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 15.1,
447
+ "learning_rate": 7.64909090909091e-06,
448
+ "loss": 0.1135,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 15.31,
453
+ "learning_rate": 7.603636363636364e-06,
454
+ "loss": 0.0613,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 15.52,
459
+ "learning_rate": 7.558181818181819e-06,
460
+ "loss": 0.0682,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 15.73,
465
+ "learning_rate": 7.512727272727273e-06,
466
+ "loss": 0.0667,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 15.94,
471
+ "learning_rate": 7.467272727272727e-06,
472
+ "loss": 0.0815,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 16.15,
477
+ "learning_rate": 7.421818181818183e-06,
478
+ "loss": 0.0599,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 16.36,
483
+ "learning_rate": 7.376363636363637e-06,
484
+ "loss": 0.0639,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 16.57,
489
+ "learning_rate": 7.330909090909092e-06,
490
+ "loss": 0.0578,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 16.78,
495
+ "learning_rate": 7.2854545454545455e-06,
496
+ "loss": 0.0448,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 16.78,
501
+ "eval_loss": 2.7543158531188965,
502
+ "eval_runtime": 502.2301,
503
+ "eval_samples_per_second": 1.268,
504
+ "eval_steps_per_second": 0.159,
505
+ "eval_wer": 5006.62100456621,
506
+ "step": 2000
507
+ }
508
+ ],
509
+ "max_steps": 6000,
510
+ "num_train_epochs": 51,
511
+ "total_flos": 9.24108168904704e+18,
512
+ "trial_name": null,
513
+ "trial_params": null
514
+ }
checkpoint-2000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cbeda0929f5bade1f2682f85820a8eb42b209173de61e856a1e94b350f2c7d1
3
+ size 4091
checkpoint-3000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
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": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.31.0",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-3000/generation_config.json ADDED
@@ -0,0 +1,263 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "max_initial_timestamp_index": 1,
164
+ "max_length": 448,
165
+ "no_timestamps_token_id": 50363,
166
+ "pad_token_id": 50257,
167
+ "return_timestamps": false,
168
+ "suppress_tokens": [
169
+ 1,
170
+ 2,
171
+ 7,
172
+ 8,
173
+ 9,
174
+ 10,
175
+ 14,
176
+ 25,
177
+ 26,
178
+ 27,
179
+ 28,
180
+ 29,
181
+ 31,
182
+ 58,
183
+ 59,
184
+ 60,
185
+ 61,
186
+ 62,
187
+ 63,
188
+ 90,
189
+ 91,
190
+ 92,
191
+ 93,
192
+ 359,
193
+ 503,
194
+ 522,
195
+ 542,
196
+ 873,
197
+ 893,
198
+ 902,
199
+ 918,
200
+ 922,
201
+ 931,
202
+ 1350,
203
+ 1853,
204
+ 1982,
205
+ 2460,
206
+ 2627,
207
+ 3246,
208
+ 3253,
209
+ 3268,
210
+ 3536,
211
+ 3846,
212
+ 3961,
213
+ 4183,
214
+ 4667,
215
+ 6585,
216
+ 6647,
217
+ 7273,
218
+ 9061,
219
+ 9383,
220
+ 10428,
221
+ 10929,
222
+ 11938,
223
+ 12033,
224
+ 12331,
225
+ 12562,
226
+ 13793,
227
+ 14157,
228
+ 14635,
229
+ 15265,
230
+ 15618,
231
+ 16553,
232
+ 16604,
233
+ 18362,
234
+ 18956,
235
+ 20075,
236
+ 21675,
237
+ 22520,
238
+ 26130,
239
+ 26161,
240
+ 26435,
241
+ 28279,
242
+ 29464,
243
+ 31650,
244
+ 32302,
245
+ 32470,
246
+ 36865,
247
+ 42863,
248
+ 47425,
249
+ 49870,
250
+ 50254,
251
+ 50258,
252
+ 50358,
253
+ 50359,
254
+ 50360,
255
+ 50361,
256
+ 50362
257
+ ],
258
+ "task_to_id": {
259
+ "transcribe": 50359,
260
+ "translate": 50358
261
+ },
262
+ "transformers_version": "4.31.0"
263
+ }
checkpoint-3000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a358416bc1c2b511ab99649add03ad63dd010215e70d12e798902b6500325b6c
3
+ size 1934161157
checkpoint-3000/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-3000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:557a5f566735537b3074e1e383be6c981d3f46afabef156ab85196a9b9a74c23
3
+ size 967102729
checkpoint-3000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3f1563c2c00cd26cfa5256937d2a2fbdd78b82707cd2486d996170731a0820c1
3
+ size 14575
checkpoint-3000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d878d446467728ac61086a89f57f53a85bc0770360a55d6fd0e018fba58c339
3
+ size 627
checkpoint-3000/trainer_state.json ADDED
@@ -0,0 +1,763 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 5006.62100456621,
3
+ "best_model_checkpoint": "./20230912-output\\checkpoint-2000",
4
+ "epoch": 25.17042475091767,
5
+ "global_step": 3000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.21,
12
+ "learning_rate": 3.8e-07,
13
+ "loss": 18.9312,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.42,
18
+ "learning_rate": 8.8e-07,
19
+ "loss": 17.0566,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.63,
24
+ "learning_rate": 1.3800000000000001e-06,
25
+ "loss": 13.4189,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.84,
30
+ "learning_rate": 1.8800000000000002e-06,
31
+ "loss": 10.4803,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 1.05,
36
+ "learning_rate": 2.38e-06,
37
+ "loss": 9.3304,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 1.26,
42
+ "learning_rate": 2.88e-06,
43
+ "loss": 8.3203,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 1.47,
48
+ "learning_rate": 3.3600000000000004e-06,
49
+ "loss": 8.9822,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 1.68,
54
+ "learning_rate": 3.86e-06,
55
+ "loss": 10.7581,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 1.89,
60
+ "learning_rate": 4.360000000000001e-06,
61
+ "loss": 13.2164,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 2.1,
66
+ "learning_rate": 4.86e-06,
67
+ "loss": 12.7052,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 2.31,
72
+ "learning_rate": 5.36e-06,
73
+ "loss": 10.3904,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 2.52,
78
+ "learning_rate": 5.86e-06,
79
+ "loss": 8.2617,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 2.73,
84
+ "learning_rate": 6.360000000000001e-06,
85
+ "loss": 6.8613,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 2.94,
90
+ "learning_rate": 6.860000000000001e-06,
91
+ "loss": 6.1093,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 3.15,
96
+ "learning_rate": 7.360000000000001e-06,
97
+ "loss": 5.8769,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 3.36,
102
+ "learning_rate": 7.860000000000001e-06,
103
+ "loss": 5.6272,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 3.57,
108
+ "learning_rate": 8.36e-06,
109
+ "loss": 5.674,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 3.78,
114
+ "learning_rate": 8.860000000000002e-06,
115
+ "loss": 5.5667,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 3.99,
120
+ "learning_rate": 9.360000000000002e-06,
121
+ "loss": 5.4705,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 4.2,
126
+ "learning_rate": 9.86e-06,
127
+ "loss": 5.1703,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 4.4,
132
+ "learning_rate": 9.967272727272728e-06,
133
+ "loss": 5.1044,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 4.61,
138
+ "learning_rate": 9.921818181818183e-06,
139
+ "loss": 4.9244,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 4.82,
144
+ "learning_rate": 9.876363636363637e-06,
145
+ "loss": 4.8602,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 5.03,
150
+ "learning_rate": 9.830909090909092e-06,
151
+ "loss": 4.6437,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 5.24,
156
+ "learning_rate": 9.785454545454546e-06,
157
+ "loss": 4.0014,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 5.45,
162
+ "learning_rate": 9.74e-06,
163
+ "loss": 3.8653,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 5.66,
168
+ "learning_rate": 9.694545454545455e-06,
169
+ "loss": 3.8299,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 5.87,
174
+ "learning_rate": 9.64909090909091e-06,
175
+ "loss": 3.6928,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 6.08,
180
+ "learning_rate": 9.603636363636365e-06,
181
+ "loss": 3.1672,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 6.29,
186
+ "learning_rate": 9.558181818181818e-06,
187
+ "loss": 2.5309,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 6.5,
192
+ "learning_rate": 9.512727272727274e-06,
193
+ "loss": 2.4801,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 6.71,
198
+ "learning_rate": 9.467272727272727e-06,
199
+ "loss": 2.2236,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 6.92,
204
+ "learning_rate": 9.421818181818183e-06,
205
+ "loss": 2.3091,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 7.13,
210
+ "learning_rate": 9.376363636363637e-06,
211
+ "loss": 1.9126,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 7.34,
216
+ "learning_rate": 9.330909090909092e-06,
217
+ "loss": 1.4071,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 7.55,
222
+ "learning_rate": 9.285454545454546e-06,
223
+ "loss": 1.5367,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 7.76,
228
+ "learning_rate": 9.240000000000001e-06,
229
+ "loss": 1.3686,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 7.97,
234
+ "learning_rate": 9.194545454545455e-06,
235
+ "loss": 1.3177,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 8.18,
240
+ "learning_rate": 9.14909090909091e-06,
241
+ "loss": 0.9167,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 8.39,
246
+ "learning_rate": 9.103636363636364e-06,
247
+ "loss": 0.9502,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 8.39,
252
+ "eval_loss": 3.2547426223754883,
253
+ "eval_runtime": 507.4588,
254
+ "eval_samples_per_second": 1.255,
255
+ "eval_steps_per_second": 0.158,
256
+ "eval_wer": 5215.41095890411,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 8.6,
261
+ "learning_rate": 9.05818181818182e-06,
262
+ "loss": 0.774,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 8.81,
267
+ "learning_rate": 9.012727272727273e-06,
268
+ "loss": 0.8792,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 9.02,
273
+ "learning_rate": 8.967272727272727e-06,
274
+ "loss": 0.8182,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 9.23,
279
+ "learning_rate": 8.921818181818183e-06,
280
+ "loss": 0.5773,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 9.44,
285
+ "learning_rate": 8.876363636363638e-06,
286
+ "loss": 0.5618,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 9.65,
291
+ "learning_rate": 8.830909090909092e-06,
292
+ "loss": 0.5499,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 9.86,
297
+ "learning_rate": 8.785454545454546e-06,
298
+ "loss": 0.5145,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 10.07,
303
+ "learning_rate": 8.740000000000001e-06,
304
+ "loss": 0.4436,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 10.28,
309
+ "learning_rate": 8.694545454545455e-06,
310
+ "loss": 0.3293,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 10.49,
315
+ "learning_rate": 8.64909090909091e-06,
316
+ "loss": 0.3603,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 10.7,
321
+ "learning_rate": 8.603636363636364e-06,
322
+ "loss": 0.3796,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 10.91,
327
+ "learning_rate": 8.55818181818182e-06,
328
+ "loss": 0.347,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 11.12,
333
+ "learning_rate": 8.512727272727273e-06,
334
+ "loss": 0.263,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 11.33,
339
+ "learning_rate": 8.467272727272729e-06,
340
+ "loss": 0.272,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 11.54,
345
+ "learning_rate": 8.421818181818182e-06,
346
+ "loss": 0.2657,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 11.75,
351
+ "learning_rate": 8.376363636363636e-06,
352
+ "loss": 0.2299,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 11.96,
357
+ "learning_rate": 8.330909090909092e-06,
358
+ "loss": 0.2563,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 12.17,
363
+ "learning_rate": 8.285454545454547e-06,
364
+ "loss": 0.1685,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 12.38,
369
+ "learning_rate": 8.24e-06,
370
+ "loss": 0.2457,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 12.59,
375
+ "learning_rate": 8.194545454545454e-06,
376
+ "loss": 0.1859,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 12.79,
381
+ "learning_rate": 8.14909090909091e-06,
382
+ "loss": 0.1883,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 13.0,
387
+ "learning_rate": 8.103636363636365e-06,
388
+ "loss": 0.1484,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 13.21,
393
+ "learning_rate": 8.058181818181819e-06,
394
+ "loss": 0.1322,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 13.42,
399
+ "learning_rate": 8.012727272727273e-06,
400
+ "loss": 0.1099,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 13.63,
405
+ "learning_rate": 7.967272727272728e-06,
406
+ "loss": 0.1455,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 13.84,
411
+ "learning_rate": 7.921818181818182e-06,
412
+ "loss": 0.1534,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 14.05,
417
+ "learning_rate": 7.876363636363637e-06,
418
+ "loss": 0.1459,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 14.26,
423
+ "learning_rate": 7.830909090909091e-06,
424
+ "loss": 0.1286,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 14.47,
429
+ "learning_rate": 7.785454545454547e-06,
430
+ "loss": 0.0906,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 14.68,
435
+ "learning_rate": 7.74e-06,
436
+ "loss": 0.1186,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 14.89,
441
+ "learning_rate": 7.694545454545456e-06,
442
+ "loss": 0.0855,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 15.1,
447
+ "learning_rate": 7.64909090909091e-06,
448
+ "loss": 0.1135,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 15.31,
453
+ "learning_rate": 7.603636363636364e-06,
454
+ "loss": 0.0613,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 15.52,
459
+ "learning_rate": 7.558181818181819e-06,
460
+ "loss": 0.0682,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 15.73,
465
+ "learning_rate": 7.512727272727273e-06,
466
+ "loss": 0.0667,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 15.94,
471
+ "learning_rate": 7.467272727272727e-06,
472
+ "loss": 0.0815,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 16.15,
477
+ "learning_rate": 7.421818181818183e-06,
478
+ "loss": 0.0599,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 16.36,
483
+ "learning_rate": 7.376363636363637e-06,
484
+ "loss": 0.0639,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 16.57,
489
+ "learning_rate": 7.330909090909092e-06,
490
+ "loss": 0.0578,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 16.78,
495
+ "learning_rate": 7.2854545454545455e-06,
496
+ "loss": 0.0448,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 16.78,
501
+ "eval_loss": 2.7543158531188965,
502
+ "eval_runtime": 502.2301,
503
+ "eval_samples_per_second": 1.268,
504
+ "eval_steps_per_second": 0.159,
505
+ "eval_wer": 5006.62100456621,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 16.99,
510
+ "learning_rate": 7.24e-06,
511
+ "loss": 0.0774,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 17.2,
516
+ "learning_rate": 7.194545454545456e-06,
517
+ "loss": 0.0488,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 17.41,
522
+ "learning_rate": 7.14909090909091e-06,
523
+ "loss": 0.0417,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 17.62,
528
+ "learning_rate": 7.103636363636364e-06,
529
+ "loss": 0.0502,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 17.83,
534
+ "learning_rate": 7.0581818181818185e-06,
535
+ "loss": 0.0548,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 18.04,
540
+ "learning_rate": 7.012727272727273e-06,
541
+ "loss": 0.055,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 18.25,
546
+ "learning_rate": 6.967272727272728e-06,
547
+ "loss": 0.0507,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 18.46,
552
+ "learning_rate": 6.921818181818183e-06,
553
+ "loss": 0.0592,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 18.67,
558
+ "learning_rate": 6.876363636363637e-06,
559
+ "loss": 0.0313,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 18.88,
564
+ "learning_rate": 6.8309090909090915e-06,
565
+ "loss": 0.028,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 19.09,
570
+ "learning_rate": 6.785454545454546e-06,
571
+ "loss": 0.0392,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 19.3,
576
+ "learning_rate": 6.740000000000001e-06,
577
+ "loss": 0.0346,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 19.51,
582
+ "learning_rate": 6.6945454545454544e-06,
583
+ "loss": 0.0347,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 19.72,
588
+ "learning_rate": 6.64909090909091e-06,
589
+ "loss": 0.0279,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 19.93,
594
+ "learning_rate": 6.6036363636363645e-06,
595
+ "loss": 0.0351,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 20.14,
600
+ "learning_rate": 6.558181818181819e-06,
601
+ "loss": 0.0241,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 20.35,
606
+ "learning_rate": 6.512727272727273e-06,
607
+ "loss": 0.0337,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 20.56,
612
+ "learning_rate": 6.467272727272727e-06,
613
+ "loss": 0.0308,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 20.77,
618
+ "learning_rate": 6.421818181818182e-06,
619
+ "loss": 0.0189,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 20.98,
624
+ "learning_rate": 6.3763636363636375e-06,
625
+ "loss": 0.0201,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 21.19,
630
+ "learning_rate": 6.330909090909091e-06,
631
+ "loss": 0.02,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 21.39,
636
+ "learning_rate": 6.285454545454546e-06,
637
+ "loss": 0.0173,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 21.6,
642
+ "learning_rate": 6.24e-06,
643
+ "loss": 0.0218,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 21.81,
648
+ "learning_rate": 6.194545454545455e-06,
649
+ "loss": 0.027,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 22.02,
654
+ "learning_rate": 6.1490909090909104e-06,
655
+ "loss": 0.0188,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 22.23,
660
+ "learning_rate": 6.103636363636364e-06,
661
+ "loss": 0.02,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 22.44,
666
+ "learning_rate": 6.058181818181819e-06,
667
+ "loss": 0.0165,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 22.65,
672
+ "learning_rate": 6.012727272727273e-06,
673
+ "loss": 0.0115,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 22.86,
678
+ "learning_rate": 5.967272727272728e-06,
679
+ "loss": 0.0143,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 23.07,
684
+ "learning_rate": 5.921818181818182e-06,
685
+ "loss": 0.0104,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 23.28,
690
+ "learning_rate": 5.876363636363636e-06,
691
+ "loss": 0.0209,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 23.49,
696
+ "learning_rate": 5.830909090909092e-06,
697
+ "loss": 0.0056,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 23.7,
702
+ "learning_rate": 5.785454545454546e-06,
703
+ "loss": 0.0133,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 23.91,
708
+ "learning_rate": 5.74e-06,
709
+ "loss": 0.0186,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 24.12,
714
+ "learning_rate": 5.694545454545455e-06,
715
+ "loss": 0.0094,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 24.33,
720
+ "learning_rate": 5.649090909090909e-06,
721
+ "loss": 0.0069,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 24.54,
726
+ "learning_rate": 5.603636363636365e-06,
727
+ "loss": 0.0115,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 24.75,
732
+ "learning_rate": 5.5581818181818185e-06,
733
+ "loss": 0.0045,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 24.96,
738
+ "learning_rate": 5.512727272727273e-06,
739
+ "loss": 0.0089,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 25.17,
744
+ "learning_rate": 5.467272727272728e-06,
745
+ "loss": 0.0089,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 25.17,
750
+ "eval_loss": 2.8150265216827393,
751
+ "eval_runtime": 495.3178,
752
+ "eval_samples_per_second": 1.286,
753
+ "eval_steps_per_second": 0.162,
754
+ "eval_wer": 5198.934550989346,
755
+ "step": 3000
756
+ }
757
+ ],
758
+ "max_steps": 6000,
759
+ "num_train_epochs": 51,
760
+ "total_flos": 1.385844809416704e+19,
761
+ "trial_name": null,
762
+ "trial_params": null
763
+ }
checkpoint-3000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cbeda0929f5bade1f2682f85820a8eb42b209173de61e856a1e94b350f2c7d1
3
+ size 4091
checkpoint-4000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
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": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.31.0",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-4000/generation_config.json ADDED
@@ -0,0 +1,263 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "max_initial_timestamp_index": 1,
164
+ "max_length": 448,
165
+ "no_timestamps_token_id": 50363,
166
+ "pad_token_id": 50257,
167
+ "return_timestamps": false,
168
+ "suppress_tokens": [
169
+ 1,
170
+ 2,
171
+ 7,
172
+ 8,
173
+ 9,
174
+ 10,
175
+ 14,
176
+ 25,
177
+ 26,
178
+ 27,
179
+ 28,
180
+ 29,
181
+ 31,
182
+ 58,
183
+ 59,
184
+ 60,
185
+ 61,
186
+ 62,
187
+ 63,
188
+ 90,
189
+ 91,
190
+ 92,
191
+ 93,
192
+ 359,
193
+ 503,
194
+ 522,
195
+ 542,
196
+ 873,
197
+ 893,
198
+ 902,
199
+ 918,
200
+ 922,
201
+ 931,
202
+ 1350,
203
+ 1853,
204
+ 1982,
205
+ 2460,
206
+ 2627,
207
+ 3246,
208
+ 3253,
209
+ 3268,
210
+ 3536,
211
+ 3846,
212
+ 3961,
213
+ 4183,
214
+ 4667,
215
+ 6585,
216
+ 6647,
217
+ 7273,
218
+ 9061,
219
+ 9383,
220
+ 10428,
221
+ 10929,
222
+ 11938,
223
+ 12033,
224
+ 12331,
225
+ 12562,
226
+ 13793,
227
+ 14157,
228
+ 14635,
229
+ 15265,
230
+ 15618,
231
+ 16553,
232
+ 16604,
233
+ 18362,
234
+ 18956,
235
+ 20075,
236
+ 21675,
237
+ 22520,
238
+ 26130,
239
+ 26161,
240
+ 26435,
241
+ 28279,
242
+ 29464,
243
+ 31650,
244
+ 32302,
245
+ 32470,
246
+ 36865,
247
+ 42863,
248
+ 47425,
249
+ 49870,
250
+ 50254,
251
+ 50258,
252
+ 50358,
253
+ 50359,
254
+ 50360,
255
+ 50361,
256
+ 50362
257
+ ],
258
+ "task_to_id": {
259
+ "transcribe": 50359,
260
+ "translate": 50358
261
+ },
262
+ "transformers_version": "4.31.0"
263
+ }
checkpoint-4000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d14b1af17fa9e379e8c1ae624c3439073a9c1c3e12124f4c68f0bb8d8e2e822f
3
+ size 1934161157
checkpoint-4000/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-4000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c951c2c8275fa32c29d4175631f26445048f1f8d4a6f2f6a7371c2bc18152356
3
+ size 967102729
checkpoint-4000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:896965bd376ebba70871003004c9e6263f08df684a308381606e0a9ca3be6641
3
+ size 14575
checkpoint-4000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b438f559819afbc3f6060004328a5b8c6b9fa6bcc044ead2d537c2083c1f5cf
3
+ size 627
checkpoint-4000/trainer_state.json ADDED
@@ -0,0 +1,1012 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 5006.62100456621,
3
+ "best_model_checkpoint": "./20230912-output\\checkpoint-2000",
4
+ "epoch": 33.560566334556896,
5
+ "global_step": 4000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.21,
12
+ "learning_rate": 3.8e-07,
13
+ "loss": 18.9312,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.42,
18
+ "learning_rate": 8.8e-07,
19
+ "loss": 17.0566,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.63,
24
+ "learning_rate": 1.3800000000000001e-06,
25
+ "loss": 13.4189,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.84,
30
+ "learning_rate": 1.8800000000000002e-06,
31
+ "loss": 10.4803,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 1.05,
36
+ "learning_rate": 2.38e-06,
37
+ "loss": 9.3304,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 1.26,
42
+ "learning_rate": 2.88e-06,
43
+ "loss": 8.3203,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 1.47,
48
+ "learning_rate": 3.3600000000000004e-06,
49
+ "loss": 8.9822,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 1.68,
54
+ "learning_rate": 3.86e-06,
55
+ "loss": 10.7581,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 1.89,
60
+ "learning_rate": 4.360000000000001e-06,
61
+ "loss": 13.2164,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 2.1,
66
+ "learning_rate": 4.86e-06,
67
+ "loss": 12.7052,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 2.31,
72
+ "learning_rate": 5.36e-06,
73
+ "loss": 10.3904,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 2.52,
78
+ "learning_rate": 5.86e-06,
79
+ "loss": 8.2617,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 2.73,
84
+ "learning_rate": 6.360000000000001e-06,
85
+ "loss": 6.8613,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 2.94,
90
+ "learning_rate": 6.860000000000001e-06,
91
+ "loss": 6.1093,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 3.15,
96
+ "learning_rate": 7.360000000000001e-06,
97
+ "loss": 5.8769,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 3.36,
102
+ "learning_rate": 7.860000000000001e-06,
103
+ "loss": 5.6272,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 3.57,
108
+ "learning_rate": 8.36e-06,
109
+ "loss": 5.674,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 3.78,
114
+ "learning_rate": 8.860000000000002e-06,
115
+ "loss": 5.5667,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 3.99,
120
+ "learning_rate": 9.360000000000002e-06,
121
+ "loss": 5.4705,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 4.2,
126
+ "learning_rate": 9.86e-06,
127
+ "loss": 5.1703,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 4.4,
132
+ "learning_rate": 9.967272727272728e-06,
133
+ "loss": 5.1044,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 4.61,
138
+ "learning_rate": 9.921818181818183e-06,
139
+ "loss": 4.9244,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 4.82,
144
+ "learning_rate": 9.876363636363637e-06,
145
+ "loss": 4.8602,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 5.03,
150
+ "learning_rate": 9.830909090909092e-06,
151
+ "loss": 4.6437,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 5.24,
156
+ "learning_rate": 9.785454545454546e-06,
157
+ "loss": 4.0014,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 5.45,
162
+ "learning_rate": 9.74e-06,
163
+ "loss": 3.8653,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 5.66,
168
+ "learning_rate": 9.694545454545455e-06,
169
+ "loss": 3.8299,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 5.87,
174
+ "learning_rate": 9.64909090909091e-06,
175
+ "loss": 3.6928,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 6.08,
180
+ "learning_rate": 9.603636363636365e-06,
181
+ "loss": 3.1672,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 6.29,
186
+ "learning_rate": 9.558181818181818e-06,
187
+ "loss": 2.5309,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 6.5,
192
+ "learning_rate": 9.512727272727274e-06,
193
+ "loss": 2.4801,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 6.71,
198
+ "learning_rate": 9.467272727272727e-06,
199
+ "loss": 2.2236,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 6.92,
204
+ "learning_rate": 9.421818181818183e-06,
205
+ "loss": 2.3091,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 7.13,
210
+ "learning_rate": 9.376363636363637e-06,
211
+ "loss": 1.9126,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 7.34,
216
+ "learning_rate": 9.330909090909092e-06,
217
+ "loss": 1.4071,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 7.55,
222
+ "learning_rate": 9.285454545454546e-06,
223
+ "loss": 1.5367,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 7.76,
228
+ "learning_rate": 9.240000000000001e-06,
229
+ "loss": 1.3686,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 7.97,
234
+ "learning_rate": 9.194545454545455e-06,
235
+ "loss": 1.3177,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 8.18,
240
+ "learning_rate": 9.14909090909091e-06,
241
+ "loss": 0.9167,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 8.39,
246
+ "learning_rate": 9.103636363636364e-06,
247
+ "loss": 0.9502,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 8.39,
252
+ "eval_loss": 3.2547426223754883,
253
+ "eval_runtime": 507.4588,
254
+ "eval_samples_per_second": 1.255,
255
+ "eval_steps_per_second": 0.158,
256
+ "eval_wer": 5215.41095890411,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 8.6,
261
+ "learning_rate": 9.05818181818182e-06,
262
+ "loss": 0.774,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 8.81,
267
+ "learning_rate": 9.012727272727273e-06,
268
+ "loss": 0.8792,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 9.02,
273
+ "learning_rate": 8.967272727272727e-06,
274
+ "loss": 0.8182,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 9.23,
279
+ "learning_rate": 8.921818181818183e-06,
280
+ "loss": 0.5773,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 9.44,
285
+ "learning_rate": 8.876363636363638e-06,
286
+ "loss": 0.5618,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 9.65,
291
+ "learning_rate": 8.830909090909092e-06,
292
+ "loss": 0.5499,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 9.86,
297
+ "learning_rate": 8.785454545454546e-06,
298
+ "loss": 0.5145,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 10.07,
303
+ "learning_rate": 8.740000000000001e-06,
304
+ "loss": 0.4436,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 10.28,
309
+ "learning_rate": 8.694545454545455e-06,
310
+ "loss": 0.3293,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 10.49,
315
+ "learning_rate": 8.64909090909091e-06,
316
+ "loss": 0.3603,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 10.7,
321
+ "learning_rate": 8.603636363636364e-06,
322
+ "loss": 0.3796,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 10.91,
327
+ "learning_rate": 8.55818181818182e-06,
328
+ "loss": 0.347,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 11.12,
333
+ "learning_rate": 8.512727272727273e-06,
334
+ "loss": 0.263,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 11.33,
339
+ "learning_rate": 8.467272727272729e-06,
340
+ "loss": 0.272,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 11.54,
345
+ "learning_rate": 8.421818181818182e-06,
346
+ "loss": 0.2657,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 11.75,
351
+ "learning_rate": 8.376363636363636e-06,
352
+ "loss": 0.2299,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 11.96,
357
+ "learning_rate": 8.330909090909092e-06,
358
+ "loss": 0.2563,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 12.17,
363
+ "learning_rate": 8.285454545454547e-06,
364
+ "loss": 0.1685,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 12.38,
369
+ "learning_rate": 8.24e-06,
370
+ "loss": 0.2457,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 12.59,
375
+ "learning_rate": 8.194545454545454e-06,
376
+ "loss": 0.1859,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 12.79,
381
+ "learning_rate": 8.14909090909091e-06,
382
+ "loss": 0.1883,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 13.0,
387
+ "learning_rate": 8.103636363636365e-06,
388
+ "loss": 0.1484,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 13.21,
393
+ "learning_rate": 8.058181818181819e-06,
394
+ "loss": 0.1322,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 13.42,
399
+ "learning_rate": 8.012727272727273e-06,
400
+ "loss": 0.1099,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 13.63,
405
+ "learning_rate": 7.967272727272728e-06,
406
+ "loss": 0.1455,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 13.84,
411
+ "learning_rate": 7.921818181818182e-06,
412
+ "loss": 0.1534,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 14.05,
417
+ "learning_rate": 7.876363636363637e-06,
418
+ "loss": 0.1459,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 14.26,
423
+ "learning_rate": 7.830909090909091e-06,
424
+ "loss": 0.1286,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 14.47,
429
+ "learning_rate": 7.785454545454547e-06,
430
+ "loss": 0.0906,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 14.68,
435
+ "learning_rate": 7.74e-06,
436
+ "loss": 0.1186,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 14.89,
441
+ "learning_rate": 7.694545454545456e-06,
442
+ "loss": 0.0855,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 15.1,
447
+ "learning_rate": 7.64909090909091e-06,
448
+ "loss": 0.1135,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 15.31,
453
+ "learning_rate": 7.603636363636364e-06,
454
+ "loss": 0.0613,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 15.52,
459
+ "learning_rate": 7.558181818181819e-06,
460
+ "loss": 0.0682,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 15.73,
465
+ "learning_rate": 7.512727272727273e-06,
466
+ "loss": 0.0667,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 15.94,
471
+ "learning_rate": 7.467272727272727e-06,
472
+ "loss": 0.0815,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 16.15,
477
+ "learning_rate": 7.421818181818183e-06,
478
+ "loss": 0.0599,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 16.36,
483
+ "learning_rate": 7.376363636363637e-06,
484
+ "loss": 0.0639,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 16.57,
489
+ "learning_rate": 7.330909090909092e-06,
490
+ "loss": 0.0578,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 16.78,
495
+ "learning_rate": 7.2854545454545455e-06,
496
+ "loss": 0.0448,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 16.78,
501
+ "eval_loss": 2.7543158531188965,
502
+ "eval_runtime": 502.2301,
503
+ "eval_samples_per_second": 1.268,
504
+ "eval_steps_per_second": 0.159,
505
+ "eval_wer": 5006.62100456621,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 16.99,
510
+ "learning_rate": 7.24e-06,
511
+ "loss": 0.0774,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 17.2,
516
+ "learning_rate": 7.194545454545456e-06,
517
+ "loss": 0.0488,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 17.41,
522
+ "learning_rate": 7.14909090909091e-06,
523
+ "loss": 0.0417,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 17.62,
528
+ "learning_rate": 7.103636363636364e-06,
529
+ "loss": 0.0502,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 17.83,
534
+ "learning_rate": 7.0581818181818185e-06,
535
+ "loss": 0.0548,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 18.04,
540
+ "learning_rate": 7.012727272727273e-06,
541
+ "loss": 0.055,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 18.25,
546
+ "learning_rate": 6.967272727272728e-06,
547
+ "loss": 0.0507,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 18.46,
552
+ "learning_rate": 6.921818181818183e-06,
553
+ "loss": 0.0592,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 18.67,
558
+ "learning_rate": 6.876363636363637e-06,
559
+ "loss": 0.0313,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 18.88,
564
+ "learning_rate": 6.8309090909090915e-06,
565
+ "loss": 0.028,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 19.09,
570
+ "learning_rate": 6.785454545454546e-06,
571
+ "loss": 0.0392,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 19.3,
576
+ "learning_rate": 6.740000000000001e-06,
577
+ "loss": 0.0346,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 19.51,
582
+ "learning_rate": 6.6945454545454544e-06,
583
+ "loss": 0.0347,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 19.72,
588
+ "learning_rate": 6.64909090909091e-06,
589
+ "loss": 0.0279,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 19.93,
594
+ "learning_rate": 6.6036363636363645e-06,
595
+ "loss": 0.0351,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 20.14,
600
+ "learning_rate": 6.558181818181819e-06,
601
+ "loss": 0.0241,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 20.35,
606
+ "learning_rate": 6.512727272727273e-06,
607
+ "loss": 0.0337,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 20.56,
612
+ "learning_rate": 6.467272727272727e-06,
613
+ "loss": 0.0308,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 20.77,
618
+ "learning_rate": 6.421818181818182e-06,
619
+ "loss": 0.0189,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 20.98,
624
+ "learning_rate": 6.3763636363636375e-06,
625
+ "loss": 0.0201,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 21.19,
630
+ "learning_rate": 6.330909090909091e-06,
631
+ "loss": 0.02,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 21.39,
636
+ "learning_rate": 6.285454545454546e-06,
637
+ "loss": 0.0173,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 21.6,
642
+ "learning_rate": 6.24e-06,
643
+ "loss": 0.0218,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 21.81,
648
+ "learning_rate": 6.194545454545455e-06,
649
+ "loss": 0.027,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 22.02,
654
+ "learning_rate": 6.1490909090909104e-06,
655
+ "loss": 0.0188,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 22.23,
660
+ "learning_rate": 6.103636363636364e-06,
661
+ "loss": 0.02,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 22.44,
666
+ "learning_rate": 6.058181818181819e-06,
667
+ "loss": 0.0165,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 22.65,
672
+ "learning_rate": 6.012727272727273e-06,
673
+ "loss": 0.0115,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 22.86,
678
+ "learning_rate": 5.967272727272728e-06,
679
+ "loss": 0.0143,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 23.07,
684
+ "learning_rate": 5.921818181818182e-06,
685
+ "loss": 0.0104,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 23.28,
690
+ "learning_rate": 5.876363636363636e-06,
691
+ "loss": 0.0209,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 23.49,
696
+ "learning_rate": 5.830909090909092e-06,
697
+ "loss": 0.0056,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 23.7,
702
+ "learning_rate": 5.785454545454546e-06,
703
+ "loss": 0.0133,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 23.91,
708
+ "learning_rate": 5.74e-06,
709
+ "loss": 0.0186,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 24.12,
714
+ "learning_rate": 5.694545454545455e-06,
715
+ "loss": 0.0094,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 24.33,
720
+ "learning_rate": 5.649090909090909e-06,
721
+ "loss": 0.0069,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 24.54,
726
+ "learning_rate": 5.603636363636365e-06,
727
+ "loss": 0.0115,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 24.75,
732
+ "learning_rate": 5.5581818181818185e-06,
733
+ "loss": 0.0045,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 24.96,
738
+ "learning_rate": 5.512727272727273e-06,
739
+ "loss": 0.0089,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 25.17,
744
+ "learning_rate": 5.467272727272728e-06,
745
+ "loss": 0.0089,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 25.17,
750
+ "eval_loss": 2.8150265216827393,
751
+ "eval_runtime": 495.3178,
752
+ "eval_samples_per_second": 1.286,
753
+ "eval_steps_per_second": 0.162,
754
+ "eval_wer": 5198.934550989346,
755
+ "step": 3000
756
+ },
757
+ {
758
+ "epoch": 25.38,
759
+ "learning_rate": 5.421818181818182e-06,
760
+ "loss": 0.0111,
761
+ "step": 3025
762
+ },
763
+ {
764
+ "epoch": 25.59,
765
+ "learning_rate": 5.376363636363638e-06,
766
+ "loss": 0.0106,
767
+ "step": 3050
768
+ },
769
+ {
770
+ "epoch": 25.8,
771
+ "learning_rate": 5.330909090909091e-06,
772
+ "loss": 0.0032,
773
+ "step": 3075
774
+ },
775
+ {
776
+ "epoch": 26.01,
777
+ "learning_rate": 5.285454545454546e-06,
778
+ "loss": 0.0041,
779
+ "step": 3100
780
+ },
781
+ {
782
+ "epoch": 26.22,
783
+ "learning_rate": 5.240000000000001e-06,
784
+ "loss": 0.0078,
785
+ "step": 3125
786
+ },
787
+ {
788
+ "epoch": 26.43,
789
+ "learning_rate": 5.194545454545455e-06,
790
+ "loss": 0.0052,
791
+ "step": 3150
792
+ },
793
+ {
794
+ "epoch": 26.64,
795
+ "learning_rate": 5.149090909090909e-06,
796
+ "loss": 0.0067,
797
+ "step": 3175
798
+ },
799
+ {
800
+ "epoch": 26.85,
801
+ "learning_rate": 5.103636363636364e-06,
802
+ "loss": 0.0044,
803
+ "step": 3200
804
+ },
805
+ {
806
+ "epoch": 27.06,
807
+ "learning_rate": 5.058181818181819e-06,
808
+ "loss": 0.0054,
809
+ "step": 3225
810
+ },
811
+ {
812
+ "epoch": 27.27,
813
+ "learning_rate": 5.012727272727274e-06,
814
+ "loss": 0.0025,
815
+ "step": 3250
816
+ },
817
+ {
818
+ "epoch": 27.48,
819
+ "learning_rate": 4.967272727272727e-06,
820
+ "loss": 0.0043,
821
+ "step": 3275
822
+ },
823
+ {
824
+ "epoch": 27.69,
825
+ "learning_rate": 4.921818181818182e-06,
826
+ "loss": 0.0038,
827
+ "step": 3300
828
+ },
829
+ {
830
+ "epoch": 27.9,
831
+ "learning_rate": 4.8763636363636366e-06,
832
+ "loss": 0.004,
833
+ "step": 3325
834
+ },
835
+ {
836
+ "epoch": 28.11,
837
+ "learning_rate": 4.830909090909091e-06,
838
+ "loss": 0.0041,
839
+ "step": 3350
840
+ },
841
+ {
842
+ "epoch": 28.32,
843
+ "learning_rate": 4.785454545454546e-06,
844
+ "loss": 0.0052,
845
+ "step": 3375
846
+ },
847
+ {
848
+ "epoch": 28.53,
849
+ "learning_rate": 4.74e-06,
850
+ "loss": 0.0059,
851
+ "step": 3400
852
+ },
853
+ {
854
+ "epoch": 28.74,
855
+ "learning_rate": 4.694545454545455e-06,
856
+ "loss": 0.0012,
857
+ "step": 3425
858
+ },
859
+ {
860
+ "epoch": 28.95,
861
+ "learning_rate": 4.6490909090909095e-06,
862
+ "loss": 0.0021,
863
+ "step": 3450
864
+ },
865
+ {
866
+ "epoch": 29.16,
867
+ "learning_rate": 4.603636363636364e-06,
868
+ "loss": 0.0034,
869
+ "step": 3475
870
+ },
871
+ {
872
+ "epoch": 29.37,
873
+ "learning_rate": 4.558181818181819e-06,
874
+ "loss": 0.0045,
875
+ "step": 3500
876
+ },
877
+ {
878
+ "epoch": 29.58,
879
+ "learning_rate": 4.512727272727273e-06,
880
+ "loss": 0.0023,
881
+ "step": 3525
882
+ },
883
+ {
884
+ "epoch": 29.79,
885
+ "learning_rate": 4.467272727272728e-06,
886
+ "loss": 0.0054,
887
+ "step": 3550
888
+ },
889
+ {
890
+ "epoch": 29.99,
891
+ "learning_rate": 4.4218181818181825e-06,
892
+ "loss": 0.0021,
893
+ "step": 3575
894
+ },
895
+ {
896
+ "epoch": 30.2,
897
+ "learning_rate": 4.376363636363637e-06,
898
+ "loss": 0.0019,
899
+ "step": 3600
900
+ },
901
+ {
902
+ "epoch": 30.41,
903
+ "learning_rate": 4.330909090909091e-06,
904
+ "loss": 0.004,
905
+ "step": 3625
906
+ },
907
+ {
908
+ "epoch": 30.62,
909
+ "learning_rate": 4.285454545454546e-06,
910
+ "loss": 0.0033,
911
+ "step": 3650
912
+ },
913
+ {
914
+ "epoch": 30.83,
915
+ "learning_rate": 4.24e-06,
916
+ "loss": 0.0015,
917
+ "step": 3675
918
+ },
919
+ {
920
+ "epoch": 31.04,
921
+ "learning_rate": 4.194545454545455e-06,
922
+ "loss": 0.0014,
923
+ "step": 3700
924
+ },
925
+ {
926
+ "epoch": 31.25,
927
+ "learning_rate": 4.149090909090909e-06,
928
+ "loss": 0.0027,
929
+ "step": 3725
930
+ },
931
+ {
932
+ "epoch": 31.46,
933
+ "learning_rate": 4.103636363636364e-06,
934
+ "loss": 0.0035,
935
+ "step": 3750
936
+ },
937
+ {
938
+ "epoch": 31.67,
939
+ "learning_rate": 4.0581818181818184e-06,
940
+ "loss": 0.0034,
941
+ "step": 3775
942
+ },
943
+ {
944
+ "epoch": 31.88,
945
+ "learning_rate": 4.012727272727273e-06,
946
+ "loss": 0.0006,
947
+ "step": 3800
948
+ },
949
+ {
950
+ "epoch": 32.09,
951
+ "learning_rate": 3.967272727272728e-06,
952
+ "loss": 0.0034,
953
+ "step": 3825
954
+ },
955
+ {
956
+ "epoch": 32.3,
957
+ "learning_rate": 3.921818181818182e-06,
958
+ "loss": 0.0011,
959
+ "step": 3850
960
+ },
961
+ {
962
+ "epoch": 32.51,
963
+ "learning_rate": 3.876363636363637e-06,
964
+ "loss": 0.001,
965
+ "step": 3875
966
+ },
967
+ {
968
+ "epoch": 32.72,
969
+ "learning_rate": 3.830909090909091e-06,
970
+ "loss": 0.002,
971
+ "step": 3900
972
+ },
973
+ {
974
+ "epoch": 32.93,
975
+ "learning_rate": 3.7854545454545456e-06,
976
+ "loss": 0.001,
977
+ "step": 3925
978
+ },
979
+ {
980
+ "epoch": 33.14,
981
+ "learning_rate": 3.74e-06,
982
+ "loss": 0.0012,
983
+ "step": 3950
984
+ },
985
+ {
986
+ "epoch": 33.35,
987
+ "learning_rate": 3.694545454545455e-06,
988
+ "loss": 0.0019,
989
+ "step": 3975
990
+ },
991
+ {
992
+ "epoch": 33.56,
993
+ "learning_rate": 3.6490909090909094e-06,
994
+ "loss": 0.0023,
995
+ "step": 4000
996
+ },
997
+ {
998
+ "epoch": 33.56,
999
+ "eval_loss": 2.811387300491333,
1000
+ "eval_runtime": 497.1526,
1001
+ "eval_samples_per_second": 1.281,
1002
+ "eval_steps_per_second": 0.161,
1003
+ "eval_wer": 5089.497716894977,
1004
+ "step": 4000
1005
+ }
1006
+ ],
1007
+ "max_steps": 6000,
1008
+ "num_train_epochs": 51,
1009
+ "total_flos": 1.847581449928704e+19,
1010
+ "trial_name": null,
1011
+ "trial_params": null
1012
+ }
checkpoint-4000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cbeda0929f5bade1f2682f85820a8eb42b209173de61e856a1e94b350f2c7d1
3
+ size 4091
checkpoint-5000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
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": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.31.0",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-5000/generation_config.json ADDED
@@ -0,0 +1,263 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "max_initial_timestamp_index": 1,
164
+ "max_length": 448,
165
+ "no_timestamps_token_id": 50363,
166
+ "pad_token_id": 50257,
167
+ "return_timestamps": false,
168
+ "suppress_tokens": [
169
+ 1,
170
+ 2,
171
+ 7,
172
+ 8,
173
+ 9,
174
+ 10,
175
+ 14,
176
+ 25,
177
+ 26,
178
+ 27,
179
+ 28,
180
+ 29,
181
+ 31,
182
+ 58,
183
+ 59,
184
+ 60,
185
+ 61,
186
+ 62,
187
+ 63,
188
+ 90,
189
+ 91,
190
+ 92,
191
+ 93,
192
+ 359,
193
+ 503,
194
+ 522,
195
+ 542,
196
+ 873,
197
+ 893,
198
+ 902,
199
+ 918,
200
+ 922,
201
+ 931,
202
+ 1350,
203
+ 1853,
204
+ 1982,
205
+ 2460,
206
+ 2627,
207
+ 3246,
208
+ 3253,
209
+ 3268,
210
+ 3536,
211
+ 3846,
212
+ 3961,
213
+ 4183,
214
+ 4667,
215
+ 6585,
216
+ 6647,
217
+ 7273,
218
+ 9061,
219
+ 9383,
220
+ 10428,
221
+ 10929,
222
+ 11938,
223
+ 12033,
224
+ 12331,
225
+ 12562,
226
+ 13793,
227
+ 14157,
228
+ 14635,
229
+ 15265,
230
+ 15618,
231
+ 16553,
232
+ 16604,
233
+ 18362,
234
+ 18956,
235
+ 20075,
236
+ 21675,
237
+ 22520,
238
+ 26130,
239
+ 26161,
240
+ 26435,
241
+ 28279,
242
+ 29464,
243
+ 31650,
244
+ 32302,
245
+ 32470,
246
+ 36865,
247
+ 42863,
248
+ 47425,
249
+ 49870,
250
+ 50254,
251
+ 50258,
252
+ 50358,
253
+ 50359,
254
+ 50360,
255
+ 50361,
256
+ 50362
257
+ ],
258
+ "task_to_id": {
259
+ "transcribe": 50359,
260
+ "translate": 50358
261
+ },
262
+ "transformers_version": "4.31.0"
263
+ }
checkpoint-5000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3ee704ec4c3e9f81cd723b82acf259133b8a87f8d91e5a55ca8a91ce63e70cbc
3
+ size 1934161157
checkpoint-5000/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-5000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1077e8688b670ae67dc690535c7d3c282011214d1fdad64d533e1a37132c1649
3
+ size 967102729
checkpoint-5000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ad437307d599d510ae848698b020a75380290dbf51f15b8bd9d86a2ff778ee6
3
+ size 14575
checkpoint-5000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b84aa930fb89633df319c38ecd296a3ce86b1f3538f526d59fdd1155bf15330e
3
+ size 627
checkpoint-5000/trainer_state.json ADDED
@@ -0,0 +1,1261 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 5006.62100456621,
3
+ "best_model_checkpoint": "./20230912-output\\checkpoint-2000",
4
+ "epoch": 41.950707918196116,
5
+ "global_step": 5000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.21,
12
+ "learning_rate": 3.8e-07,
13
+ "loss": 18.9312,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.42,
18
+ "learning_rate": 8.8e-07,
19
+ "loss": 17.0566,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.63,
24
+ "learning_rate": 1.3800000000000001e-06,
25
+ "loss": 13.4189,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.84,
30
+ "learning_rate": 1.8800000000000002e-06,
31
+ "loss": 10.4803,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 1.05,
36
+ "learning_rate": 2.38e-06,
37
+ "loss": 9.3304,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 1.26,
42
+ "learning_rate": 2.88e-06,
43
+ "loss": 8.3203,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 1.47,
48
+ "learning_rate": 3.3600000000000004e-06,
49
+ "loss": 8.9822,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 1.68,
54
+ "learning_rate": 3.86e-06,
55
+ "loss": 10.7581,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 1.89,
60
+ "learning_rate": 4.360000000000001e-06,
61
+ "loss": 13.2164,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 2.1,
66
+ "learning_rate": 4.86e-06,
67
+ "loss": 12.7052,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 2.31,
72
+ "learning_rate": 5.36e-06,
73
+ "loss": 10.3904,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 2.52,
78
+ "learning_rate": 5.86e-06,
79
+ "loss": 8.2617,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 2.73,
84
+ "learning_rate": 6.360000000000001e-06,
85
+ "loss": 6.8613,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 2.94,
90
+ "learning_rate": 6.860000000000001e-06,
91
+ "loss": 6.1093,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 3.15,
96
+ "learning_rate": 7.360000000000001e-06,
97
+ "loss": 5.8769,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 3.36,
102
+ "learning_rate": 7.860000000000001e-06,
103
+ "loss": 5.6272,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 3.57,
108
+ "learning_rate": 8.36e-06,
109
+ "loss": 5.674,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 3.78,
114
+ "learning_rate": 8.860000000000002e-06,
115
+ "loss": 5.5667,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 3.99,
120
+ "learning_rate": 9.360000000000002e-06,
121
+ "loss": 5.4705,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 4.2,
126
+ "learning_rate": 9.86e-06,
127
+ "loss": 5.1703,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 4.4,
132
+ "learning_rate": 9.967272727272728e-06,
133
+ "loss": 5.1044,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 4.61,
138
+ "learning_rate": 9.921818181818183e-06,
139
+ "loss": 4.9244,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 4.82,
144
+ "learning_rate": 9.876363636363637e-06,
145
+ "loss": 4.8602,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 5.03,
150
+ "learning_rate": 9.830909090909092e-06,
151
+ "loss": 4.6437,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 5.24,
156
+ "learning_rate": 9.785454545454546e-06,
157
+ "loss": 4.0014,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 5.45,
162
+ "learning_rate": 9.74e-06,
163
+ "loss": 3.8653,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 5.66,
168
+ "learning_rate": 9.694545454545455e-06,
169
+ "loss": 3.8299,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 5.87,
174
+ "learning_rate": 9.64909090909091e-06,
175
+ "loss": 3.6928,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 6.08,
180
+ "learning_rate": 9.603636363636365e-06,
181
+ "loss": 3.1672,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 6.29,
186
+ "learning_rate": 9.558181818181818e-06,
187
+ "loss": 2.5309,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 6.5,
192
+ "learning_rate": 9.512727272727274e-06,
193
+ "loss": 2.4801,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 6.71,
198
+ "learning_rate": 9.467272727272727e-06,
199
+ "loss": 2.2236,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 6.92,
204
+ "learning_rate": 9.421818181818183e-06,
205
+ "loss": 2.3091,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 7.13,
210
+ "learning_rate": 9.376363636363637e-06,
211
+ "loss": 1.9126,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 7.34,
216
+ "learning_rate": 9.330909090909092e-06,
217
+ "loss": 1.4071,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 7.55,
222
+ "learning_rate": 9.285454545454546e-06,
223
+ "loss": 1.5367,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 7.76,
228
+ "learning_rate": 9.240000000000001e-06,
229
+ "loss": 1.3686,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 7.97,
234
+ "learning_rate": 9.194545454545455e-06,
235
+ "loss": 1.3177,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 8.18,
240
+ "learning_rate": 9.14909090909091e-06,
241
+ "loss": 0.9167,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 8.39,
246
+ "learning_rate": 9.103636363636364e-06,
247
+ "loss": 0.9502,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 8.39,
252
+ "eval_loss": 3.2547426223754883,
253
+ "eval_runtime": 507.4588,
254
+ "eval_samples_per_second": 1.255,
255
+ "eval_steps_per_second": 0.158,
256
+ "eval_wer": 5215.41095890411,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 8.6,
261
+ "learning_rate": 9.05818181818182e-06,
262
+ "loss": 0.774,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 8.81,
267
+ "learning_rate": 9.012727272727273e-06,
268
+ "loss": 0.8792,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 9.02,
273
+ "learning_rate": 8.967272727272727e-06,
274
+ "loss": 0.8182,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 9.23,
279
+ "learning_rate": 8.921818181818183e-06,
280
+ "loss": 0.5773,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 9.44,
285
+ "learning_rate": 8.876363636363638e-06,
286
+ "loss": 0.5618,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 9.65,
291
+ "learning_rate": 8.830909090909092e-06,
292
+ "loss": 0.5499,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 9.86,
297
+ "learning_rate": 8.785454545454546e-06,
298
+ "loss": 0.5145,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 10.07,
303
+ "learning_rate": 8.740000000000001e-06,
304
+ "loss": 0.4436,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 10.28,
309
+ "learning_rate": 8.694545454545455e-06,
310
+ "loss": 0.3293,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 10.49,
315
+ "learning_rate": 8.64909090909091e-06,
316
+ "loss": 0.3603,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 10.7,
321
+ "learning_rate": 8.603636363636364e-06,
322
+ "loss": 0.3796,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 10.91,
327
+ "learning_rate": 8.55818181818182e-06,
328
+ "loss": 0.347,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 11.12,
333
+ "learning_rate": 8.512727272727273e-06,
334
+ "loss": 0.263,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 11.33,
339
+ "learning_rate": 8.467272727272729e-06,
340
+ "loss": 0.272,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 11.54,
345
+ "learning_rate": 8.421818181818182e-06,
346
+ "loss": 0.2657,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 11.75,
351
+ "learning_rate": 8.376363636363636e-06,
352
+ "loss": 0.2299,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 11.96,
357
+ "learning_rate": 8.330909090909092e-06,
358
+ "loss": 0.2563,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 12.17,
363
+ "learning_rate": 8.285454545454547e-06,
364
+ "loss": 0.1685,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 12.38,
369
+ "learning_rate": 8.24e-06,
370
+ "loss": 0.2457,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 12.59,
375
+ "learning_rate": 8.194545454545454e-06,
376
+ "loss": 0.1859,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 12.79,
381
+ "learning_rate": 8.14909090909091e-06,
382
+ "loss": 0.1883,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 13.0,
387
+ "learning_rate": 8.103636363636365e-06,
388
+ "loss": 0.1484,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 13.21,
393
+ "learning_rate": 8.058181818181819e-06,
394
+ "loss": 0.1322,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 13.42,
399
+ "learning_rate": 8.012727272727273e-06,
400
+ "loss": 0.1099,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 13.63,
405
+ "learning_rate": 7.967272727272728e-06,
406
+ "loss": 0.1455,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 13.84,
411
+ "learning_rate": 7.921818181818182e-06,
412
+ "loss": 0.1534,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 14.05,
417
+ "learning_rate": 7.876363636363637e-06,
418
+ "loss": 0.1459,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 14.26,
423
+ "learning_rate": 7.830909090909091e-06,
424
+ "loss": 0.1286,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 14.47,
429
+ "learning_rate": 7.785454545454547e-06,
430
+ "loss": 0.0906,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 14.68,
435
+ "learning_rate": 7.74e-06,
436
+ "loss": 0.1186,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 14.89,
441
+ "learning_rate": 7.694545454545456e-06,
442
+ "loss": 0.0855,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 15.1,
447
+ "learning_rate": 7.64909090909091e-06,
448
+ "loss": 0.1135,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 15.31,
453
+ "learning_rate": 7.603636363636364e-06,
454
+ "loss": 0.0613,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 15.52,
459
+ "learning_rate": 7.558181818181819e-06,
460
+ "loss": 0.0682,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 15.73,
465
+ "learning_rate": 7.512727272727273e-06,
466
+ "loss": 0.0667,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 15.94,
471
+ "learning_rate": 7.467272727272727e-06,
472
+ "loss": 0.0815,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 16.15,
477
+ "learning_rate": 7.421818181818183e-06,
478
+ "loss": 0.0599,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 16.36,
483
+ "learning_rate": 7.376363636363637e-06,
484
+ "loss": 0.0639,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 16.57,
489
+ "learning_rate": 7.330909090909092e-06,
490
+ "loss": 0.0578,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 16.78,
495
+ "learning_rate": 7.2854545454545455e-06,
496
+ "loss": 0.0448,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 16.78,
501
+ "eval_loss": 2.7543158531188965,
502
+ "eval_runtime": 502.2301,
503
+ "eval_samples_per_second": 1.268,
504
+ "eval_steps_per_second": 0.159,
505
+ "eval_wer": 5006.62100456621,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 16.99,
510
+ "learning_rate": 7.24e-06,
511
+ "loss": 0.0774,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 17.2,
516
+ "learning_rate": 7.194545454545456e-06,
517
+ "loss": 0.0488,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 17.41,
522
+ "learning_rate": 7.14909090909091e-06,
523
+ "loss": 0.0417,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 17.62,
528
+ "learning_rate": 7.103636363636364e-06,
529
+ "loss": 0.0502,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 17.83,
534
+ "learning_rate": 7.0581818181818185e-06,
535
+ "loss": 0.0548,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 18.04,
540
+ "learning_rate": 7.012727272727273e-06,
541
+ "loss": 0.055,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 18.25,
546
+ "learning_rate": 6.967272727272728e-06,
547
+ "loss": 0.0507,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 18.46,
552
+ "learning_rate": 6.921818181818183e-06,
553
+ "loss": 0.0592,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 18.67,
558
+ "learning_rate": 6.876363636363637e-06,
559
+ "loss": 0.0313,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 18.88,
564
+ "learning_rate": 6.8309090909090915e-06,
565
+ "loss": 0.028,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 19.09,
570
+ "learning_rate": 6.785454545454546e-06,
571
+ "loss": 0.0392,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 19.3,
576
+ "learning_rate": 6.740000000000001e-06,
577
+ "loss": 0.0346,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 19.51,
582
+ "learning_rate": 6.6945454545454544e-06,
583
+ "loss": 0.0347,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 19.72,
588
+ "learning_rate": 6.64909090909091e-06,
589
+ "loss": 0.0279,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 19.93,
594
+ "learning_rate": 6.6036363636363645e-06,
595
+ "loss": 0.0351,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 20.14,
600
+ "learning_rate": 6.558181818181819e-06,
601
+ "loss": 0.0241,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 20.35,
606
+ "learning_rate": 6.512727272727273e-06,
607
+ "loss": 0.0337,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 20.56,
612
+ "learning_rate": 6.467272727272727e-06,
613
+ "loss": 0.0308,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 20.77,
618
+ "learning_rate": 6.421818181818182e-06,
619
+ "loss": 0.0189,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 20.98,
624
+ "learning_rate": 6.3763636363636375e-06,
625
+ "loss": 0.0201,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 21.19,
630
+ "learning_rate": 6.330909090909091e-06,
631
+ "loss": 0.02,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 21.39,
636
+ "learning_rate": 6.285454545454546e-06,
637
+ "loss": 0.0173,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 21.6,
642
+ "learning_rate": 6.24e-06,
643
+ "loss": 0.0218,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 21.81,
648
+ "learning_rate": 6.194545454545455e-06,
649
+ "loss": 0.027,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 22.02,
654
+ "learning_rate": 6.1490909090909104e-06,
655
+ "loss": 0.0188,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 22.23,
660
+ "learning_rate": 6.103636363636364e-06,
661
+ "loss": 0.02,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 22.44,
666
+ "learning_rate": 6.058181818181819e-06,
667
+ "loss": 0.0165,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 22.65,
672
+ "learning_rate": 6.012727272727273e-06,
673
+ "loss": 0.0115,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 22.86,
678
+ "learning_rate": 5.967272727272728e-06,
679
+ "loss": 0.0143,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 23.07,
684
+ "learning_rate": 5.921818181818182e-06,
685
+ "loss": 0.0104,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 23.28,
690
+ "learning_rate": 5.876363636363636e-06,
691
+ "loss": 0.0209,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 23.49,
696
+ "learning_rate": 5.830909090909092e-06,
697
+ "loss": 0.0056,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 23.7,
702
+ "learning_rate": 5.785454545454546e-06,
703
+ "loss": 0.0133,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 23.91,
708
+ "learning_rate": 5.74e-06,
709
+ "loss": 0.0186,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 24.12,
714
+ "learning_rate": 5.694545454545455e-06,
715
+ "loss": 0.0094,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 24.33,
720
+ "learning_rate": 5.649090909090909e-06,
721
+ "loss": 0.0069,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 24.54,
726
+ "learning_rate": 5.603636363636365e-06,
727
+ "loss": 0.0115,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 24.75,
732
+ "learning_rate": 5.5581818181818185e-06,
733
+ "loss": 0.0045,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 24.96,
738
+ "learning_rate": 5.512727272727273e-06,
739
+ "loss": 0.0089,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 25.17,
744
+ "learning_rate": 5.467272727272728e-06,
745
+ "loss": 0.0089,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 25.17,
750
+ "eval_loss": 2.8150265216827393,
751
+ "eval_runtime": 495.3178,
752
+ "eval_samples_per_second": 1.286,
753
+ "eval_steps_per_second": 0.162,
754
+ "eval_wer": 5198.934550989346,
755
+ "step": 3000
756
+ },
757
+ {
758
+ "epoch": 25.38,
759
+ "learning_rate": 5.421818181818182e-06,
760
+ "loss": 0.0111,
761
+ "step": 3025
762
+ },
763
+ {
764
+ "epoch": 25.59,
765
+ "learning_rate": 5.376363636363638e-06,
766
+ "loss": 0.0106,
767
+ "step": 3050
768
+ },
769
+ {
770
+ "epoch": 25.8,
771
+ "learning_rate": 5.330909090909091e-06,
772
+ "loss": 0.0032,
773
+ "step": 3075
774
+ },
775
+ {
776
+ "epoch": 26.01,
777
+ "learning_rate": 5.285454545454546e-06,
778
+ "loss": 0.0041,
779
+ "step": 3100
780
+ },
781
+ {
782
+ "epoch": 26.22,
783
+ "learning_rate": 5.240000000000001e-06,
784
+ "loss": 0.0078,
785
+ "step": 3125
786
+ },
787
+ {
788
+ "epoch": 26.43,
789
+ "learning_rate": 5.194545454545455e-06,
790
+ "loss": 0.0052,
791
+ "step": 3150
792
+ },
793
+ {
794
+ "epoch": 26.64,
795
+ "learning_rate": 5.149090909090909e-06,
796
+ "loss": 0.0067,
797
+ "step": 3175
798
+ },
799
+ {
800
+ "epoch": 26.85,
801
+ "learning_rate": 5.103636363636364e-06,
802
+ "loss": 0.0044,
803
+ "step": 3200
804
+ },
805
+ {
806
+ "epoch": 27.06,
807
+ "learning_rate": 5.058181818181819e-06,
808
+ "loss": 0.0054,
809
+ "step": 3225
810
+ },
811
+ {
812
+ "epoch": 27.27,
813
+ "learning_rate": 5.012727272727274e-06,
814
+ "loss": 0.0025,
815
+ "step": 3250
816
+ },
817
+ {
818
+ "epoch": 27.48,
819
+ "learning_rate": 4.967272727272727e-06,
820
+ "loss": 0.0043,
821
+ "step": 3275
822
+ },
823
+ {
824
+ "epoch": 27.69,
825
+ "learning_rate": 4.921818181818182e-06,
826
+ "loss": 0.0038,
827
+ "step": 3300
828
+ },
829
+ {
830
+ "epoch": 27.9,
831
+ "learning_rate": 4.8763636363636366e-06,
832
+ "loss": 0.004,
833
+ "step": 3325
834
+ },
835
+ {
836
+ "epoch": 28.11,
837
+ "learning_rate": 4.830909090909091e-06,
838
+ "loss": 0.0041,
839
+ "step": 3350
840
+ },
841
+ {
842
+ "epoch": 28.32,
843
+ "learning_rate": 4.785454545454546e-06,
844
+ "loss": 0.0052,
845
+ "step": 3375
846
+ },
847
+ {
848
+ "epoch": 28.53,
849
+ "learning_rate": 4.74e-06,
850
+ "loss": 0.0059,
851
+ "step": 3400
852
+ },
853
+ {
854
+ "epoch": 28.74,
855
+ "learning_rate": 4.694545454545455e-06,
856
+ "loss": 0.0012,
857
+ "step": 3425
858
+ },
859
+ {
860
+ "epoch": 28.95,
861
+ "learning_rate": 4.6490909090909095e-06,
862
+ "loss": 0.0021,
863
+ "step": 3450
864
+ },
865
+ {
866
+ "epoch": 29.16,
867
+ "learning_rate": 4.603636363636364e-06,
868
+ "loss": 0.0034,
869
+ "step": 3475
870
+ },
871
+ {
872
+ "epoch": 29.37,
873
+ "learning_rate": 4.558181818181819e-06,
874
+ "loss": 0.0045,
875
+ "step": 3500
876
+ },
877
+ {
878
+ "epoch": 29.58,
879
+ "learning_rate": 4.512727272727273e-06,
880
+ "loss": 0.0023,
881
+ "step": 3525
882
+ },
883
+ {
884
+ "epoch": 29.79,
885
+ "learning_rate": 4.467272727272728e-06,
886
+ "loss": 0.0054,
887
+ "step": 3550
888
+ },
889
+ {
890
+ "epoch": 29.99,
891
+ "learning_rate": 4.4218181818181825e-06,
892
+ "loss": 0.0021,
893
+ "step": 3575
894
+ },
895
+ {
896
+ "epoch": 30.2,
897
+ "learning_rate": 4.376363636363637e-06,
898
+ "loss": 0.0019,
899
+ "step": 3600
900
+ },
901
+ {
902
+ "epoch": 30.41,
903
+ "learning_rate": 4.330909090909091e-06,
904
+ "loss": 0.004,
905
+ "step": 3625
906
+ },
907
+ {
908
+ "epoch": 30.62,
909
+ "learning_rate": 4.285454545454546e-06,
910
+ "loss": 0.0033,
911
+ "step": 3650
912
+ },
913
+ {
914
+ "epoch": 30.83,
915
+ "learning_rate": 4.24e-06,
916
+ "loss": 0.0015,
917
+ "step": 3675
918
+ },
919
+ {
920
+ "epoch": 31.04,
921
+ "learning_rate": 4.194545454545455e-06,
922
+ "loss": 0.0014,
923
+ "step": 3700
924
+ },
925
+ {
926
+ "epoch": 31.25,
927
+ "learning_rate": 4.149090909090909e-06,
928
+ "loss": 0.0027,
929
+ "step": 3725
930
+ },
931
+ {
932
+ "epoch": 31.46,
933
+ "learning_rate": 4.103636363636364e-06,
934
+ "loss": 0.0035,
935
+ "step": 3750
936
+ },
937
+ {
938
+ "epoch": 31.67,
939
+ "learning_rate": 4.0581818181818184e-06,
940
+ "loss": 0.0034,
941
+ "step": 3775
942
+ },
943
+ {
944
+ "epoch": 31.88,
945
+ "learning_rate": 4.012727272727273e-06,
946
+ "loss": 0.0006,
947
+ "step": 3800
948
+ },
949
+ {
950
+ "epoch": 32.09,
951
+ "learning_rate": 3.967272727272728e-06,
952
+ "loss": 0.0034,
953
+ "step": 3825
954
+ },
955
+ {
956
+ "epoch": 32.3,
957
+ "learning_rate": 3.921818181818182e-06,
958
+ "loss": 0.0011,
959
+ "step": 3850
960
+ },
961
+ {
962
+ "epoch": 32.51,
963
+ "learning_rate": 3.876363636363637e-06,
964
+ "loss": 0.001,
965
+ "step": 3875
966
+ },
967
+ {
968
+ "epoch": 32.72,
969
+ "learning_rate": 3.830909090909091e-06,
970
+ "loss": 0.002,
971
+ "step": 3900
972
+ },
973
+ {
974
+ "epoch": 32.93,
975
+ "learning_rate": 3.7854545454545456e-06,
976
+ "loss": 0.001,
977
+ "step": 3925
978
+ },
979
+ {
980
+ "epoch": 33.14,
981
+ "learning_rate": 3.74e-06,
982
+ "loss": 0.0012,
983
+ "step": 3950
984
+ },
985
+ {
986
+ "epoch": 33.35,
987
+ "learning_rate": 3.694545454545455e-06,
988
+ "loss": 0.0019,
989
+ "step": 3975
990
+ },
991
+ {
992
+ "epoch": 33.56,
993
+ "learning_rate": 3.6490909090909094e-06,
994
+ "loss": 0.0023,
995
+ "step": 4000
996
+ },
997
+ {
998
+ "epoch": 33.56,
999
+ "eval_loss": 2.811387300491333,
1000
+ "eval_runtime": 497.1526,
1001
+ "eval_samples_per_second": 1.281,
1002
+ "eval_steps_per_second": 0.161,
1003
+ "eval_wer": 5089.497716894977,
1004
+ "step": 4000
1005
+ },
1006
+ {
1007
+ "epoch": 33.77,
1008
+ "learning_rate": 3.603636363636364e-06,
1009
+ "loss": 0.0021,
1010
+ "step": 4025
1011
+ },
1012
+ {
1013
+ "epoch": 33.98,
1014
+ "learning_rate": 3.5581818181818186e-06,
1015
+ "loss": 0.0007,
1016
+ "step": 4050
1017
+ },
1018
+ {
1019
+ "epoch": 34.19,
1020
+ "learning_rate": 3.512727272727273e-06,
1021
+ "loss": 0.001,
1022
+ "step": 4075
1023
+ },
1024
+ {
1025
+ "epoch": 34.4,
1026
+ "learning_rate": 3.4672727272727273e-06,
1027
+ "loss": 0.0008,
1028
+ "step": 4100
1029
+ },
1030
+ {
1031
+ "epoch": 34.61,
1032
+ "learning_rate": 3.4218181818181824e-06,
1033
+ "loss": 0.0013,
1034
+ "step": 4125
1035
+ },
1036
+ {
1037
+ "epoch": 34.82,
1038
+ "learning_rate": 3.3763636363636365e-06,
1039
+ "loss": 0.0017,
1040
+ "step": 4150
1041
+ },
1042
+ {
1043
+ "epoch": 35.03,
1044
+ "learning_rate": 3.330909090909091e-06,
1045
+ "loss": 0.0013,
1046
+ "step": 4175
1047
+ },
1048
+ {
1049
+ "epoch": 35.24,
1050
+ "learning_rate": 3.2854545454545457e-06,
1051
+ "loss": 0.0015,
1052
+ "step": 4200
1053
+ },
1054
+ {
1055
+ "epoch": 35.45,
1056
+ "learning_rate": 3.2400000000000003e-06,
1057
+ "loss": 0.0009,
1058
+ "step": 4225
1059
+ },
1060
+ {
1061
+ "epoch": 35.66,
1062
+ "learning_rate": 3.1945454545454545e-06,
1063
+ "loss": 0.0016,
1064
+ "step": 4250
1065
+ },
1066
+ {
1067
+ "epoch": 35.87,
1068
+ "learning_rate": 3.1490909090909095e-06,
1069
+ "loss": 0.0013,
1070
+ "step": 4275
1071
+ },
1072
+ {
1073
+ "epoch": 36.08,
1074
+ "learning_rate": 3.1036363636363637e-06,
1075
+ "loss": 0.0024,
1076
+ "step": 4300
1077
+ },
1078
+ {
1079
+ "epoch": 36.29,
1080
+ "learning_rate": 3.0581818181818183e-06,
1081
+ "loss": 0.0004,
1082
+ "step": 4325
1083
+ },
1084
+ {
1085
+ "epoch": 36.5,
1086
+ "learning_rate": 3.012727272727273e-06,
1087
+ "loss": 0.0012,
1088
+ "step": 4350
1089
+ },
1090
+ {
1091
+ "epoch": 36.71,
1092
+ "learning_rate": 2.9672727272727275e-06,
1093
+ "loss": 0.0004,
1094
+ "step": 4375
1095
+ },
1096
+ {
1097
+ "epoch": 36.92,
1098
+ "learning_rate": 2.9218181818181825e-06,
1099
+ "loss": 0.0011,
1100
+ "step": 4400
1101
+ },
1102
+ {
1103
+ "epoch": 37.13,
1104
+ "learning_rate": 2.8763636363636367e-06,
1105
+ "loss": 0.0012,
1106
+ "step": 4425
1107
+ },
1108
+ {
1109
+ "epoch": 37.34,
1110
+ "learning_rate": 2.8309090909090913e-06,
1111
+ "loss": 0.0009,
1112
+ "step": 4450
1113
+ },
1114
+ {
1115
+ "epoch": 37.55,
1116
+ "learning_rate": 2.7854545454545454e-06,
1117
+ "loss": 0.0009,
1118
+ "step": 4475
1119
+ },
1120
+ {
1121
+ "epoch": 37.76,
1122
+ "learning_rate": 2.7400000000000004e-06,
1123
+ "loss": 0.0004,
1124
+ "step": 4500
1125
+ },
1126
+ {
1127
+ "epoch": 37.97,
1128
+ "learning_rate": 2.6945454545454546e-06,
1129
+ "loss": 0.0003,
1130
+ "step": 4525
1131
+ },
1132
+ {
1133
+ "epoch": 38.18,
1134
+ "learning_rate": 2.6490909090909096e-06,
1135
+ "loss": 0.0011,
1136
+ "step": 4550
1137
+ },
1138
+ {
1139
+ "epoch": 38.38,
1140
+ "learning_rate": 2.603636363636364e-06,
1141
+ "loss": 0.0003,
1142
+ "step": 4575
1143
+ },
1144
+ {
1145
+ "epoch": 38.59,
1146
+ "learning_rate": 2.5581818181818184e-06,
1147
+ "loss": 0.0003,
1148
+ "step": 4600
1149
+ },
1150
+ {
1151
+ "epoch": 38.8,
1152
+ "learning_rate": 2.512727272727273e-06,
1153
+ "loss": 0.0013,
1154
+ "step": 4625
1155
+ },
1156
+ {
1157
+ "epoch": 39.01,
1158
+ "learning_rate": 2.4672727272727276e-06,
1159
+ "loss": 0.0003,
1160
+ "step": 4650
1161
+ },
1162
+ {
1163
+ "epoch": 39.22,
1164
+ "learning_rate": 2.421818181818182e-06,
1165
+ "loss": 0.0002,
1166
+ "step": 4675
1167
+ },
1168
+ {
1169
+ "epoch": 39.43,
1170
+ "learning_rate": 2.376363636363637e-06,
1171
+ "loss": 0.0003,
1172
+ "step": 4700
1173
+ },
1174
+ {
1175
+ "epoch": 39.64,
1176
+ "learning_rate": 2.3309090909090914e-06,
1177
+ "loss": 0.0015,
1178
+ "step": 4725
1179
+ },
1180
+ {
1181
+ "epoch": 39.85,
1182
+ "learning_rate": 2.2854545454545456e-06,
1183
+ "loss": 0.0006,
1184
+ "step": 4750
1185
+ },
1186
+ {
1187
+ "epoch": 40.06,
1188
+ "learning_rate": 2.24e-06,
1189
+ "loss": 0.001,
1190
+ "step": 4775
1191
+ },
1192
+ {
1193
+ "epoch": 40.27,
1194
+ "learning_rate": 2.1945454545454547e-06,
1195
+ "loss": 0.0003,
1196
+ "step": 4800
1197
+ },
1198
+ {
1199
+ "epoch": 40.48,
1200
+ "learning_rate": 2.1490909090909093e-06,
1201
+ "loss": 0.0004,
1202
+ "step": 4825
1203
+ },
1204
+ {
1205
+ "epoch": 40.69,
1206
+ "learning_rate": 2.103636363636364e-06,
1207
+ "loss": 0.0003,
1208
+ "step": 4850
1209
+ },
1210
+ {
1211
+ "epoch": 40.9,
1212
+ "learning_rate": 2.0581818181818185e-06,
1213
+ "loss": 0.001,
1214
+ "step": 4875
1215
+ },
1216
+ {
1217
+ "epoch": 41.11,
1218
+ "learning_rate": 2.0127272727272727e-06,
1219
+ "loss": 0.0003,
1220
+ "step": 4900
1221
+ },
1222
+ {
1223
+ "epoch": 41.32,
1224
+ "learning_rate": 1.9672727272727273e-06,
1225
+ "loss": 0.0009,
1226
+ "step": 4925
1227
+ },
1228
+ {
1229
+ "epoch": 41.53,
1230
+ "learning_rate": 1.921818181818182e-06,
1231
+ "loss": 0.0004,
1232
+ "step": 4950
1233
+ },
1234
+ {
1235
+ "epoch": 41.74,
1236
+ "learning_rate": 1.8763636363636365e-06,
1237
+ "loss": 0.0003,
1238
+ "step": 4975
1239
+ },
1240
+ {
1241
+ "epoch": 41.95,
1242
+ "learning_rate": 1.8309090909090909e-06,
1243
+ "loss": 0.0006,
1244
+ "step": 5000
1245
+ },
1246
+ {
1247
+ "epoch": 41.95,
1248
+ "eval_loss": 2.853930950164795,
1249
+ "eval_runtime": 495.4476,
1250
+ "eval_samples_per_second": 1.286,
1251
+ "eval_steps_per_second": 0.161,
1252
+ "eval_wer": 5118.645357686453,
1253
+ "step": 5000
1254
+ }
1255
+ ],
1256
+ "max_steps": 6000,
1257
+ "num_train_epochs": 51,
1258
+ "total_flos": 2.309318090440704e+19,
1259
+ "trial_name": null,
1260
+ "trial_params": null
1261
+ }
checkpoint-5000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cbeda0929f5bade1f2682f85820a8eb42b209173de61e856a1e94b350f2c7d1
3
+ size 4091
checkpoint-6000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
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": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.31.0",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-6000/generation_config.json ADDED
@@ -0,0 +1,263 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "max_initial_timestamp_index": 1,
164
+ "max_length": 448,
165
+ "no_timestamps_token_id": 50363,
166
+ "pad_token_id": 50257,
167
+ "return_timestamps": false,
168
+ "suppress_tokens": [
169
+ 1,
170
+ 2,
171
+ 7,
172
+ 8,
173
+ 9,
174
+ 10,
175
+ 14,
176
+ 25,
177
+ 26,
178
+ 27,
179
+ 28,
180
+ 29,
181
+ 31,
182
+ 58,
183
+ 59,
184
+ 60,
185
+ 61,
186
+ 62,
187
+ 63,
188
+ 90,
189
+ 91,
190
+ 92,
191
+ 93,
192
+ 359,
193
+ 503,
194
+ 522,
195
+ 542,
196
+ 873,
197
+ 893,
198
+ 902,
199
+ 918,
200
+ 922,
201
+ 931,
202
+ 1350,
203
+ 1853,
204
+ 1982,
205
+ 2460,
206
+ 2627,
207
+ 3246,
208
+ 3253,
209
+ 3268,
210
+ 3536,
211
+ 3846,
212
+ 3961,
213
+ 4183,
214
+ 4667,
215
+ 6585,
216
+ 6647,
217
+ 7273,
218
+ 9061,
219
+ 9383,
220
+ 10428,
221
+ 10929,
222
+ 11938,
223
+ 12033,
224
+ 12331,
225
+ 12562,
226
+ 13793,
227
+ 14157,
228
+ 14635,
229
+ 15265,
230
+ 15618,
231
+ 16553,
232
+ 16604,
233
+ 18362,
234
+ 18956,
235
+ 20075,
236
+ 21675,
237
+ 22520,
238
+ 26130,
239
+ 26161,
240
+ 26435,
241
+ 28279,
242
+ 29464,
243
+ 31650,
244
+ 32302,
245
+ 32470,
246
+ 36865,
247
+ 42863,
248
+ 47425,
249
+ 49870,
250
+ 50254,
251
+ 50258,
252
+ 50358,
253
+ 50359,
254
+ 50360,
255
+ 50361,
256
+ 50362
257
+ ],
258
+ "task_to_id": {
259
+ "transcribe": 50359,
260
+ "translate": 50358
261
+ },
262
+ "transformers_version": "4.31.0"
263
+ }
checkpoint-6000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7481569d94f47af5286680f0c949a423af156ac3e07e3e70a77ecb6486869529
3
+ size 1934161157
checkpoint-6000/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-6000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54424e948419e87a2da35133a7aa3823e2b5a3e892c05c825a0f9ed005359e2a
3
+ size 967102729