kiwakwok commited on
Commit
f9a7e45
·
verified ·
1 Parent(s): 74fa44b

Upload 9 files

Browse files
config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "alvanlii/whisper-small-cantonese",
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": "4.38.0.dev0",
49
+ "use_cache": false,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "max_length": 448,
10
+ "pad_token_id": 50257,
11
+ "suppress_tokens": [],
12
+ "transformers_version": "4.38.0.dev0",
13
+ "use_cache": false
14
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b3023e1d7f795bced16de615212bc5d8e76921bc85c91387253c64d3aa32c65
3
+ size 966995080
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d99afb1d9e8a4061524d172aaa6c39dc93f0070711d38f1a421d24501f4737df
3
+ size 1925050668
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
+ }
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4c4f4c420b0288d3e50b2a82f1ed43cb8730e5b6fc6892dbf1466ff178e84a3
3
+ size 13990
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab62ae4e323d390b9f8a08f27282e49fe629f96684e5dae68eb44982a3def160
3
+ size 1064
trainer_state.json ADDED
@@ -0,0 +1,975 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 100.0,
3
+ "best_model_checkpoint": "/Users/kevinkwok/Desktop/whisper-small-cantonese/checkpoint-1200",
4
+ "epoch": 6.83111954459203,
5
+ "eval_steps": 400,
6
+ "global_step": 3600,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.05,
13
+ "learning_rate": 5e-06,
14
+ "loss": 0.7513,
15
+ "step": 25
16
+ },
17
+ {
18
+ "epoch": 0.09,
19
+ "learning_rate": 1e-05,
20
+ "loss": 0.7588,
21
+ "step": 50
22
+ },
23
+ {
24
+ "epoch": 0.14,
25
+ "learning_rate": 1e-05,
26
+ "loss": 0.7222,
27
+ "step": 75
28
+ },
29
+ {
30
+ "epoch": 0.19,
31
+ "learning_rate": 1e-05,
32
+ "loss": 0.7234,
33
+ "step": 100
34
+ },
35
+ {
36
+ "epoch": 0.24,
37
+ "learning_rate": 1e-05,
38
+ "loss": 0.6998,
39
+ "step": 125
40
+ },
41
+ {
42
+ "epoch": 0.28,
43
+ "learning_rate": 1e-05,
44
+ "loss": 0.7416,
45
+ "step": 150
46
+ },
47
+ {
48
+ "epoch": 0.33,
49
+ "learning_rate": 1e-05,
50
+ "loss": 0.7151,
51
+ "step": 175
52
+ },
53
+ {
54
+ "epoch": 0.38,
55
+ "learning_rate": 1e-05,
56
+ "loss": 0.7279,
57
+ "step": 200
58
+ },
59
+ {
60
+ "epoch": 0.43,
61
+ "learning_rate": 1e-05,
62
+ "loss": 0.7443,
63
+ "step": 225
64
+ },
65
+ {
66
+ "epoch": 0.47,
67
+ "learning_rate": 1e-05,
68
+ "loss": 0.7879,
69
+ "step": 250
70
+ },
71
+ {
72
+ "epoch": 0.52,
73
+ "learning_rate": 1e-05,
74
+ "loss": 0.951,
75
+ "step": 275
76
+ },
77
+ {
78
+ "epoch": 0.57,
79
+ "learning_rate": 1e-05,
80
+ "loss": 1.0695,
81
+ "step": 300
82
+ },
83
+ {
84
+ "epoch": 0.62,
85
+ "learning_rate": 1e-05,
86
+ "loss": 1.2673,
87
+ "step": 325
88
+ },
89
+ {
90
+ "epoch": 0.66,
91
+ "learning_rate": 1e-05,
92
+ "loss": 1.3683,
93
+ "step": 350
94
+ },
95
+ {
96
+ "epoch": 0.71,
97
+ "learning_rate": 1e-05,
98
+ "loss": 1.3847,
99
+ "step": 375
100
+ },
101
+ {
102
+ "epoch": 0.76,
103
+ "learning_rate": 1e-05,
104
+ "loss": 1.6241,
105
+ "step": 400
106
+ },
107
+ {
108
+ "epoch": 0.76,
109
+ "eval_loss": 1.6757683753967285,
110
+ "eval_runtime": 3755.6247,
111
+ "eval_samples_per_second": 1.489,
112
+ "eval_steps_per_second": 0.186,
113
+ "eval_wer": 102.7664256523106,
114
+ "eval_wer_ortho": 99.64387464387464,
115
+ "step": 400
116
+ },
117
+ {
118
+ "epoch": 0.81,
119
+ "learning_rate": 1e-05,
120
+ "loss": 1.6914,
121
+ "step": 425
122
+ },
123
+ {
124
+ "epoch": 0.85,
125
+ "learning_rate": 1e-05,
126
+ "loss": 1.6943,
127
+ "step": 450
128
+ },
129
+ {
130
+ "epoch": 0.9,
131
+ "learning_rate": 1e-05,
132
+ "loss": 1.7515,
133
+ "step": 475
134
+ },
135
+ {
136
+ "epoch": 0.95,
137
+ "learning_rate": 1e-05,
138
+ "loss": 1.8382,
139
+ "step": 500
140
+ },
141
+ {
142
+ "epoch": 1.0,
143
+ "learning_rate": 1e-05,
144
+ "loss": 1.8555,
145
+ "step": 525
146
+ },
147
+ {
148
+ "epoch": 1.04,
149
+ "learning_rate": 1e-05,
150
+ "loss": 1.9791,
151
+ "step": 550
152
+ },
153
+ {
154
+ "epoch": 1.09,
155
+ "learning_rate": 1e-05,
156
+ "loss": 1.8566,
157
+ "step": 575
158
+ },
159
+ {
160
+ "epoch": 1.14,
161
+ "learning_rate": 1e-05,
162
+ "loss": 2.0313,
163
+ "step": 600
164
+ },
165
+ {
166
+ "epoch": 1.19,
167
+ "learning_rate": 1e-05,
168
+ "loss": 2.019,
169
+ "step": 625
170
+ },
171
+ {
172
+ "epoch": 1.23,
173
+ "learning_rate": 1e-05,
174
+ "loss": 1.9514,
175
+ "step": 650
176
+ },
177
+ {
178
+ "epoch": 1.28,
179
+ "learning_rate": 1e-05,
180
+ "loss": 1.8461,
181
+ "step": 675
182
+ },
183
+ {
184
+ "epoch": 1.33,
185
+ "learning_rate": 1e-05,
186
+ "loss": 2.0464,
187
+ "step": 700
188
+ },
189
+ {
190
+ "epoch": 1.38,
191
+ "learning_rate": 1e-05,
192
+ "loss": 2.0249,
193
+ "step": 725
194
+ },
195
+ {
196
+ "epoch": 1.42,
197
+ "learning_rate": 1e-05,
198
+ "loss": 1.948,
199
+ "step": 750
200
+ },
201
+ {
202
+ "epoch": 1.47,
203
+ "learning_rate": 1e-05,
204
+ "loss": 2.0429,
205
+ "step": 775
206
+ },
207
+ {
208
+ "epoch": 1.52,
209
+ "learning_rate": 1e-05,
210
+ "loss": 2.0215,
211
+ "step": 800
212
+ },
213
+ {
214
+ "epoch": 1.52,
215
+ "eval_loss": 2.010716438293457,
216
+ "eval_runtime": 3081.2923,
217
+ "eval_samples_per_second": 1.815,
218
+ "eval_steps_per_second": 0.227,
219
+ "eval_wer": 100.17290160326941,
220
+ "eval_wer_ortho": 100.0,
221
+ "step": 800
222
+ },
223
+ {
224
+ "epoch": 1.57,
225
+ "learning_rate": 1e-05,
226
+ "loss": 2.1012,
227
+ "step": 825
228
+ },
229
+ {
230
+ "epoch": 1.61,
231
+ "learning_rate": 1e-05,
232
+ "loss": 1.9747,
233
+ "step": 850
234
+ },
235
+ {
236
+ "epoch": 1.66,
237
+ "learning_rate": 1e-05,
238
+ "loss": 1.9878,
239
+ "step": 875
240
+ },
241
+ {
242
+ "epoch": 1.71,
243
+ "learning_rate": 1e-05,
244
+ "loss": 1.9765,
245
+ "step": 900
246
+ },
247
+ {
248
+ "epoch": 1.76,
249
+ "learning_rate": 1e-05,
250
+ "loss": 2.1184,
251
+ "step": 925
252
+ },
253
+ {
254
+ "epoch": 1.8,
255
+ "learning_rate": 1e-05,
256
+ "loss": 1.9324,
257
+ "step": 950
258
+ },
259
+ {
260
+ "epoch": 1.85,
261
+ "learning_rate": 1e-05,
262
+ "loss": 1.9585,
263
+ "step": 975
264
+ },
265
+ {
266
+ "epoch": 1.9,
267
+ "learning_rate": 1e-05,
268
+ "loss": 1.9719,
269
+ "step": 1000
270
+ },
271
+ {
272
+ "epoch": 1.94,
273
+ "learning_rate": 1e-05,
274
+ "loss": 2.0917,
275
+ "step": 1025
276
+ },
277
+ {
278
+ "epoch": 1.99,
279
+ "learning_rate": 1e-05,
280
+ "loss": 2.0203,
281
+ "step": 1050
282
+ },
283
+ {
284
+ "epoch": 2.04,
285
+ "learning_rate": 1e-05,
286
+ "loss": 1.9908,
287
+ "step": 1075
288
+ },
289
+ {
290
+ "epoch": 2.09,
291
+ "learning_rate": 1e-05,
292
+ "loss": 1.9275,
293
+ "step": 1100
294
+ },
295
+ {
296
+ "epoch": 2.13,
297
+ "learning_rate": 1e-05,
298
+ "loss": 1.9532,
299
+ "step": 1125
300
+ },
301
+ {
302
+ "epoch": 2.18,
303
+ "learning_rate": 1e-05,
304
+ "loss": 1.8964,
305
+ "step": 1150
306
+ },
307
+ {
308
+ "epoch": 2.23,
309
+ "learning_rate": 1e-05,
310
+ "loss": 1.904,
311
+ "step": 1175
312
+ },
313
+ {
314
+ "epoch": 2.28,
315
+ "learning_rate": 1e-05,
316
+ "loss": 1.9827,
317
+ "step": 1200
318
+ },
319
+ {
320
+ "epoch": 2.28,
321
+ "eval_loss": 1.9757949113845825,
322
+ "eval_runtime": 2892.4065,
323
+ "eval_samples_per_second": 1.934,
324
+ "eval_steps_per_second": 0.242,
325
+ "eval_wer": 100.0,
326
+ "eval_wer_ortho": 100.0,
327
+ "step": 1200
328
+ },
329
+ {
330
+ "epoch": 2.32,
331
+ "learning_rate": 1e-05,
332
+ "loss": 2.0301,
333
+ "step": 1225
334
+ },
335
+ {
336
+ "epoch": 2.37,
337
+ "learning_rate": 1e-05,
338
+ "loss": 2.1268,
339
+ "step": 1250
340
+ },
341
+ {
342
+ "epoch": 2.42,
343
+ "learning_rate": 1e-05,
344
+ "loss": 2.0044,
345
+ "step": 1275
346
+ },
347
+ {
348
+ "epoch": 2.47,
349
+ "learning_rate": 1e-05,
350
+ "loss": 2.1126,
351
+ "step": 1300
352
+ },
353
+ {
354
+ "epoch": 2.51,
355
+ "learning_rate": 1e-05,
356
+ "loss": 2.02,
357
+ "step": 1325
358
+ },
359
+ {
360
+ "epoch": 2.56,
361
+ "learning_rate": 1e-05,
362
+ "loss": 1.9804,
363
+ "step": 1350
364
+ },
365
+ {
366
+ "epoch": 2.61,
367
+ "learning_rate": 1e-05,
368
+ "loss": 1.9873,
369
+ "step": 1375
370
+ },
371
+ {
372
+ "epoch": 2.66,
373
+ "learning_rate": 1e-05,
374
+ "loss": 2.0043,
375
+ "step": 1400
376
+ },
377
+ {
378
+ "epoch": 2.7,
379
+ "learning_rate": 1e-05,
380
+ "loss": 1.965,
381
+ "step": 1425
382
+ },
383
+ {
384
+ "epoch": 2.75,
385
+ "learning_rate": 1e-05,
386
+ "loss": 1.9411,
387
+ "step": 1450
388
+ },
389
+ {
390
+ "epoch": 2.8,
391
+ "learning_rate": 1e-05,
392
+ "loss": 2.0059,
393
+ "step": 1475
394
+ },
395
+ {
396
+ "epoch": 2.85,
397
+ "learning_rate": 1e-05,
398
+ "loss": 1.8808,
399
+ "step": 1500
400
+ },
401
+ {
402
+ "epoch": 2.89,
403
+ "learning_rate": 1e-05,
404
+ "loss": 1.9022,
405
+ "step": 1525
406
+ },
407
+ {
408
+ "epoch": 2.94,
409
+ "learning_rate": 1e-05,
410
+ "loss": 1.9633,
411
+ "step": 1550
412
+ },
413
+ {
414
+ "epoch": 2.99,
415
+ "learning_rate": 1e-05,
416
+ "loss": 1.8834,
417
+ "step": 1575
418
+ },
419
+ {
420
+ "epoch": 3.04,
421
+ "learning_rate": 1e-05,
422
+ "loss": 2.0119,
423
+ "step": 1600
424
+ },
425
+ {
426
+ "epoch": 3.04,
427
+ "eval_loss": 1.9379667043685913,
428
+ "eval_runtime": 9420.2123,
429
+ "eval_samples_per_second": 0.594,
430
+ "eval_steps_per_second": 0.074,
431
+ "eval_wer": 100.0,
432
+ "eval_wer_ortho": 100.0,
433
+ "step": 1600
434
+ },
435
+ {
436
+ "epoch": 3.08,
437
+ "learning_rate": 1e-05,
438
+ "loss": 1.9068,
439
+ "step": 1625
440
+ },
441
+ {
442
+ "epoch": 3.13,
443
+ "learning_rate": 1e-05,
444
+ "loss": 1.841,
445
+ "step": 1650
446
+ },
447
+ {
448
+ "epoch": 3.18,
449
+ "learning_rate": 1e-05,
450
+ "loss": 1.9803,
451
+ "step": 1675
452
+ },
453
+ {
454
+ "epoch": 3.23,
455
+ "learning_rate": 1e-05,
456
+ "loss": 1.9377,
457
+ "step": 1700
458
+ },
459
+ {
460
+ "epoch": 3.27,
461
+ "learning_rate": 1e-05,
462
+ "loss": 1.9931,
463
+ "step": 1725
464
+ },
465
+ {
466
+ "epoch": 3.32,
467
+ "learning_rate": 1e-05,
468
+ "loss": 1.9932,
469
+ "step": 1750
470
+ },
471
+ {
472
+ "epoch": 3.37,
473
+ "learning_rate": 1e-05,
474
+ "loss": 1.8911,
475
+ "step": 1775
476
+ },
477
+ {
478
+ "epoch": 3.42,
479
+ "learning_rate": 1e-05,
480
+ "loss": 1.8134,
481
+ "step": 1800
482
+ },
483
+ {
484
+ "epoch": 3.46,
485
+ "learning_rate": 1e-05,
486
+ "loss": 1.9463,
487
+ "step": 1825
488
+ },
489
+ {
490
+ "epoch": 3.51,
491
+ "learning_rate": 1e-05,
492
+ "loss": 1.7938,
493
+ "step": 1850
494
+ },
495
+ {
496
+ "epoch": 3.56,
497
+ "learning_rate": 1e-05,
498
+ "loss": 1.8559,
499
+ "step": 1875
500
+ },
501
+ {
502
+ "epoch": 3.61,
503
+ "learning_rate": 1e-05,
504
+ "loss": 1.9709,
505
+ "step": 1900
506
+ },
507
+ {
508
+ "epoch": 3.65,
509
+ "learning_rate": 1e-05,
510
+ "loss": 1.9548,
511
+ "step": 1925
512
+ },
513
+ {
514
+ "epoch": 3.7,
515
+ "learning_rate": 1e-05,
516
+ "loss": 1.8475,
517
+ "step": 1950
518
+ },
519
+ {
520
+ "epoch": 3.75,
521
+ "learning_rate": 1e-05,
522
+ "loss": 1.99,
523
+ "step": 1975
524
+ },
525
+ {
526
+ "epoch": 3.8,
527
+ "learning_rate": 1e-05,
528
+ "loss": 1.8693,
529
+ "step": 2000
530
+ },
531
+ {
532
+ "epoch": 3.8,
533
+ "eval_loss": 1.9047924280166626,
534
+ "eval_runtime": 2877.2302,
535
+ "eval_samples_per_second": 1.944,
536
+ "eval_steps_per_second": 0.243,
537
+ "eval_wer": 100.0,
538
+ "eval_wer_ortho": 100.0,
539
+ "step": 2000
540
+ },
541
+ {
542
+ "epoch": 3.84,
543
+ "learning_rate": 1e-05,
544
+ "loss": 1.925,
545
+ "step": 2025
546
+ },
547
+ {
548
+ "epoch": 3.89,
549
+ "learning_rate": 1e-05,
550
+ "loss": 1.9491,
551
+ "step": 2050
552
+ },
553
+ {
554
+ "epoch": 3.94,
555
+ "learning_rate": 1e-05,
556
+ "loss": 1.9243,
557
+ "step": 2075
558
+ },
559
+ {
560
+ "epoch": 3.98,
561
+ "learning_rate": 1e-05,
562
+ "loss": 1.8867,
563
+ "step": 2100
564
+ },
565
+ {
566
+ "epoch": 4.03,
567
+ "learning_rate": 1e-05,
568
+ "loss": 1.9149,
569
+ "step": 2125
570
+ },
571
+ {
572
+ "epoch": 4.08,
573
+ "learning_rate": 1e-05,
574
+ "loss": 2.0169,
575
+ "step": 2150
576
+ },
577
+ {
578
+ "epoch": 4.13,
579
+ "learning_rate": 1e-05,
580
+ "loss": 1.8339,
581
+ "step": 2175
582
+ },
583
+ {
584
+ "epoch": 4.17,
585
+ "learning_rate": 1e-05,
586
+ "loss": 1.8081,
587
+ "step": 2200
588
+ },
589
+ {
590
+ "epoch": 4.22,
591
+ "learning_rate": 1e-05,
592
+ "loss": 1.9069,
593
+ "step": 2225
594
+ },
595
+ {
596
+ "epoch": 4.27,
597
+ "learning_rate": 1e-05,
598
+ "loss": 1.8723,
599
+ "step": 2250
600
+ },
601
+ {
602
+ "epoch": 4.32,
603
+ "learning_rate": 1e-05,
604
+ "loss": 1.8821,
605
+ "step": 2275
606
+ },
607
+ {
608
+ "epoch": 4.36,
609
+ "learning_rate": 1e-05,
610
+ "loss": 1.8486,
611
+ "step": 2300
612
+ },
613
+ {
614
+ "epoch": 4.41,
615
+ "learning_rate": 1e-05,
616
+ "loss": 1.9235,
617
+ "step": 2325
618
+ },
619
+ {
620
+ "epoch": 4.46,
621
+ "learning_rate": 1e-05,
622
+ "loss": 1.838,
623
+ "step": 2350
624
+ },
625
+ {
626
+ "epoch": 4.51,
627
+ "learning_rate": 1e-05,
628
+ "loss": 1.8834,
629
+ "step": 2375
630
+ },
631
+ {
632
+ "epoch": 4.55,
633
+ "learning_rate": 1e-05,
634
+ "loss": 1.9805,
635
+ "step": 2400
636
+ },
637
+ {
638
+ "epoch": 4.55,
639
+ "eval_loss": 1.8733848333358765,
640
+ "eval_runtime": 2872.4313,
641
+ "eval_samples_per_second": 1.947,
642
+ "eval_steps_per_second": 0.244,
643
+ "eval_wer": 100.01571832756994,
644
+ "eval_wer_ortho": 100.0,
645
+ "step": 2400
646
+ },
647
+ {
648
+ "epoch": 4.6,
649
+ "learning_rate": 1e-05,
650
+ "loss": 1.9637,
651
+ "step": 2425
652
+ },
653
+ {
654
+ "epoch": 4.65,
655
+ "learning_rate": 1e-05,
656
+ "loss": 1.836,
657
+ "step": 2450
658
+ },
659
+ {
660
+ "epoch": 4.7,
661
+ "learning_rate": 1e-05,
662
+ "loss": 1.8557,
663
+ "step": 2475
664
+ },
665
+ {
666
+ "epoch": 4.74,
667
+ "learning_rate": 1e-05,
668
+ "loss": 1.8696,
669
+ "step": 2500
670
+ },
671
+ {
672
+ "epoch": 4.79,
673
+ "learning_rate": 1e-05,
674
+ "loss": 1.8709,
675
+ "step": 2525
676
+ },
677
+ {
678
+ "epoch": 4.84,
679
+ "learning_rate": 1e-05,
680
+ "loss": 1.9039,
681
+ "step": 2550
682
+ },
683
+ {
684
+ "epoch": 4.89,
685
+ "learning_rate": 1e-05,
686
+ "loss": 1.9168,
687
+ "step": 2575
688
+ },
689
+ {
690
+ "epoch": 4.93,
691
+ "learning_rate": 1e-05,
692
+ "loss": 1.7614,
693
+ "step": 2600
694
+ },
695
+ {
696
+ "epoch": 4.98,
697
+ "learning_rate": 1e-05,
698
+ "loss": 1.9177,
699
+ "step": 2625
700
+ },
701
+ {
702
+ "epoch": 5.03,
703
+ "learning_rate": 1e-05,
704
+ "loss": 1.8272,
705
+ "step": 2650
706
+ },
707
+ {
708
+ "epoch": 5.08,
709
+ "learning_rate": 1e-05,
710
+ "loss": 1.8066,
711
+ "step": 2675
712
+ },
713
+ {
714
+ "epoch": 5.12,
715
+ "learning_rate": 1e-05,
716
+ "loss": 1.8952,
717
+ "step": 2700
718
+ },
719
+ {
720
+ "epoch": 5.17,
721
+ "learning_rate": 1e-05,
722
+ "loss": 1.857,
723
+ "step": 2725
724
+ },
725
+ {
726
+ "epoch": 5.22,
727
+ "learning_rate": 1e-05,
728
+ "loss": 1.8715,
729
+ "step": 2750
730
+ },
731
+ {
732
+ "epoch": 5.27,
733
+ "learning_rate": 1e-05,
734
+ "loss": 1.8985,
735
+ "step": 2775
736
+ },
737
+ {
738
+ "epoch": 5.31,
739
+ "learning_rate": 1e-05,
740
+ "loss": 1.8382,
741
+ "step": 2800
742
+ },
743
+ {
744
+ "epoch": 5.31,
745
+ "eval_loss": 1.8439116477966309,
746
+ "eval_runtime": 2891.6652,
747
+ "eval_samples_per_second": 1.935,
748
+ "eval_steps_per_second": 0.242,
749
+ "eval_wer": 100.01571832756994,
750
+ "eval_wer_ortho": 100.0,
751
+ "step": 2800
752
+ },
753
+ {
754
+ "epoch": 5.36,
755
+ "learning_rate": 1e-05,
756
+ "loss": 1.9262,
757
+ "step": 2825
758
+ },
759
+ {
760
+ "epoch": 5.41,
761
+ "learning_rate": 1e-05,
762
+ "loss": 1.9521,
763
+ "step": 2850
764
+ },
765
+ {
766
+ "epoch": 5.46,
767
+ "learning_rate": 1e-05,
768
+ "loss": 1.7268,
769
+ "step": 2875
770
+ },
771
+ {
772
+ "epoch": 5.5,
773
+ "learning_rate": 1e-05,
774
+ "loss": 1.7545,
775
+ "step": 2900
776
+ },
777
+ {
778
+ "epoch": 5.55,
779
+ "learning_rate": 1e-05,
780
+ "loss": 1.904,
781
+ "step": 2925
782
+ },
783
+ {
784
+ "epoch": 5.6,
785
+ "learning_rate": 1e-05,
786
+ "loss": 1.8546,
787
+ "step": 2950
788
+ },
789
+ {
790
+ "epoch": 5.65,
791
+ "learning_rate": 1e-05,
792
+ "loss": 1.7786,
793
+ "step": 2975
794
+ },
795
+ {
796
+ "epoch": 5.69,
797
+ "learning_rate": 1e-05,
798
+ "loss": 1.7294,
799
+ "step": 3000
800
+ },
801
+ {
802
+ "epoch": 5.74,
803
+ "learning_rate": 1e-05,
804
+ "loss": 1.758,
805
+ "step": 3025
806
+ },
807
+ {
808
+ "epoch": 5.79,
809
+ "learning_rate": 1e-05,
810
+ "loss": 1.7159,
811
+ "step": 3050
812
+ },
813
+ {
814
+ "epoch": 5.83,
815
+ "learning_rate": 1e-05,
816
+ "loss": 1.8201,
817
+ "step": 3075
818
+ },
819
+ {
820
+ "epoch": 5.88,
821
+ "learning_rate": 1e-05,
822
+ "loss": 1.8107,
823
+ "step": 3100
824
+ },
825
+ {
826
+ "epoch": 5.93,
827
+ "learning_rate": 1e-05,
828
+ "loss": 1.8754,
829
+ "step": 3125
830
+ },
831
+ {
832
+ "epoch": 5.98,
833
+ "learning_rate": 1e-05,
834
+ "loss": 1.9456,
835
+ "step": 3150
836
+ },
837
+ {
838
+ "epoch": 6.02,
839
+ "learning_rate": 1e-05,
840
+ "loss": 1.8856,
841
+ "step": 3175
842
+ },
843
+ {
844
+ "epoch": 6.07,
845
+ "learning_rate": 1e-05,
846
+ "loss": 1.818,
847
+ "step": 3200
848
+ },
849
+ {
850
+ "epoch": 6.07,
851
+ "eval_loss": 1.8148053884506226,
852
+ "eval_runtime": 2870.6751,
853
+ "eval_samples_per_second": 1.949,
854
+ "eval_steps_per_second": 0.244,
855
+ "eval_wer": 100.0314366551399,
856
+ "eval_wer_ortho": 100.0,
857
+ "step": 3200
858
+ },
859
+ {
860
+ "epoch": 6.12,
861
+ "learning_rate": 1e-05,
862
+ "loss": 1.7602,
863
+ "step": 3225
864
+ },
865
+ {
866
+ "epoch": 6.17,
867
+ "learning_rate": 1e-05,
868
+ "loss": 1.8412,
869
+ "step": 3250
870
+ },
871
+ {
872
+ "epoch": 6.21,
873
+ "learning_rate": 1e-05,
874
+ "loss": 1.7344,
875
+ "step": 3275
876
+ },
877
+ {
878
+ "epoch": 6.26,
879
+ "learning_rate": 1e-05,
880
+ "loss": 1.7457,
881
+ "step": 3300
882
+ },
883
+ {
884
+ "epoch": 6.31,
885
+ "learning_rate": 1e-05,
886
+ "loss": 1.7492,
887
+ "step": 3325
888
+ },
889
+ {
890
+ "epoch": 6.36,
891
+ "learning_rate": 1e-05,
892
+ "loss": 1.8071,
893
+ "step": 3350
894
+ },
895
+ {
896
+ "epoch": 6.4,
897
+ "learning_rate": 1e-05,
898
+ "loss": 1.8119,
899
+ "step": 3375
900
+ },
901
+ {
902
+ "epoch": 6.45,
903
+ "learning_rate": 1e-05,
904
+ "loss": 1.9135,
905
+ "step": 3400
906
+ },
907
+ {
908
+ "epoch": 6.5,
909
+ "learning_rate": 1e-05,
910
+ "loss": 1.8219,
911
+ "step": 3425
912
+ },
913
+ {
914
+ "epoch": 6.55,
915
+ "learning_rate": 1e-05,
916
+ "loss": 1.8165,
917
+ "step": 3450
918
+ },
919
+ {
920
+ "epoch": 6.59,
921
+ "learning_rate": 1e-05,
922
+ "loss": 1.8399,
923
+ "step": 3475
924
+ },
925
+ {
926
+ "epoch": 6.64,
927
+ "learning_rate": 1e-05,
928
+ "loss": 1.8227,
929
+ "step": 3500
930
+ },
931
+ {
932
+ "epoch": 6.69,
933
+ "learning_rate": 1e-05,
934
+ "loss": 1.6899,
935
+ "step": 3525
936
+ },
937
+ {
938
+ "epoch": 6.74,
939
+ "learning_rate": 1e-05,
940
+ "loss": 1.7632,
941
+ "step": 3550
942
+ },
943
+ {
944
+ "epoch": 6.78,
945
+ "learning_rate": 1e-05,
946
+ "loss": 1.7686,
947
+ "step": 3575
948
+ },
949
+ {
950
+ "epoch": 6.83,
951
+ "learning_rate": 1e-05,
952
+ "loss": 1.8491,
953
+ "step": 3600
954
+ },
955
+ {
956
+ "epoch": 6.83,
957
+ "eval_loss": 1.7864172458648682,
958
+ "eval_runtime": 3734.5543,
959
+ "eval_samples_per_second": 1.498,
960
+ "eval_steps_per_second": 0.187,
961
+ "eval_wer": 100.0,
962
+ "eval_wer_ortho": 100.0,
963
+ "step": 3600
964
+ }
965
+ ],
966
+ "logging_steps": 25,
967
+ "max_steps": 20000,
968
+ "num_input_tokens_seen": 0,
969
+ "num_train_epochs": 38,
970
+ "save_steps": 400,
971
+ "total_flos": 1.661732452122624e+19,
972
+ "train_batch_size": 8,
973
+ "trial_name": null,
974
+ "trial_params": null
975
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a226a83354ab508a1eda244aa9b0b84448dd5da345f4638b4ac7f34496c30d67
3
+ size 4920