soaring0616 commited on
Commit
37dc5ca
1 Parent(s): 7f36220

Upload folder using huggingface_hub

Browse files
checkpoint-1000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "soaring0616/whisper-small-vietnamese",
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": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "lang_to_id": {
10
+ "<|id|>": 50275,
11
+ "<|vi|>": 50278
12
+ },
13
+ "language": "indonesian",
14
+ "max_length": 448,
15
+ "pad_token_id": 50257,
16
+ "suppress_tokens": [],
17
+ "transformers_version": "4.39.1"
18
+ }
checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:409b131b2b6b89b1a619cce45c0ef4d23c303ee2d6a154a7fb195e6dc9acffca
3
+ size 966995080
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b5908c0d8823cef594cd13852f55721e5f728e301254489c2122a9941926448
3
+ size 1925064044
checkpoint-1000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c29334f857fb39218b3eb2c67acb056e10392cedb3f61bdf7b6329defc27b559
3
+ size 14244
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b05e09481e16be6920637a3fff45223880ebd6057d17611017bd0d9e08ae606b
3
+ size 1064
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,310 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 17.814782082887575,
3
+ "best_model_checkpoint": "./whisper-small-in_from_vi\\checkpoint-1000",
4
+ "epoch": 1.9305019305019306,
5
+ "eval_steps": 1000,
6
+ "global_step": 1000,
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
+ "grad_norm": 218.94273376464844,
14
+ "learning_rate": 3.8e-07,
15
+ "loss": 5.6914,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.1,
20
+ "grad_norm": 35.336517333984375,
21
+ "learning_rate": 8.8e-07,
22
+ "loss": 2.5237,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.14,
27
+ "grad_norm": 14.206599235534668,
28
+ "learning_rate": 1.3800000000000001e-06,
29
+ "loss": 0.8446,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.19,
34
+ "grad_norm": 12.359480857849121,
35
+ "learning_rate": 1.8800000000000002e-06,
36
+ "loss": 0.579,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.24,
41
+ "grad_norm": 8.259902954101562,
42
+ "learning_rate": 2.38e-06,
43
+ "loss": 0.5636,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.29,
48
+ "grad_norm": 11.162034034729004,
49
+ "learning_rate": 2.88e-06,
50
+ "loss": 0.5191,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 0.34,
55
+ "grad_norm": 8.917764663696289,
56
+ "learning_rate": 3.3800000000000007e-06,
57
+ "loss": 0.4879,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 0.39,
62
+ "grad_norm": 8.865296363830566,
63
+ "learning_rate": 3.88e-06,
64
+ "loss": 0.4333,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 0.43,
69
+ "grad_norm": 8.579310417175293,
70
+ "learning_rate": 4.38e-06,
71
+ "loss": 0.4679,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 0.48,
76
+ "grad_norm": 8.361818313598633,
77
+ "learning_rate": 4.880000000000001e-06,
78
+ "loss": 0.3948,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 0.53,
83
+ "grad_norm": 6.773808479309082,
84
+ "learning_rate": 5.380000000000001e-06,
85
+ "loss": 0.4284,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 0.58,
90
+ "grad_norm": 7.8365678787231445,
91
+ "learning_rate": 5.8800000000000005e-06,
92
+ "loss": 0.3829,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 0.63,
97
+ "grad_norm": 10.097782135009766,
98
+ "learning_rate": 6.380000000000001e-06,
99
+ "loss": 0.4091,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 0.68,
104
+ "grad_norm": 8.221065521240234,
105
+ "learning_rate": 6.88e-06,
106
+ "loss": 0.4067,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 0.72,
111
+ "grad_norm": 7.281307697296143,
112
+ "learning_rate": 7.3800000000000005e-06,
113
+ "loss": 0.3501,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 0.77,
118
+ "grad_norm": 7.796460151672363,
119
+ "learning_rate": 7.88e-06,
120
+ "loss": 0.3754,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 0.82,
125
+ "grad_norm": 6.829456329345703,
126
+ "learning_rate": 8.380000000000001e-06,
127
+ "loss": 0.3789,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 0.87,
132
+ "grad_norm": 7.42793607711792,
133
+ "learning_rate": 8.880000000000001e-06,
134
+ "loss": 0.3709,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 0.92,
139
+ "grad_norm": 6.301506519317627,
140
+ "learning_rate": 9.38e-06,
141
+ "loss": 0.355,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 0.97,
146
+ "grad_norm": 7.611047744750977,
147
+ "learning_rate": 9.88e-06,
148
+ "loss": 0.3347,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 1.01,
153
+ "grad_norm": 6.571840286254883,
154
+ "learning_rate": 9.945714285714286e-06,
155
+ "loss": 0.3231,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 1.06,
160
+ "grad_norm": 6.700900554656982,
161
+ "learning_rate": 9.874285714285715e-06,
162
+ "loss": 0.2356,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 1.11,
167
+ "grad_norm": 6.217418670654297,
168
+ "learning_rate": 9.802857142857144e-06,
169
+ "loss": 0.2177,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 1.16,
174
+ "grad_norm": 5.774504661560059,
175
+ "learning_rate": 9.731428571428573e-06,
176
+ "loss": 0.2103,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 1.21,
181
+ "grad_norm": 5.770651340484619,
182
+ "learning_rate": 9.66e-06,
183
+ "loss": 0.2529,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 1.25,
188
+ "grad_norm": 5.579860687255859,
189
+ "learning_rate": 9.58857142857143e-06,
190
+ "loss": 0.2435,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 1.3,
195
+ "grad_norm": 5.943060398101807,
196
+ "learning_rate": 9.517142857142859e-06,
197
+ "loss": 0.2064,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 1.35,
202
+ "grad_norm": 7.721786975860596,
203
+ "learning_rate": 9.445714285714288e-06,
204
+ "loss": 0.2273,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 1.4,
209
+ "grad_norm": 6.976935386657715,
210
+ "learning_rate": 9.374285714285715e-06,
211
+ "loss": 0.2373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 1.45,
216
+ "grad_norm": 6.749694347381592,
217
+ "learning_rate": 9.302857142857144e-06,
218
+ "loss": 0.2306,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 1.5,
223
+ "grad_norm": 5.087153434753418,
224
+ "learning_rate": 9.231428571428573e-06,
225
+ "loss": 0.2292,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 1.54,
230
+ "grad_norm": 6.032232761383057,
231
+ "learning_rate": 9.16e-06,
232
+ "loss": 0.2096,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 1.59,
237
+ "grad_norm": 5.749306678771973,
238
+ "learning_rate": 9.08857142857143e-06,
239
+ "loss": 0.2217,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 1.64,
244
+ "grad_norm": 7.6291184425354,
245
+ "learning_rate": 9.017142857142858e-06,
246
+ "loss": 0.2023,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 1.69,
251
+ "grad_norm": 4.76064395904541,
252
+ "learning_rate": 8.945714285714286e-06,
253
+ "loss": 0.2099,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 1.74,
258
+ "grad_norm": 5.0226898193359375,
259
+ "learning_rate": 8.874285714285715e-06,
260
+ "loss": 0.2017,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 1.79,
265
+ "grad_norm": 5.6098313331604,
266
+ "learning_rate": 8.802857142857144e-06,
267
+ "loss": 0.2126,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 1.83,
272
+ "grad_norm": 5.58658504486084,
273
+ "learning_rate": 8.731428571428571e-06,
274
+ "loss": 0.2209,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 1.88,
279
+ "grad_norm": 7.067436218261719,
280
+ "learning_rate": 8.66e-06,
281
+ "loss": 0.225,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 1.93,
286
+ "grad_norm": 4.616860389709473,
287
+ "learning_rate": 8.588571428571429e-06,
288
+ "loss": 0.2059,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 1.93,
293
+ "eval_loss": 0.25656187534332275,
294
+ "eval_runtime": 2791.9455,
295
+ "eval_samples_per_second": 1.296,
296
+ "eval_steps_per_second": 0.162,
297
+ "eval_wer": 17.814782082887575,
298
+ "step": 1000
299
+ }
300
+ ],
301
+ "logging_steps": 25,
302
+ "max_steps": 4000,
303
+ "num_input_tokens_seen": 0,
304
+ "num_train_epochs": 8,
305
+ "save_steps": 1000,
306
+ "total_flos": 4.61332620951552e+18,
307
+ "train_batch_size": 16,
308
+ "trial_name": null,
309
+ "trial_params": null
310
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d97a7d674ba5458016027e8fcbeac271e6fe45c463b2f89dfb01782b9ae4928d
3
+ size 5112
checkpoint-2000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "soaring0616/whisper-small-vietnamese",
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": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-2000/generation_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "lang_to_id": {
10
+ "<|id|>": 50275,
11
+ "<|vi|>": 50278
12
+ },
13
+ "language": "indonesian",
14
+ "max_length": 448,
15
+ "pad_token_id": 50257,
16
+ "suppress_tokens": [],
17
+ "transformers_version": "4.39.1"
18
+ }
checkpoint-2000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9bdf8dd5e50b4d747fd632fcb05a6eb5f0aa1e71789bd2279a8601856f952cd0
3
+ size 966995080
checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d731de9f880a4937bbca96b4b05942ee9956648112b9950c1e0e271edf92c7b
3
+ size 1925064044
checkpoint-2000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-2000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f7a343a22814950dc78f0eee3ac380f9abc6236ec1a214b3aa30607b846d41fb
3
+ size 14244
checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a943107e4f3199cccea374c098af12bfc7ca97eb99b7cc85cdbea246fbef66a9
3
+ size 1064
checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,599 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 17.470580026978,
3
+ "best_model_checkpoint": "./whisper-small-in_from_vi\\checkpoint-2000",
4
+ "epoch": 3.861003861003861,
5
+ "eval_steps": 1000,
6
+ "global_step": 2000,
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
+ "grad_norm": 218.94273376464844,
14
+ "learning_rate": 3.8e-07,
15
+ "loss": 5.6914,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.1,
20
+ "grad_norm": 35.336517333984375,
21
+ "learning_rate": 8.8e-07,
22
+ "loss": 2.5237,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.14,
27
+ "grad_norm": 14.206599235534668,
28
+ "learning_rate": 1.3800000000000001e-06,
29
+ "loss": 0.8446,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.19,
34
+ "grad_norm": 12.359480857849121,
35
+ "learning_rate": 1.8800000000000002e-06,
36
+ "loss": 0.579,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.24,
41
+ "grad_norm": 8.259902954101562,
42
+ "learning_rate": 2.38e-06,
43
+ "loss": 0.5636,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.29,
48
+ "grad_norm": 11.162034034729004,
49
+ "learning_rate": 2.88e-06,
50
+ "loss": 0.5191,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 0.34,
55
+ "grad_norm": 8.917764663696289,
56
+ "learning_rate": 3.3800000000000007e-06,
57
+ "loss": 0.4879,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 0.39,
62
+ "grad_norm": 8.865296363830566,
63
+ "learning_rate": 3.88e-06,
64
+ "loss": 0.4333,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 0.43,
69
+ "grad_norm": 8.579310417175293,
70
+ "learning_rate": 4.38e-06,
71
+ "loss": 0.4679,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 0.48,
76
+ "grad_norm": 8.361818313598633,
77
+ "learning_rate": 4.880000000000001e-06,
78
+ "loss": 0.3948,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 0.53,
83
+ "grad_norm": 6.773808479309082,
84
+ "learning_rate": 5.380000000000001e-06,
85
+ "loss": 0.4284,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 0.58,
90
+ "grad_norm": 7.8365678787231445,
91
+ "learning_rate": 5.8800000000000005e-06,
92
+ "loss": 0.3829,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 0.63,
97
+ "grad_norm": 10.097782135009766,
98
+ "learning_rate": 6.380000000000001e-06,
99
+ "loss": 0.4091,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 0.68,
104
+ "grad_norm": 8.221065521240234,
105
+ "learning_rate": 6.88e-06,
106
+ "loss": 0.4067,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 0.72,
111
+ "grad_norm": 7.281307697296143,
112
+ "learning_rate": 7.3800000000000005e-06,
113
+ "loss": 0.3501,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 0.77,
118
+ "grad_norm": 7.796460151672363,
119
+ "learning_rate": 7.88e-06,
120
+ "loss": 0.3754,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 0.82,
125
+ "grad_norm": 6.829456329345703,
126
+ "learning_rate": 8.380000000000001e-06,
127
+ "loss": 0.3789,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 0.87,
132
+ "grad_norm": 7.42793607711792,
133
+ "learning_rate": 8.880000000000001e-06,
134
+ "loss": 0.3709,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 0.92,
139
+ "grad_norm": 6.301506519317627,
140
+ "learning_rate": 9.38e-06,
141
+ "loss": 0.355,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 0.97,
146
+ "grad_norm": 7.611047744750977,
147
+ "learning_rate": 9.88e-06,
148
+ "loss": 0.3347,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 1.01,
153
+ "grad_norm": 6.571840286254883,
154
+ "learning_rate": 9.945714285714286e-06,
155
+ "loss": 0.3231,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 1.06,
160
+ "grad_norm": 6.700900554656982,
161
+ "learning_rate": 9.874285714285715e-06,
162
+ "loss": 0.2356,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 1.11,
167
+ "grad_norm": 6.217418670654297,
168
+ "learning_rate": 9.802857142857144e-06,
169
+ "loss": 0.2177,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 1.16,
174
+ "grad_norm": 5.774504661560059,
175
+ "learning_rate": 9.731428571428573e-06,
176
+ "loss": 0.2103,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 1.21,
181
+ "grad_norm": 5.770651340484619,
182
+ "learning_rate": 9.66e-06,
183
+ "loss": 0.2529,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 1.25,
188
+ "grad_norm": 5.579860687255859,
189
+ "learning_rate": 9.58857142857143e-06,
190
+ "loss": 0.2435,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 1.3,
195
+ "grad_norm": 5.943060398101807,
196
+ "learning_rate": 9.517142857142859e-06,
197
+ "loss": 0.2064,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 1.35,
202
+ "grad_norm": 7.721786975860596,
203
+ "learning_rate": 9.445714285714288e-06,
204
+ "loss": 0.2273,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 1.4,
209
+ "grad_norm": 6.976935386657715,
210
+ "learning_rate": 9.374285714285715e-06,
211
+ "loss": 0.2373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 1.45,
216
+ "grad_norm": 6.749694347381592,
217
+ "learning_rate": 9.302857142857144e-06,
218
+ "loss": 0.2306,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 1.5,
223
+ "grad_norm": 5.087153434753418,
224
+ "learning_rate": 9.231428571428573e-06,
225
+ "loss": 0.2292,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 1.54,
230
+ "grad_norm": 6.032232761383057,
231
+ "learning_rate": 9.16e-06,
232
+ "loss": 0.2096,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 1.59,
237
+ "grad_norm": 5.749306678771973,
238
+ "learning_rate": 9.08857142857143e-06,
239
+ "loss": 0.2217,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 1.64,
244
+ "grad_norm": 7.6291184425354,
245
+ "learning_rate": 9.017142857142858e-06,
246
+ "loss": 0.2023,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 1.69,
251
+ "grad_norm": 4.76064395904541,
252
+ "learning_rate": 8.945714285714286e-06,
253
+ "loss": 0.2099,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 1.74,
258
+ "grad_norm": 5.0226898193359375,
259
+ "learning_rate": 8.874285714285715e-06,
260
+ "loss": 0.2017,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 1.79,
265
+ "grad_norm": 5.6098313331604,
266
+ "learning_rate": 8.802857142857144e-06,
267
+ "loss": 0.2126,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 1.83,
272
+ "grad_norm": 5.58658504486084,
273
+ "learning_rate": 8.731428571428571e-06,
274
+ "loss": 0.2209,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 1.88,
279
+ "grad_norm": 7.067436218261719,
280
+ "learning_rate": 8.66e-06,
281
+ "loss": 0.225,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 1.93,
286
+ "grad_norm": 4.616860389709473,
287
+ "learning_rate": 8.588571428571429e-06,
288
+ "loss": 0.2059,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 1.93,
293
+ "eval_loss": 0.25656187534332275,
294
+ "eval_runtime": 2791.9455,
295
+ "eval_samples_per_second": 1.296,
296
+ "eval_steps_per_second": 0.162,
297
+ "eval_wer": 17.814782082887575,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 1.98,
302
+ "grad_norm": 4.517664909362793,
303
+ "learning_rate": 8.517142857142858e-06,
304
+ "loss": 0.2132,
305
+ "step": 1025
306
+ },
307
+ {
308
+ "epoch": 2.03,
309
+ "grad_norm": 3.016611099243164,
310
+ "learning_rate": 8.445714285714285e-06,
311
+ "loss": 0.1498,
312
+ "step": 1050
313
+ },
314
+ {
315
+ "epoch": 2.08,
316
+ "grad_norm": 4.563220977783203,
317
+ "learning_rate": 8.374285714285714e-06,
318
+ "loss": 0.0848,
319
+ "step": 1075
320
+ },
321
+ {
322
+ "epoch": 2.12,
323
+ "grad_norm": 4.658745288848877,
324
+ "learning_rate": 8.302857142857143e-06,
325
+ "loss": 0.0877,
326
+ "step": 1100
327
+ },
328
+ {
329
+ "epoch": 2.17,
330
+ "grad_norm": 3.2172155380249023,
331
+ "learning_rate": 8.231428571428572e-06,
332
+ "loss": 0.0858,
333
+ "step": 1125
334
+ },
335
+ {
336
+ "epoch": 2.22,
337
+ "grad_norm": 3.33174991607666,
338
+ "learning_rate": 8.16e-06,
339
+ "loss": 0.08,
340
+ "step": 1150
341
+ },
342
+ {
343
+ "epoch": 2.27,
344
+ "grad_norm": 4.132905960083008,
345
+ "learning_rate": 8.088571428571429e-06,
346
+ "loss": 0.0945,
347
+ "step": 1175
348
+ },
349
+ {
350
+ "epoch": 2.32,
351
+ "grad_norm": 3.5062320232391357,
352
+ "learning_rate": 8.017142857142858e-06,
353
+ "loss": 0.0849,
354
+ "step": 1200
355
+ },
356
+ {
357
+ "epoch": 2.36,
358
+ "grad_norm": 2.746927261352539,
359
+ "learning_rate": 7.945714285714287e-06,
360
+ "loss": 0.0814,
361
+ "step": 1225
362
+ },
363
+ {
364
+ "epoch": 2.41,
365
+ "grad_norm": 2.8374855518341064,
366
+ "learning_rate": 7.874285714285716e-06,
367
+ "loss": 0.0898,
368
+ "step": 1250
369
+ },
370
+ {
371
+ "epoch": 2.46,
372
+ "grad_norm": 2.8154654502868652,
373
+ "learning_rate": 7.802857142857143e-06,
374
+ "loss": 0.091,
375
+ "step": 1275
376
+ },
377
+ {
378
+ "epoch": 2.51,
379
+ "grad_norm": 3.5242350101470947,
380
+ "learning_rate": 7.731428571428572e-06,
381
+ "loss": 0.085,
382
+ "step": 1300
383
+ },
384
+ {
385
+ "epoch": 2.56,
386
+ "grad_norm": 3.3596408367156982,
387
+ "learning_rate": 7.660000000000001e-06,
388
+ "loss": 0.093,
389
+ "step": 1325
390
+ },
391
+ {
392
+ "epoch": 2.61,
393
+ "grad_norm": 3.4869582653045654,
394
+ "learning_rate": 7.588571428571429e-06,
395
+ "loss": 0.0836,
396
+ "step": 1350
397
+ },
398
+ {
399
+ "epoch": 2.65,
400
+ "grad_norm": 2.8120055198669434,
401
+ "learning_rate": 7.5171428571428575e-06,
402
+ "loss": 0.0769,
403
+ "step": 1375
404
+ },
405
+ {
406
+ "epoch": 2.7,
407
+ "grad_norm": 3.5809738636016846,
408
+ "learning_rate": 7.445714285714286e-06,
409
+ "loss": 0.0916,
410
+ "step": 1400
411
+ },
412
+ {
413
+ "epoch": 2.75,
414
+ "grad_norm": 2.8839168548583984,
415
+ "learning_rate": 7.374285714285715e-06,
416
+ "loss": 0.09,
417
+ "step": 1425
418
+ },
419
+ {
420
+ "epoch": 2.8,
421
+ "grad_norm": 4.721149921417236,
422
+ "learning_rate": 7.302857142857144e-06,
423
+ "loss": 0.0915,
424
+ "step": 1450
425
+ },
426
+ {
427
+ "epoch": 2.85,
428
+ "grad_norm": 3.5091800689697266,
429
+ "learning_rate": 7.231428571428573e-06,
430
+ "loss": 0.0794,
431
+ "step": 1475
432
+ },
433
+ {
434
+ "epoch": 2.9,
435
+ "grad_norm": 3.820974349975586,
436
+ "learning_rate": 7.16e-06,
437
+ "loss": 0.0934,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 2.94,
442
+ "grad_norm": 4.247288227081299,
443
+ "learning_rate": 7.088571428571429e-06,
444
+ "loss": 0.0822,
445
+ "step": 1525
446
+ },
447
+ {
448
+ "epoch": 2.99,
449
+ "grad_norm": 3.3485350608825684,
450
+ "learning_rate": 7.017142857142858e-06,
451
+ "loss": 0.0971,
452
+ "step": 1550
453
+ },
454
+ {
455
+ "epoch": 3.04,
456
+ "grad_norm": 1.4603108167648315,
457
+ "learning_rate": 6.945714285714287e-06,
458
+ "loss": 0.0352,
459
+ "step": 1575
460
+ },
461
+ {
462
+ "epoch": 3.09,
463
+ "grad_norm": 1.7609468698501587,
464
+ "learning_rate": 6.874285714285714e-06,
465
+ "loss": 0.0402,
466
+ "step": 1600
467
+ },
468
+ {
469
+ "epoch": 3.14,
470
+ "grad_norm": 2.6159989833831787,
471
+ "learning_rate": 6.8028571428571434e-06,
472
+ "loss": 0.0317,
473
+ "step": 1625
474
+ },
475
+ {
476
+ "epoch": 3.19,
477
+ "grad_norm": 1.857428789138794,
478
+ "learning_rate": 6.7314285714285724e-06,
479
+ "loss": 0.039,
480
+ "step": 1650
481
+ },
482
+ {
483
+ "epoch": 3.23,
484
+ "grad_norm": 3.3788986206054688,
485
+ "learning_rate": 6.660000000000001e-06,
486
+ "loss": 0.0347,
487
+ "step": 1675
488
+ },
489
+ {
490
+ "epoch": 3.28,
491
+ "grad_norm": 1.8846665620803833,
492
+ "learning_rate": 6.588571428571429e-06,
493
+ "loss": 0.0321,
494
+ "step": 1700
495
+ },
496
+ {
497
+ "epoch": 3.33,
498
+ "grad_norm": 1.2387264966964722,
499
+ "learning_rate": 6.517142857142858e-06,
500
+ "loss": 0.0327,
501
+ "step": 1725
502
+ },
503
+ {
504
+ "epoch": 3.38,
505
+ "grad_norm": 1.4381897449493408,
506
+ "learning_rate": 6.445714285714286e-06,
507
+ "loss": 0.0354,
508
+ "step": 1750
509
+ },
510
+ {
511
+ "epoch": 3.43,
512
+ "grad_norm": 2.2566747665405273,
513
+ "learning_rate": 6.374285714285715e-06,
514
+ "loss": 0.0368,
515
+ "step": 1775
516
+ },
517
+ {
518
+ "epoch": 3.47,
519
+ "grad_norm": 1.6017179489135742,
520
+ "learning_rate": 6.302857142857144e-06,
521
+ "loss": 0.0351,
522
+ "step": 1800
523
+ },
524
+ {
525
+ "epoch": 3.52,
526
+ "grad_norm": 3.2970995903015137,
527
+ "learning_rate": 6.231428571428571e-06,
528
+ "loss": 0.0332,
529
+ "step": 1825
530
+ },
531
+ {
532
+ "epoch": 3.57,
533
+ "grad_norm": 2.8597681522369385,
534
+ "learning_rate": 6.16e-06,
535
+ "loss": 0.0355,
536
+ "step": 1850
537
+ },
538
+ {
539
+ "epoch": 3.62,
540
+ "grad_norm": 2.8335232734680176,
541
+ "learning_rate": 6.088571428571429e-06,
542
+ "loss": 0.0338,
543
+ "step": 1875
544
+ },
545
+ {
546
+ "epoch": 3.67,
547
+ "grad_norm": 3.405618667602539,
548
+ "learning_rate": 6.017142857142858e-06,
549
+ "loss": 0.0376,
550
+ "step": 1900
551
+ },
552
+ {
553
+ "epoch": 3.72,
554
+ "grad_norm": 2.4519917964935303,
555
+ "learning_rate": 5.945714285714286e-06,
556
+ "loss": 0.0359,
557
+ "step": 1925
558
+ },
559
+ {
560
+ "epoch": 3.76,
561
+ "grad_norm": 3.2067463397979736,
562
+ "learning_rate": 5.874285714285715e-06,
563
+ "loss": 0.029,
564
+ "step": 1950
565
+ },
566
+ {
567
+ "epoch": 3.81,
568
+ "grad_norm": 2.5410799980163574,
569
+ "learning_rate": 5.802857142857144e-06,
570
+ "loss": 0.0313,
571
+ "step": 1975
572
+ },
573
+ {
574
+ "epoch": 3.86,
575
+ "grad_norm": 1.2326685190200806,
576
+ "learning_rate": 5.731428571428572e-06,
577
+ "loss": 0.0379,
578
+ "step": 2000
579
+ },
580
+ {
581
+ "epoch": 3.86,
582
+ "eval_loss": 0.28133508563041687,
583
+ "eval_runtime": 3847.5516,
584
+ "eval_samples_per_second": 0.94,
585
+ "eval_steps_per_second": 0.118,
586
+ "eval_wer": 17.470580026978,
587
+ "step": 2000
588
+ }
589
+ ],
590
+ "logging_steps": 25,
591
+ "max_steps": 4000,
592
+ "num_input_tokens_seen": 0,
593
+ "num_train_epochs": 8,
594
+ "save_steps": 1000,
595
+ "total_flos": 9.22261222342656e+18,
596
+ "train_batch_size": 16,
597
+ "trial_name": null,
598
+ "trial_params": null
599
+ }
checkpoint-2000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d97a7d674ba5458016027e8fcbeac271e6fe45c463b2f89dfb01782b9ae4928d
3
+ size 5112
checkpoint-3000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "soaring0616/whisper-small-vietnamese",
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": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-3000/generation_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "lang_to_id": {
10
+ "<|id|>": 50275,
11
+ "<|vi|>": 50278
12
+ },
13
+ "language": "indonesian",
14
+ "max_length": 448,
15
+ "pad_token_id": 50257,
16
+ "suppress_tokens": [],
17
+ "transformers_version": "4.39.1"
18
+ }
checkpoint-3000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b2ec79a801b50df28d9dcf71f3ccb414ede7323b8ad5deca0727f6df7a6ce20
3
+ size 966995080
checkpoint-3000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11c4ea31347a7fcc610d51ca4993379c662430065ecdba232a62dee37bef2242
3
+ size 1925064044
checkpoint-3000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-3000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f95c8c9f4042f7d715cc8da1a8d12e4c9b1216536db0a30c619f6cf232f2504e
3
+ size 14244
checkpoint-3000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e8c76b5c6475a914c815628b47a0a7cb6f2916cc52428671ba7c3d32633cb5e
3
+ size 1064
checkpoint-3000/trainer_state.json ADDED
@@ -0,0 +1,888 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 17.18219452067538,
3
+ "best_model_checkpoint": "./whisper-small-in_from_vi\\checkpoint-3000",
4
+ "epoch": 5.7915057915057915,
5
+ "eval_steps": 1000,
6
+ "global_step": 3000,
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
+ "grad_norm": 218.94273376464844,
14
+ "learning_rate": 3.8e-07,
15
+ "loss": 5.6914,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.1,
20
+ "grad_norm": 35.336517333984375,
21
+ "learning_rate": 8.8e-07,
22
+ "loss": 2.5237,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.14,
27
+ "grad_norm": 14.206599235534668,
28
+ "learning_rate": 1.3800000000000001e-06,
29
+ "loss": 0.8446,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.19,
34
+ "grad_norm": 12.359480857849121,
35
+ "learning_rate": 1.8800000000000002e-06,
36
+ "loss": 0.579,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.24,
41
+ "grad_norm": 8.259902954101562,
42
+ "learning_rate": 2.38e-06,
43
+ "loss": 0.5636,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.29,
48
+ "grad_norm": 11.162034034729004,
49
+ "learning_rate": 2.88e-06,
50
+ "loss": 0.5191,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 0.34,
55
+ "grad_norm": 8.917764663696289,
56
+ "learning_rate": 3.3800000000000007e-06,
57
+ "loss": 0.4879,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 0.39,
62
+ "grad_norm": 8.865296363830566,
63
+ "learning_rate": 3.88e-06,
64
+ "loss": 0.4333,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 0.43,
69
+ "grad_norm": 8.579310417175293,
70
+ "learning_rate": 4.38e-06,
71
+ "loss": 0.4679,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 0.48,
76
+ "grad_norm": 8.361818313598633,
77
+ "learning_rate": 4.880000000000001e-06,
78
+ "loss": 0.3948,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 0.53,
83
+ "grad_norm": 6.773808479309082,
84
+ "learning_rate": 5.380000000000001e-06,
85
+ "loss": 0.4284,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 0.58,
90
+ "grad_norm": 7.8365678787231445,
91
+ "learning_rate": 5.8800000000000005e-06,
92
+ "loss": 0.3829,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 0.63,
97
+ "grad_norm": 10.097782135009766,
98
+ "learning_rate": 6.380000000000001e-06,
99
+ "loss": 0.4091,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 0.68,
104
+ "grad_norm": 8.221065521240234,
105
+ "learning_rate": 6.88e-06,
106
+ "loss": 0.4067,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 0.72,
111
+ "grad_norm": 7.281307697296143,
112
+ "learning_rate": 7.3800000000000005e-06,
113
+ "loss": 0.3501,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 0.77,
118
+ "grad_norm": 7.796460151672363,
119
+ "learning_rate": 7.88e-06,
120
+ "loss": 0.3754,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 0.82,
125
+ "grad_norm": 6.829456329345703,
126
+ "learning_rate": 8.380000000000001e-06,
127
+ "loss": 0.3789,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 0.87,
132
+ "grad_norm": 7.42793607711792,
133
+ "learning_rate": 8.880000000000001e-06,
134
+ "loss": 0.3709,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 0.92,
139
+ "grad_norm": 6.301506519317627,
140
+ "learning_rate": 9.38e-06,
141
+ "loss": 0.355,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 0.97,
146
+ "grad_norm": 7.611047744750977,
147
+ "learning_rate": 9.88e-06,
148
+ "loss": 0.3347,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 1.01,
153
+ "grad_norm": 6.571840286254883,
154
+ "learning_rate": 9.945714285714286e-06,
155
+ "loss": 0.3231,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 1.06,
160
+ "grad_norm": 6.700900554656982,
161
+ "learning_rate": 9.874285714285715e-06,
162
+ "loss": 0.2356,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 1.11,
167
+ "grad_norm": 6.217418670654297,
168
+ "learning_rate": 9.802857142857144e-06,
169
+ "loss": 0.2177,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 1.16,
174
+ "grad_norm": 5.774504661560059,
175
+ "learning_rate": 9.731428571428573e-06,
176
+ "loss": 0.2103,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 1.21,
181
+ "grad_norm": 5.770651340484619,
182
+ "learning_rate": 9.66e-06,
183
+ "loss": 0.2529,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 1.25,
188
+ "grad_norm": 5.579860687255859,
189
+ "learning_rate": 9.58857142857143e-06,
190
+ "loss": 0.2435,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 1.3,
195
+ "grad_norm": 5.943060398101807,
196
+ "learning_rate": 9.517142857142859e-06,
197
+ "loss": 0.2064,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 1.35,
202
+ "grad_norm": 7.721786975860596,
203
+ "learning_rate": 9.445714285714288e-06,
204
+ "loss": 0.2273,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 1.4,
209
+ "grad_norm": 6.976935386657715,
210
+ "learning_rate": 9.374285714285715e-06,
211
+ "loss": 0.2373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 1.45,
216
+ "grad_norm": 6.749694347381592,
217
+ "learning_rate": 9.302857142857144e-06,
218
+ "loss": 0.2306,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 1.5,
223
+ "grad_norm": 5.087153434753418,
224
+ "learning_rate": 9.231428571428573e-06,
225
+ "loss": 0.2292,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 1.54,
230
+ "grad_norm": 6.032232761383057,
231
+ "learning_rate": 9.16e-06,
232
+ "loss": 0.2096,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 1.59,
237
+ "grad_norm": 5.749306678771973,
238
+ "learning_rate": 9.08857142857143e-06,
239
+ "loss": 0.2217,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 1.64,
244
+ "grad_norm": 7.6291184425354,
245
+ "learning_rate": 9.017142857142858e-06,
246
+ "loss": 0.2023,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 1.69,
251
+ "grad_norm": 4.76064395904541,
252
+ "learning_rate": 8.945714285714286e-06,
253
+ "loss": 0.2099,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 1.74,
258
+ "grad_norm": 5.0226898193359375,
259
+ "learning_rate": 8.874285714285715e-06,
260
+ "loss": 0.2017,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 1.79,
265
+ "grad_norm": 5.6098313331604,
266
+ "learning_rate": 8.802857142857144e-06,
267
+ "loss": 0.2126,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 1.83,
272
+ "grad_norm": 5.58658504486084,
273
+ "learning_rate": 8.731428571428571e-06,
274
+ "loss": 0.2209,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 1.88,
279
+ "grad_norm": 7.067436218261719,
280
+ "learning_rate": 8.66e-06,
281
+ "loss": 0.225,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 1.93,
286
+ "grad_norm": 4.616860389709473,
287
+ "learning_rate": 8.588571428571429e-06,
288
+ "loss": 0.2059,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 1.93,
293
+ "eval_loss": 0.25656187534332275,
294
+ "eval_runtime": 2791.9455,
295
+ "eval_samples_per_second": 1.296,
296
+ "eval_steps_per_second": 0.162,
297
+ "eval_wer": 17.814782082887575,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 1.98,
302
+ "grad_norm": 4.517664909362793,
303
+ "learning_rate": 8.517142857142858e-06,
304
+ "loss": 0.2132,
305
+ "step": 1025
306
+ },
307
+ {
308
+ "epoch": 2.03,
309
+ "grad_norm": 3.016611099243164,
310
+ "learning_rate": 8.445714285714285e-06,
311
+ "loss": 0.1498,
312
+ "step": 1050
313
+ },
314
+ {
315
+ "epoch": 2.08,
316
+ "grad_norm": 4.563220977783203,
317
+ "learning_rate": 8.374285714285714e-06,
318
+ "loss": 0.0848,
319
+ "step": 1075
320
+ },
321
+ {
322
+ "epoch": 2.12,
323
+ "grad_norm": 4.658745288848877,
324
+ "learning_rate": 8.302857142857143e-06,
325
+ "loss": 0.0877,
326
+ "step": 1100
327
+ },
328
+ {
329
+ "epoch": 2.17,
330
+ "grad_norm": 3.2172155380249023,
331
+ "learning_rate": 8.231428571428572e-06,
332
+ "loss": 0.0858,
333
+ "step": 1125
334
+ },
335
+ {
336
+ "epoch": 2.22,
337
+ "grad_norm": 3.33174991607666,
338
+ "learning_rate": 8.16e-06,
339
+ "loss": 0.08,
340
+ "step": 1150
341
+ },
342
+ {
343
+ "epoch": 2.27,
344
+ "grad_norm": 4.132905960083008,
345
+ "learning_rate": 8.088571428571429e-06,
346
+ "loss": 0.0945,
347
+ "step": 1175
348
+ },
349
+ {
350
+ "epoch": 2.32,
351
+ "grad_norm": 3.5062320232391357,
352
+ "learning_rate": 8.017142857142858e-06,
353
+ "loss": 0.0849,
354
+ "step": 1200
355
+ },
356
+ {
357
+ "epoch": 2.36,
358
+ "grad_norm": 2.746927261352539,
359
+ "learning_rate": 7.945714285714287e-06,
360
+ "loss": 0.0814,
361
+ "step": 1225
362
+ },
363
+ {
364
+ "epoch": 2.41,
365
+ "grad_norm": 2.8374855518341064,
366
+ "learning_rate": 7.874285714285716e-06,
367
+ "loss": 0.0898,
368
+ "step": 1250
369
+ },
370
+ {
371
+ "epoch": 2.46,
372
+ "grad_norm": 2.8154654502868652,
373
+ "learning_rate": 7.802857142857143e-06,
374
+ "loss": 0.091,
375
+ "step": 1275
376
+ },
377
+ {
378
+ "epoch": 2.51,
379
+ "grad_norm": 3.5242350101470947,
380
+ "learning_rate": 7.731428571428572e-06,
381
+ "loss": 0.085,
382
+ "step": 1300
383
+ },
384
+ {
385
+ "epoch": 2.56,
386
+ "grad_norm": 3.3596408367156982,
387
+ "learning_rate": 7.660000000000001e-06,
388
+ "loss": 0.093,
389
+ "step": 1325
390
+ },
391
+ {
392
+ "epoch": 2.61,
393
+ "grad_norm": 3.4869582653045654,
394
+ "learning_rate": 7.588571428571429e-06,
395
+ "loss": 0.0836,
396
+ "step": 1350
397
+ },
398
+ {
399
+ "epoch": 2.65,
400
+ "grad_norm": 2.8120055198669434,
401
+ "learning_rate": 7.5171428571428575e-06,
402
+ "loss": 0.0769,
403
+ "step": 1375
404
+ },
405
+ {
406
+ "epoch": 2.7,
407
+ "grad_norm": 3.5809738636016846,
408
+ "learning_rate": 7.445714285714286e-06,
409
+ "loss": 0.0916,
410
+ "step": 1400
411
+ },
412
+ {
413
+ "epoch": 2.75,
414
+ "grad_norm": 2.8839168548583984,
415
+ "learning_rate": 7.374285714285715e-06,
416
+ "loss": 0.09,
417
+ "step": 1425
418
+ },
419
+ {
420
+ "epoch": 2.8,
421
+ "grad_norm": 4.721149921417236,
422
+ "learning_rate": 7.302857142857144e-06,
423
+ "loss": 0.0915,
424
+ "step": 1450
425
+ },
426
+ {
427
+ "epoch": 2.85,
428
+ "grad_norm": 3.5091800689697266,
429
+ "learning_rate": 7.231428571428573e-06,
430
+ "loss": 0.0794,
431
+ "step": 1475
432
+ },
433
+ {
434
+ "epoch": 2.9,
435
+ "grad_norm": 3.820974349975586,
436
+ "learning_rate": 7.16e-06,
437
+ "loss": 0.0934,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 2.94,
442
+ "grad_norm": 4.247288227081299,
443
+ "learning_rate": 7.088571428571429e-06,
444
+ "loss": 0.0822,
445
+ "step": 1525
446
+ },
447
+ {
448
+ "epoch": 2.99,
449
+ "grad_norm": 3.3485350608825684,
450
+ "learning_rate": 7.017142857142858e-06,
451
+ "loss": 0.0971,
452
+ "step": 1550
453
+ },
454
+ {
455
+ "epoch": 3.04,
456
+ "grad_norm": 1.4603108167648315,
457
+ "learning_rate": 6.945714285714287e-06,
458
+ "loss": 0.0352,
459
+ "step": 1575
460
+ },
461
+ {
462
+ "epoch": 3.09,
463
+ "grad_norm": 1.7609468698501587,
464
+ "learning_rate": 6.874285714285714e-06,
465
+ "loss": 0.0402,
466
+ "step": 1600
467
+ },
468
+ {
469
+ "epoch": 3.14,
470
+ "grad_norm": 2.6159989833831787,
471
+ "learning_rate": 6.8028571428571434e-06,
472
+ "loss": 0.0317,
473
+ "step": 1625
474
+ },
475
+ {
476
+ "epoch": 3.19,
477
+ "grad_norm": 1.857428789138794,
478
+ "learning_rate": 6.7314285714285724e-06,
479
+ "loss": 0.039,
480
+ "step": 1650
481
+ },
482
+ {
483
+ "epoch": 3.23,
484
+ "grad_norm": 3.3788986206054688,
485
+ "learning_rate": 6.660000000000001e-06,
486
+ "loss": 0.0347,
487
+ "step": 1675
488
+ },
489
+ {
490
+ "epoch": 3.28,
491
+ "grad_norm": 1.8846665620803833,
492
+ "learning_rate": 6.588571428571429e-06,
493
+ "loss": 0.0321,
494
+ "step": 1700
495
+ },
496
+ {
497
+ "epoch": 3.33,
498
+ "grad_norm": 1.2387264966964722,
499
+ "learning_rate": 6.517142857142858e-06,
500
+ "loss": 0.0327,
501
+ "step": 1725
502
+ },
503
+ {
504
+ "epoch": 3.38,
505
+ "grad_norm": 1.4381897449493408,
506
+ "learning_rate": 6.445714285714286e-06,
507
+ "loss": 0.0354,
508
+ "step": 1750
509
+ },
510
+ {
511
+ "epoch": 3.43,
512
+ "grad_norm": 2.2566747665405273,
513
+ "learning_rate": 6.374285714285715e-06,
514
+ "loss": 0.0368,
515
+ "step": 1775
516
+ },
517
+ {
518
+ "epoch": 3.47,
519
+ "grad_norm": 1.6017179489135742,
520
+ "learning_rate": 6.302857142857144e-06,
521
+ "loss": 0.0351,
522
+ "step": 1800
523
+ },
524
+ {
525
+ "epoch": 3.52,
526
+ "grad_norm": 3.2970995903015137,
527
+ "learning_rate": 6.231428571428571e-06,
528
+ "loss": 0.0332,
529
+ "step": 1825
530
+ },
531
+ {
532
+ "epoch": 3.57,
533
+ "grad_norm": 2.8597681522369385,
534
+ "learning_rate": 6.16e-06,
535
+ "loss": 0.0355,
536
+ "step": 1850
537
+ },
538
+ {
539
+ "epoch": 3.62,
540
+ "grad_norm": 2.8335232734680176,
541
+ "learning_rate": 6.088571428571429e-06,
542
+ "loss": 0.0338,
543
+ "step": 1875
544
+ },
545
+ {
546
+ "epoch": 3.67,
547
+ "grad_norm": 3.405618667602539,
548
+ "learning_rate": 6.017142857142858e-06,
549
+ "loss": 0.0376,
550
+ "step": 1900
551
+ },
552
+ {
553
+ "epoch": 3.72,
554
+ "grad_norm": 2.4519917964935303,
555
+ "learning_rate": 5.945714285714286e-06,
556
+ "loss": 0.0359,
557
+ "step": 1925
558
+ },
559
+ {
560
+ "epoch": 3.76,
561
+ "grad_norm": 3.2067463397979736,
562
+ "learning_rate": 5.874285714285715e-06,
563
+ "loss": 0.029,
564
+ "step": 1950
565
+ },
566
+ {
567
+ "epoch": 3.81,
568
+ "grad_norm": 2.5410799980163574,
569
+ "learning_rate": 5.802857142857144e-06,
570
+ "loss": 0.0313,
571
+ "step": 1975
572
+ },
573
+ {
574
+ "epoch": 3.86,
575
+ "grad_norm": 1.2326685190200806,
576
+ "learning_rate": 5.731428571428572e-06,
577
+ "loss": 0.0379,
578
+ "step": 2000
579
+ },
580
+ {
581
+ "epoch": 3.86,
582
+ "eval_loss": 0.28133508563041687,
583
+ "eval_runtime": 3847.5516,
584
+ "eval_samples_per_second": 0.94,
585
+ "eval_steps_per_second": 0.118,
586
+ "eval_wer": 17.470580026978,
587
+ "step": 2000
588
+ },
589
+ {
590
+ "epoch": 3.91,
591
+ "grad_norm": 2.7244298458099365,
592
+ "learning_rate": 5.66e-06,
593
+ "loss": 0.0356,
594
+ "step": 2025
595
+ },
596
+ {
597
+ "epoch": 3.96,
598
+ "grad_norm": 2.446558952331543,
599
+ "learning_rate": 5.588571428571429e-06,
600
+ "loss": 0.0273,
601
+ "step": 2050
602
+ },
603
+ {
604
+ "epoch": 4.01,
605
+ "grad_norm": 0.7038655281066895,
606
+ "learning_rate": 5.517142857142857e-06,
607
+ "loss": 0.0298,
608
+ "step": 2075
609
+ },
610
+ {
611
+ "epoch": 4.05,
612
+ "grad_norm": 0.8899651765823364,
613
+ "learning_rate": 5.445714285714286e-06,
614
+ "loss": 0.0162,
615
+ "step": 2100
616
+ },
617
+ {
618
+ "epoch": 4.1,
619
+ "grad_norm": 2.154207944869995,
620
+ "learning_rate": 5.374285714285715e-06,
621
+ "loss": 0.0143,
622
+ "step": 2125
623
+ },
624
+ {
625
+ "epoch": 4.15,
626
+ "grad_norm": 1.203642725944519,
627
+ "learning_rate": 5.3028571428571425e-06,
628
+ "loss": 0.0142,
629
+ "step": 2150
630
+ },
631
+ {
632
+ "epoch": 4.2,
633
+ "grad_norm": 0.45312538743019104,
634
+ "learning_rate": 5.2314285714285716e-06,
635
+ "loss": 0.0133,
636
+ "step": 2175
637
+ },
638
+ {
639
+ "epoch": 4.25,
640
+ "grad_norm": 1.146144151687622,
641
+ "learning_rate": 5.1600000000000006e-06,
642
+ "loss": 0.0142,
643
+ "step": 2200
644
+ },
645
+ {
646
+ "epoch": 4.3,
647
+ "grad_norm": 2.2396395206451416,
648
+ "learning_rate": 5.08857142857143e-06,
649
+ "loss": 0.0163,
650
+ "step": 2225
651
+ },
652
+ {
653
+ "epoch": 4.34,
654
+ "grad_norm": 0.6650336384773254,
655
+ "learning_rate": 5.017142857142857e-06,
656
+ "loss": 0.0119,
657
+ "step": 2250
658
+ },
659
+ {
660
+ "epoch": 4.39,
661
+ "grad_norm": 0.5963819026947021,
662
+ "learning_rate": 4.945714285714286e-06,
663
+ "loss": 0.0173,
664
+ "step": 2275
665
+ },
666
+ {
667
+ "epoch": 4.44,
668
+ "grad_norm": 1.5120543241500854,
669
+ "learning_rate": 4.874285714285715e-06,
670
+ "loss": 0.0148,
671
+ "step": 2300
672
+ },
673
+ {
674
+ "epoch": 4.49,
675
+ "grad_norm": 0.7102583050727844,
676
+ "learning_rate": 4.802857142857143e-06,
677
+ "loss": 0.0121,
678
+ "step": 2325
679
+ },
680
+ {
681
+ "epoch": 4.54,
682
+ "grad_norm": 1.2483290433883667,
683
+ "learning_rate": 4.731428571428572e-06,
684
+ "loss": 0.0169,
685
+ "step": 2350
686
+ },
687
+ {
688
+ "epoch": 4.58,
689
+ "grad_norm": 1.8999782800674438,
690
+ "learning_rate": 4.66e-06,
691
+ "loss": 0.013,
692
+ "step": 2375
693
+ },
694
+ {
695
+ "epoch": 4.63,
696
+ "grad_norm": 0.893521249294281,
697
+ "learning_rate": 4.588571428571429e-06,
698
+ "loss": 0.0149,
699
+ "step": 2400
700
+ },
701
+ {
702
+ "epoch": 4.68,
703
+ "grad_norm": 1.4212769269943237,
704
+ "learning_rate": 4.5171428571428575e-06,
705
+ "loss": 0.0132,
706
+ "step": 2425
707
+ },
708
+ {
709
+ "epoch": 4.73,
710
+ "grad_norm": 0.5899932980537415,
711
+ "learning_rate": 4.445714285714286e-06,
712
+ "loss": 0.018,
713
+ "step": 2450
714
+ },
715
+ {
716
+ "epoch": 4.78,
717
+ "grad_norm": 0.5150299668312073,
718
+ "learning_rate": 4.374285714285715e-06,
719
+ "loss": 0.012,
720
+ "step": 2475
721
+ },
722
+ {
723
+ "epoch": 4.83,
724
+ "grad_norm": 1.5570229291915894,
725
+ "learning_rate": 4.302857142857143e-06,
726
+ "loss": 0.0121,
727
+ "step": 2500
728
+ },
729
+ {
730
+ "epoch": 4.87,
731
+ "grad_norm": 1.727569818496704,
732
+ "learning_rate": 4.231428571428572e-06,
733
+ "loss": 0.0117,
734
+ "step": 2525
735
+ },
736
+ {
737
+ "epoch": 4.92,
738
+ "grad_norm": 0.8942755460739136,
739
+ "learning_rate": 4.16e-06,
740
+ "loss": 0.0125,
741
+ "step": 2550
742
+ },
743
+ {
744
+ "epoch": 4.97,
745
+ "grad_norm": 2.9615914821624756,
746
+ "learning_rate": 4.088571428571429e-06,
747
+ "loss": 0.0129,
748
+ "step": 2575
749
+ },
750
+ {
751
+ "epoch": 5.02,
752
+ "grad_norm": 0.27788445353507996,
753
+ "learning_rate": 4.017142857142857e-06,
754
+ "loss": 0.0101,
755
+ "step": 2600
756
+ },
757
+ {
758
+ "epoch": 5.07,
759
+ "grad_norm": 0.5583022832870483,
760
+ "learning_rate": 3.945714285714286e-06,
761
+ "loss": 0.007,
762
+ "step": 2625
763
+ },
764
+ {
765
+ "epoch": 5.12,
766
+ "grad_norm": 0.392986536026001,
767
+ "learning_rate": 3.874285714285715e-06,
768
+ "loss": 0.0065,
769
+ "step": 2650
770
+ },
771
+ {
772
+ "epoch": 5.16,
773
+ "grad_norm": 0.7827091813087463,
774
+ "learning_rate": 3.802857142857143e-06,
775
+ "loss": 0.0054,
776
+ "step": 2675
777
+ },
778
+ {
779
+ "epoch": 5.21,
780
+ "grad_norm": 0.5473342537879944,
781
+ "learning_rate": 3.731428571428572e-06,
782
+ "loss": 0.0068,
783
+ "step": 2700
784
+ },
785
+ {
786
+ "epoch": 5.26,
787
+ "grad_norm": 0.545080840587616,
788
+ "learning_rate": 3.66e-06,
789
+ "loss": 0.0055,
790
+ "step": 2725
791
+ },
792
+ {
793
+ "epoch": 5.31,
794
+ "grad_norm": 0.44736358523368835,
795
+ "learning_rate": 3.588571428571429e-06,
796
+ "loss": 0.0085,
797
+ "step": 2750
798
+ },
799
+ {
800
+ "epoch": 5.36,
801
+ "grad_norm": 0.7332315444946289,
802
+ "learning_rate": 3.5171428571428573e-06,
803
+ "loss": 0.0071,
804
+ "step": 2775
805
+ },
806
+ {
807
+ "epoch": 5.41,
808
+ "grad_norm": 0.6373933553695679,
809
+ "learning_rate": 3.4457142857142863e-06,
810
+ "loss": 0.0058,
811
+ "step": 2800
812
+ },
813
+ {
814
+ "epoch": 5.45,
815
+ "grad_norm": 0.47440409660339355,
816
+ "learning_rate": 3.3742857142857145e-06,
817
+ "loss": 0.0056,
818
+ "step": 2825
819
+ },
820
+ {
821
+ "epoch": 5.5,
822
+ "grad_norm": 0.23941631615161896,
823
+ "learning_rate": 3.302857142857143e-06,
824
+ "loss": 0.0046,
825
+ "step": 2850
826
+ },
827
+ {
828
+ "epoch": 5.55,
829
+ "grad_norm": 0.5772097110748291,
830
+ "learning_rate": 3.2314285714285716e-06,
831
+ "loss": 0.0055,
832
+ "step": 2875
833
+ },
834
+ {
835
+ "epoch": 5.6,
836
+ "grad_norm": 0.22144152224063873,
837
+ "learning_rate": 3.1600000000000002e-06,
838
+ "loss": 0.0048,
839
+ "step": 2900
840
+ },
841
+ {
842
+ "epoch": 5.65,
843
+ "grad_norm": 0.28436514735221863,
844
+ "learning_rate": 3.0885714285714284e-06,
845
+ "loss": 0.0052,
846
+ "step": 2925
847
+ },
848
+ {
849
+ "epoch": 5.69,
850
+ "grad_norm": 0.2404721975326538,
851
+ "learning_rate": 3.0171428571428574e-06,
852
+ "loss": 0.0056,
853
+ "step": 2950
854
+ },
855
+ {
856
+ "epoch": 5.74,
857
+ "grad_norm": 0.49395838379859924,
858
+ "learning_rate": 2.945714285714286e-06,
859
+ "loss": 0.005,
860
+ "step": 2975
861
+ },
862
+ {
863
+ "epoch": 5.79,
864
+ "grad_norm": 0.32905882596969604,
865
+ "learning_rate": 2.8742857142857146e-06,
866
+ "loss": 0.006,
867
+ "step": 3000
868
+ },
869
+ {
870
+ "epoch": 5.79,
871
+ "eval_loss": 0.32524359226226807,
872
+ "eval_runtime": 2778.5628,
873
+ "eval_samples_per_second": 1.302,
874
+ "eval_steps_per_second": 0.163,
875
+ "eval_wer": 17.18219452067538,
876
+ "step": 3000
877
+ }
878
+ ],
879
+ "logging_steps": 25,
880
+ "max_steps": 4000,
881
+ "num_input_tokens_seen": 0,
882
+ "num_train_epochs": 8,
883
+ "save_steps": 1000,
884
+ "total_flos": 1.38318982373376e+19,
885
+ "train_batch_size": 16,
886
+ "trial_name": null,
887
+ "trial_params": null
888
+ }
checkpoint-3000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d97a7d674ba5458016027e8fcbeac271e6fe45c463b2f89dfb01782b9ae4928d
3
+ size 5112
checkpoint-4000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "soaring0616/whisper-small-vietnamese",
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": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-4000/generation_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "lang_to_id": {
10
+ "<|id|>": 50275,
11
+ "<|vi|>": 50278
12
+ },
13
+ "language": "indonesian",
14
+ "max_length": 448,
15
+ "pad_token_id": 50257,
16
+ "suppress_tokens": [],
17
+ "transformers_version": "4.39.1"
18
+ }
checkpoint-4000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc40f62bc219565ea4fe7f38a6a349c133b6dfc2a5c89c40c0abf2be56c48320
3
+ size 966995080
checkpoint-4000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ff92fd5980e080d2c3378e5b65d7be86c9462f60ffe5c96be0656923426c8567
3
+ size 1925064044
checkpoint-4000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-4000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b05194b2f99c0ed6f16f3ed9d04212c1494104bdaea1339f2285ef60727128a
3
+ size 14244
checkpoint-4000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2855e8906b1a9cae8fe1084d27dc3bbefc0b436cc2ae5fffa97d518a5d5d732d
3
+ size 1064
checkpoint-4000/trainer_state.json ADDED
@@ -0,0 +1,1177 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 17.051955904925812,
3
+ "best_model_checkpoint": "./whisper-small-in_from_vi\\checkpoint-4000",
4
+ "epoch": 7.722007722007722,
5
+ "eval_steps": 1000,
6
+ "global_step": 4000,
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
+ "grad_norm": 218.94273376464844,
14
+ "learning_rate": 3.8e-07,
15
+ "loss": 5.6914,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.1,
20
+ "grad_norm": 35.336517333984375,
21
+ "learning_rate": 8.8e-07,
22
+ "loss": 2.5237,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.14,
27
+ "grad_norm": 14.206599235534668,
28
+ "learning_rate": 1.3800000000000001e-06,
29
+ "loss": 0.8446,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.19,
34
+ "grad_norm": 12.359480857849121,
35
+ "learning_rate": 1.8800000000000002e-06,
36
+ "loss": 0.579,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.24,
41
+ "grad_norm": 8.259902954101562,
42
+ "learning_rate": 2.38e-06,
43
+ "loss": 0.5636,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.29,
48
+ "grad_norm": 11.162034034729004,
49
+ "learning_rate": 2.88e-06,
50
+ "loss": 0.5191,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 0.34,
55
+ "grad_norm": 8.917764663696289,
56
+ "learning_rate": 3.3800000000000007e-06,
57
+ "loss": 0.4879,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 0.39,
62
+ "grad_norm": 8.865296363830566,
63
+ "learning_rate": 3.88e-06,
64
+ "loss": 0.4333,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 0.43,
69
+ "grad_norm": 8.579310417175293,
70
+ "learning_rate": 4.38e-06,
71
+ "loss": 0.4679,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 0.48,
76
+ "grad_norm": 8.361818313598633,
77
+ "learning_rate": 4.880000000000001e-06,
78
+ "loss": 0.3948,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 0.53,
83
+ "grad_norm": 6.773808479309082,
84
+ "learning_rate": 5.380000000000001e-06,
85
+ "loss": 0.4284,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 0.58,
90
+ "grad_norm": 7.8365678787231445,
91
+ "learning_rate": 5.8800000000000005e-06,
92
+ "loss": 0.3829,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 0.63,
97
+ "grad_norm": 10.097782135009766,
98
+ "learning_rate": 6.380000000000001e-06,
99
+ "loss": 0.4091,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 0.68,
104
+ "grad_norm": 8.221065521240234,
105
+ "learning_rate": 6.88e-06,
106
+ "loss": 0.4067,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 0.72,
111
+ "grad_norm": 7.281307697296143,
112
+ "learning_rate": 7.3800000000000005e-06,
113
+ "loss": 0.3501,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 0.77,
118
+ "grad_norm": 7.796460151672363,
119
+ "learning_rate": 7.88e-06,
120
+ "loss": 0.3754,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 0.82,
125
+ "grad_norm": 6.829456329345703,
126
+ "learning_rate": 8.380000000000001e-06,
127
+ "loss": 0.3789,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 0.87,
132
+ "grad_norm": 7.42793607711792,
133
+ "learning_rate": 8.880000000000001e-06,
134
+ "loss": 0.3709,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 0.92,
139
+ "grad_norm": 6.301506519317627,
140
+ "learning_rate": 9.38e-06,
141
+ "loss": 0.355,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 0.97,
146
+ "grad_norm": 7.611047744750977,
147
+ "learning_rate": 9.88e-06,
148
+ "loss": 0.3347,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 1.01,
153
+ "grad_norm": 6.571840286254883,
154
+ "learning_rate": 9.945714285714286e-06,
155
+ "loss": 0.3231,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 1.06,
160
+ "grad_norm": 6.700900554656982,
161
+ "learning_rate": 9.874285714285715e-06,
162
+ "loss": 0.2356,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 1.11,
167
+ "grad_norm": 6.217418670654297,
168
+ "learning_rate": 9.802857142857144e-06,
169
+ "loss": 0.2177,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 1.16,
174
+ "grad_norm": 5.774504661560059,
175
+ "learning_rate": 9.731428571428573e-06,
176
+ "loss": 0.2103,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 1.21,
181
+ "grad_norm": 5.770651340484619,
182
+ "learning_rate": 9.66e-06,
183
+ "loss": 0.2529,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 1.25,
188
+ "grad_norm": 5.579860687255859,
189
+ "learning_rate": 9.58857142857143e-06,
190
+ "loss": 0.2435,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 1.3,
195
+ "grad_norm": 5.943060398101807,
196
+ "learning_rate": 9.517142857142859e-06,
197
+ "loss": 0.2064,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 1.35,
202
+ "grad_norm": 7.721786975860596,
203
+ "learning_rate": 9.445714285714288e-06,
204
+ "loss": 0.2273,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 1.4,
209
+ "grad_norm": 6.976935386657715,
210
+ "learning_rate": 9.374285714285715e-06,
211
+ "loss": 0.2373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 1.45,
216
+ "grad_norm": 6.749694347381592,
217
+ "learning_rate": 9.302857142857144e-06,
218
+ "loss": 0.2306,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 1.5,
223
+ "grad_norm": 5.087153434753418,
224
+ "learning_rate": 9.231428571428573e-06,
225
+ "loss": 0.2292,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 1.54,
230
+ "grad_norm": 6.032232761383057,
231
+ "learning_rate": 9.16e-06,
232
+ "loss": 0.2096,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 1.59,
237
+ "grad_norm": 5.749306678771973,
238
+ "learning_rate": 9.08857142857143e-06,
239
+ "loss": 0.2217,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 1.64,
244
+ "grad_norm": 7.6291184425354,
245
+ "learning_rate": 9.017142857142858e-06,
246
+ "loss": 0.2023,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 1.69,
251
+ "grad_norm": 4.76064395904541,
252
+ "learning_rate": 8.945714285714286e-06,
253
+ "loss": 0.2099,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 1.74,
258
+ "grad_norm": 5.0226898193359375,
259
+ "learning_rate": 8.874285714285715e-06,
260
+ "loss": 0.2017,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 1.79,
265
+ "grad_norm": 5.6098313331604,
266
+ "learning_rate": 8.802857142857144e-06,
267
+ "loss": 0.2126,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 1.83,
272
+ "grad_norm": 5.58658504486084,
273
+ "learning_rate": 8.731428571428571e-06,
274
+ "loss": 0.2209,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 1.88,
279
+ "grad_norm": 7.067436218261719,
280
+ "learning_rate": 8.66e-06,
281
+ "loss": 0.225,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 1.93,
286
+ "grad_norm": 4.616860389709473,
287
+ "learning_rate": 8.588571428571429e-06,
288
+ "loss": 0.2059,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 1.93,
293
+ "eval_loss": 0.25656187534332275,
294
+ "eval_runtime": 2791.9455,
295
+ "eval_samples_per_second": 1.296,
296
+ "eval_steps_per_second": 0.162,
297
+ "eval_wer": 17.814782082887575,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 1.98,
302
+ "grad_norm": 4.517664909362793,
303
+ "learning_rate": 8.517142857142858e-06,
304
+ "loss": 0.2132,
305
+ "step": 1025
306
+ },
307
+ {
308
+ "epoch": 2.03,
309
+ "grad_norm": 3.016611099243164,
310
+ "learning_rate": 8.445714285714285e-06,
311
+ "loss": 0.1498,
312
+ "step": 1050
313
+ },
314
+ {
315
+ "epoch": 2.08,
316
+ "grad_norm": 4.563220977783203,
317
+ "learning_rate": 8.374285714285714e-06,
318
+ "loss": 0.0848,
319
+ "step": 1075
320
+ },
321
+ {
322
+ "epoch": 2.12,
323
+ "grad_norm": 4.658745288848877,
324
+ "learning_rate": 8.302857142857143e-06,
325
+ "loss": 0.0877,
326
+ "step": 1100
327
+ },
328
+ {
329
+ "epoch": 2.17,
330
+ "grad_norm": 3.2172155380249023,
331
+ "learning_rate": 8.231428571428572e-06,
332
+ "loss": 0.0858,
333
+ "step": 1125
334
+ },
335
+ {
336
+ "epoch": 2.22,
337
+ "grad_norm": 3.33174991607666,
338
+ "learning_rate": 8.16e-06,
339
+ "loss": 0.08,
340
+ "step": 1150
341
+ },
342
+ {
343
+ "epoch": 2.27,
344
+ "grad_norm": 4.132905960083008,
345
+ "learning_rate": 8.088571428571429e-06,
346
+ "loss": 0.0945,
347
+ "step": 1175
348
+ },
349
+ {
350
+ "epoch": 2.32,
351
+ "grad_norm": 3.5062320232391357,
352
+ "learning_rate": 8.017142857142858e-06,
353
+ "loss": 0.0849,
354
+ "step": 1200
355
+ },
356
+ {
357
+ "epoch": 2.36,
358
+ "grad_norm": 2.746927261352539,
359
+ "learning_rate": 7.945714285714287e-06,
360
+ "loss": 0.0814,
361
+ "step": 1225
362
+ },
363
+ {
364
+ "epoch": 2.41,
365
+ "grad_norm": 2.8374855518341064,
366
+ "learning_rate": 7.874285714285716e-06,
367
+ "loss": 0.0898,
368
+ "step": 1250
369
+ },
370
+ {
371
+ "epoch": 2.46,
372
+ "grad_norm": 2.8154654502868652,
373
+ "learning_rate": 7.802857142857143e-06,
374
+ "loss": 0.091,
375
+ "step": 1275
376
+ },
377
+ {
378
+ "epoch": 2.51,
379
+ "grad_norm": 3.5242350101470947,
380
+ "learning_rate": 7.731428571428572e-06,
381
+ "loss": 0.085,
382
+ "step": 1300
383
+ },
384
+ {
385
+ "epoch": 2.56,
386
+ "grad_norm": 3.3596408367156982,
387
+ "learning_rate": 7.660000000000001e-06,
388
+ "loss": 0.093,
389
+ "step": 1325
390
+ },
391
+ {
392
+ "epoch": 2.61,
393
+ "grad_norm": 3.4869582653045654,
394
+ "learning_rate": 7.588571428571429e-06,
395
+ "loss": 0.0836,
396
+ "step": 1350
397
+ },
398
+ {
399
+ "epoch": 2.65,
400
+ "grad_norm": 2.8120055198669434,
401
+ "learning_rate": 7.5171428571428575e-06,
402
+ "loss": 0.0769,
403
+ "step": 1375
404
+ },
405
+ {
406
+ "epoch": 2.7,
407
+ "grad_norm": 3.5809738636016846,
408
+ "learning_rate": 7.445714285714286e-06,
409
+ "loss": 0.0916,
410
+ "step": 1400
411
+ },
412
+ {
413
+ "epoch": 2.75,
414
+ "grad_norm": 2.8839168548583984,
415
+ "learning_rate": 7.374285714285715e-06,
416
+ "loss": 0.09,
417
+ "step": 1425
418
+ },
419
+ {
420
+ "epoch": 2.8,
421
+ "grad_norm": 4.721149921417236,
422
+ "learning_rate": 7.302857142857144e-06,
423
+ "loss": 0.0915,
424
+ "step": 1450
425
+ },
426
+ {
427
+ "epoch": 2.85,
428
+ "grad_norm": 3.5091800689697266,
429
+ "learning_rate": 7.231428571428573e-06,
430
+ "loss": 0.0794,
431
+ "step": 1475
432
+ },
433
+ {
434
+ "epoch": 2.9,
435
+ "grad_norm": 3.820974349975586,
436
+ "learning_rate": 7.16e-06,
437
+ "loss": 0.0934,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 2.94,
442
+ "grad_norm": 4.247288227081299,
443
+ "learning_rate": 7.088571428571429e-06,
444
+ "loss": 0.0822,
445
+ "step": 1525
446
+ },
447
+ {
448
+ "epoch": 2.99,
449
+ "grad_norm": 3.3485350608825684,
450
+ "learning_rate": 7.017142857142858e-06,
451
+ "loss": 0.0971,
452
+ "step": 1550
453
+ },
454
+ {
455
+ "epoch": 3.04,
456
+ "grad_norm": 1.4603108167648315,
457
+ "learning_rate": 6.945714285714287e-06,
458
+ "loss": 0.0352,
459
+ "step": 1575
460
+ },
461
+ {
462
+ "epoch": 3.09,
463
+ "grad_norm": 1.7609468698501587,
464
+ "learning_rate": 6.874285714285714e-06,
465
+ "loss": 0.0402,
466
+ "step": 1600
467
+ },
468
+ {
469
+ "epoch": 3.14,
470
+ "grad_norm": 2.6159989833831787,
471
+ "learning_rate": 6.8028571428571434e-06,
472
+ "loss": 0.0317,
473
+ "step": 1625
474
+ },
475
+ {
476
+ "epoch": 3.19,
477
+ "grad_norm": 1.857428789138794,
478
+ "learning_rate": 6.7314285714285724e-06,
479
+ "loss": 0.039,
480
+ "step": 1650
481
+ },
482
+ {
483
+ "epoch": 3.23,
484
+ "grad_norm": 3.3788986206054688,
485
+ "learning_rate": 6.660000000000001e-06,
486
+ "loss": 0.0347,
487
+ "step": 1675
488
+ },
489
+ {
490
+ "epoch": 3.28,
491
+ "grad_norm": 1.8846665620803833,
492
+ "learning_rate": 6.588571428571429e-06,
493
+ "loss": 0.0321,
494
+ "step": 1700
495
+ },
496
+ {
497
+ "epoch": 3.33,
498
+ "grad_norm": 1.2387264966964722,
499
+ "learning_rate": 6.517142857142858e-06,
500
+ "loss": 0.0327,
501
+ "step": 1725
502
+ },
503
+ {
504
+ "epoch": 3.38,
505
+ "grad_norm": 1.4381897449493408,
506
+ "learning_rate": 6.445714285714286e-06,
507
+ "loss": 0.0354,
508
+ "step": 1750
509
+ },
510
+ {
511
+ "epoch": 3.43,
512
+ "grad_norm": 2.2566747665405273,
513
+ "learning_rate": 6.374285714285715e-06,
514
+ "loss": 0.0368,
515
+ "step": 1775
516
+ },
517
+ {
518
+ "epoch": 3.47,
519
+ "grad_norm": 1.6017179489135742,
520
+ "learning_rate": 6.302857142857144e-06,
521
+ "loss": 0.0351,
522
+ "step": 1800
523
+ },
524
+ {
525
+ "epoch": 3.52,
526
+ "grad_norm": 3.2970995903015137,
527
+ "learning_rate": 6.231428571428571e-06,
528
+ "loss": 0.0332,
529
+ "step": 1825
530
+ },
531
+ {
532
+ "epoch": 3.57,
533
+ "grad_norm": 2.8597681522369385,
534
+ "learning_rate": 6.16e-06,
535
+ "loss": 0.0355,
536
+ "step": 1850
537
+ },
538
+ {
539
+ "epoch": 3.62,
540
+ "grad_norm": 2.8335232734680176,
541
+ "learning_rate": 6.088571428571429e-06,
542
+ "loss": 0.0338,
543
+ "step": 1875
544
+ },
545
+ {
546
+ "epoch": 3.67,
547
+ "grad_norm": 3.405618667602539,
548
+ "learning_rate": 6.017142857142858e-06,
549
+ "loss": 0.0376,
550
+ "step": 1900
551
+ },
552
+ {
553
+ "epoch": 3.72,
554
+ "grad_norm": 2.4519917964935303,
555
+ "learning_rate": 5.945714285714286e-06,
556
+ "loss": 0.0359,
557
+ "step": 1925
558
+ },
559
+ {
560
+ "epoch": 3.76,
561
+ "grad_norm": 3.2067463397979736,
562
+ "learning_rate": 5.874285714285715e-06,
563
+ "loss": 0.029,
564
+ "step": 1950
565
+ },
566
+ {
567
+ "epoch": 3.81,
568
+ "grad_norm": 2.5410799980163574,
569
+ "learning_rate": 5.802857142857144e-06,
570
+ "loss": 0.0313,
571
+ "step": 1975
572
+ },
573
+ {
574
+ "epoch": 3.86,
575
+ "grad_norm": 1.2326685190200806,
576
+ "learning_rate": 5.731428571428572e-06,
577
+ "loss": 0.0379,
578
+ "step": 2000
579
+ },
580
+ {
581
+ "epoch": 3.86,
582
+ "eval_loss": 0.28133508563041687,
583
+ "eval_runtime": 3847.5516,
584
+ "eval_samples_per_second": 0.94,
585
+ "eval_steps_per_second": 0.118,
586
+ "eval_wer": 17.470580026978,
587
+ "step": 2000
588
+ },
589
+ {
590
+ "epoch": 3.91,
591
+ "grad_norm": 2.7244298458099365,
592
+ "learning_rate": 5.66e-06,
593
+ "loss": 0.0356,
594
+ "step": 2025
595
+ },
596
+ {
597
+ "epoch": 3.96,
598
+ "grad_norm": 2.446558952331543,
599
+ "learning_rate": 5.588571428571429e-06,
600
+ "loss": 0.0273,
601
+ "step": 2050
602
+ },
603
+ {
604
+ "epoch": 4.01,
605
+ "grad_norm": 0.7038655281066895,
606
+ "learning_rate": 5.517142857142857e-06,
607
+ "loss": 0.0298,
608
+ "step": 2075
609
+ },
610
+ {
611
+ "epoch": 4.05,
612
+ "grad_norm": 0.8899651765823364,
613
+ "learning_rate": 5.445714285714286e-06,
614
+ "loss": 0.0162,
615
+ "step": 2100
616
+ },
617
+ {
618
+ "epoch": 4.1,
619
+ "grad_norm": 2.154207944869995,
620
+ "learning_rate": 5.374285714285715e-06,
621
+ "loss": 0.0143,
622
+ "step": 2125
623
+ },
624
+ {
625
+ "epoch": 4.15,
626
+ "grad_norm": 1.203642725944519,
627
+ "learning_rate": 5.3028571428571425e-06,
628
+ "loss": 0.0142,
629
+ "step": 2150
630
+ },
631
+ {
632
+ "epoch": 4.2,
633
+ "grad_norm": 0.45312538743019104,
634
+ "learning_rate": 5.2314285714285716e-06,
635
+ "loss": 0.0133,
636
+ "step": 2175
637
+ },
638
+ {
639
+ "epoch": 4.25,
640
+ "grad_norm": 1.146144151687622,
641
+ "learning_rate": 5.1600000000000006e-06,
642
+ "loss": 0.0142,
643
+ "step": 2200
644
+ },
645
+ {
646
+ "epoch": 4.3,
647
+ "grad_norm": 2.2396395206451416,
648
+ "learning_rate": 5.08857142857143e-06,
649
+ "loss": 0.0163,
650
+ "step": 2225
651
+ },
652
+ {
653
+ "epoch": 4.34,
654
+ "grad_norm": 0.6650336384773254,
655
+ "learning_rate": 5.017142857142857e-06,
656
+ "loss": 0.0119,
657
+ "step": 2250
658
+ },
659
+ {
660
+ "epoch": 4.39,
661
+ "grad_norm": 0.5963819026947021,
662
+ "learning_rate": 4.945714285714286e-06,
663
+ "loss": 0.0173,
664
+ "step": 2275
665
+ },
666
+ {
667
+ "epoch": 4.44,
668
+ "grad_norm": 1.5120543241500854,
669
+ "learning_rate": 4.874285714285715e-06,
670
+ "loss": 0.0148,
671
+ "step": 2300
672
+ },
673
+ {
674
+ "epoch": 4.49,
675
+ "grad_norm": 0.7102583050727844,
676
+ "learning_rate": 4.802857142857143e-06,
677
+ "loss": 0.0121,
678
+ "step": 2325
679
+ },
680
+ {
681
+ "epoch": 4.54,
682
+ "grad_norm": 1.2483290433883667,
683
+ "learning_rate": 4.731428571428572e-06,
684
+ "loss": 0.0169,
685
+ "step": 2350
686
+ },
687
+ {
688
+ "epoch": 4.58,
689
+ "grad_norm": 1.8999782800674438,
690
+ "learning_rate": 4.66e-06,
691
+ "loss": 0.013,
692
+ "step": 2375
693
+ },
694
+ {
695
+ "epoch": 4.63,
696
+ "grad_norm": 0.893521249294281,
697
+ "learning_rate": 4.588571428571429e-06,
698
+ "loss": 0.0149,
699
+ "step": 2400
700
+ },
701
+ {
702
+ "epoch": 4.68,
703
+ "grad_norm": 1.4212769269943237,
704
+ "learning_rate": 4.5171428571428575e-06,
705
+ "loss": 0.0132,
706
+ "step": 2425
707
+ },
708
+ {
709
+ "epoch": 4.73,
710
+ "grad_norm": 0.5899932980537415,
711
+ "learning_rate": 4.445714285714286e-06,
712
+ "loss": 0.018,
713
+ "step": 2450
714
+ },
715
+ {
716
+ "epoch": 4.78,
717
+ "grad_norm": 0.5150299668312073,
718
+ "learning_rate": 4.374285714285715e-06,
719
+ "loss": 0.012,
720
+ "step": 2475
721
+ },
722
+ {
723
+ "epoch": 4.83,
724
+ "grad_norm": 1.5570229291915894,
725
+ "learning_rate": 4.302857142857143e-06,
726
+ "loss": 0.0121,
727
+ "step": 2500
728
+ },
729
+ {
730
+ "epoch": 4.87,
731
+ "grad_norm": 1.727569818496704,
732
+ "learning_rate": 4.231428571428572e-06,
733
+ "loss": 0.0117,
734
+ "step": 2525
735
+ },
736
+ {
737
+ "epoch": 4.92,
738
+ "grad_norm": 0.8942755460739136,
739
+ "learning_rate": 4.16e-06,
740
+ "loss": 0.0125,
741
+ "step": 2550
742
+ },
743
+ {
744
+ "epoch": 4.97,
745
+ "grad_norm": 2.9615914821624756,
746
+ "learning_rate": 4.088571428571429e-06,
747
+ "loss": 0.0129,
748
+ "step": 2575
749
+ },
750
+ {
751
+ "epoch": 5.02,
752
+ "grad_norm": 0.27788445353507996,
753
+ "learning_rate": 4.017142857142857e-06,
754
+ "loss": 0.0101,
755
+ "step": 2600
756
+ },
757
+ {
758
+ "epoch": 5.07,
759
+ "grad_norm": 0.5583022832870483,
760
+ "learning_rate": 3.945714285714286e-06,
761
+ "loss": 0.007,
762
+ "step": 2625
763
+ },
764
+ {
765
+ "epoch": 5.12,
766
+ "grad_norm": 0.392986536026001,
767
+ "learning_rate": 3.874285714285715e-06,
768
+ "loss": 0.0065,
769
+ "step": 2650
770
+ },
771
+ {
772
+ "epoch": 5.16,
773
+ "grad_norm": 0.7827091813087463,
774
+ "learning_rate": 3.802857142857143e-06,
775
+ "loss": 0.0054,
776
+ "step": 2675
777
+ },
778
+ {
779
+ "epoch": 5.21,
780
+ "grad_norm": 0.5473342537879944,
781
+ "learning_rate": 3.731428571428572e-06,
782
+ "loss": 0.0068,
783
+ "step": 2700
784
+ },
785
+ {
786
+ "epoch": 5.26,
787
+ "grad_norm": 0.545080840587616,
788
+ "learning_rate": 3.66e-06,
789
+ "loss": 0.0055,
790
+ "step": 2725
791
+ },
792
+ {
793
+ "epoch": 5.31,
794
+ "grad_norm": 0.44736358523368835,
795
+ "learning_rate": 3.588571428571429e-06,
796
+ "loss": 0.0085,
797
+ "step": 2750
798
+ },
799
+ {
800
+ "epoch": 5.36,
801
+ "grad_norm": 0.7332315444946289,
802
+ "learning_rate": 3.5171428571428573e-06,
803
+ "loss": 0.0071,
804
+ "step": 2775
805
+ },
806
+ {
807
+ "epoch": 5.41,
808
+ "grad_norm": 0.6373933553695679,
809
+ "learning_rate": 3.4457142857142863e-06,
810
+ "loss": 0.0058,
811
+ "step": 2800
812
+ },
813
+ {
814
+ "epoch": 5.45,
815
+ "grad_norm": 0.47440409660339355,
816
+ "learning_rate": 3.3742857142857145e-06,
817
+ "loss": 0.0056,
818
+ "step": 2825
819
+ },
820
+ {
821
+ "epoch": 5.5,
822
+ "grad_norm": 0.23941631615161896,
823
+ "learning_rate": 3.302857142857143e-06,
824
+ "loss": 0.0046,
825
+ "step": 2850
826
+ },
827
+ {
828
+ "epoch": 5.55,
829
+ "grad_norm": 0.5772097110748291,
830
+ "learning_rate": 3.2314285714285716e-06,
831
+ "loss": 0.0055,
832
+ "step": 2875
833
+ },
834
+ {
835
+ "epoch": 5.6,
836
+ "grad_norm": 0.22144152224063873,
837
+ "learning_rate": 3.1600000000000002e-06,
838
+ "loss": 0.0048,
839
+ "step": 2900
840
+ },
841
+ {
842
+ "epoch": 5.65,
843
+ "grad_norm": 0.28436514735221863,
844
+ "learning_rate": 3.0885714285714284e-06,
845
+ "loss": 0.0052,
846
+ "step": 2925
847
+ },
848
+ {
849
+ "epoch": 5.69,
850
+ "grad_norm": 0.2404721975326538,
851
+ "learning_rate": 3.0171428571428574e-06,
852
+ "loss": 0.0056,
853
+ "step": 2950
854
+ },
855
+ {
856
+ "epoch": 5.74,
857
+ "grad_norm": 0.49395838379859924,
858
+ "learning_rate": 2.945714285714286e-06,
859
+ "loss": 0.005,
860
+ "step": 2975
861
+ },
862
+ {
863
+ "epoch": 5.79,
864
+ "grad_norm": 0.32905882596969604,
865
+ "learning_rate": 2.8742857142857146e-06,
866
+ "loss": 0.006,
867
+ "step": 3000
868
+ },
869
+ {
870
+ "epoch": 5.79,
871
+ "eval_loss": 0.32524359226226807,
872
+ "eval_runtime": 2778.5628,
873
+ "eval_samples_per_second": 1.302,
874
+ "eval_steps_per_second": 0.163,
875
+ "eval_wer": 17.18219452067538,
876
+ "step": 3000
877
+ },
878
+ {
879
+ "epoch": 5.84,
880
+ "grad_norm": 0.21157915890216827,
881
+ "learning_rate": 2.802857142857143e-06,
882
+ "loss": 0.0053,
883
+ "step": 3025
884
+ },
885
+ {
886
+ "epoch": 5.89,
887
+ "grad_norm": 1.544253945350647,
888
+ "learning_rate": 2.7314285714285714e-06,
889
+ "loss": 0.0048,
890
+ "step": 3050
891
+ },
892
+ {
893
+ "epoch": 5.94,
894
+ "grad_norm": 1.376765251159668,
895
+ "learning_rate": 2.6600000000000004e-06,
896
+ "loss": 0.0059,
897
+ "step": 3075
898
+ },
899
+ {
900
+ "epoch": 5.98,
901
+ "grad_norm": 0.9934395551681519,
902
+ "learning_rate": 2.5885714285714285e-06,
903
+ "loss": 0.0063,
904
+ "step": 3100
905
+ },
906
+ {
907
+ "epoch": 6.03,
908
+ "grad_norm": 0.12821950018405914,
909
+ "learning_rate": 2.5171428571428575e-06,
910
+ "loss": 0.0071,
911
+ "step": 3125
912
+ },
913
+ {
914
+ "epoch": 6.08,
915
+ "grad_norm": 0.1806168109178543,
916
+ "learning_rate": 2.445714285714286e-06,
917
+ "loss": 0.0033,
918
+ "step": 3150
919
+ },
920
+ {
921
+ "epoch": 6.13,
922
+ "grad_norm": 0.17940457165241241,
923
+ "learning_rate": 2.3742857142857147e-06,
924
+ "loss": 0.0036,
925
+ "step": 3175
926
+ },
927
+ {
928
+ "epoch": 6.18,
929
+ "grad_norm": 0.16006788611412048,
930
+ "learning_rate": 2.302857142857143e-06,
931
+ "loss": 0.0034,
932
+ "step": 3200
933
+ },
934
+ {
935
+ "epoch": 6.23,
936
+ "grad_norm": 0.14261183142662048,
937
+ "learning_rate": 2.2314285714285715e-06,
938
+ "loss": 0.003,
939
+ "step": 3225
940
+ },
941
+ {
942
+ "epoch": 6.27,
943
+ "grad_norm": 0.17255297303199768,
944
+ "learning_rate": 2.16e-06,
945
+ "loss": 0.003,
946
+ "step": 3250
947
+ },
948
+ {
949
+ "epoch": 6.32,
950
+ "grad_norm": 0.17864187061786652,
951
+ "learning_rate": 2.0885714285714287e-06,
952
+ "loss": 0.0045,
953
+ "step": 3275
954
+ },
955
+ {
956
+ "epoch": 6.37,
957
+ "grad_norm": 0.1522483378648758,
958
+ "learning_rate": 2.0171428571428573e-06,
959
+ "loss": 0.0032,
960
+ "step": 3300
961
+ },
962
+ {
963
+ "epoch": 6.42,
964
+ "grad_norm": 0.137981116771698,
965
+ "learning_rate": 1.945714285714286e-06,
966
+ "loss": 0.0037,
967
+ "step": 3325
968
+ },
969
+ {
970
+ "epoch": 6.47,
971
+ "grad_norm": 0.14591948688030243,
972
+ "learning_rate": 1.8742857142857142e-06,
973
+ "loss": 0.0033,
974
+ "step": 3350
975
+ },
976
+ {
977
+ "epoch": 6.52,
978
+ "grad_norm": 0.18055008351802826,
979
+ "learning_rate": 1.8028571428571432e-06,
980
+ "loss": 0.0031,
981
+ "step": 3375
982
+ },
983
+ {
984
+ "epoch": 6.56,
985
+ "grad_norm": 0.5128812789916992,
986
+ "learning_rate": 1.7314285714285716e-06,
987
+ "loss": 0.0033,
988
+ "step": 3400
989
+ },
990
+ {
991
+ "epoch": 6.61,
992
+ "grad_norm": 0.3612448275089264,
993
+ "learning_rate": 1.6600000000000002e-06,
994
+ "loss": 0.0035,
995
+ "step": 3425
996
+ },
997
+ {
998
+ "epoch": 6.66,
999
+ "grad_norm": 0.2181573212146759,
1000
+ "learning_rate": 1.5885714285714288e-06,
1001
+ "loss": 0.0031,
1002
+ "step": 3450
1003
+ },
1004
+ {
1005
+ "epoch": 6.71,
1006
+ "grad_norm": 0.14946699142456055,
1007
+ "learning_rate": 1.5171428571428574e-06,
1008
+ "loss": 0.003,
1009
+ "step": 3475
1010
+ },
1011
+ {
1012
+ "epoch": 6.76,
1013
+ "grad_norm": 0.1762605756521225,
1014
+ "learning_rate": 1.4457142857142858e-06,
1015
+ "loss": 0.0029,
1016
+ "step": 3500
1017
+ },
1018
+ {
1019
+ "epoch": 6.81,
1020
+ "grad_norm": 0.15365581214427948,
1021
+ "learning_rate": 1.3742857142857143e-06,
1022
+ "loss": 0.0029,
1023
+ "step": 3525
1024
+ },
1025
+ {
1026
+ "epoch": 6.85,
1027
+ "grad_norm": 0.16057215631008148,
1028
+ "learning_rate": 1.302857142857143e-06,
1029
+ "loss": 0.0029,
1030
+ "step": 3550
1031
+ },
1032
+ {
1033
+ "epoch": 6.9,
1034
+ "grad_norm": 0.16833922266960144,
1035
+ "learning_rate": 1.2314285714285715e-06,
1036
+ "loss": 0.0028,
1037
+ "step": 3575
1038
+ },
1039
+ {
1040
+ "epoch": 6.95,
1041
+ "grad_norm": 0.12415596097707748,
1042
+ "learning_rate": 1.1600000000000001e-06,
1043
+ "loss": 0.0034,
1044
+ "step": 3600
1045
+ },
1046
+ {
1047
+ "epoch": 7.0,
1048
+ "grad_norm": 0.1122882068157196,
1049
+ "learning_rate": 1.0885714285714287e-06,
1050
+ "loss": 0.0039,
1051
+ "step": 3625
1052
+ },
1053
+ {
1054
+ "epoch": 7.05,
1055
+ "grad_norm": 0.1392960548400879,
1056
+ "learning_rate": 1.0171428571428573e-06,
1057
+ "loss": 0.0025,
1058
+ "step": 3650
1059
+ },
1060
+ {
1061
+ "epoch": 7.09,
1062
+ "grad_norm": 0.12869219481945038,
1063
+ "learning_rate": 9.457142857142858e-07,
1064
+ "loss": 0.0024,
1065
+ "step": 3675
1066
+ },
1067
+ {
1068
+ "epoch": 7.14,
1069
+ "grad_norm": 0.11602462083101273,
1070
+ "learning_rate": 8.742857142857144e-07,
1071
+ "loss": 0.0025,
1072
+ "step": 3700
1073
+ },
1074
+ {
1075
+ "epoch": 7.19,
1076
+ "grad_norm": 0.14960308372974396,
1077
+ "learning_rate": 8.028571428571429e-07,
1078
+ "loss": 0.0025,
1079
+ "step": 3725
1080
+ },
1081
+ {
1082
+ "epoch": 7.24,
1083
+ "grad_norm": 0.1800779402256012,
1084
+ "learning_rate": 7.314285714285715e-07,
1085
+ "loss": 0.0025,
1086
+ "step": 3750
1087
+ },
1088
+ {
1089
+ "epoch": 7.29,
1090
+ "grad_norm": 0.09803508967161179,
1091
+ "learning_rate": 6.6e-07,
1092
+ "loss": 0.0026,
1093
+ "step": 3775
1094
+ },
1095
+ {
1096
+ "epoch": 7.34,
1097
+ "grad_norm": 0.1335836946964264,
1098
+ "learning_rate": 5.885714285714286e-07,
1099
+ "loss": 0.0029,
1100
+ "step": 3800
1101
+ },
1102
+ {
1103
+ "epoch": 7.38,
1104
+ "grad_norm": 0.13514183461666107,
1105
+ "learning_rate": 5.171428571428572e-07,
1106
+ "loss": 0.0025,
1107
+ "step": 3825
1108
+ },
1109
+ {
1110
+ "epoch": 7.43,
1111
+ "grad_norm": 0.13820233941078186,
1112
+ "learning_rate": 4.457142857142858e-07,
1113
+ "loss": 0.0025,
1114
+ "step": 3850
1115
+ },
1116
+ {
1117
+ "epoch": 7.48,
1118
+ "grad_norm": 0.12157788872718811,
1119
+ "learning_rate": 3.7428571428571434e-07,
1120
+ "loss": 0.0027,
1121
+ "step": 3875
1122
+ },
1123
+ {
1124
+ "epoch": 7.53,
1125
+ "grad_norm": 0.13221661746501923,
1126
+ "learning_rate": 3.028571428571429e-07,
1127
+ "loss": 0.0024,
1128
+ "step": 3900
1129
+ },
1130
+ {
1131
+ "epoch": 7.58,
1132
+ "grad_norm": 0.1175406202673912,
1133
+ "learning_rate": 2.3142857142857144e-07,
1134
+ "loss": 0.0025,
1135
+ "step": 3925
1136
+ },
1137
+ {
1138
+ "epoch": 7.63,
1139
+ "grad_norm": 0.14412976801395416,
1140
+ "learning_rate": 1.6e-07,
1141
+ "loss": 0.0029,
1142
+ "step": 3950
1143
+ },
1144
+ {
1145
+ "epoch": 7.67,
1146
+ "grad_norm": 0.13637402653694153,
1147
+ "learning_rate": 8.857142857142858e-08,
1148
+ "loss": 0.0024,
1149
+ "step": 3975
1150
+ },
1151
+ {
1152
+ "epoch": 7.72,
1153
+ "grad_norm": 0.1199258491396904,
1154
+ "learning_rate": 1.7142857142857143e-08,
1155
+ "loss": 0.0024,
1156
+ "step": 4000
1157
+ },
1158
+ {
1159
+ "epoch": 7.72,
1160
+ "eval_loss": 0.33461278676986694,
1161
+ "eval_runtime": 2683.6147,
1162
+ "eval_samples_per_second": 1.348,
1163
+ "eval_steps_per_second": 0.169,
1164
+ "eval_wer": 17.051955904925812,
1165
+ "step": 4000
1166
+ }
1167
+ ],
1168
+ "logging_steps": 25,
1169
+ "max_steps": 4000,
1170
+ "num_input_tokens_seen": 0,
1171
+ "num_train_epochs": 8,
1172
+ "save_steps": 1000,
1173
+ "total_flos": 1.844118425124864e+19,
1174
+ "train_batch_size": 16,
1175
+ "trial_name": null,
1176
+ "trial_params": null
1177
+ }
checkpoint-4000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d97a7d674ba5458016027e8fcbeac271e6fe45c463b2f89dfb01782b9ae4928d
3
+ size 5112