Hugging Face's logo

zionia
/
whisper-small-isizulu-noisy

Automatic Speech Recognition
TensorBoard
Safetensors
Zulu
whisper
south-african-languages
Model card Files
xet
Metrics Community
zionia commited on
Commit
e0f597c
·
verified ·
1 Parent(s): 5ca8f29

Training complete - 2 datasets combined

Browse files
Files changed (30) hide show
  1. checkpoint-330/config.json +60 -0
  2. checkpoint-330/generation_config.json +255 -0
  3. checkpoint-330/model.safetensors +3 -0
  4. checkpoint-330/optimizer.pt +3 -0
  5. checkpoint-330/preprocessor_config.json +15 -0
  6. checkpoint-330/rng_state.pth +3 -0
  7. checkpoint-330/scaler.pt +3 -0
  8. checkpoint-330/scheduler.pt +3 -0
  9. checkpoint-330/trainer_state.json +355 -0
  10. checkpoint-330/training_args.bin +3 -0
  11. checkpoint-462/config.json +60 -0
  12. checkpoint-462/generation_config.json +255 -0
  13. checkpoint-462/model.safetensors +3 -0
  14. checkpoint-462/optimizer.pt +3 -0
  15. checkpoint-462/preprocessor_config.json +15 -0
  16. checkpoint-462/rng_state.pth +3 -0
  17. checkpoint-462/scaler.pt +3 -0
  18. checkpoint-462/scheduler.pt +3 -0
  19. checkpoint-462/trainer_state.json +482 -0
  20. checkpoint-462/training_args.bin +3 -0
  21. checkpoint-495/config.json +60 -0
  22. checkpoint-495/generation_config.json +255 -0
  23. checkpoint-495/model.safetensors +3 -0
  24. checkpoint-495/optimizer.pt +3 -0
  25. checkpoint-495/preprocessor_config.json +15 -0
  26. checkpoint-495/rng_state.pth +3 -0
  27. checkpoint-495/scaler.pt +3 -0
  28. checkpoint-495/scheduler.pt +3 -0
  29. checkpoint-495/trainer_state.json +512 -0
  30. checkpoint-495/training_args.bin +3 -0
checkpoint-330/config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "activation_function": "gelu",
4
+ "apply_spec_augment": false,
5
+ "architectures": [
6
+ "WhisperForConditionalGeneration"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "begin_suppress_tokens": null,
10
+ "bos_token_id": 50257,
11
+ "classifier_proj_size": 256,
12
+ "d_model": 768,
13
+ "decoder_attention_heads": 12,
14
+ "decoder_ffn_dim": 3072,
15
+ "decoder_layerdrop": 0.0,
16
+ "decoder_layers": 12,
17
+ "decoder_start_token_id": 50258,
18
+ "dropout": 0.0,
19
+ "dtype": "float32",
20
+ "encoder_attention_heads": 12,
21
+ "encoder_ffn_dim": 3072,
22
+ "encoder_layerdrop": 0.0,
23
+ "encoder_layers": 12,
24
+ "eos_token_id": 50257,
25
+ "forced_decoder_ids": [
26
+ [
27
+ 1,
28
+ 50259
29
+ ],
30
+ [
31
+ 2,
32
+ 50359
33
+ ],
34
+ [
35
+ 3,
36
+ 50363
37
+ ]
38
+ ],
39
+ "init_std": 0.02,
40
+ "is_encoder_decoder": true,
41
+ "mask_feature_length": 10,
42
+ "mask_feature_min_masks": 0,
43
+ "mask_feature_prob": 0.0,
44
+ "mask_time_length": 10,
45
+ "mask_time_min_masks": 2,
46
+ "mask_time_prob": 0.05,
47
+ "max_length": null,
48
+ "max_source_positions": 1500,
49
+ "max_target_positions": 448,
50
+ "median_filter_width": 7,
51
+ "model_type": "whisper",
52
+ "num_hidden_layers": 12,
53
+ "num_mel_bins": 80,
54
+ "pad_token_id": 50257,
55
+ "scale_embedding": false,
56
+ "transformers_version": "4.57.0",
57
+ "use_cache": true,
58
+ "use_weighted_layer_sum": false,
59
+ "vocab_size": 51865
60
+ }
checkpoint-330/generation_config.json ADDED
@@ -0,0 +1,255 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": null,
52
+ "is_multilingual": true,
53
+ "lang_to_id": {
54
+ "<|af|>": 50327,
55
+ "<|am|>": 50334,
56
+ "<|ar|>": 50272,
57
+ "<|as|>": 50350,
58
+ "<|az|>": 50304,
59
+ "<|ba|>": 50355,
60
+ "<|be|>": 50330,
61
+ "<|bg|>": 50292,
62
+ "<|bn|>": 50302,
63
+ "<|bo|>": 50347,
64
+ "<|br|>": 50309,
65
+ "<|bs|>": 50315,
66
+ "<|ca|>": 50270,
67
+ "<|cs|>": 50283,
68
+ "<|cy|>": 50297,
69
+ "<|da|>": 50285,
70
+ "<|de|>": 50261,
71
+ "<|el|>": 50281,
72
+ "<|en|>": 50259,
73
+ "<|es|>": 50262,
74
+ "<|et|>": 50307,
75
+ "<|eu|>": 50310,
76
+ "<|fa|>": 50300,
77
+ "<|fi|>": 50277,
78
+ "<|fo|>": 50338,
79
+ "<|fr|>": 50265,
80
+ "<|gl|>": 50319,
81
+ "<|gu|>": 50333,
82
+ "<|haw|>": 50352,
83
+ "<|ha|>": 50354,
84
+ "<|he|>": 50279,
85
+ "<|hi|>": 50276,
86
+ "<|hr|>": 50291,
87
+ "<|ht|>": 50339,
88
+ "<|hu|>": 50286,
89
+ "<|hy|>": 50312,
90
+ "<|id|>": 50275,
91
+ "<|is|>": 50311,
92
+ "<|it|>": 50274,
93
+ "<|ja|>": 50266,
94
+ "<|jw|>": 50356,
95
+ "<|ka|>": 50329,
96
+ "<|kk|>": 50316,
97
+ "<|km|>": 50323,
98
+ "<|kn|>": 50306,
99
+ "<|ko|>": 50264,
100
+ "<|la|>": 50294,
101
+ "<|lb|>": 50345,
102
+ "<|ln|>": 50353,
103
+ "<|lo|>": 50336,
104
+ "<|lt|>": 50293,
105
+ "<|lv|>": 50301,
106
+ "<|mg|>": 50349,
107
+ "<|mi|>": 50295,
108
+ "<|mk|>": 50308,
109
+ "<|ml|>": 50296,
110
+ "<|mn|>": 50314,
111
+ "<|mr|>": 50320,
112
+ "<|ms|>": 50282,
113
+ "<|mt|>": 50343,
114
+ "<|my|>": 50346,
115
+ "<|ne|>": 50313,
116
+ "<|nl|>": 50271,
117
+ "<|nn|>": 50342,
118
+ "<|no|>": 50288,
119
+ "<|oc|>": 50328,
120
+ "<|pa|>": 50321,
121
+ "<|pl|>": 50269,
122
+ "<|ps|>": 50340,
123
+ "<|pt|>": 50267,
124
+ "<|ro|>": 50284,
125
+ "<|ru|>": 50263,
126
+ "<|sa|>": 50344,
127
+ "<|sd|>": 50332,
128
+ "<|si|>": 50322,
129
+ "<|sk|>": 50298,
130
+ "<|sl|>": 50305,
131
+ "<|sn|>": 50324,
132
+ "<|so|>": 50326,
133
+ "<|sq|>": 50317,
134
+ "<|sr|>": 50303,
135
+ "<|su|>": 50357,
136
+ "<|sv|>": 50273,
137
+ "<|sw|>": 50318,
138
+ "<|ta|>": 50287,
139
+ "<|te|>": 50299,
140
+ "<|tg|>": 50331,
141
+ "<|th|>": 50289,
142
+ "<|tk|>": 50341,
143
+ "<|tl|>": 50348,
144
+ "<|tr|>": 50268,
145
+ "<|tt|>": 50351,
146
+ "<|uk|>": 50280,
147
+ "<|ur|>": 50290,
148
+ "<|uz|>": 50337,
149
+ "<|vi|>": 50278,
150
+ "<|yi|>": 50335,
151
+ "<|yo|>": 50325,
152
+ "<|zh|>": 50260
153
+ },
154
+ "language": "swahili",
155
+ "max_initial_timestamp_index": 50,
156
+ "max_length": 448,
157
+ "no_timestamps_token_id": 50363,
158
+ "pad_token_id": 50257,
159
+ "prev_sot_token_id": 50361,
160
+ "return_timestamps": false,
161
+ "suppress_tokens": [
162
+ 1,
163
+ 2,
164
+ 7,
165
+ 8,
166
+ 9,
167
+ 10,
168
+ 14,
169
+ 25,
170
+ 26,
171
+ 27,
172
+ 28,
173
+ 29,
174
+ 31,
175
+ 58,
176
+ 59,
177
+ 60,
178
+ 61,
179
+ 62,
180
+ 63,
181
+ 90,
182
+ 91,
183
+ 92,
184
+ 93,
185
+ 359,
186
+ 503,
187
+ 522,
188
+ 542,
189
+ 873,
190
+ 893,
191
+ 902,
192
+ 918,
193
+ 922,
194
+ 931,
195
+ 1350,
196
+ 1853,
197
+ 1982,
198
+ 2460,
199
+ 2627,
200
+ 3246,
201
+ 3253,
202
+ 3268,
203
+ 3536,
204
+ 3846,
205
+ 3961,
206
+ 4183,
207
+ 4667,
208
+ 6585,
209
+ 6647,
210
+ 7273,
211
+ 9061,
212
+ 9383,
213
+ 10428,
214
+ 10929,
215
+ 11938,
216
+ 12033,
217
+ 12331,
218
+ 12562,
219
+ 13793,
220
+ 14157,
221
+ 14635,
222
+ 15265,
223
+ 15618,
224
+ 16553,
225
+ 16604,
226
+ 18362,
227
+ 18956,
228
+ 20075,
229
+ 21675,
230
+ 22520,
231
+ 26130,
232
+ 26161,
233
+ 26435,
234
+ 28279,
235
+ 29464,
236
+ 31650,
237
+ 32302,
238
+ 32470,
239
+ 36865,
240
+ 42863,
241
+ 47425,
242
+ 49870,
243
+ 50254,
244
+ 50258,
245
+ 50360,
246
+ 50361,
247
+ 50362
248
+ ],
249
+ "task": "transcribe",
250
+ "task_to_id": {
251
+ "transcribe": 50359,
252
+ "translate": 50358
253
+ },
254
+ "transformers_version": "4.57.0"
255
+ }
checkpoint-330/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ae7cf39540e7551f6769b8ffd8f90adf34b48224d390f5946203b778cd55582
3
+ size 966995080
checkpoint-330/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6efca3ad6ac170c17ad91d3b092a580c98c88526093d6442ab5bd1100a306f2b
3
+ size 1925071165
checkpoint-330/preprocessor_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "dither": 0.0,
4
+ "feature_extractor_type": "WhisperFeatureExtractor",
5
+ "feature_size": 80,
6
+ "hop_length": 160,
7
+ "n_fft": 400,
8
+ "n_samples": 480000,
9
+ "nb_max_frames": 3000,
10
+ "padding_side": "right",
11
+ "padding_value": 0.0,
12
+ "processor_class": "WhisperProcessor",
13
+ "return_attention_mask": false,
14
+ "sampling_rate": 16000
15
+ }
checkpoint-330/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14fa9c1575447e15beea91192a01c08b828f8387f17428b3b445a1f25e5a5760
3
+ size 14645
checkpoint-330/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74a0d07f4422fd1f47be90701ea6b19f0237c56d29cfd24a3f4d2735744dd1dc
3
+ size 1383
checkpoint-330/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ecf57c6467ddfcfc69c8bbab6e182eb403b7e08d0079e50755e25a11f34d2ceb
3
+ size 1465
checkpoint-330/trainer_state.json ADDED
@@ -0,0 +1,355 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 330,
3
+ "best_metric": 63.56830941983781,
4
+ "best_model_checkpoint": "./whisper-small-isizulu-noisy/checkpoint-330",
5
+ "epoch": 10.0,
6
+ "eval_steps": 500,
7
+ "global_step": 330,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.30303030303030304,
14
+ "grad_norm": 26.396528244018555,
15
+ "learning_rate": 9.000000000000001e-07,
16
+ "loss": 3.5466,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.6060606060606061,
21
+ "grad_norm": 19.094928741455078,
22
+ "learning_rate": 1.9000000000000002e-06,
23
+ "loss": 3.2038,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.9090909090909091,
28
+ "grad_norm": 14.402777671813965,
29
+ "learning_rate": 2.9e-06,
30
+ "loss": 2.7094,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 1.0,
35
+ "eval_loss": 2.2550225257873535,
36
+ "eval_runtime": 60.9646,
37
+ "eval_samples_per_second": 4.314,
38
+ "eval_steps_per_second": 0.279,
39
+ "eval_wer": 108.79600748596381,
40
+ "step": 33
41
+ },
42
+ {
43
+ "epoch": 1.2121212121212122,
44
+ "grad_norm": 12.169499397277832,
45
+ "learning_rate": 3.900000000000001e-06,
46
+ "loss": 2.188,
47
+ "step": 40
48
+ },
49
+ {
50
+ "epoch": 1.5151515151515151,
51
+ "grad_norm": 12.002925872802734,
52
+ "learning_rate": 4.9000000000000005e-06,
53
+ "loss": 1.8277,
54
+ "step": 50
55
+ },
56
+ {
57
+ "epoch": 1.8181818181818183,
58
+ "grad_norm": 10.814740180969238,
59
+ "learning_rate": 5.9e-06,
60
+ "loss": 1.5701,
61
+ "step": 60
62
+ },
63
+ {
64
+ "epoch": 2.0,
65
+ "eval_loss": 1.405279517173767,
66
+ "eval_runtime": 44.4698,
67
+ "eval_samples_per_second": 5.914,
68
+ "eval_steps_per_second": 0.382,
69
+ "eval_wer": 90.1434809731753,
70
+ "step": 66
71
+ },
72
+ {
73
+ "epoch": 2.121212121212121,
74
+ "grad_norm": 9.393446922302246,
75
+ "learning_rate": 6.9e-06,
76
+ "loss": 1.4165,
77
+ "step": 70
78
+ },
79
+ {
80
+ "epoch": 2.4242424242424243,
81
+ "grad_norm": 9.702391624450684,
82
+ "learning_rate": 7.9e-06,
83
+ "loss": 1.0921,
84
+ "step": 80
85
+ },
86
+ {
87
+ "epoch": 2.7272727272727275,
88
+ "grad_norm": 8.38471508026123,
89
+ "learning_rate": 8.900000000000001e-06,
90
+ "loss": 1.0347,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 3.0,
95
+ "eval_loss": 1.0759814977645874,
96
+ "eval_runtime": 53.0857,
97
+ "eval_samples_per_second": 4.954,
98
+ "eval_steps_per_second": 0.32,
99
+ "eval_wer": 103.74298190892077,
100
+ "step": 99
101
+ },
102
+ {
103
+ "epoch": 3.0303030303030303,
104
+ "grad_norm": 6.726572036743164,
105
+ "learning_rate": 9.9e-06,
106
+ "loss": 0.8997,
107
+ "step": 100
108
+ },
109
+ {
110
+ "epoch": 3.3333333333333335,
111
+ "grad_norm": 6.715061187744141,
112
+ "learning_rate": 9.772151898734179e-06,
113
+ "loss": 0.6359,
114
+ "step": 110
115
+ },
116
+ {
117
+ "epoch": 3.6363636363636362,
118
+ "grad_norm": 6.393348217010498,
119
+ "learning_rate": 9.518987341772152e-06,
120
+ "loss": 0.5551,
121
+ "step": 120
122
+ },
123
+ {
124
+ "epoch": 3.9393939393939394,
125
+ "grad_norm": 7.166343688964844,
126
+ "learning_rate": 9.265822784810128e-06,
127
+ "loss": 0.5594,
128
+ "step": 130
129
+ },
130
+ {
131
+ "epoch": 4.0,
132
+ "eval_loss": 0.9267733693122864,
133
+ "eval_runtime": 48.7558,
134
+ "eval_samples_per_second": 5.394,
135
+ "eval_steps_per_second": 0.349,
136
+ "eval_wer": 100.62383031815347,
137
+ "step": 132
138
+ },
139
+ {
140
+ "epoch": 4.242424242424242,
141
+ "grad_norm": 6.093661308288574,
142
+ "learning_rate": 9.012658227848103e-06,
143
+ "loss": 0.3714,
144
+ "step": 140
145
+ },
146
+ {
147
+ "epoch": 4.545454545454545,
148
+ "grad_norm": 7.345926761627197,
149
+ "learning_rate": 8.759493670886077e-06,
150
+ "loss": 0.2967,
151
+ "step": 150
152
+ },
153
+ {
154
+ "epoch": 4.848484848484849,
155
+ "grad_norm": 4.838716506958008,
156
+ "learning_rate": 8.50632911392405e-06,
157
+ "loss": 0.2742,
158
+ "step": 160
159
+ },
160
+ {
161
+ "epoch": 5.0,
162
+ "eval_loss": 0.9051285982131958,
163
+ "eval_runtime": 46.5026,
164
+ "eval_samples_per_second": 5.656,
165
+ "eval_steps_per_second": 0.366,
166
+ "eval_wer": 80.22457891453524,
167
+ "step": 165
168
+ },
169
+ {
170
+ "epoch": 5.151515151515151,
171
+ "grad_norm": 3.3409740924835205,
172
+ "learning_rate": 8.253164556962026e-06,
173
+ "loss": 0.2169,
174
+ "step": 170
175
+ },
176
+ {
177
+ "epoch": 5.454545454545454,
178
+ "grad_norm": 3.894425868988037,
179
+ "learning_rate": 8.000000000000001e-06,
180
+ "loss": 0.1534,
181
+ "step": 180
182
+ },
183
+ {
184
+ "epoch": 5.757575757575758,
185
+ "grad_norm": 4.993983745574951,
186
+ "learning_rate": 7.746835443037975e-06,
187
+ "loss": 0.1504,
188
+ "step": 190
189
+ },
190
+ {
191
+ "epoch": 6.0,
192
+ "eval_loss": 0.8892419934272766,
193
+ "eval_runtime": 50.604,
194
+ "eval_samples_per_second": 5.197,
195
+ "eval_steps_per_second": 0.336,
196
+ "eval_wer": 89.58203368683718,
197
+ "step": 198
198
+ },
199
+ {
200
+ "epoch": 6.0606060606060606,
201
+ "grad_norm": 2.509334087371826,
202
+ "learning_rate": 7.49367088607595e-06,
203
+ "loss": 0.1417,
204
+ "step": 200
205
+ },
206
+ {
207
+ "epoch": 6.363636363636363,
208
+ "grad_norm": 3.116098403930664,
209
+ "learning_rate": 7.240506329113924e-06,
210
+ "loss": 0.0785,
211
+ "step": 210
212
+ },
213
+ {
214
+ "epoch": 6.666666666666667,
215
+ "grad_norm": 2.244513750076294,
216
+ "learning_rate": 6.987341772151899e-06,
217
+ "loss": 0.072,
218
+ "step": 220
219
+ },
220
+ {
221
+ "epoch": 6.96969696969697,
222
+ "grad_norm": 2.546046495437622,
223
+ "learning_rate": 6.734177215189874e-06,
224
+ "loss": 0.065,
225
+ "step": 230
226
+ },
227
+ {
228
+ "epoch": 7.0,
229
+ "eval_loss": 0.9148803949356079,
230
+ "eval_runtime": 46.4915,
231
+ "eval_samples_per_second": 5.657,
232
+ "eval_steps_per_second": 0.366,
233
+ "eval_wer": 81.16032439176544,
234
+ "step": 231
235
+ },
236
+ {
237
+ "epoch": 7.2727272727272725,
238
+ "grad_norm": 1.2738147974014282,
239
+ "learning_rate": 6.481012658227849e-06,
240
+ "loss": 0.0444,
241
+ "step": 240
242
+ },
243
+ {
244
+ "epoch": 7.575757575757576,
245
+ "grad_norm": 1.1117249727249146,
246
+ "learning_rate": 6.227848101265823e-06,
247
+ "loss": 0.0374,
248
+ "step": 250
249
+ },
250
+ {
251
+ "epoch": 7.878787878787879,
252
+ "grad_norm": 1.59920072555542,
253
+ "learning_rate": 5.974683544303798e-06,
254
+ "loss": 0.0369,
255
+ "step": 260
256
+ },
257
+ {
258
+ "epoch": 8.0,
259
+ "eval_loss": 0.9126399755477905,
260
+ "eval_runtime": 44.4651,
261
+ "eval_samples_per_second": 5.915,
262
+ "eval_steps_per_second": 0.382,
263
+ "eval_wer": 66.126013724267,
264
+ "step": 264
265
+ },
266
+ {
267
+ "epoch": 8.181818181818182,
268
+ "grad_norm": 0.7837973237037659,
269
+ "learning_rate": 5.721518987341773e-06,
270
+ "loss": 0.0269,
271
+ "step": 270
272
+ },
273
+ {
274
+ "epoch": 8.484848484848484,
275
+ "grad_norm": 0.7720461487770081,
276
+ "learning_rate": 5.468354430379748e-06,
277
+ "loss": 0.0198,
278
+ "step": 280
279
+ },
280
+ {
281
+ "epoch": 8.787878787878787,
282
+ "grad_norm": 2.480095386505127,
283
+ "learning_rate": 5.2151898734177216e-06,
284
+ "loss": 0.0215,
285
+ "step": 290
286
+ },
287
+ {
288
+ "epoch": 9.0,
289
+ "eval_loss": 0.9265649318695068,
290
+ "eval_runtime": 44.2469,
291
+ "eval_samples_per_second": 5.944,
292
+ "eval_steps_per_second": 0.384,
293
+ "eval_wer": 66.9369931378665,
294
+ "step": 297
295
+ },
296
+ {
297
+ "epoch": 9.090909090909092,
298
+ "grad_norm": 0.400156170129776,
299
+ "learning_rate": 4.962025316455696e-06,
300
+ "loss": 0.0186,
301
+ "step": 300
302
+ },
303
+ {
304
+ "epoch": 9.393939393939394,
305
+ "grad_norm": 0.3897418975830078,
306
+ "learning_rate": 4.7088607594936715e-06,
307
+ "loss": 0.0107,
308
+ "step": 310
309
+ },
310
+ {
311
+ "epoch": 9.696969696969697,
312
+ "grad_norm": 4.099902153015137,
313
+ "learning_rate": 4.455696202531646e-06,
314
+ "loss": 0.0136,
315
+ "step": 320
316
+ },
317
+ {
318
+ "epoch": 10.0,
319
+ "grad_norm": 0.33050861954689026,
320
+ "learning_rate": 4.202531645569621e-06,
321
+ "loss": 0.0129,
322
+ "step": 330
323
+ },
324
+ {
325
+ "epoch": 10.0,
326
+ "eval_loss": 0.9320704936981201,
327
+ "eval_runtime": 44.2449,
328
+ "eval_samples_per_second": 5.944,
329
+ "eval_steps_per_second": 0.384,
330
+ "eval_wer": 63.56830941983781,
331
+ "step": 330
332
+ }
333
+ ],
334
+ "logging_steps": 10,
335
+ "max_steps": 495,
336
+ "num_input_tokens_seen": 0,
337
+ "num_train_epochs": 15,
338
+ "save_steps": 500,
339
+ "stateful_callbacks": {
340
+ "TrainerControl": {
341
+ "args": {
342
+ "should_epoch_stop": false,
343
+ "should_evaluate": false,
344
+ "should_log": false,
345
+ "should_save": true,
346
+ "should_training_stop": false
347
+ },
348
+ "attributes": {}
349
+ }
350
+ },
351
+ "total_flos": 3.03014670336e+18,
352
+ "train_batch_size": 32,
353
+ "trial_name": null,
354
+ "trial_params": null
355
+ }
checkpoint-330/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68d60f46e270ca874141fea17fa80ade9cf0002e14e72e1e1cd28d219ddaf599
3
+ size 6033
checkpoint-462/config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "activation_function": "gelu",
4
+ "apply_spec_augment": false,
5
+ "architectures": [
6
+ "WhisperForConditionalGeneration"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "begin_suppress_tokens": null,
10
+ "bos_token_id": 50257,
11
+ "classifier_proj_size": 256,
12
+ "d_model": 768,
13
+ "decoder_attention_heads": 12,
14
+ "decoder_ffn_dim": 3072,
15
+ "decoder_layerdrop": 0.0,
16
+ "decoder_layers": 12,
17
+ "decoder_start_token_id": 50258,
18
+ "dropout": 0.0,
19
+ "dtype": "float32",
20
+ "encoder_attention_heads": 12,
21
+ "encoder_ffn_dim": 3072,
22
+ "encoder_layerdrop": 0.0,
23
+ "encoder_layers": 12,
24
+ "eos_token_id": 50257,
25
+ "forced_decoder_ids": [
26
+ [
27
+ 1,
28
+ 50259
29
+ ],
30
+ [
31
+ 2,
32
+ 50359
33
+ ],
34
+ [
35
+ 3,
36
+ 50363
37
+ ]
38
+ ],
39
+ "init_std": 0.02,
40
+ "is_encoder_decoder": true,
41
+ "mask_feature_length": 10,
42
+ "mask_feature_min_masks": 0,
43
+ "mask_feature_prob": 0.0,
44
+ "mask_time_length": 10,
45
+ "mask_time_min_masks": 2,
46
+ "mask_time_prob": 0.05,
47
+ "max_length": null,
48
+ "max_source_positions": 1500,
49
+ "max_target_positions": 448,
50
+ "median_filter_width": 7,
51
+ "model_type": "whisper",
52
+ "num_hidden_layers": 12,
53
+ "num_mel_bins": 80,
54
+ "pad_token_id": 50257,
55
+ "scale_embedding": false,
56
+ "transformers_version": "4.57.0",
57
+ "use_cache": true,
58
+ "use_weighted_layer_sum": false,
59
+ "vocab_size": 51865
60
+ }
checkpoint-462/generation_config.json ADDED
@@ -0,0 +1,255 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": null,
52
+ "is_multilingual": true,
53
+ "lang_to_id": {
54
+ "<|af|>": 50327,
55
+ "<|am|>": 50334,
56
+ "<|ar|>": 50272,
57
+ "<|as|>": 50350,
58
+ "<|az|>": 50304,
59
+ "<|ba|>": 50355,
60
+ "<|be|>": 50330,
61
+ "<|bg|>": 50292,
62
+ "<|bn|>": 50302,
63
+ "<|bo|>": 50347,
64
+ "<|br|>": 50309,
65
+ "<|bs|>": 50315,
66
+ "<|ca|>": 50270,
67
+ "<|cs|>": 50283,
68
+ "<|cy|>": 50297,
69
+ "<|da|>": 50285,
70
+ "<|de|>": 50261,
71
+ "<|el|>": 50281,
72
+ "<|en|>": 50259,
73
+ "<|es|>": 50262,
74
+ "<|et|>": 50307,
75
+ "<|eu|>": 50310,
76
+ "<|fa|>": 50300,
77
+ "<|fi|>": 50277,
78
+ "<|fo|>": 50338,
79
+ "<|fr|>": 50265,
80
+ "<|gl|>": 50319,
81
+ "<|gu|>": 50333,
82
+ "<|haw|>": 50352,
83
+ "<|ha|>": 50354,
84
+ "<|he|>": 50279,
85
+ "<|hi|>": 50276,
86
+ "<|hr|>": 50291,
87
+ "<|ht|>": 50339,
88
+ "<|hu|>": 50286,
89
+ "<|hy|>": 50312,
90
+ "<|id|>": 50275,
91
+ "<|is|>": 50311,
92
+ "<|it|>": 50274,
93
+ "<|ja|>": 50266,
94
+ "<|jw|>": 50356,
95
+ "<|ka|>": 50329,
96
+ "<|kk|>": 50316,
97
+ "<|km|>": 50323,
98
+ "<|kn|>": 50306,
99
+ "<|ko|>": 50264,
100
+ "<|la|>": 50294,
101
+ "<|lb|>": 50345,
102
+ "<|ln|>": 50353,
103
+ "<|lo|>": 50336,
104
+ "<|lt|>": 50293,
105
+ "<|lv|>": 50301,
106
+ "<|mg|>": 50349,
107
+ "<|mi|>": 50295,
108
+ "<|mk|>": 50308,
109
+ "<|ml|>": 50296,
110
+ "<|mn|>": 50314,
111
+ "<|mr|>": 50320,
112
+ "<|ms|>": 50282,
113
+ "<|mt|>": 50343,
114
+ "<|my|>": 50346,
115
+ "<|ne|>": 50313,
116
+ "<|nl|>": 50271,
117
+ "<|nn|>": 50342,
118
+ "<|no|>": 50288,
119
+ "<|oc|>": 50328,
120
+ "<|pa|>": 50321,
121
+ "<|pl|>": 50269,
122
+ "<|ps|>": 50340,
123
+ "<|pt|>": 50267,
124
+ "<|ro|>": 50284,
125
+ "<|ru|>": 50263,
126
+ "<|sa|>": 50344,
127
+ "<|sd|>": 50332,
128
+ "<|si|>": 50322,
129
+ "<|sk|>": 50298,
130
+ "<|sl|>": 50305,
131
+ "<|sn|>": 50324,
132
+ "<|so|>": 50326,
133
+ "<|sq|>": 50317,
134
+ "<|sr|>": 50303,
135
+ "<|su|>": 50357,
136
+ "<|sv|>": 50273,
137
+ "<|sw|>": 50318,
138
+ "<|ta|>": 50287,
139
+ "<|te|>": 50299,
140
+ "<|tg|>": 50331,
141
+ "<|th|>": 50289,
142
+ "<|tk|>": 50341,
143
+ "<|tl|>": 50348,
144
+ "<|tr|>": 50268,
145
+ "<|tt|>": 50351,
146
+ "<|uk|>": 50280,
147
+ "<|ur|>": 50290,
148
+ "<|uz|>": 50337,
149
+ "<|vi|>": 50278,
150
+ "<|yi|>": 50335,
151
+ "<|yo|>": 50325,
152
+ "<|zh|>": 50260
153
+ },
154
+ "language": "swahili",
155
+ "max_initial_timestamp_index": 50,
156
+ "max_length": 448,
157
+ "no_timestamps_token_id": 50363,
158
+ "pad_token_id": 50257,
159
+ "prev_sot_token_id": 50361,
160
+ "return_timestamps": false,
161
+ "suppress_tokens": [
162
+ 1,
163
+ 2,
164
+ 7,
165
+ 8,
166
+ 9,
167
+ 10,
168
+ 14,
169
+ 25,
170
+ 26,
171
+ 27,
172
+ 28,
173
+ 29,
174
+ 31,
175
+ 58,
176
+ 59,
177
+ 60,
178
+ 61,
179
+ 62,
180
+ 63,
181
+ 90,
182
+ 91,
183
+ 92,
184
+ 93,
185
+ 359,
186
+ 503,
187
+ 522,
188
+ 542,
189
+ 873,
190
+ 893,
191
+ 902,
192
+ 918,
193
+ 922,
194
+ 931,
195
+ 1350,
196
+ 1853,
197
+ 1982,
198
+ 2460,
199
+ 2627,
200
+ 3246,
201
+ 3253,
202
+ 3268,
203
+ 3536,
204
+ 3846,
205
+ 3961,
206
+ 4183,
207
+ 4667,
208
+ 6585,
209
+ 6647,
210
+ 7273,
211
+ 9061,
212
+ 9383,
213
+ 10428,
214
+ 10929,
215
+ 11938,
216
+ 12033,
217
+ 12331,
218
+ 12562,
219
+ 13793,
220
+ 14157,
221
+ 14635,
222
+ 15265,
223
+ 15618,
224
+ 16553,
225
+ 16604,
226
+ 18362,
227
+ 18956,
228
+ 20075,
229
+ 21675,
230
+ 22520,
231
+ 26130,
232
+ 26161,
233
+ 26435,
234
+ 28279,
235
+ 29464,
236
+ 31650,
237
+ 32302,
238
+ 32470,
239
+ 36865,
240
+ 42863,
241
+ 47425,
242
+ 49870,
243
+ 50254,
244
+ 50258,
245
+ 50360,
246
+ 50361,
247
+ 50362
248
+ ],
249
+ "task": "transcribe",
250
+ "task_to_id": {
251
+ "transcribe": 50359,
252
+ "translate": 50358
253
+ },
254
+ "transformers_version": "4.57.0"
255
+ }
checkpoint-462/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a27c82f343b256875defe60fb86fe521b77543c851a8c1f87893db10324c5511
3
+ size 966995080
checkpoint-462/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3ada5a8959035c289347e8f61646b6769d51be6beb4e4828fccaa9e5ee90f523
3
+ size 1925071165
checkpoint-462/preprocessor_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "dither": 0.0,
4
+ "feature_extractor_type": "WhisperFeatureExtractor",
5
+ "feature_size": 80,
6
+ "hop_length": 160,
7
+ "n_fft": 400,
8
+ "n_samples": 480000,
9
+ "nb_max_frames": 3000,
10
+ "padding_side": "right",
11
+ "padding_value": 0.0,
12
+ "processor_class": "WhisperProcessor",
13
+ "return_attention_mask": false,
14
+ "sampling_rate": 16000
15
+ }
checkpoint-462/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f25cc922e8fe8191af798b08e0f25672338bd151d99fc34e8369b96b22fe7ff8
3
+ size 14645
checkpoint-462/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:438326cd53f1f84b9354e3c18c94beba3f14ebc180971ee17184c0a4cc7d9c56
3
+ size 1383
checkpoint-462/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a82e3eaea369430f7f78bc8804ef4b7bc1e6c7e4984c7e5fd60c07772d18b991
3
+ size 1465
checkpoint-462/trainer_state.json ADDED
@@ -0,0 +1,482 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 330,
3
+ "best_metric": 63.56830941983781,
4
+ "best_model_checkpoint": "./whisper-small-isizulu-noisy/checkpoint-330",
5
+ "epoch": 14.0,
6
+ "eval_steps": 500,
7
+ "global_step": 462,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.30303030303030304,
14
+ "grad_norm": 26.396528244018555,
15
+ "learning_rate": 9.000000000000001e-07,
16
+ "loss": 3.5466,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.6060606060606061,
21
+ "grad_norm": 19.094928741455078,
22
+ "learning_rate": 1.9000000000000002e-06,
23
+ "loss": 3.2038,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.9090909090909091,
28
+ "grad_norm": 14.402777671813965,
29
+ "learning_rate": 2.9e-06,
30
+ "loss": 2.7094,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 1.0,
35
+ "eval_loss": 2.2550225257873535,
36
+ "eval_runtime": 60.9646,
37
+ "eval_samples_per_second": 4.314,
38
+ "eval_steps_per_second": 0.279,
39
+ "eval_wer": 108.79600748596381,
40
+ "step": 33
41
+ },
42
+ {
43
+ "epoch": 1.2121212121212122,
44
+ "grad_norm": 12.169499397277832,
45
+ "learning_rate": 3.900000000000001e-06,
46
+ "loss": 2.188,
47
+ "step": 40
48
+ },
49
+ {
50
+ "epoch": 1.5151515151515151,
51
+ "grad_norm": 12.002925872802734,
52
+ "learning_rate": 4.9000000000000005e-06,
53
+ "loss": 1.8277,
54
+ "step": 50
55
+ },
56
+ {
57
+ "epoch": 1.8181818181818183,
58
+ "grad_norm": 10.814740180969238,
59
+ "learning_rate": 5.9e-06,
60
+ "loss": 1.5701,
61
+ "step": 60
62
+ },
63
+ {
64
+ "epoch": 2.0,
65
+ "eval_loss": 1.405279517173767,
66
+ "eval_runtime": 44.4698,
67
+ "eval_samples_per_second": 5.914,
68
+ "eval_steps_per_second": 0.382,
69
+ "eval_wer": 90.1434809731753,
70
+ "step": 66
71
+ },
72
+ {
73
+ "epoch": 2.121212121212121,
74
+ "grad_norm": 9.393446922302246,
75
+ "learning_rate": 6.9e-06,
76
+ "loss": 1.4165,
77
+ "step": 70
78
+ },
79
+ {
80
+ "epoch": 2.4242424242424243,
81
+ "grad_norm": 9.702391624450684,
82
+ "learning_rate": 7.9e-06,
83
+ "loss": 1.0921,
84
+ "step": 80
85
+ },
86
+ {
87
+ "epoch": 2.7272727272727275,
88
+ "grad_norm": 8.38471508026123,
89
+ "learning_rate": 8.900000000000001e-06,
90
+ "loss": 1.0347,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 3.0,
95
+ "eval_loss": 1.0759814977645874,
96
+ "eval_runtime": 53.0857,
97
+ "eval_samples_per_second": 4.954,
98
+ "eval_steps_per_second": 0.32,
99
+ "eval_wer": 103.74298190892077,
100
+ "step": 99
101
+ },
102
+ {
103
+ "epoch": 3.0303030303030303,
104
+ "grad_norm": 6.726572036743164,
105
+ "learning_rate": 9.9e-06,
106
+ "loss": 0.8997,
107
+ "step": 100
108
+ },
109
+ {
110
+ "epoch": 3.3333333333333335,
111
+ "grad_norm": 6.715061187744141,
112
+ "learning_rate": 9.772151898734179e-06,
113
+ "loss": 0.6359,
114
+ "step": 110
115
+ },
116
+ {
117
+ "epoch": 3.6363636363636362,
118
+ "grad_norm": 6.393348217010498,
119
+ "learning_rate": 9.518987341772152e-06,
120
+ "loss": 0.5551,
121
+ "step": 120
122
+ },
123
+ {
124
+ "epoch": 3.9393939393939394,
125
+ "grad_norm": 7.166343688964844,
126
+ "learning_rate": 9.265822784810128e-06,
127
+ "loss": 0.5594,
128
+ "step": 130
129
+ },
130
+ {
131
+ "epoch": 4.0,
132
+ "eval_loss": 0.9267733693122864,
133
+ "eval_runtime": 48.7558,
134
+ "eval_samples_per_second": 5.394,
135
+ "eval_steps_per_second": 0.349,
136
+ "eval_wer": 100.62383031815347,
137
+ "step": 132
138
+ },
139
+ {
140
+ "epoch": 4.242424242424242,
141
+ "grad_norm": 6.093661308288574,
142
+ "learning_rate": 9.012658227848103e-06,
143
+ "loss": 0.3714,
144
+ "step": 140
145
+ },
146
+ {
147
+ "epoch": 4.545454545454545,
148
+ "grad_norm": 7.345926761627197,
149
+ "learning_rate": 8.759493670886077e-06,
150
+ "loss": 0.2967,
151
+ "step": 150
152
+ },
153
+ {
154
+ "epoch": 4.848484848484849,
155
+ "grad_norm": 4.838716506958008,
156
+ "learning_rate": 8.50632911392405e-06,
157
+ "loss": 0.2742,
158
+ "step": 160
159
+ },
160
+ {
161
+ "epoch": 5.0,
162
+ "eval_loss": 0.9051285982131958,
163
+ "eval_runtime": 46.5026,
164
+ "eval_samples_per_second": 5.656,
165
+ "eval_steps_per_second": 0.366,
166
+ "eval_wer": 80.22457891453524,
167
+ "step": 165
168
+ },
169
+ {
170
+ "epoch": 5.151515151515151,
171
+ "grad_norm": 3.3409740924835205,
172
+ "learning_rate": 8.253164556962026e-06,
173
+ "loss": 0.2169,
174
+ "step": 170
175
+ },
176
+ {
177
+ "epoch": 5.454545454545454,
178
+ "grad_norm": 3.894425868988037,
179
+ "learning_rate": 8.000000000000001e-06,
180
+ "loss": 0.1534,
181
+ "step": 180
182
+ },
183
+ {
184
+ "epoch": 5.757575757575758,
185
+ "grad_norm": 4.993983745574951,
186
+ "learning_rate": 7.746835443037975e-06,
187
+ "loss": 0.1504,
188
+ "step": 190
189
+ },
190
+ {
191
+ "epoch": 6.0,
192
+ "eval_loss": 0.8892419934272766,
193
+ "eval_runtime": 50.604,
194
+ "eval_samples_per_second": 5.197,
195
+ "eval_steps_per_second": 0.336,
196
+ "eval_wer": 89.58203368683718,
197
+ "step": 198
198
+ },
199
+ {
200
+ "epoch": 6.0606060606060606,
201
+ "grad_norm": 2.509334087371826,
202
+ "learning_rate": 7.49367088607595e-06,
203
+ "loss": 0.1417,
204
+ "step": 200
205
+ },
206
+ {
207
+ "epoch": 6.363636363636363,
208
+ "grad_norm": 3.116098403930664,
209
+ "learning_rate": 7.240506329113924e-06,
210
+ "loss": 0.0785,
211
+ "step": 210
212
+ },
213
+ {
214
+ "epoch": 6.666666666666667,
215
+ "grad_norm": 2.244513750076294,
216
+ "learning_rate": 6.987341772151899e-06,
217
+ "loss": 0.072,
218
+ "step": 220
219
+ },
220
+ {
221
+ "epoch": 6.96969696969697,
222
+ "grad_norm": 2.546046495437622,
223
+ "learning_rate": 6.734177215189874e-06,
224
+ "loss": 0.065,
225
+ "step": 230
226
+ },
227
+ {
228
+ "epoch": 7.0,
229
+ "eval_loss": 0.9148803949356079,
230
+ "eval_runtime": 46.4915,
231
+ "eval_samples_per_second": 5.657,
232
+ "eval_steps_per_second": 0.366,
233
+ "eval_wer": 81.16032439176544,
234
+ "step": 231
235
+ },
236
+ {
237
+ "epoch": 7.2727272727272725,
238
+ "grad_norm": 1.2738147974014282,
239
+ "learning_rate": 6.481012658227849e-06,
240
+ "loss": 0.0444,
241
+ "step": 240
242
+ },
243
+ {
244
+ "epoch": 7.575757575757576,
245
+ "grad_norm": 1.1117249727249146,
246
+ "learning_rate": 6.227848101265823e-06,
247
+ "loss": 0.0374,
248
+ "step": 250
249
+ },
250
+ {
251
+ "epoch": 7.878787878787879,
252
+ "grad_norm": 1.59920072555542,
253
+ "learning_rate": 5.974683544303798e-06,
254
+ "loss": 0.0369,
255
+ "step": 260
256
+ },
257
+ {
258
+ "epoch": 8.0,
259
+ "eval_loss": 0.9126399755477905,
260
+ "eval_runtime": 44.4651,
261
+ "eval_samples_per_second": 5.915,
262
+ "eval_steps_per_second": 0.382,
263
+ "eval_wer": 66.126013724267,
264
+ "step": 264
265
+ },
266
+ {
267
+ "epoch": 8.181818181818182,
268
+ "grad_norm": 0.7837973237037659,
269
+ "learning_rate": 5.721518987341773e-06,
270
+ "loss": 0.0269,
271
+ "step": 270
272
+ },
273
+ {
274
+ "epoch": 8.484848484848484,
275
+ "grad_norm": 0.7720461487770081,
276
+ "learning_rate": 5.468354430379748e-06,
277
+ "loss": 0.0198,
278
+ "step": 280
279
+ },
280
+ {
281
+ "epoch": 8.787878787878787,
282
+ "grad_norm": 2.480095386505127,
283
+ "learning_rate": 5.2151898734177216e-06,
284
+ "loss": 0.0215,
285
+ "step": 290
286
+ },
287
+ {
288
+ "epoch": 9.0,
289
+ "eval_loss": 0.9265649318695068,
290
+ "eval_runtime": 44.2469,
291
+ "eval_samples_per_second": 5.944,
292
+ "eval_steps_per_second": 0.384,
293
+ "eval_wer": 66.9369931378665,
294
+ "step": 297
295
+ },
296
+ {
297
+ "epoch": 9.090909090909092,
298
+ "grad_norm": 0.400156170129776,
299
+ "learning_rate": 4.962025316455696e-06,
300
+ "loss": 0.0186,
301
+ "step": 300
302
+ },
303
+ {
304
+ "epoch": 9.393939393939394,
305
+ "grad_norm": 0.3897418975830078,
306
+ "learning_rate": 4.7088607594936715e-06,
307
+ "loss": 0.0107,
308
+ "step": 310
309
+ },
310
+ {
311
+ "epoch": 9.696969696969697,
312
+ "grad_norm": 4.099902153015137,
313
+ "learning_rate": 4.455696202531646e-06,
314
+ "loss": 0.0136,
315
+ "step": 320
316
+ },
317
+ {
318
+ "epoch": 10.0,
319
+ "grad_norm": 0.33050861954689026,
320
+ "learning_rate": 4.202531645569621e-06,
321
+ "loss": 0.0129,
322
+ "step": 330
323
+ },
324
+ {
325
+ "epoch": 10.0,
326
+ "eval_loss": 0.9320704936981201,
327
+ "eval_runtime": 44.2449,
328
+ "eval_samples_per_second": 5.944,
329
+ "eval_steps_per_second": 0.384,
330
+ "eval_wer": 63.56830941983781,
331
+ "step": 330
332
+ },
333
+ {
334
+ "epoch": 10.303030303030303,
335
+ "grad_norm": 0.24591486155986786,
336
+ "learning_rate": 3.949367088607595e-06,
337
+ "loss": 0.0077,
338
+ "step": 340
339
+ },
340
+ {
341
+ "epoch": 10.606060606060606,
342
+ "grad_norm": 0.2317628711462021,
343
+ "learning_rate": 3.6962025316455702e-06,
344
+ "loss": 0.0084,
345
+ "step": 350
346
+ },
347
+ {
348
+ "epoch": 10.909090909090908,
349
+ "grad_norm": 0.188068225979805,
350
+ "learning_rate": 3.443037974683545e-06,
351
+ "loss": 0.0076,
352
+ "step": 360
353
+ },
354
+ {
355
+ "epoch": 11.0,
356
+ "eval_loss": 0.9407175779342651,
357
+ "eval_runtime": 44.2857,
358
+ "eval_samples_per_second": 5.939,
359
+ "eval_steps_per_second": 0.384,
360
+ "eval_wer": 66.18839675608234,
361
+ "step": 363
362
+ },
363
+ {
364
+ "epoch": 11.212121212121213,
365
+ "grad_norm": 0.19238056242465973,
366
+ "learning_rate": 3.189873417721519e-06,
367
+ "loss": 0.0069,
368
+ "step": 370
369
+ },
370
+ {
371
+ "epoch": 11.515151515151516,
372
+ "grad_norm": 0.16192281246185303,
373
+ "learning_rate": 2.936708860759494e-06,
374
+ "loss": 0.007,
375
+ "step": 380
376
+ },
377
+ {
378
+ "epoch": 11.818181818181818,
379
+ "grad_norm": 0.15943264961242676,
380
+ "learning_rate": 2.6835443037974685e-06,
381
+ "loss": 0.006,
382
+ "step": 390
383
+ },
384
+ {
385
+ "epoch": 12.0,
386
+ "eval_loss": 0.949792206287384,
387
+ "eval_runtime": 44.3925,
388
+ "eval_samples_per_second": 5.924,
389
+ "eval_steps_per_second": 0.383,
390
+ "eval_wer": 64.37928883343731,
391
+ "step": 396
392
+ },
393
+ {
394
+ "epoch": 12.121212121212121,
395
+ "grad_norm": 0.14337708055973053,
396
+ "learning_rate": 2.430379746835443e-06,
397
+ "loss": 0.007,
398
+ "step": 400
399
+ },
400
+ {
401
+ "epoch": 12.424242424242424,
402
+ "grad_norm": 0.13037937879562378,
403
+ "learning_rate": 2.177215189873418e-06,
404
+ "loss": 0.0049,
405
+ "step": 410
406
+ },
407
+ {
408
+ "epoch": 12.727272727272727,
409
+ "grad_norm": 0.14935876429080963,
410
+ "learning_rate": 1.9240506329113926e-06,
411
+ "loss": 0.0051,
412
+ "step": 420
413
+ },
414
+ {
415
+ "epoch": 13.0,
416
+ "eval_loss": 0.9536267518997192,
417
+ "eval_runtime": 46.3986,
418
+ "eval_samples_per_second": 5.668,
419
+ "eval_steps_per_second": 0.366,
420
+ "eval_wer": 77.72925764192141,
421
+ "step": 429
422
+ },
423
+ {
424
+ "epoch": 13.030303030303031,
425
+ "grad_norm": 0.12029508501291275,
426
+ "learning_rate": 1.6708860759493672e-06,
427
+ "loss": 0.0054,
428
+ "step": 430
429
+ },
430
+ {
431
+ "epoch": 13.333333333333334,
432
+ "grad_norm": 0.1245550662279129,
433
+ "learning_rate": 1.417721518987342e-06,
434
+ "loss": 0.0047,
435
+ "step": 440
436
+ },
437
+ {
438
+ "epoch": 13.636363636363637,
439
+ "grad_norm": 0.11999326199293137,
440
+ "learning_rate": 1.1645569620253165e-06,
441
+ "loss": 0.0048,
442
+ "step": 450
443
+ },
444
+ {
445
+ "epoch": 13.93939393939394,
446
+ "grad_norm": 0.13922753930091858,
447
+ "learning_rate": 9.113924050632912e-07,
448
+ "loss": 0.0046,
449
+ "step": 460
450
+ },
451
+ {
452
+ "epoch": 14.0,
453
+ "eval_loss": 0.9570201635360718,
454
+ "eval_runtime": 46.2646,
455
+ "eval_samples_per_second": 5.685,
456
+ "eval_steps_per_second": 0.367,
457
+ "eval_wer": 77.91640673736744,
458
+ "step": 462
459
+ }
460
+ ],
461
+ "logging_steps": 10,
462
+ "max_steps": 495,
463
+ "num_input_tokens_seen": 0,
464
+ "num_train_epochs": 15,
465
+ "save_steps": 500,
466
+ "stateful_callbacks": {
467
+ "TrainerControl": {
468
+ "args": {
469
+ "should_epoch_stop": false,
470
+ "should_evaluate": false,
471
+ "should_log": false,
472
+ "should_save": true,
473
+ "should_training_stop": false
474
+ },
475
+ "attributes": {}
476
+ }
477
+ },
478
+ "total_flos": 4.242205384704e+18,
479
+ "train_batch_size": 32,
480
+ "trial_name": null,
481
+ "trial_params": null
482
+ }
checkpoint-462/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68d60f46e270ca874141fea17fa80ade9cf0002e14e72e1e1cd28d219ddaf599
3
+ size 6033
checkpoint-495/config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "activation_function": "gelu",
4
+ "apply_spec_augment": false,
5
+ "architectures": [
6
+ "WhisperForConditionalGeneration"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "begin_suppress_tokens": null,
10
+ "bos_token_id": 50257,
11
+ "classifier_proj_size": 256,
12
+ "d_model": 768,
13
+ "decoder_attention_heads": 12,
14
+ "decoder_ffn_dim": 3072,
15
+ "decoder_layerdrop": 0.0,
16
+ "decoder_layers": 12,
17
+ "decoder_start_token_id": 50258,
18
+ "dropout": 0.0,
19
+ "dtype": "float32",
20
+ "encoder_attention_heads": 12,
21
+ "encoder_ffn_dim": 3072,
22
+ "encoder_layerdrop": 0.0,
23
+ "encoder_layers": 12,
24
+ "eos_token_id": 50257,
25
+ "forced_decoder_ids": [
26
+ [
27
+ 1,
28
+ 50259
29
+ ],
30
+ [
31
+ 2,
32
+ 50359
33
+ ],
34
+ [
35
+ 3,
36
+ 50363
37
+ ]
38
+ ],
39
+ "init_std": 0.02,
40
+ "is_encoder_decoder": true,
41
+ "mask_feature_length": 10,
42
+ "mask_feature_min_masks": 0,
43
+ "mask_feature_prob": 0.0,
44
+ "mask_time_length": 10,
45
+ "mask_time_min_masks": 2,
46
+ "mask_time_prob": 0.05,
47
+ "max_length": null,
48
+ "max_source_positions": 1500,
49
+ "max_target_positions": 448,
50
+ "median_filter_width": 7,
51
+ "model_type": "whisper",
52
+ "num_hidden_layers": 12,
53
+ "num_mel_bins": 80,
54
+ "pad_token_id": 50257,
55
+ "scale_embedding": false,
56
+ "transformers_version": "4.57.0",
57
+ "use_cache": true,
58
+ "use_weighted_layer_sum": false,
59
+ "vocab_size": 51865
60
+ }
checkpoint-495/generation_config.json ADDED
@@ -0,0 +1,255 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": null,
52
+ "is_multilingual": true,
53
+ "lang_to_id": {
54
+ "<|af|>": 50327,
55
+ "<|am|>": 50334,
56
+ "<|ar|>": 50272,
57
+ "<|as|>": 50350,
58
+ "<|az|>": 50304,
59
+ "<|ba|>": 50355,
60
+ "<|be|>": 50330,
61
+ "<|bg|>": 50292,
62
+ "<|bn|>": 50302,
63
+ "<|bo|>": 50347,
64
+ "<|br|>": 50309,
65
+ "<|bs|>": 50315,
66
+ "<|ca|>": 50270,
67
+ "<|cs|>": 50283,
68
+ "<|cy|>": 50297,
69
+ "<|da|>": 50285,
70
+ "<|de|>": 50261,
71
+ "<|el|>": 50281,
72
+ "<|en|>": 50259,
73
+ "<|es|>": 50262,
74
+ "<|et|>": 50307,
75
+ "<|eu|>": 50310,
76
+ "<|fa|>": 50300,
77
+ "<|fi|>": 50277,
78
+ "<|fo|>": 50338,
79
+ "<|fr|>": 50265,
80
+ "<|gl|>": 50319,
81
+ "<|gu|>": 50333,
82
+ "<|haw|>": 50352,
83
+ "<|ha|>": 50354,
84
+ "<|he|>": 50279,
85
+ "<|hi|>": 50276,
86
+ "<|hr|>": 50291,
87
+ "<|ht|>": 50339,
88
+ "<|hu|>": 50286,
89
+ "<|hy|>": 50312,
90
+ "<|id|>": 50275,
91
+ "<|is|>": 50311,
92
+ "<|it|>": 50274,
93
+ "<|ja|>": 50266,
94
+ "<|jw|>": 50356,
95
+ "<|ka|>": 50329,
96
+ "<|kk|>": 50316,
97
+ "<|km|>": 50323,
98
+ "<|kn|>": 50306,
99
+ "<|ko|>": 50264,
100
+ "<|la|>": 50294,
101
+ "<|lb|>": 50345,
102
+ "<|ln|>": 50353,
103
+ "<|lo|>": 50336,
104
+ "<|lt|>": 50293,
105
+ "<|lv|>": 50301,
106
+ "<|mg|>": 50349,
107
+ "<|mi|>": 50295,
108
+ "<|mk|>": 50308,
109
+ "<|ml|>": 50296,
110
+ "<|mn|>": 50314,
111
+ "<|mr|>": 50320,
112
+ "<|ms|>": 50282,
113
+ "<|mt|>": 50343,
114
+ "<|my|>": 50346,
115
+ "<|ne|>": 50313,
116
+ "<|nl|>": 50271,
117
+ "<|nn|>": 50342,
118
+ "<|no|>": 50288,
119
+ "<|oc|>": 50328,
120
+ "<|pa|>": 50321,
121
+ "<|pl|>": 50269,
122
+ "<|ps|>": 50340,
123
+ "<|pt|>": 50267,
124
+ "<|ro|>": 50284,
125
+ "<|ru|>": 50263,
126
+ "<|sa|>": 50344,
127
+ "<|sd|>": 50332,
128
+ "<|si|>": 50322,
129
+ "<|sk|>": 50298,
130
+ "<|sl|>": 50305,
131
+ "<|sn|>": 50324,
132
+ "<|so|>": 50326,
133
+ "<|sq|>": 50317,
134
+ "<|sr|>": 50303,
135
+ "<|su|>": 50357,
136
+ "<|sv|>": 50273,
137
+ "<|sw|>": 50318,
138
+ "<|ta|>": 50287,
139
+ "<|te|>": 50299,
140
+ "<|tg|>": 50331,
141
+ "<|th|>": 50289,
142
+ "<|tk|>": 50341,
143
+ "<|tl|>": 50348,
144
+ "<|tr|>": 50268,
145
+ "<|tt|>": 50351,
146
+ "<|uk|>": 50280,
147
+ "<|ur|>": 50290,
148
+ "<|uz|>": 50337,
149
+ "<|vi|>": 50278,
150
+ "<|yi|>": 50335,
151
+ "<|yo|>": 50325,
152
+ "<|zh|>": 50260
153
+ },
154
+ "language": "swahili",
155
+ "max_initial_timestamp_index": 50,
156
+ "max_length": 448,
157
+ "no_timestamps_token_id": 50363,
158
+ "pad_token_id": 50257,
159
+ "prev_sot_token_id": 50361,
160
+ "return_timestamps": false,
161
+ "suppress_tokens": [
162
+ 1,
163
+ 2,
164
+ 7,
165
+ 8,
166
+ 9,
167
+ 10,
168
+ 14,
169
+ 25,
170
+ 26,
171
+ 27,
172
+ 28,
173
+ 29,
174
+ 31,
175
+ 58,
176
+ 59,
177
+ 60,
178
+ 61,
179
+ 62,
180
+ 63,
181
+ 90,
182
+ 91,
183
+ 92,
184
+ 93,
185
+ 359,
186
+ 503,
187
+ 522,
188
+ 542,
189
+ 873,
190
+ 893,
191
+ 902,
192
+ 918,
193
+ 922,
194
+ 931,
195
+ 1350,
196
+ 1853,
197
+ 1982,
198
+ 2460,
199
+ 2627,
200
+ 3246,
201
+ 3253,
202
+ 3268,
203
+ 3536,
204
+ 3846,
205
+ 3961,
206
+ 4183,
207
+ 4667,
208
+ 6585,
209
+ 6647,
210
+ 7273,
211
+ 9061,
212
+ 9383,
213
+ 10428,
214
+ 10929,
215
+ 11938,
216
+ 12033,
217
+ 12331,
218
+ 12562,
219
+ 13793,
220
+ 14157,
221
+ 14635,
222
+ 15265,
223
+ 15618,
224
+ 16553,
225
+ 16604,
226
+ 18362,
227
+ 18956,
228
+ 20075,
229
+ 21675,
230
+ 22520,
231
+ 26130,
232
+ 26161,
233
+ 26435,
234
+ 28279,
235
+ 29464,
236
+ 31650,
237
+ 32302,
238
+ 32470,
239
+ 36865,
240
+ 42863,
241
+ 47425,
242
+ 49870,
243
+ 50254,
244
+ 50258,
245
+ 50360,
246
+ 50361,
247
+ 50362
248
+ ],
249
+ "task": "transcribe",
250
+ "task_to_id": {
251
+ "transcribe": 50359,
252
+ "translate": 50358
253
+ },
254
+ "transformers_version": "4.57.0"
255
+ }
checkpoint-495/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d086270422dfc886cead91bf17e01b4ea1a77449ac8704e7ffe1d4bcc18cab07
3
+ size 966995080
checkpoint-495/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f4ddd0d6b4f86a065f63fd3f83f38e7f92426c3ee3470c1217daf843eff04b6
3
+ size 1925071165
checkpoint-495/preprocessor_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "dither": 0.0,
4
+ "feature_extractor_type": "WhisperFeatureExtractor",
5
+ "feature_size": 80,
6
+ "hop_length": 160,
7
+ "n_fft": 400,
8
+ "n_samples": 480000,
9
+ "nb_max_frames": 3000,
10
+ "padding_side": "right",
11
+ "padding_value": 0.0,
12
+ "processor_class": "WhisperProcessor",
13
+ "return_attention_mask": false,
14
+ "sampling_rate": 16000
15
+ }
checkpoint-495/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eac8e1035e3ace912e246a804ac15115efe637605b66e8cec73a238d9682b232
3
+ size 14645
checkpoint-495/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e18cc2e853c7630494f53b2a8e61edff7fd40d1bf21fd1f5a060025ac1dd14ca
3
+ size 1383
checkpoint-495/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af0955b8f8fb0a68f911429720ca9f668c998c95c2c0c383afc8a44c6207193f
3
+ size 1465
checkpoint-495/trainer_state.json ADDED
@@ -0,0 +1,512 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 330,
3
+ "best_metric": 63.56830941983781,
4
+ "best_model_checkpoint": "./whisper-small-isizulu-noisy/checkpoint-330",
5
+ "epoch": 15.0,
6
+ "eval_steps": 500,
7
+ "global_step": 495,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.30303030303030304,
14
+ "grad_norm": 26.396528244018555,
15
+ "learning_rate": 9.000000000000001e-07,
16
+ "loss": 3.5466,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.6060606060606061,
21
+ "grad_norm": 19.094928741455078,
22
+ "learning_rate": 1.9000000000000002e-06,
23
+ "loss": 3.2038,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.9090909090909091,
28
+ "grad_norm": 14.402777671813965,
29
+ "learning_rate": 2.9e-06,
30
+ "loss": 2.7094,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 1.0,
35
+ "eval_loss": 2.2550225257873535,
36
+ "eval_runtime": 60.9646,
37
+ "eval_samples_per_second": 4.314,
38
+ "eval_steps_per_second": 0.279,
39
+ "eval_wer": 108.79600748596381,
40
+ "step": 33
41
+ },
42
+ {
43
+ "epoch": 1.2121212121212122,
44
+ "grad_norm": 12.169499397277832,
45
+ "learning_rate": 3.900000000000001e-06,
46
+ "loss": 2.188,
47
+ "step": 40
48
+ },
49
+ {
50
+ "epoch": 1.5151515151515151,
51
+ "grad_norm": 12.002925872802734,
52
+ "learning_rate": 4.9000000000000005e-06,
53
+ "loss": 1.8277,
54
+ "step": 50
55
+ },
56
+ {
57
+ "epoch": 1.8181818181818183,
58
+ "grad_norm": 10.814740180969238,
59
+ "learning_rate": 5.9e-06,
60
+ "loss": 1.5701,
61
+ "step": 60
62
+ },
63
+ {
64
+ "epoch": 2.0,
65
+ "eval_loss": 1.405279517173767,
66
+ "eval_runtime": 44.4698,
67
+ "eval_samples_per_second": 5.914,
68
+ "eval_steps_per_second": 0.382,
69
+ "eval_wer": 90.1434809731753,
70
+ "step": 66
71
+ },
72
+ {
73
+ "epoch": 2.121212121212121,
74
+ "grad_norm": 9.393446922302246,
75
+ "learning_rate": 6.9e-06,
76
+ "loss": 1.4165,
77
+ "step": 70
78
+ },
79
+ {
80
+ "epoch": 2.4242424242424243,
81
+ "grad_norm": 9.702391624450684,
82
+ "learning_rate": 7.9e-06,
83
+ "loss": 1.0921,
84
+ "step": 80
85
+ },
86
+ {
87
+ "epoch": 2.7272727272727275,
88
+ "grad_norm": 8.38471508026123,
89
+ "learning_rate": 8.900000000000001e-06,
90
+ "loss": 1.0347,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 3.0,
95
+ "eval_loss": 1.0759814977645874,
96
+ "eval_runtime": 53.0857,
97
+ "eval_samples_per_second": 4.954,
98
+ "eval_steps_per_second": 0.32,
99
+ "eval_wer": 103.74298190892077,
100
+ "step": 99
101
+ },
102
+ {
103
+ "epoch": 3.0303030303030303,
104
+ "grad_norm": 6.726572036743164,
105
+ "learning_rate": 9.9e-06,
106
+ "loss": 0.8997,
107
+ "step": 100
108
+ },
109
+ {
110
+ "epoch": 3.3333333333333335,
111
+ "grad_norm": 6.715061187744141,
112
+ "learning_rate": 9.772151898734179e-06,
113
+ "loss": 0.6359,
114
+ "step": 110
115
+ },
116
+ {
117
+ "epoch": 3.6363636363636362,
118
+ "grad_norm": 6.393348217010498,
119
+ "learning_rate": 9.518987341772152e-06,
120
+ "loss": 0.5551,
121
+ "step": 120
122
+ },
123
+ {
124
+ "epoch": 3.9393939393939394,
125
+ "grad_norm": 7.166343688964844,
126
+ "learning_rate": 9.265822784810128e-06,
127
+ "loss": 0.5594,
128
+ "step": 130
129
+ },
130
+ {
131
+ "epoch": 4.0,
132
+ "eval_loss": 0.9267733693122864,
133
+ "eval_runtime": 48.7558,
134
+ "eval_samples_per_second": 5.394,
135
+ "eval_steps_per_second": 0.349,
136
+ "eval_wer": 100.62383031815347,
137
+ "step": 132
138
+ },
139
+ {
140
+ "epoch": 4.242424242424242,
141
+ "grad_norm": 6.093661308288574,
142
+ "learning_rate": 9.012658227848103e-06,
143
+ "loss": 0.3714,
144
+ "step": 140
145
+ },
146
+ {
147
+ "epoch": 4.545454545454545,
148
+ "grad_norm": 7.345926761627197,
149
+ "learning_rate": 8.759493670886077e-06,
150
+ "loss": 0.2967,
151
+ "step": 150
152
+ },
153
+ {
154
+ "epoch": 4.848484848484849,
155
+ "grad_norm": 4.838716506958008,
156
+ "learning_rate": 8.50632911392405e-06,
157
+ "loss": 0.2742,
158
+ "step": 160
159
+ },
160
+ {
161
+ "epoch": 5.0,
162
+ "eval_loss": 0.9051285982131958,
163
+ "eval_runtime": 46.5026,
164
+ "eval_samples_per_second": 5.656,
165
+ "eval_steps_per_second": 0.366,
166
+ "eval_wer": 80.22457891453524,
167
+ "step": 165
168
+ },
169
+ {
170
+ "epoch": 5.151515151515151,
171
+ "grad_norm": 3.3409740924835205,
172
+ "learning_rate": 8.253164556962026e-06,
173
+ "loss": 0.2169,
174
+ "step": 170
175
+ },
176
+ {
177
+ "epoch": 5.454545454545454,
178
+ "grad_norm": 3.894425868988037,
179
+ "learning_rate": 8.000000000000001e-06,
180
+ "loss": 0.1534,
181
+ "step": 180
182
+ },
183
+ {
184
+ "epoch": 5.757575757575758,
185
+ "grad_norm": 4.993983745574951,
186
+ "learning_rate": 7.746835443037975e-06,
187
+ "loss": 0.1504,
188
+ "step": 190
189
+ },
190
+ {
191
+ "epoch": 6.0,
192
+ "eval_loss": 0.8892419934272766,
193
+ "eval_runtime": 50.604,
194
+ "eval_samples_per_second": 5.197,
195
+ "eval_steps_per_second": 0.336,
196
+ "eval_wer": 89.58203368683718,
197
+ "step": 198
198
+ },
199
+ {
200
+ "epoch": 6.0606060606060606,
201
+ "grad_norm": 2.509334087371826,
202
+ "learning_rate": 7.49367088607595e-06,
203
+ "loss": 0.1417,
204
+ "step": 200
205
+ },
206
+ {
207
+ "epoch": 6.363636363636363,
208
+ "grad_norm": 3.116098403930664,
209
+ "learning_rate": 7.240506329113924e-06,
210
+ "loss": 0.0785,
211
+ "step": 210
212
+ },
213
+ {
214
+ "epoch": 6.666666666666667,
215
+ "grad_norm": 2.244513750076294,
216
+ "learning_rate": 6.987341772151899e-06,
217
+ "loss": 0.072,
218
+ "step": 220
219
+ },
220
+ {
221
+ "epoch": 6.96969696969697,
222
+ "grad_norm": 2.546046495437622,
223
+ "learning_rate": 6.734177215189874e-06,
224
+ "loss": 0.065,
225
+ "step": 230
226
+ },
227
+ {
228
+ "epoch": 7.0,
229
+ "eval_loss": 0.9148803949356079,
230
+ "eval_runtime": 46.4915,
231
+ "eval_samples_per_second": 5.657,
232
+ "eval_steps_per_second": 0.366,
233
+ "eval_wer": 81.16032439176544,
234
+ "step": 231
235
+ },
236
+ {
237
+ "epoch": 7.2727272727272725,
238
+ "grad_norm": 1.2738147974014282,
239
+ "learning_rate": 6.481012658227849e-06,
240
+ "loss": 0.0444,
241
+ "step": 240
242
+ },
243
+ {
244
+ "epoch": 7.575757575757576,
245
+ "grad_norm": 1.1117249727249146,
246
+ "learning_rate": 6.227848101265823e-06,
247
+ "loss": 0.0374,
248
+ "step": 250
249
+ },
250
+ {
251
+ "epoch": 7.878787878787879,
252
+ "grad_norm": 1.59920072555542,
253
+ "learning_rate": 5.974683544303798e-06,
254
+ "loss": 0.0369,
255
+ "step": 260
256
+ },
257
+ {
258
+ "epoch": 8.0,
259
+ "eval_loss": 0.9126399755477905,
260
+ "eval_runtime": 44.4651,
261
+ "eval_samples_per_second": 5.915,
262
+ "eval_steps_per_second": 0.382,
263
+ "eval_wer": 66.126013724267,
264
+ "step": 264
265
+ },
266
+ {
267
+ "epoch": 8.181818181818182,
268
+ "grad_norm": 0.7837973237037659,
269
+ "learning_rate": 5.721518987341773e-06,
270
+ "loss": 0.0269,
271
+ "step": 270
272
+ },
273
+ {
274
+ "epoch": 8.484848484848484,
275
+ "grad_norm": 0.7720461487770081,
276
+ "learning_rate": 5.468354430379748e-06,
277
+ "loss": 0.0198,
278
+ "step": 280
279
+ },
280
+ {
281
+ "epoch": 8.787878787878787,
282
+ "grad_norm": 2.480095386505127,
283
+ "learning_rate": 5.2151898734177216e-06,
284
+ "loss": 0.0215,
285
+ "step": 290
286
+ },
287
+ {
288
+ "epoch": 9.0,
289
+ "eval_loss": 0.9265649318695068,
290
+ "eval_runtime": 44.2469,
291
+ "eval_samples_per_second": 5.944,
292
+ "eval_steps_per_second": 0.384,
293
+ "eval_wer": 66.9369931378665,
294
+ "step": 297
295
+ },
296
+ {
297
+ "epoch": 9.090909090909092,
298
+ "grad_norm": 0.400156170129776,
299
+ "learning_rate": 4.962025316455696e-06,
300
+ "loss": 0.0186,
301
+ "step": 300
302
+ },
303
+ {
304
+ "epoch": 9.393939393939394,
305
+ "grad_norm": 0.3897418975830078,
306
+ "learning_rate": 4.7088607594936715e-06,
307
+ "loss": 0.0107,
308
+ "step": 310
309
+ },
310
+ {
311
+ "epoch": 9.696969696969697,
312
+ "grad_norm": 4.099902153015137,
313
+ "learning_rate": 4.455696202531646e-06,
314
+ "loss": 0.0136,
315
+ "step": 320
316
+ },
317
+ {
318
+ "epoch": 10.0,
319
+ "grad_norm": 0.33050861954689026,
320
+ "learning_rate": 4.202531645569621e-06,
321
+ "loss": 0.0129,
322
+ "step": 330
323
+ },
324
+ {
325
+ "epoch": 10.0,
326
+ "eval_loss": 0.9320704936981201,
327
+ "eval_runtime": 44.2449,
328
+ "eval_samples_per_second": 5.944,
329
+ "eval_steps_per_second": 0.384,
330
+ "eval_wer": 63.56830941983781,
331
+ "step": 330
332
+ },
333
+ {
334
+ "epoch": 10.303030303030303,
335
+ "grad_norm": 0.24591486155986786,
336
+ "learning_rate": 3.949367088607595e-06,
337
+ "loss": 0.0077,
338
+ "step": 340
339
+ },
340
+ {
341
+ "epoch": 10.606060606060606,
342
+ "grad_norm": 0.2317628711462021,
343
+ "learning_rate": 3.6962025316455702e-06,
344
+ "loss": 0.0084,
345
+ "step": 350
346
+ },
347
+ {
348
+ "epoch": 10.909090909090908,
349
+ "grad_norm": 0.188068225979805,
350
+ "learning_rate": 3.443037974683545e-06,
351
+ "loss": 0.0076,
352
+ "step": 360
353
+ },
354
+ {
355
+ "epoch": 11.0,
356
+ "eval_loss": 0.9407175779342651,
357
+ "eval_runtime": 44.2857,
358
+ "eval_samples_per_second": 5.939,
359
+ "eval_steps_per_second": 0.384,
360
+ "eval_wer": 66.18839675608234,
361
+ "step": 363
362
+ },
363
+ {
364
+ "epoch": 11.212121212121213,
365
+ "grad_norm": 0.19238056242465973,
366
+ "learning_rate": 3.189873417721519e-06,
367
+ "loss": 0.0069,
368
+ "step": 370
369
+ },
370
+ {
371
+ "epoch": 11.515151515151516,
372
+ "grad_norm": 0.16192281246185303,
373
+ "learning_rate": 2.936708860759494e-06,
374
+ "loss": 0.007,
375
+ "step": 380
376
+ },
377
+ {
378
+ "epoch": 11.818181818181818,
379
+ "grad_norm": 0.15943264961242676,
380
+ "learning_rate": 2.6835443037974685e-06,
381
+ "loss": 0.006,
382
+ "step": 390
383
+ },
384
+ {
385
+ "epoch": 12.0,
386
+ "eval_loss": 0.949792206287384,
387
+ "eval_runtime": 44.3925,
388
+ "eval_samples_per_second": 5.924,
389
+ "eval_steps_per_second": 0.383,
390
+ "eval_wer": 64.37928883343731,
391
+ "step": 396
392
+ },
393
+ {
394
+ "epoch": 12.121212121212121,
395
+ "grad_norm": 0.14337708055973053,
396
+ "learning_rate": 2.430379746835443e-06,
397
+ "loss": 0.007,
398
+ "step": 400
399
+ },
400
+ {
401
+ "epoch": 12.424242424242424,
402
+ "grad_norm": 0.13037937879562378,
403
+ "learning_rate": 2.177215189873418e-06,
404
+ "loss": 0.0049,
405
+ "step": 410
406
+ },
407
+ {
408
+ "epoch": 12.727272727272727,
409
+ "grad_norm": 0.14935876429080963,
410
+ "learning_rate": 1.9240506329113926e-06,
411
+ "loss": 0.0051,
412
+ "step": 420
413
+ },
414
+ {
415
+ "epoch": 13.0,
416
+ "eval_loss": 0.9536267518997192,
417
+ "eval_runtime": 46.3986,
418
+ "eval_samples_per_second": 5.668,
419
+ "eval_steps_per_second": 0.366,
420
+ "eval_wer": 77.72925764192141,
421
+ "step": 429
422
+ },
423
+ {
424
+ "epoch": 13.030303030303031,
425
+ "grad_norm": 0.12029508501291275,
426
+ "learning_rate": 1.6708860759493672e-06,
427
+ "loss": 0.0054,
428
+ "step": 430
429
+ },
430
+ {
431
+ "epoch": 13.333333333333334,
432
+ "grad_norm": 0.1245550662279129,
433
+ "learning_rate": 1.417721518987342e-06,
434
+ "loss": 0.0047,
435
+ "step": 440
436
+ },
437
+ {
438
+ "epoch": 13.636363636363637,
439
+ "grad_norm": 0.11999326199293137,
440
+ "learning_rate": 1.1645569620253165e-06,
441
+ "loss": 0.0048,
442
+ "step": 450
443
+ },
444
+ {
445
+ "epoch": 13.93939393939394,
446
+ "grad_norm": 0.13922753930091858,
447
+ "learning_rate": 9.113924050632912e-07,
448
+ "loss": 0.0046,
449
+ "step": 460
450
+ },
451
+ {
452
+ "epoch": 14.0,
453
+ "eval_loss": 0.9570201635360718,
454
+ "eval_runtime": 46.2646,
455
+ "eval_samples_per_second": 5.685,
456
+ "eval_steps_per_second": 0.367,
457
+ "eval_wer": 77.91640673736744,
458
+ "step": 462
459
+ },
460
+ {
461
+ "epoch": 14.242424242424242,
462
+ "grad_norm": 0.12097723037004471,
463
+ "learning_rate": 6.582278481012659e-07,
464
+ "loss": 0.0045,
465
+ "step": 470
466
+ },
467
+ {
468
+ "epoch": 14.545454545454545,
469
+ "grad_norm": 0.10953503847122192,
470
+ "learning_rate": 4.0506329113924055e-07,
471
+ "loss": 0.0044,
472
+ "step": 480
473
+ },
474
+ {
475
+ "epoch": 14.848484848484848,
476
+ "grad_norm": 0.12444847822189331,
477
+ "learning_rate": 1.518987341772152e-07,
478
+ "loss": 0.0044,
479
+ "step": 490
480
+ },
481
+ {
482
+ "epoch": 15.0,
483
+ "eval_loss": 0.9577784538269043,
484
+ "eval_runtime": 46.5765,
485
+ "eval_samples_per_second": 5.647,
486
+ "eval_steps_per_second": 0.365,
487
+ "eval_wer": 77.91640673736744,
488
+ "step": 495
489
+ }
490
+ ],
491
+ "logging_steps": 10,
492
+ "max_steps": 495,
493
+ "num_input_tokens_seen": 0,
494
+ "num_train_epochs": 15,
495
+ "save_steps": 500,
496
+ "stateful_callbacks": {
497
+ "TrainerControl": {
498
+ "args": {
499
+ "should_epoch_stop": false,
500
+ "should_evaluate": false,
501
+ "should_log": false,
502
+ "should_save": true,
503
+ "should_training_stop": true
504
+ },
505
+ "attributes": {}
506
+ }
507
+ },
508
+ "total_flos": 4.54522005504e+18,
509
+ "train_batch_size": 32,
510
+ "trial_name": null,
511
+ "trial_params": null
512
+ }
checkpoint-495/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68d60f46e270ca874141fea17fa80ade9cf0002e14e72e1e1cd28d219ddaf599
3
+ size 6033