LennyBijan commited on
Commit
acc421a
·
verified ·
1 Parent(s): c807a3c

Upload folder using huggingface_hub

Browse files
checkpoint-5000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-large-v2",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 1280,
17
+ "decoder_attention_heads": 20,
18
+ "decoder_ffn_dim": 5120,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 32,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 20,
24
+ "encoder_ffn_dim": 5120,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 32,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": 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": 32,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.38.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-5000/generation_config.json ADDED
@@ -0,0 +1,317 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 10,
5
+ 12
6
+ ],
7
+ [
8
+ 13,
9
+ 17
10
+ ],
11
+ [
12
+ 16,
13
+ 11
14
+ ],
15
+ [
16
+ 16,
17
+ 12
18
+ ],
19
+ [
20
+ 16,
21
+ 13
22
+ ],
23
+ [
24
+ 17,
25
+ 15
26
+ ],
27
+ [
28
+ 17,
29
+ 16
30
+ ],
31
+ [
32
+ 18,
33
+ 4
34
+ ],
35
+ [
36
+ 18,
37
+ 11
38
+ ],
39
+ [
40
+ 18,
41
+ 19
42
+ ],
43
+ [
44
+ 19,
45
+ 11
46
+ ],
47
+ [
48
+ 21,
49
+ 2
50
+ ],
51
+ [
52
+ 21,
53
+ 3
54
+ ],
55
+ [
56
+ 22,
57
+ 3
58
+ ],
59
+ [
60
+ 22,
61
+ 9
62
+ ],
63
+ [
64
+ 22,
65
+ 12
66
+ ],
67
+ [
68
+ 23,
69
+ 5
70
+ ],
71
+ [
72
+ 23,
73
+ 7
74
+ ],
75
+ [
76
+ 23,
77
+ 13
78
+ ],
79
+ [
80
+ 25,
81
+ 5
82
+ ],
83
+ [
84
+ 26,
85
+ 1
86
+ ],
87
+ [
88
+ 26,
89
+ 12
90
+ ],
91
+ [
92
+ 27,
93
+ 15
94
+ ]
95
+ ],
96
+ "begin_suppress_tokens": [
97
+ 220,
98
+ 50257
99
+ ],
100
+ "bos_token_id": 50257,
101
+ "decoder_start_token_id": 50258,
102
+ "eos_token_id": 50257,
103
+ "forced_decoder_ids": [
104
+ [
105
+ 1,
106
+ null
107
+ ],
108
+ [
109
+ 2,
110
+ 50359
111
+ ]
112
+ ],
113
+ "is_multilingual": true,
114
+ "lang_to_id": {
115
+ "<|af|>": 50327,
116
+ "<|am|>": 50334,
117
+ "<|ar|>": 50272,
118
+ "<|as|>": 50350,
119
+ "<|az|>": 50304,
120
+ "<|ba|>": 50355,
121
+ "<|be|>": 50330,
122
+ "<|bg|>": 50292,
123
+ "<|bn|>": 50302,
124
+ "<|bo|>": 50347,
125
+ "<|br|>": 50309,
126
+ "<|bs|>": 50315,
127
+ "<|ca|>": 50270,
128
+ "<|cs|>": 50283,
129
+ "<|cy|>": 50297,
130
+ "<|da|>": 50285,
131
+ "<|de|>": 50261,
132
+ "<|el|>": 50281,
133
+ "<|en|>": 50259,
134
+ "<|es|>": 50262,
135
+ "<|et|>": 50307,
136
+ "<|eu|>": 50310,
137
+ "<|fa|>": 50300,
138
+ "<|fi|>": 50277,
139
+ "<|fo|>": 50338,
140
+ "<|fr|>": 50265,
141
+ "<|gl|>": 50319,
142
+ "<|gu|>": 50333,
143
+ "<|haw|>": 50352,
144
+ "<|ha|>": 50354,
145
+ "<|he|>": 50279,
146
+ "<|hi|>": 50276,
147
+ "<|hr|>": 50291,
148
+ "<|ht|>": 50339,
149
+ "<|hu|>": 50286,
150
+ "<|hy|>": 50312,
151
+ "<|id|>": 50275,
152
+ "<|is|>": 50311,
153
+ "<|it|>": 50274,
154
+ "<|ja|>": 50266,
155
+ "<|jw|>": 50356,
156
+ "<|ka|>": 50329,
157
+ "<|kk|>": 50316,
158
+ "<|km|>": 50323,
159
+ "<|kn|>": 50306,
160
+ "<|ko|>": 50264,
161
+ "<|la|>": 50294,
162
+ "<|lb|>": 50345,
163
+ "<|ln|>": 50353,
164
+ "<|lo|>": 50336,
165
+ "<|lt|>": 50293,
166
+ "<|lv|>": 50301,
167
+ "<|mg|>": 50349,
168
+ "<|mi|>": 50295,
169
+ "<|mk|>": 50308,
170
+ "<|ml|>": 50296,
171
+ "<|mn|>": 50314,
172
+ "<|mr|>": 50320,
173
+ "<|ms|>": 50282,
174
+ "<|mt|>": 50343,
175
+ "<|my|>": 50346,
176
+ "<|ne|>": 50313,
177
+ "<|nl|>": 50271,
178
+ "<|nn|>": 50342,
179
+ "<|no|>": 50288,
180
+ "<|oc|>": 50328,
181
+ "<|pa|>": 50321,
182
+ "<|pl|>": 50269,
183
+ "<|ps|>": 50340,
184
+ "<|pt|>": 50267,
185
+ "<|ro|>": 50284,
186
+ "<|ru|>": 50263,
187
+ "<|sa|>": 50344,
188
+ "<|sd|>": 50332,
189
+ "<|si|>": 50322,
190
+ "<|sk|>": 50298,
191
+ "<|sl|>": 50305,
192
+ "<|sn|>": 50324,
193
+ "<|so|>": 50326,
194
+ "<|sq|>": 50317,
195
+ "<|sr|>": 50303,
196
+ "<|su|>": 50357,
197
+ "<|sv|>": 50273,
198
+ "<|sw|>": 50318,
199
+ "<|ta|>": 50287,
200
+ "<|te|>": 50299,
201
+ "<|tg|>": 50331,
202
+ "<|th|>": 50289,
203
+ "<|tk|>": 50341,
204
+ "<|tl|>": 50348,
205
+ "<|tr|>": 50268,
206
+ "<|tt|>": 50351,
207
+ "<|uk|>": 50280,
208
+ "<|ur|>": 50290,
209
+ "<|uz|>": 50337,
210
+ "<|vi|>": 50278,
211
+ "<|yi|>": 50335,
212
+ "<|yo|>": 50325,
213
+ "<|zh|>": 50260
214
+ },
215
+ "language": "de",
216
+ "max_initial_timestamp_index": 50,
217
+ "max_length": 448,
218
+ "no_timestamps_token_id": 50363,
219
+ "pad_token_id": 50257,
220
+ "prev_sot_token_id": 50361,
221
+ "return_timestamps": false,
222
+ "suppress_tokens": [
223
+ 1,
224
+ 2,
225
+ 7,
226
+ 8,
227
+ 9,
228
+ 10,
229
+ 14,
230
+ 25,
231
+ 26,
232
+ 27,
233
+ 28,
234
+ 29,
235
+ 31,
236
+ 58,
237
+ 59,
238
+ 60,
239
+ 61,
240
+ 62,
241
+ 63,
242
+ 90,
243
+ 91,
244
+ 92,
245
+ 93,
246
+ 359,
247
+ 503,
248
+ 522,
249
+ 542,
250
+ 873,
251
+ 893,
252
+ 902,
253
+ 918,
254
+ 922,
255
+ 931,
256
+ 1350,
257
+ 1853,
258
+ 1982,
259
+ 2460,
260
+ 2627,
261
+ 3246,
262
+ 3253,
263
+ 3268,
264
+ 3536,
265
+ 3846,
266
+ 3961,
267
+ 4183,
268
+ 4667,
269
+ 6585,
270
+ 6647,
271
+ 7273,
272
+ 9061,
273
+ 9383,
274
+ 10428,
275
+ 10929,
276
+ 11938,
277
+ 12033,
278
+ 12331,
279
+ 12562,
280
+ 13793,
281
+ 14157,
282
+ 14635,
283
+ 15265,
284
+ 15618,
285
+ 16553,
286
+ 16604,
287
+ 18362,
288
+ 18956,
289
+ 20075,
290
+ 21675,
291
+ 22520,
292
+ 26130,
293
+ 26161,
294
+ 26435,
295
+ 28279,
296
+ 29464,
297
+ 31650,
298
+ 32302,
299
+ 32470,
300
+ 36865,
301
+ 42863,
302
+ 47425,
303
+ 49870,
304
+ 50254,
305
+ 50258,
306
+ 50358,
307
+ 50359,
308
+ 50360,
309
+ 50361,
310
+ 50362
311
+ ],
312
+ "task_to_id": {
313
+ "transcribe": 50359,
314
+ "translate": 50358
315
+ },
316
+ "transformers_version": "4.38.1"
317
+ }
checkpoint-5000/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d8a671554d0c61ad38cedb6e579d1bf18f6a38e42e0b8387f33dec065b1c051
3
+ size 4992706480
checkpoint-5000/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f0d6794aa012f64b171db65cb53517b40998d2813f42446f60f98b3afc1a424
3
+ size 1180663192
checkpoint-5000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
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/trainer_state.json ADDED
@@ -0,0 +1,421 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 22.042796384901646,
3
+ "best_model_checkpoint": "BA_Model_V2/checkpoint-3000",
4
+ "epoch": 17.985611510791365,
5
+ "eval_steps": 1000,
6
+ "global_step": 5000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.36,
13
+ "grad_norm": 8.64929485321045,
14
+ "learning_rate": 1.9600000000000003e-06,
15
+ "loss": 1.2156,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.72,
20
+ "grad_norm": 3.87185001373291,
21
+ "learning_rate": 3.96e-06,
22
+ "loss": 0.3891,
23
+ "step": 200
24
+ },
25
+ {
26
+ "epoch": 1.08,
27
+ "grad_norm": 3.1216392517089844,
28
+ "learning_rate": 5.9600000000000005e-06,
29
+ "loss": 0.2699,
30
+ "step": 300
31
+ },
32
+ {
33
+ "epoch": 1.44,
34
+ "grad_norm": 6.549575328826904,
35
+ "learning_rate": 7.960000000000002e-06,
36
+ "loss": 0.2151,
37
+ "step": 400
38
+ },
39
+ {
40
+ "epoch": 1.8,
41
+ "grad_norm": 2.978250026702881,
42
+ "learning_rate": 9.960000000000001e-06,
43
+ "loss": 0.2213,
44
+ "step": 500
45
+ },
46
+ {
47
+ "epoch": 2.16,
48
+ "grad_norm": 8.912956237792969,
49
+ "learning_rate": 9.782222222222222e-06,
50
+ "loss": 0.1881,
51
+ "step": 600
52
+ },
53
+ {
54
+ "epoch": 2.52,
55
+ "grad_norm": 2.4266467094421387,
56
+ "learning_rate": 9.56e-06,
57
+ "loss": 0.1264,
58
+ "step": 700
59
+ },
60
+ {
61
+ "epoch": 2.88,
62
+ "grad_norm": 2.5170629024505615,
63
+ "learning_rate": 9.33777777777778e-06,
64
+ "loss": 0.1319,
65
+ "step": 800
66
+ },
67
+ {
68
+ "epoch": 3.24,
69
+ "grad_norm": 1.9321565628051758,
70
+ "learning_rate": 9.115555555555556e-06,
71
+ "loss": 0.0872,
72
+ "step": 900
73
+ },
74
+ {
75
+ "epoch": 3.6,
76
+ "grad_norm": 2.928337335586548,
77
+ "learning_rate": 8.893333333333333e-06,
78
+ "loss": 0.0687,
79
+ "step": 1000
80
+ },
81
+ {
82
+ "epoch": 3.6,
83
+ "eval_cer": 12.433448002116961,
84
+ "eval_loss": 0.37324240803718567,
85
+ "eval_runtime": 946.036,
86
+ "eval_samples_per_second": 1.568,
87
+ "eval_steps_per_second": 0.197,
88
+ "eval_wer": 22.268740031897927,
89
+ "step": 1000
90
+ },
91
+ {
92
+ "epoch": 3.96,
93
+ "grad_norm": 2.3376998901367188,
94
+ "learning_rate": 8.671111111111113e-06,
95
+ "loss": 0.0724,
96
+ "step": 1100
97
+ },
98
+ {
99
+ "epoch": 4.32,
100
+ "grad_norm": 1.9221140146255493,
101
+ "learning_rate": 8.448888888888889e-06,
102
+ "loss": 0.0423,
103
+ "step": 1200
104
+ },
105
+ {
106
+ "epoch": 4.68,
107
+ "grad_norm": 1.7310447692871094,
108
+ "learning_rate": 8.226666666666667e-06,
109
+ "loss": 0.0423,
110
+ "step": 1300
111
+ },
112
+ {
113
+ "epoch": 5.04,
114
+ "grad_norm": 1.6635299921035767,
115
+ "learning_rate": 8.004444444444445e-06,
116
+ "loss": 0.0425,
117
+ "step": 1400
118
+ },
119
+ {
120
+ "epoch": 5.4,
121
+ "grad_norm": 1.2995408773422241,
122
+ "learning_rate": 7.782222222222223e-06,
123
+ "loss": 0.0272,
124
+ "step": 1500
125
+ },
126
+ {
127
+ "epoch": 5.76,
128
+ "grad_norm": 2.0113301277160645,
129
+ "learning_rate": 7.5600000000000005e-06,
130
+ "loss": 0.0307,
131
+ "step": 1600
132
+ },
133
+ {
134
+ "epoch": 6.12,
135
+ "grad_norm": 1.9505223035812378,
136
+ "learning_rate": 7.337777777777778e-06,
137
+ "loss": 0.0244,
138
+ "step": 1700
139
+ },
140
+ {
141
+ "epoch": 6.47,
142
+ "grad_norm": 1.1860969066619873,
143
+ "learning_rate": 7.115555555555557e-06,
144
+ "loss": 0.018,
145
+ "step": 1800
146
+ },
147
+ {
148
+ "epoch": 6.83,
149
+ "grad_norm": 0.8492623567581177,
150
+ "learning_rate": 6.893333333333334e-06,
151
+ "loss": 0.0196,
152
+ "step": 1900
153
+ },
154
+ {
155
+ "epoch": 7.19,
156
+ "grad_norm": 1.2874751091003418,
157
+ "learning_rate": 6.671111111111112e-06,
158
+ "loss": 0.0151,
159
+ "step": 2000
160
+ },
161
+ {
162
+ "epoch": 7.19,
163
+ "eval_cer": 12.361471288700715,
164
+ "eval_loss": 0.4464028477668762,
165
+ "eval_runtime": 940.6773,
166
+ "eval_samples_per_second": 1.577,
167
+ "eval_steps_per_second": 0.198,
168
+ "eval_wer": 22.610978203083466,
169
+ "step": 2000
170
+ },
171
+ {
172
+ "epoch": 7.55,
173
+ "grad_norm": 0.9422910809516907,
174
+ "learning_rate": 6.448888888888889e-06,
175
+ "loss": 0.0117,
176
+ "step": 2100
177
+ },
178
+ {
179
+ "epoch": 7.91,
180
+ "grad_norm": 0.8238167762756348,
181
+ "learning_rate": 6.2266666666666675e-06,
182
+ "loss": 0.0119,
183
+ "step": 2200
184
+ },
185
+ {
186
+ "epoch": 8.27,
187
+ "grad_norm": 0.7173901796340942,
188
+ "learning_rate": 6.004444444444445e-06,
189
+ "loss": 0.0094,
190
+ "step": 2300
191
+ },
192
+ {
193
+ "epoch": 8.63,
194
+ "grad_norm": 0.7112721800804138,
195
+ "learning_rate": 5.782222222222222e-06,
196
+ "loss": 0.0103,
197
+ "step": 2400
198
+ },
199
+ {
200
+ "epoch": 8.99,
201
+ "grad_norm": 0.7713698148727417,
202
+ "learning_rate": 5.560000000000001e-06,
203
+ "loss": 0.0087,
204
+ "step": 2500
205
+ },
206
+ {
207
+ "epoch": 9.35,
208
+ "grad_norm": 0.798090934753418,
209
+ "learning_rate": 5.337777777777779e-06,
210
+ "loss": 0.0045,
211
+ "step": 2600
212
+ },
213
+ {
214
+ "epoch": 9.71,
215
+ "grad_norm": 0.2931395471096039,
216
+ "learning_rate": 5.115555555555556e-06,
217
+ "loss": 0.0062,
218
+ "step": 2700
219
+ },
220
+ {
221
+ "epoch": 10.07,
222
+ "grad_norm": 0.447310209274292,
223
+ "learning_rate": 4.893333333333334e-06,
224
+ "loss": 0.0044,
225
+ "step": 2800
226
+ },
227
+ {
228
+ "epoch": 10.43,
229
+ "grad_norm": 0.07779108732938766,
230
+ "learning_rate": 4.6711111111111115e-06,
231
+ "loss": 0.0045,
232
+ "step": 2900
233
+ },
234
+ {
235
+ "epoch": 10.79,
236
+ "grad_norm": 0.874796986579895,
237
+ "learning_rate": 4.448888888888889e-06,
238
+ "loss": 0.0039,
239
+ "step": 3000
240
+ },
241
+ {
242
+ "epoch": 10.79,
243
+ "eval_cer": 12.351944958983859,
244
+ "eval_loss": 0.4558783173561096,
245
+ "eval_runtime": 949.821,
246
+ "eval_samples_per_second": 1.561,
247
+ "eval_steps_per_second": 0.196,
248
+ "eval_wer": 22.042796384901646,
249
+ "step": 3000
250
+ },
251
+ {
252
+ "epoch": 11.15,
253
+ "grad_norm": 0.058852240443229675,
254
+ "learning_rate": 4.226666666666667e-06,
255
+ "loss": 0.0029,
256
+ "step": 3100
257
+ },
258
+ {
259
+ "epoch": 11.51,
260
+ "grad_norm": 0.043637294322252274,
261
+ "learning_rate": 4.004444444444445e-06,
262
+ "loss": 0.002,
263
+ "step": 3200
264
+ },
265
+ {
266
+ "epoch": 11.87,
267
+ "grad_norm": 0.03788485378026962,
268
+ "learning_rate": 3.782222222222223e-06,
269
+ "loss": 0.0028,
270
+ "step": 3300
271
+ },
272
+ {
273
+ "epoch": 12.23,
274
+ "grad_norm": 0.0388493537902832,
275
+ "learning_rate": 3.5600000000000002e-06,
276
+ "loss": 0.0021,
277
+ "step": 3400
278
+ },
279
+ {
280
+ "epoch": 12.59,
281
+ "grad_norm": 0.36381232738494873,
282
+ "learning_rate": 3.337777777777778e-06,
283
+ "loss": 0.0014,
284
+ "step": 3500
285
+ },
286
+ {
287
+ "epoch": 12.95,
288
+ "grad_norm": 0.0187509935349226,
289
+ "learning_rate": 3.1155555555555555e-06,
290
+ "loss": 0.0015,
291
+ "step": 3600
292
+ },
293
+ {
294
+ "epoch": 13.31,
295
+ "grad_norm": 0.017583340406417847,
296
+ "learning_rate": 2.8933333333333337e-06,
297
+ "loss": 0.0016,
298
+ "step": 3700
299
+ },
300
+ {
301
+ "epoch": 13.67,
302
+ "grad_norm": 0.019498176872730255,
303
+ "learning_rate": 2.6711111111111116e-06,
304
+ "loss": 0.0012,
305
+ "step": 3800
306
+ },
307
+ {
308
+ "epoch": 14.03,
309
+ "grad_norm": 0.019207799807190895,
310
+ "learning_rate": 2.448888888888889e-06,
311
+ "loss": 0.0008,
312
+ "step": 3900
313
+ },
314
+ {
315
+ "epoch": 14.39,
316
+ "grad_norm": 0.016761431470513344,
317
+ "learning_rate": 2.226666666666667e-06,
318
+ "loss": 0.0008,
319
+ "step": 4000
320
+ },
321
+ {
322
+ "epoch": 14.39,
323
+ "eval_cer": 12.13548557819529,
324
+ "eval_loss": 0.5066311359405518,
325
+ "eval_runtime": 938.9637,
326
+ "eval_samples_per_second": 1.579,
327
+ "eval_steps_per_second": 0.198,
328
+ "eval_wer": 22.155768208399788,
329
+ "step": 4000
330
+ },
331
+ {
332
+ "epoch": 14.75,
333
+ "grad_norm": 0.014156104065477848,
334
+ "learning_rate": 2.0044444444444446e-06,
335
+ "loss": 0.0009,
336
+ "step": 4100
337
+ },
338
+ {
339
+ "epoch": 15.11,
340
+ "grad_norm": 0.01545019168406725,
341
+ "learning_rate": 1.7822222222222225e-06,
342
+ "loss": 0.0008,
343
+ "step": 4200
344
+ },
345
+ {
346
+ "epoch": 15.47,
347
+ "grad_norm": 0.011677294038236141,
348
+ "learning_rate": 1.56e-06,
349
+ "loss": 0.0006,
350
+ "step": 4300
351
+ },
352
+ {
353
+ "epoch": 15.83,
354
+ "grad_norm": 0.018884949386119843,
355
+ "learning_rate": 1.337777777777778e-06,
356
+ "loss": 0.0006,
357
+ "step": 4400
358
+ },
359
+ {
360
+ "epoch": 16.19,
361
+ "grad_norm": 0.013283302076160908,
362
+ "learning_rate": 1.1155555555555558e-06,
363
+ "loss": 0.0007,
364
+ "step": 4500
365
+ },
366
+ {
367
+ "epoch": 16.55,
368
+ "grad_norm": 0.016208523884415627,
369
+ "learning_rate": 8.933333333333334e-07,
370
+ "loss": 0.0007,
371
+ "step": 4600
372
+ },
373
+ {
374
+ "epoch": 16.91,
375
+ "grad_norm": 0.017564095556735992,
376
+ "learning_rate": 6.711111111111111e-07,
377
+ "loss": 0.0005,
378
+ "step": 4700
379
+ },
380
+ {
381
+ "epoch": 17.27,
382
+ "grad_norm": 0.011613059788942337,
383
+ "learning_rate": 4.488888888888889e-07,
384
+ "loss": 0.0005,
385
+ "step": 4800
386
+ },
387
+ {
388
+ "epoch": 17.63,
389
+ "grad_norm": 0.014883192256093025,
390
+ "learning_rate": 2.266666666666667e-07,
391
+ "loss": 0.0006,
392
+ "step": 4900
393
+ },
394
+ {
395
+ "epoch": 17.99,
396
+ "grad_norm": 0.009823828004300594,
397
+ "learning_rate": 4.444444444444445e-09,
398
+ "loss": 0.0005,
399
+ "step": 5000
400
+ },
401
+ {
402
+ "epoch": 17.99,
403
+ "eval_cer": 12.508070918232336,
404
+ "eval_loss": 0.5289328098297119,
405
+ "eval_runtime": 942.463,
406
+ "eval_samples_per_second": 1.574,
407
+ "eval_steps_per_second": 0.197,
408
+ "eval_wer": 22.760499734183945,
409
+ "step": 5000
410
+ }
411
+ ],
412
+ "logging_steps": 100,
413
+ "max_steps": 5000,
414
+ "num_input_tokens_seen": 0,
415
+ "num_train_epochs": 18,
416
+ "save_steps": 1000,
417
+ "total_flos": 1.696010243125248e+20,
418
+ "train_batch_size": 8,
419
+ "trial_name": null,
420
+ "trial_params": null
421
+ }
checkpoint-5000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95d0d507d3167e641752772bd5dedb1d042269df03c68017b9a606876cb384cf
3
+ size 5048
runs/Mar02_17-38-28_whisperlab/events.out.tfevents.1709401111.whisperlab.549227.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b31f3eb7250d1b8989ee3020461724ab9ecb15620b44e65b0e3ec2eb3631e280
3
- size 15295
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10d0e812e13e107de299b2138b10d405233f361c61130fd9413b8e1b3e732c13
3
+ size 18124