Bui commited on
Commit
6811f13
1 Parent(s): b1ceeeb

Upload 33 files

Browse files
checkpoint-1000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,265 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "language": "vi",
164
+ "max_initial_timestamp_index": 50,
165
+ "max_length": 448,
166
+ "no_timestamps_token_id": 50363,
167
+ "pad_token_id": 50257,
168
+ "prev_sot_token_id": 50361,
169
+ "return_timestamps": false,
170
+ "suppress_tokens": [
171
+ 1,
172
+ 2,
173
+ 7,
174
+ 8,
175
+ 9,
176
+ 10,
177
+ 14,
178
+ 25,
179
+ 26,
180
+ 27,
181
+ 28,
182
+ 29,
183
+ 31,
184
+ 58,
185
+ 59,
186
+ 60,
187
+ 61,
188
+ 62,
189
+ 63,
190
+ 90,
191
+ 91,
192
+ 92,
193
+ 93,
194
+ 359,
195
+ 503,
196
+ 522,
197
+ 542,
198
+ 873,
199
+ 893,
200
+ 902,
201
+ 918,
202
+ 922,
203
+ 931,
204
+ 1350,
205
+ 1853,
206
+ 1982,
207
+ 2460,
208
+ 2627,
209
+ 3246,
210
+ 3253,
211
+ 3268,
212
+ 3536,
213
+ 3846,
214
+ 3961,
215
+ 4183,
216
+ 4667,
217
+ 6585,
218
+ 6647,
219
+ 7273,
220
+ 9061,
221
+ 9383,
222
+ 10428,
223
+ 10929,
224
+ 11938,
225
+ 12033,
226
+ 12331,
227
+ 12562,
228
+ 13793,
229
+ 14157,
230
+ 14635,
231
+ 15265,
232
+ 15618,
233
+ 16553,
234
+ 16604,
235
+ 18362,
236
+ 18956,
237
+ 20075,
238
+ 21675,
239
+ 22520,
240
+ 26130,
241
+ 26161,
242
+ 26435,
243
+ 28279,
244
+ 29464,
245
+ 31650,
246
+ 32302,
247
+ 32470,
248
+ 36865,
249
+ 42863,
250
+ 47425,
251
+ 49870,
252
+ 50254,
253
+ 50258,
254
+ 50358,
255
+ 50359,
256
+ 50360,
257
+ 50361,
258
+ 50362
259
+ ],
260
+ "task_to_id": {
261
+ "transcribe": 50359,
262
+ "translate": 50358
263
+ },
264
+ "transformers_version": "4.39.1"
265
+ }
checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef009ac15c9f46ab93a9a3b1d0135990f19f8bce36a61cbb26afe0b3de4c9bb6
3
+ size 780140544
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:ea92135d7d3b7301cd76e858a0c9f132e30f7003974e5662b77d48112023688d
3
+ size 14244
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b4597d96578ca53c4b23a2960c30792c952c57e798cedfec3114ae9938daeb4
3
+ size 1064
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,310 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 29.213729575611364,
3
+ "best_model_checkpoint": "./whisper-small-vietnamese\\checkpoint-1000",
4
+ "epoch": 5.747126436781609,
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": 0.14,
13
+ "grad_norm": 118.77095794677734,
14
+ "learning_rate": 4.2000000000000006e-07,
15
+ "loss": 5.0409,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.29,
20
+ "grad_norm": 29.974971771240234,
21
+ "learning_rate": 9.200000000000001e-07,
22
+ "loss": 4.0132,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.43,
27
+ "grad_norm": 25.654932022094727,
28
+ "learning_rate": 1.42e-06,
29
+ "loss": 2.9199,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.57,
34
+ "grad_norm": 16.862260818481445,
35
+ "learning_rate": 1.9200000000000003e-06,
36
+ "loss": 1.7066,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.72,
41
+ "grad_norm": 13.715407371520996,
42
+ "learning_rate": 2.42e-06,
43
+ "loss": 1.2867,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.86,
48
+ "grad_norm": 16.711305618286133,
49
+ "learning_rate": 2.92e-06,
50
+ "loss": 1.1578,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 1.01,
55
+ "grad_norm": 14.38102912902832,
56
+ "learning_rate": 3.4200000000000007e-06,
57
+ "loss": 1.0035,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 1.15,
62
+ "grad_norm": 15.285762786865234,
63
+ "learning_rate": 3.920000000000001e-06,
64
+ "loss": 0.8052,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 1.29,
69
+ "grad_norm": 8.91514778137207,
70
+ "learning_rate": 4.42e-06,
71
+ "loss": 0.286,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 1.44,
76
+ "grad_norm": 7.789880275726318,
77
+ "learning_rate": 4.92e-06,
78
+ "loss": 0.2305,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 1.58,
83
+ "grad_norm": 8.093725204467773,
84
+ "learning_rate": 5.420000000000001e-06,
85
+ "loss": 0.2184,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 1.72,
90
+ "grad_norm": 4.7622785568237305,
91
+ "learning_rate": 5.92e-06,
92
+ "loss": 0.2219,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 1.87,
97
+ "grad_norm": 7.469940662384033,
98
+ "learning_rate": 6.42e-06,
99
+ "loss": 0.2067,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 2.01,
104
+ "grad_norm": 4.024804592132568,
105
+ "learning_rate": 6.92e-06,
106
+ "loss": 0.2128,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 2.16,
111
+ "grad_norm": 4.569382667541504,
112
+ "learning_rate": 7.420000000000001e-06,
113
+ "loss": 0.1294,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 2.3,
118
+ "grad_norm": 5.233477592468262,
119
+ "learning_rate": 7.92e-06,
120
+ "loss": 0.1217,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 2.44,
125
+ "grad_norm": 6.401954174041748,
126
+ "learning_rate": 8.42e-06,
127
+ "loss": 0.1272,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 2.59,
132
+ "grad_norm": 4.058222770690918,
133
+ "learning_rate": 8.920000000000001e-06,
134
+ "loss": 0.1188,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 2.73,
139
+ "grad_norm": 4.228504657745361,
140
+ "learning_rate": 9.42e-06,
141
+ "loss": 0.1294,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 2.87,
146
+ "grad_norm": 5.160878658294678,
147
+ "learning_rate": 9.920000000000002e-06,
148
+ "loss": 0.1241,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 3.02,
153
+ "grad_norm": 3.9591662883758545,
154
+ "learning_rate": 9.940000000000001e-06,
155
+ "loss": 0.1264,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 3.16,
160
+ "grad_norm": 4.170319080352783,
161
+ "learning_rate": 9.86857142857143e-06,
162
+ "loss": 0.0793,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 3.3,
167
+ "grad_norm": 3.4333674907684326,
168
+ "learning_rate": 9.797142857142858e-06,
169
+ "loss": 0.0643,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 3.45,
174
+ "grad_norm": 4.089048862457275,
175
+ "learning_rate": 9.725714285714287e-06,
176
+ "loss": 0.0668,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 3.59,
181
+ "grad_norm": 3.6274781227111816,
182
+ "learning_rate": 9.654285714285716e-06,
183
+ "loss": 0.069,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 3.74,
188
+ "grad_norm": 3.2134461402893066,
189
+ "learning_rate": 9.582857142857143e-06,
190
+ "loss": 0.0759,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 3.88,
195
+ "grad_norm": 3.4892189502716064,
196
+ "learning_rate": 9.511428571428572e-06,
197
+ "loss": 0.0698,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 4.02,
202
+ "grad_norm": 2.3172402381896973,
203
+ "learning_rate": 9.440000000000001e-06,
204
+ "loss": 0.0651,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 4.17,
209
+ "grad_norm": 2.6511449813842773,
210
+ "learning_rate": 9.368571428571428e-06,
211
+ "loss": 0.0373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 4.31,
216
+ "grad_norm": 3.1017539501190186,
217
+ "learning_rate": 9.297142857142857e-06,
218
+ "loss": 0.0415,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 4.45,
223
+ "grad_norm": 3.3337509632110596,
224
+ "learning_rate": 9.225714285714286e-06,
225
+ "loss": 0.0388,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 4.6,
230
+ "grad_norm": 2.3944547176361084,
231
+ "learning_rate": 9.154285714285715e-06,
232
+ "loss": 0.0411,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 4.74,
237
+ "grad_norm": 3.1757237911224365,
238
+ "learning_rate": 9.082857142857143e-06,
239
+ "loss": 0.0381,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 4.89,
244
+ "grad_norm": 4.307478427886963,
245
+ "learning_rate": 9.011428571428572e-06,
246
+ "loss": 0.0434,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 5.03,
251
+ "grad_norm": 1.4969714879989624,
252
+ "learning_rate": 8.94e-06,
253
+ "loss": 0.0384,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 5.17,
258
+ "grad_norm": 1.416281819343567,
259
+ "learning_rate": 8.86857142857143e-06,
260
+ "loss": 0.0207,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 5.32,
265
+ "grad_norm": 4.1161322593688965,
266
+ "learning_rate": 8.797142857142857e-06,
267
+ "loss": 0.0222,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 5.46,
272
+ "grad_norm": 0.9343141317367554,
273
+ "learning_rate": 8.725714285714286e-06,
274
+ "loss": 0.0213,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 5.6,
279
+ "grad_norm": 3.1798384189605713,
280
+ "learning_rate": 8.654285714285715e-06,
281
+ "loss": 0.0239,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 5.75,
286
+ "grad_norm": 1.7415159940719604,
287
+ "learning_rate": 8.582857142857144e-06,
288
+ "loss": 0.0243,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 5.75,
293
+ "eval_loss": 0.6310169696807861,
294
+ "eval_runtime": 482.8872,
295
+ "eval_samples_per_second": 2.562,
296
+ "eval_steps_per_second": 0.321,
297
+ "eval_wer": 29.213729575611364,
298
+ "step": 1000
299
+ }
300
+ ],
301
+ "logging_steps": 25,
302
+ "max_steps": 4000,
303
+ "num_input_tokens_seen": 0,
304
+ "num_train_epochs": 23,
305
+ "save_steps": 1000,
306
+ "total_flos": 4.6014942081024e+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 sha256:3d3010299f45b67cddc7e6d6f4bc607816f3110b6d0b4165041ce3a03b7b1d04
3
+ size 5112
checkpoint-2000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-2000/generation_config.json ADDED
@@ -0,0 +1,265 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "language": "vi",
164
+ "max_initial_timestamp_index": 50,
165
+ "max_length": 448,
166
+ "no_timestamps_token_id": 50363,
167
+ "pad_token_id": 50257,
168
+ "prev_sot_token_id": 50361,
169
+ "return_timestamps": false,
170
+ "suppress_tokens": [
171
+ 1,
172
+ 2,
173
+ 7,
174
+ 8,
175
+ 9,
176
+ 10,
177
+ 14,
178
+ 25,
179
+ 26,
180
+ 27,
181
+ 28,
182
+ 29,
183
+ 31,
184
+ 58,
185
+ 59,
186
+ 60,
187
+ 61,
188
+ 62,
189
+ 63,
190
+ 90,
191
+ 91,
192
+ 92,
193
+ 93,
194
+ 359,
195
+ 503,
196
+ 522,
197
+ 542,
198
+ 873,
199
+ 893,
200
+ 902,
201
+ 918,
202
+ 922,
203
+ 931,
204
+ 1350,
205
+ 1853,
206
+ 1982,
207
+ 2460,
208
+ 2627,
209
+ 3246,
210
+ 3253,
211
+ 3268,
212
+ 3536,
213
+ 3846,
214
+ 3961,
215
+ 4183,
216
+ 4667,
217
+ 6585,
218
+ 6647,
219
+ 7273,
220
+ 9061,
221
+ 9383,
222
+ 10428,
223
+ 10929,
224
+ 11938,
225
+ 12033,
226
+ 12331,
227
+ 12562,
228
+ 13793,
229
+ 14157,
230
+ 14635,
231
+ 15265,
232
+ 15618,
233
+ 16553,
234
+ 16604,
235
+ 18362,
236
+ 18956,
237
+ 20075,
238
+ 21675,
239
+ 22520,
240
+ 26130,
241
+ 26161,
242
+ 26435,
243
+ 28279,
244
+ 29464,
245
+ 31650,
246
+ 32302,
247
+ 32470,
248
+ 36865,
249
+ 42863,
250
+ 47425,
251
+ 49870,
252
+ 50254,
253
+ 50258,
254
+ 50358,
255
+ 50359,
256
+ 50360,
257
+ 50361,
258
+ 50362
259
+ ],
260
+ "task_to_id": {
261
+ "transcribe": 50359,
262
+ "translate": 50358
263
+ },
264
+ "transformers_version": "4.39.1"
265
+ }
checkpoint-2000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:022fb348400b2b4b8e3d6c5fec5adcf4ad642ac4f245eabb49f9f09d4208cdca
3
+ size 966995080
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 sha256:0c02d0d7e3a4c46211a8c5a423c9a6fce7bff98bddc0ec07c01827474fc6cacd
3
+ size 14244
checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e1e6cd32324d2508802718e9ae4fb9fd3a2432ba75cb2c720557bc95b430e30
3
+ size 1064
checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,599 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 27.557846255071826,
3
+ "best_model_checkpoint": "./whisper-small-vietnamese\\checkpoint-2000",
4
+ "epoch": 11.494252873563218,
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": 0.14,
13
+ "grad_norm": 118.77095794677734,
14
+ "learning_rate": 4.2000000000000006e-07,
15
+ "loss": 5.0409,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.29,
20
+ "grad_norm": 29.974971771240234,
21
+ "learning_rate": 9.200000000000001e-07,
22
+ "loss": 4.0132,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.43,
27
+ "grad_norm": 25.654932022094727,
28
+ "learning_rate": 1.42e-06,
29
+ "loss": 2.9199,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.57,
34
+ "grad_norm": 16.862260818481445,
35
+ "learning_rate": 1.9200000000000003e-06,
36
+ "loss": 1.7066,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.72,
41
+ "grad_norm": 13.715407371520996,
42
+ "learning_rate": 2.42e-06,
43
+ "loss": 1.2867,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.86,
48
+ "grad_norm": 16.711305618286133,
49
+ "learning_rate": 2.92e-06,
50
+ "loss": 1.1578,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 1.01,
55
+ "grad_norm": 14.38102912902832,
56
+ "learning_rate": 3.4200000000000007e-06,
57
+ "loss": 1.0035,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 1.15,
62
+ "grad_norm": 15.285762786865234,
63
+ "learning_rate": 3.920000000000001e-06,
64
+ "loss": 0.8052,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 1.29,
69
+ "grad_norm": 8.91514778137207,
70
+ "learning_rate": 4.42e-06,
71
+ "loss": 0.286,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 1.44,
76
+ "grad_norm": 7.789880275726318,
77
+ "learning_rate": 4.92e-06,
78
+ "loss": 0.2305,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 1.58,
83
+ "grad_norm": 8.093725204467773,
84
+ "learning_rate": 5.420000000000001e-06,
85
+ "loss": 0.2184,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 1.72,
90
+ "grad_norm": 4.7622785568237305,
91
+ "learning_rate": 5.92e-06,
92
+ "loss": 0.2219,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 1.87,
97
+ "grad_norm": 7.469940662384033,
98
+ "learning_rate": 6.42e-06,
99
+ "loss": 0.2067,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 2.01,
104
+ "grad_norm": 4.024804592132568,
105
+ "learning_rate": 6.92e-06,
106
+ "loss": 0.2128,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 2.16,
111
+ "grad_norm": 4.569382667541504,
112
+ "learning_rate": 7.420000000000001e-06,
113
+ "loss": 0.1294,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 2.3,
118
+ "grad_norm": 5.233477592468262,
119
+ "learning_rate": 7.92e-06,
120
+ "loss": 0.1217,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 2.44,
125
+ "grad_norm": 6.401954174041748,
126
+ "learning_rate": 8.42e-06,
127
+ "loss": 0.1272,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 2.59,
132
+ "grad_norm": 4.058222770690918,
133
+ "learning_rate": 8.920000000000001e-06,
134
+ "loss": 0.1188,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 2.73,
139
+ "grad_norm": 4.228504657745361,
140
+ "learning_rate": 9.42e-06,
141
+ "loss": 0.1294,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 2.87,
146
+ "grad_norm": 5.160878658294678,
147
+ "learning_rate": 9.920000000000002e-06,
148
+ "loss": 0.1241,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 3.02,
153
+ "grad_norm": 3.9591662883758545,
154
+ "learning_rate": 9.940000000000001e-06,
155
+ "loss": 0.1264,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 3.16,
160
+ "grad_norm": 4.170319080352783,
161
+ "learning_rate": 9.86857142857143e-06,
162
+ "loss": 0.0793,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 3.3,
167
+ "grad_norm": 3.4333674907684326,
168
+ "learning_rate": 9.797142857142858e-06,
169
+ "loss": 0.0643,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 3.45,
174
+ "grad_norm": 4.089048862457275,
175
+ "learning_rate": 9.725714285714287e-06,
176
+ "loss": 0.0668,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 3.59,
181
+ "grad_norm": 3.6274781227111816,
182
+ "learning_rate": 9.654285714285716e-06,
183
+ "loss": 0.069,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 3.74,
188
+ "grad_norm": 3.2134461402893066,
189
+ "learning_rate": 9.582857142857143e-06,
190
+ "loss": 0.0759,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 3.88,
195
+ "grad_norm": 3.4892189502716064,
196
+ "learning_rate": 9.511428571428572e-06,
197
+ "loss": 0.0698,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 4.02,
202
+ "grad_norm": 2.3172402381896973,
203
+ "learning_rate": 9.440000000000001e-06,
204
+ "loss": 0.0651,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 4.17,
209
+ "grad_norm": 2.6511449813842773,
210
+ "learning_rate": 9.368571428571428e-06,
211
+ "loss": 0.0373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 4.31,
216
+ "grad_norm": 3.1017539501190186,
217
+ "learning_rate": 9.297142857142857e-06,
218
+ "loss": 0.0415,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 4.45,
223
+ "grad_norm": 3.3337509632110596,
224
+ "learning_rate": 9.225714285714286e-06,
225
+ "loss": 0.0388,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 4.6,
230
+ "grad_norm": 2.3944547176361084,
231
+ "learning_rate": 9.154285714285715e-06,
232
+ "loss": 0.0411,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 4.74,
237
+ "grad_norm": 3.1757237911224365,
238
+ "learning_rate": 9.082857142857143e-06,
239
+ "loss": 0.0381,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 4.89,
244
+ "grad_norm": 4.307478427886963,
245
+ "learning_rate": 9.011428571428572e-06,
246
+ "loss": 0.0434,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 5.03,
251
+ "grad_norm": 1.4969714879989624,
252
+ "learning_rate": 8.94e-06,
253
+ "loss": 0.0384,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 5.17,
258
+ "grad_norm": 1.416281819343567,
259
+ "learning_rate": 8.86857142857143e-06,
260
+ "loss": 0.0207,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 5.32,
265
+ "grad_norm": 4.1161322593688965,
266
+ "learning_rate": 8.797142857142857e-06,
267
+ "loss": 0.0222,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 5.46,
272
+ "grad_norm": 0.9343141317367554,
273
+ "learning_rate": 8.725714285714286e-06,
274
+ "loss": 0.0213,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 5.6,
279
+ "grad_norm": 3.1798384189605713,
280
+ "learning_rate": 8.654285714285715e-06,
281
+ "loss": 0.0239,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 5.75,
286
+ "grad_norm": 1.7415159940719604,
287
+ "learning_rate": 8.582857142857144e-06,
288
+ "loss": 0.0243,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 5.75,
293
+ "eval_loss": 0.6310169696807861,
294
+ "eval_runtime": 482.8872,
295
+ "eval_samples_per_second": 2.562,
296
+ "eval_steps_per_second": 0.321,
297
+ "eval_wer": 29.213729575611364,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 5.89,
302
+ "grad_norm": 2.3477253913879395,
303
+ "learning_rate": 8.511428571428571e-06,
304
+ "loss": 0.0215,
305
+ "step": 1025
306
+ },
307
+ {
308
+ "epoch": 6.03,
309
+ "grad_norm": 0.7934111952781677,
310
+ "learning_rate": 8.44e-06,
311
+ "loss": 0.019,
312
+ "step": 1050
313
+ },
314
+ {
315
+ "epoch": 6.18,
316
+ "grad_norm": 1.1651448011398315,
317
+ "learning_rate": 8.36857142857143e-06,
318
+ "loss": 0.0096,
319
+ "step": 1075
320
+ },
321
+ {
322
+ "epoch": 6.32,
323
+ "grad_norm": 1.3079229593276978,
324
+ "learning_rate": 8.297142857142859e-06,
325
+ "loss": 0.0117,
326
+ "step": 1100
327
+ },
328
+ {
329
+ "epoch": 6.47,
330
+ "grad_norm": 2.6112120151519775,
331
+ "learning_rate": 8.225714285714288e-06,
332
+ "loss": 0.0097,
333
+ "step": 1125
334
+ },
335
+ {
336
+ "epoch": 6.61,
337
+ "grad_norm": 0.8934247493743896,
338
+ "learning_rate": 8.154285714285715e-06,
339
+ "loss": 0.0085,
340
+ "step": 1150
341
+ },
342
+ {
343
+ "epoch": 6.75,
344
+ "grad_norm": 1.6696081161499023,
345
+ "learning_rate": 8.082857142857144e-06,
346
+ "loss": 0.011,
347
+ "step": 1175
348
+ },
349
+ {
350
+ "epoch": 6.9,
351
+ "grad_norm": 1.9043184518814087,
352
+ "learning_rate": 8.011428571428573e-06,
353
+ "loss": 0.0159,
354
+ "step": 1200
355
+ },
356
+ {
357
+ "epoch": 7.04,
358
+ "grad_norm": 0.7788882255554199,
359
+ "learning_rate": 7.94e-06,
360
+ "loss": 0.0081,
361
+ "step": 1225
362
+ },
363
+ {
364
+ "epoch": 7.18,
365
+ "grad_norm": 0.44921088218688965,
366
+ "learning_rate": 7.86857142857143e-06,
367
+ "loss": 0.0075,
368
+ "step": 1250
369
+ },
370
+ {
371
+ "epoch": 7.33,
372
+ "grad_norm": 1.6392821073532104,
373
+ "learning_rate": 7.797142857142858e-06,
374
+ "loss": 0.0055,
375
+ "step": 1275
376
+ },
377
+ {
378
+ "epoch": 7.47,
379
+ "grad_norm": 2.472712278366089,
380
+ "learning_rate": 7.725714285714286e-06,
381
+ "loss": 0.0059,
382
+ "step": 1300
383
+ },
384
+ {
385
+ "epoch": 7.61,
386
+ "grad_norm": 0.16271165013313293,
387
+ "learning_rate": 7.654285714285715e-06,
388
+ "loss": 0.0046,
389
+ "step": 1325
390
+ },
391
+ {
392
+ "epoch": 7.76,
393
+ "grad_norm": 0.6483273506164551,
394
+ "learning_rate": 7.5828571428571444e-06,
395
+ "loss": 0.0044,
396
+ "step": 1350
397
+ },
398
+ {
399
+ "epoch": 7.9,
400
+ "grad_norm": 1.0499032735824585,
401
+ "learning_rate": 7.511428571428572e-06,
402
+ "loss": 0.0046,
403
+ "step": 1375
404
+ },
405
+ {
406
+ "epoch": 8.05,
407
+ "grad_norm": 0.6188054084777832,
408
+ "learning_rate": 7.440000000000001e-06,
409
+ "loss": 0.0039,
410
+ "step": 1400
411
+ },
412
+ {
413
+ "epoch": 8.19,
414
+ "grad_norm": 0.11255456507205963,
415
+ "learning_rate": 7.36857142857143e-06,
416
+ "loss": 0.0035,
417
+ "step": 1425
418
+ },
419
+ {
420
+ "epoch": 8.33,
421
+ "grad_norm": 0.21570441126823425,
422
+ "learning_rate": 7.297142857142858e-06,
423
+ "loss": 0.0044,
424
+ "step": 1450
425
+ },
426
+ {
427
+ "epoch": 8.48,
428
+ "grad_norm": 0.2393558919429779,
429
+ "learning_rate": 7.225714285714286e-06,
430
+ "loss": 0.0025,
431
+ "step": 1475
432
+ },
433
+ {
434
+ "epoch": 8.62,
435
+ "grad_norm": 0.8370733857154846,
436
+ "learning_rate": 7.154285714285715e-06,
437
+ "loss": 0.0022,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 8.76,
442
+ "grad_norm": 0.25669318437576294,
443
+ "learning_rate": 7.082857142857143e-06,
444
+ "loss": 0.002,
445
+ "step": 1525
446
+ },
447
+ {
448
+ "epoch": 8.91,
449
+ "grad_norm": 0.2188798040151596,
450
+ "learning_rate": 7.011428571428572e-06,
451
+ "loss": 0.002,
452
+ "step": 1550
453
+ },
454
+ {
455
+ "epoch": 9.05,
456
+ "grad_norm": 0.06543828547000885,
457
+ "learning_rate": 6.9400000000000005e-06,
458
+ "loss": 0.0035,
459
+ "step": 1575
460
+ },
461
+ {
462
+ "epoch": 9.2,
463
+ "grad_norm": 0.05953472480177879,
464
+ "learning_rate": 6.868571428571429e-06,
465
+ "loss": 0.0012,
466
+ "step": 1600
467
+ },
468
+ {
469
+ "epoch": 9.34,
470
+ "grad_norm": 0.05562310665845871,
471
+ "learning_rate": 6.797142857142858e-06,
472
+ "loss": 0.0012,
473
+ "step": 1625
474
+ },
475
+ {
476
+ "epoch": 9.48,
477
+ "grad_norm": 0.06505300104618073,
478
+ "learning_rate": 6.725714285714287e-06,
479
+ "loss": 0.0011,
480
+ "step": 1650
481
+ },
482
+ {
483
+ "epoch": 9.63,
484
+ "grad_norm": 0.05420365184545517,
485
+ "learning_rate": 6.654285714285716e-06,
486
+ "loss": 0.001,
487
+ "step": 1675
488
+ },
489
+ {
490
+ "epoch": 9.77,
491
+ "grad_norm": 0.0459221787750721,
492
+ "learning_rate": 6.582857142857143e-06,
493
+ "loss": 0.0009,
494
+ "step": 1700
495
+ },
496
+ {
497
+ "epoch": 9.91,
498
+ "grad_norm": 0.05463138595223427,
499
+ "learning_rate": 6.511428571428572e-06,
500
+ "loss": 0.0009,
501
+ "step": 1725
502
+ },
503
+ {
504
+ "epoch": 10.06,
505
+ "grad_norm": 0.2885764539241791,
506
+ "learning_rate": 6.440000000000001e-06,
507
+ "loss": 0.0009,
508
+ "step": 1750
509
+ },
510
+ {
511
+ "epoch": 10.2,
512
+ "grad_norm": 0.03790697455406189,
513
+ "learning_rate": 6.368571428571429e-06,
514
+ "loss": 0.0007,
515
+ "step": 1775
516
+ },
517
+ {
518
+ "epoch": 10.34,
519
+ "grad_norm": 0.04002631828188896,
520
+ "learning_rate": 6.297142857142857e-06,
521
+ "loss": 0.0007,
522
+ "step": 1800
523
+ },
524
+ {
525
+ "epoch": 10.49,
526
+ "grad_norm": 0.04201885685324669,
527
+ "learning_rate": 6.225714285714286e-06,
528
+ "loss": 0.0006,
529
+ "step": 1825
530
+ },
531
+ {
532
+ "epoch": 10.63,
533
+ "grad_norm": 0.042038705199956894,
534
+ "learning_rate": 6.1542857142857145e-06,
535
+ "loss": 0.0007,
536
+ "step": 1850
537
+ },
538
+ {
539
+ "epoch": 10.78,
540
+ "grad_norm": 0.037008922547101974,
541
+ "learning_rate": 6.0828571428571435e-06,
542
+ "loss": 0.0006,
543
+ "step": 1875
544
+ },
545
+ {
546
+ "epoch": 10.92,
547
+ "grad_norm": 0.03480172157287598,
548
+ "learning_rate": 6.011428571428572e-06,
549
+ "loss": 0.0006,
550
+ "step": 1900
551
+ },
552
+ {
553
+ "epoch": 11.06,
554
+ "grad_norm": 0.0292174331843853,
555
+ "learning_rate": 5.94e-06,
556
+ "loss": 0.0006,
557
+ "step": 1925
558
+ },
559
+ {
560
+ "epoch": 11.21,
561
+ "grad_norm": 0.03173697739839554,
562
+ "learning_rate": 5.868571428571429e-06,
563
+ "loss": 0.0005,
564
+ "step": 1950
565
+ },
566
+ {
567
+ "epoch": 11.35,
568
+ "grad_norm": 0.026889758184552193,
569
+ "learning_rate": 5.797142857142858e-06,
570
+ "loss": 0.0005,
571
+ "step": 1975
572
+ },
573
+ {
574
+ "epoch": 11.49,
575
+ "grad_norm": 0.02639606036245823,
576
+ "learning_rate": 5.725714285714287e-06,
577
+ "loss": 0.0005,
578
+ "step": 2000
579
+ },
580
+ {
581
+ "epoch": 11.49,
582
+ "eval_loss": 0.6665691137313843,
583
+ "eval_runtime": 480.6597,
584
+ "eval_samples_per_second": 2.574,
585
+ "eval_steps_per_second": 0.322,
586
+ "eval_wer": 27.557846255071826,
587
+ "step": 2000
588
+ }
589
+ ],
590
+ "logging_steps": 25,
591
+ "max_steps": 4000,
592
+ "num_input_tokens_seen": 0,
593
+ "num_train_epochs": 23,
594
+ "save_steps": 1000,
595
+ "total_flos": 9.19981397680128e+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 sha256:3d3010299f45b67cddc7e6d6f4bc607816f3110b6d0b4165041ce3a03b7b1d04
3
+ size 5112
checkpoint-3000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-3000/generation_config.json ADDED
@@ -0,0 +1,265 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "language": "vi",
164
+ "max_initial_timestamp_index": 50,
165
+ "max_length": 448,
166
+ "no_timestamps_token_id": 50363,
167
+ "pad_token_id": 50257,
168
+ "prev_sot_token_id": 50361,
169
+ "return_timestamps": false,
170
+ "suppress_tokens": [
171
+ 1,
172
+ 2,
173
+ 7,
174
+ 8,
175
+ 9,
176
+ 10,
177
+ 14,
178
+ 25,
179
+ 26,
180
+ 27,
181
+ 28,
182
+ 29,
183
+ 31,
184
+ 58,
185
+ 59,
186
+ 60,
187
+ 61,
188
+ 62,
189
+ 63,
190
+ 90,
191
+ 91,
192
+ 92,
193
+ 93,
194
+ 359,
195
+ 503,
196
+ 522,
197
+ 542,
198
+ 873,
199
+ 893,
200
+ 902,
201
+ 918,
202
+ 922,
203
+ 931,
204
+ 1350,
205
+ 1853,
206
+ 1982,
207
+ 2460,
208
+ 2627,
209
+ 3246,
210
+ 3253,
211
+ 3268,
212
+ 3536,
213
+ 3846,
214
+ 3961,
215
+ 4183,
216
+ 4667,
217
+ 6585,
218
+ 6647,
219
+ 7273,
220
+ 9061,
221
+ 9383,
222
+ 10428,
223
+ 10929,
224
+ 11938,
225
+ 12033,
226
+ 12331,
227
+ 12562,
228
+ 13793,
229
+ 14157,
230
+ 14635,
231
+ 15265,
232
+ 15618,
233
+ 16553,
234
+ 16604,
235
+ 18362,
236
+ 18956,
237
+ 20075,
238
+ 21675,
239
+ 22520,
240
+ 26130,
241
+ 26161,
242
+ 26435,
243
+ 28279,
244
+ 29464,
245
+ 31650,
246
+ 32302,
247
+ 32470,
248
+ 36865,
249
+ 42863,
250
+ 47425,
251
+ 49870,
252
+ 50254,
253
+ 50258,
254
+ 50358,
255
+ 50359,
256
+ 50360,
257
+ 50361,
258
+ 50362
259
+ ],
260
+ "task_to_id": {
261
+ "transcribe": 50359,
262
+ "translate": 50358
263
+ },
264
+ "transformers_version": "4.39.1"
265
+ }
checkpoint-3000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74ab86a4653257da390ec85581ae6cb74f4511ec10935e7a9843525f5465595b
3
+ size 966995080
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 sha256:39b954dacf6b235c6eacf11111ce2606905b3e7e6dcb3cbb497c586be260dd87
3
+ size 14244
checkpoint-3000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72e251e028e13771444b524afe8498cbdfc5b9d543c0270681ff0bd416ade2da
3
+ size 1064
checkpoint-3000/trainer_state.json ADDED
@@ -0,0 +1,888 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 27.557846255071826,
3
+ "best_model_checkpoint": "./whisper-small-vietnamese\\checkpoint-2000",
4
+ "epoch": 17.24137931034483,
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": 0.14,
13
+ "grad_norm": 118.77095794677734,
14
+ "learning_rate": 4.2000000000000006e-07,
15
+ "loss": 5.0409,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.29,
20
+ "grad_norm": 29.974971771240234,
21
+ "learning_rate": 9.200000000000001e-07,
22
+ "loss": 4.0132,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.43,
27
+ "grad_norm": 25.654932022094727,
28
+ "learning_rate": 1.42e-06,
29
+ "loss": 2.9199,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.57,
34
+ "grad_norm": 16.862260818481445,
35
+ "learning_rate": 1.9200000000000003e-06,
36
+ "loss": 1.7066,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.72,
41
+ "grad_norm": 13.715407371520996,
42
+ "learning_rate": 2.42e-06,
43
+ "loss": 1.2867,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.86,
48
+ "grad_norm": 16.711305618286133,
49
+ "learning_rate": 2.92e-06,
50
+ "loss": 1.1578,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 1.01,
55
+ "grad_norm": 14.38102912902832,
56
+ "learning_rate": 3.4200000000000007e-06,
57
+ "loss": 1.0035,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 1.15,
62
+ "grad_norm": 15.285762786865234,
63
+ "learning_rate": 3.920000000000001e-06,
64
+ "loss": 0.8052,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 1.29,
69
+ "grad_norm": 8.91514778137207,
70
+ "learning_rate": 4.42e-06,
71
+ "loss": 0.286,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 1.44,
76
+ "grad_norm": 7.789880275726318,
77
+ "learning_rate": 4.92e-06,
78
+ "loss": 0.2305,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 1.58,
83
+ "grad_norm": 8.093725204467773,
84
+ "learning_rate": 5.420000000000001e-06,
85
+ "loss": 0.2184,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 1.72,
90
+ "grad_norm": 4.7622785568237305,
91
+ "learning_rate": 5.92e-06,
92
+ "loss": 0.2219,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 1.87,
97
+ "grad_norm": 7.469940662384033,
98
+ "learning_rate": 6.42e-06,
99
+ "loss": 0.2067,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 2.01,
104
+ "grad_norm": 4.024804592132568,
105
+ "learning_rate": 6.92e-06,
106
+ "loss": 0.2128,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 2.16,
111
+ "grad_norm": 4.569382667541504,
112
+ "learning_rate": 7.420000000000001e-06,
113
+ "loss": 0.1294,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 2.3,
118
+ "grad_norm": 5.233477592468262,
119
+ "learning_rate": 7.92e-06,
120
+ "loss": 0.1217,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 2.44,
125
+ "grad_norm": 6.401954174041748,
126
+ "learning_rate": 8.42e-06,
127
+ "loss": 0.1272,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 2.59,
132
+ "grad_norm": 4.058222770690918,
133
+ "learning_rate": 8.920000000000001e-06,
134
+ "loss": 0.1188,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 2.73,
139
+ "grad_norm": 4.228504657745361,
140
+ "learning_rate": 9.42e-06,
141
+ "loss": 0.1294,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 2.87,
146
+ "grad_norm": 5.160878658294678,
147
+ "learning_rate": 9.920000000000002e-06,
148
+ "loss": 0.1241,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 3.02,
153
+ "grad_norm": 3.9591662883758545,
154
+ "learning_rate": 9.940000000000001e-06,
155
+ "loss": 0.1264,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 3.16,
160
+ "grad_norm": 4.170319080352783,
161
+ "learning_rate": 9.86857142857143e-06,
162
+ "loss": 0.0793,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 3.3,
167
+ "grad_norm": 3.4333674907684326,
168
+ "learning_rate": 9.797142857142858e-06,
169
+ "loss": 0.0643,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 3.45,
174
+ "grad_norm": 4.089048862457275,
175
+ "learning_rate": 9.725714285714287e-06,
176
+ "loss": 0.0668,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 3.59,
181
+ "grad_norm": 3.6274781227111816,
182
+ "learning_rate": 9.654285714285716e-06,
183
+ "loss": 0.069,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 3.74,
188
+ "grad_norm": 3.2134461402893066,
189
+ "learning_rate": 9.582857142857143e-06,
190
+ "loss": 0.0759,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 3.88,
195
+ "grad_norm": 3.4892189502716064,
196
+ "learning_rate": 9.511428571428572e-06,
197
+ "loss": 0.0698,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 4.02,
202
+ "grad_norm": 2.3172402381896973,
203
+ "learning_rate": 9.440000000000001e-06,
204
+ "loss": 0.0651,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 4.17,
209
+ "grad_norm": 2.6511449813842773,
210
+ "learning_rate": 9.368571428571428e-06,
211
+ "loss": 0.0373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 4.31,
216
+ "grad_norm": 3.1017539501190186,
217
+ "learning_rate": 9.297142857142857e-06,
218
+ "loss": 0.0415,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 4.45,
223
+ "grad_norm": 3.3337509632110596,
224
+ "learning_rate": 9.225714285714286e-06,
225
+ "loss": 0.0388,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 4.6,
230
+ "grad_norm": 2.3944547176361084,
231
+ "learning_rate": 9.154285714285715e-06,
232
+ "loss": 0.0411,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 4.74,
237
+ "grad_norm": 3.1757237911224365,
238
+ "learning_rate": 9.082857142857143e-06,
239
+ "loss": 0.0381,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 4.89,
244
+ "grad_norm": 4.307478427886963,
245
+ "learning_rate": 9.011428571428572e-06,
246
+ "loss": 0.0434,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 5.03,
251
+ "grad_norm": 1.4969714879989624,
252
+ "learning_rate": 8.94e-06,
253
+ "loss": 0.0384,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 5.17,
258
+ "grad_norm": 1.416281819343567,
259
+ "learning_rate": 8.86857142857143e-06,
260
+ "loss": 0.0207,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 5.32,
265
+ "grad_norm": 4.1161322593688965,
266
+ "learning_rate": 8.797142857142857e-06,
267
+ "loss": 0.0222,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 5.46,
272
+ "grad_norm": 0.9343141317367554,
273
+ "learning_rate": 8.725714285714286e-06,
274
+ "loss": 0.0213,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 5.6,
279
+ "grad_norm": 3.1798384189605713,
280
+ "learning_rate": 8.654285714285715e-06,
281
+ "loss": 0.0239,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 5.75,
286
+ "grad_norm": 1.7415159940719604,
287
+ "learning_rate": 8.582857142857144e-06,
288
+ "loss": 0.0243,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 5.75,
293
+ "eval_loss": 0.6310169696807861,
294
+ "eval_runtime": 482.8872,
295
+ "eval_samples_per_second": 2.562,
296
+ "eval_steps_per_second": 0.321,
297
+ "eval_wer": 29.213729575611364,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 5.89,
302
+ "grad_norm": 2.3477253913879395,
303
+ "learning_rate": 8.511428571428571e-06,
304
+ "loss": 0.0215,
305
+ "step": 1025
306
+ },
307
+ {
308
+ "epoch": 6.03,
309
+ "grad_norm": 0.7934111952781677,
310
+ "learning_rate": 8.44e-06,
311
+ "loss": 0.019,
312
+ "step": 1050
313
+ },
314
+ {
315
+ "epoch": 6.18,
316
+ "grad_norm": 1.1651448011398315,
317
+ "learning_rate": 8.36857142857143e-06,
318
+ "loss": 0.0096,
319
+ "step": 1075
320
+ },
321
+ {
322
+ "epoch": 6.32,
323
+ "grad_norm": 1.3079229593276978,
324
+ "learning_rate": 8.297142857142859e-06,
325
+ "loss": 0.0117,
326
+ "step": 1100
327
+ },
328
+ {
329
+ "epoch": 6.47,
330
+ "grad_norm": 2.6112120151519775,
331
+ "learning_rate": 8.225714285714288e-06,
332
+ "loss": 0.0097,
333
+ "step": 1125
334
+ },
335
+ {
336
+ "epoch": 6.61,
337
+ "grad_norm": 0.8934247493743896,
338
+ "learning_rate": 8.154285714285715e-06,
339
+ "loss": 0.0085,
340
+ "step": 1150
341
+ },
342
+ {
343
+ "epoch": 6.75,
344
+ "grad_norm": 1.6696081161499023,
345
+ "learning_rate": 8.082857142857144e-06,
346
+ "loss": 0.011,
347
+ "step": 1175
348
+ },
349
+ {
350
+ "epoch": 6.9,
351
+ "grad_norm": 1.9043184518814087,
352
+ "learning_rate": 8.011428571428573e-06,
353
+ "loss": 0.0159,
354
+ "step": 1200
355
+ },
356
+ {
357
+ "epoch": 7.04,
358
+ "grad_norm": 0.7788882255554199,
359
+ "learning_rate": 7.94e-06,
360
+ "loss": 0.0081,
361
+ "step": 1225
362
+ },
363
+ {
364
+ "epoch": 7.18,
365
+ "grad_norm": 0.44921088218688965,
366
+ "learning_rate": 7.86857142857143e-06,
367
+ "loss": 0.0075,
368
+ "step": 1250
369
+ },
370
+ {
371
+ "epoch": 7.33,
372
+ "grad_norm": 1.6392821073532104,
373
+ "learning_rate": 7.797142857142858e-06,
374
+ "loss": 0.0055,
375
+ "step": 1275
376
+ },
377
+ {
378
+ "epoch": 7.47,
379
+ "grad_norm": 2.472712278366089,
380
+ "learning_rate": 7.725714285714286e-06,
381
+ "loss": 0.0059,
382
+ "step": 1300
383
+ },
384
+ {
385
+ "epoch": 7.61,
386
+ "grad_norm": 0.16271165013313293,
387
+ "learning_rate": 7.654285714285715e-06,
388
+ "loss": 0.0046,
389
+ "step": 1325
390
+ },
391
+ {
392
+ "epoch": 7.76,
393
+ "grad_norm": 0.6483273506164551,
394
+ "learning_rate": 7.5828571428571444e-06,
395
+ "loss": 0.0044,
396
+ "step": 1350
397
+ },
398
+ {
399
+ "epoch": 7.9,
400
+ "grad_norm": 1.0499032735824585,
401
+ "learning_rate": 7.511428571428572e-06,
402
+ "loss": 0.0046,
403
+ "step": 1375
404
+ },
405
+ {
406
+ "epoch": 8.05,
407
+ "grad_norm": 0.6188054084777832,
408
+ "learning_rate": 7.440000000000001e-06,
409
+ "loss": 0.0039,
410
+ "step": 1400
411
+ },
412
+ {
413
+ "epoch": 8.19,
414
+ "grad_norm": 0.11255456507205963,
415
+ "learning_rate": 7.36857142857143e-06,
416
+ "loss": 0.0035,
417
+ "step": 1425
418
+ },
419
+ {
420
+ "epoch": 8.33,
421
+ "grad_norm": 0.21570441126823425,
422
+ "learning_rate": 7.297142857142858e-06,
423
+ "loss": 0.0044,
424
+ "step": 1450
425
+ },
426
+ {
427
+ "epoch": 8.48,
428
+ "grad_norm": 0.2393558919429779,
429
+ "learning_rate": 7.225714285714286e-06,
430
+ "loss": 0.0025,
431
+ "step": 1475
432
+ },
433
+ {
434
+ "epoch": 8.62,
435
+ "grad_norm": 0.8370733857154846,
436
+ "learning_rate": 7.154285714285715e-06,
437
+ "loss": 0.0022,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 8.76,
442
+ "grad_norm": 0.25669318437576294,
443
+ "learning_rate": 7.082857142857143e-06,
444
+ "loss": 0.002,
445
+ "step": 1525
446
+ },
447
+ {
448
+ "epoch": 8.91,
449
+ "grad_norm": 0.2188798040151596,
450
+ "learning_rate": 7.011428571428572e-06,
451
+ "loss": 0.002,
452
+ "step": 1550
453
+ },
454
+ {
455
+ "epoch": 9.05,
456
+ "grad_norm": 0.06543828547000885,
457
+ "learning_rate": 6.9400000000000005e-06,
458
+ "loss": 0.0035,
459
+ "step": 1575
460
+ },
461
+ {
462
+ "epoch": 9.2,
463
+ "grad_norm": 0.05953472480177879,
464
+ "learning_rate": 6.868571428571429e-06,
465
+ "loss": 0.0012,
466
+ "step": 1600
467
+ },
468
+ {
469
+ "epoch": 9.34,
470
+ "grad_norm": 0.05562310665845871,
471
+ "learning_rate": 6.797142857142858e-06,
472
+ "loss": 0.0012,
473
+ "step": 1625
474
+ },
475
+ {
476
+ "epoch": 9.48,
477
+ "grad_norm": 0.06505300104618073,
478
+ "learning_rate": 6.725714285714287e-06,
479
+ "loss": 0.0011,
480
+ "step": 1650
481
+ },
482
+ {
483
+ "epoch": 9.63,
484
+ "grad_norm": 0.05420365184545517,
485
+ "learning_rate": 6.654285714285716e-06,
486
+ "loss": 0.001,
487
+ "step": 1675
488
+ },
489
+ {
490
+ "epoch": 9.77,
491
+ "grad_norm": 0.0459221787750721,
492
+ "learning_rate": 6.582857142857143e-06,
493
+ "loss": 0.0009,
494
+ "step": 1700
495
+ },
496
+ {
497
+ "epoch": 9.91,
498
+ "grad_norm": 0.05463138595223427,
499
+ "learning_rate": 6.511428571428572e-06,
500
+ "loss": 0.0009,
501
+ "step": 1725
502
+ },
503
+ {
504
+ "epoch": 10.06,
505
+ "grad_norm": 0.2885764539241791,
506
+ "learning_rate": 6.440000000000001e-06,
507
+ "loss": 0.0009,
508
+ "step": 1750
509
+ },
510
+ {
511
+ "epoch": 10.2,
512
+ "grad_norm": 0.03790697455406189,
513
+ "learning_rate": 6.368571428571429e-06,
514
+ "loss": 0.0007,
515
+ "step": 1775
516
+ },
517
+ {
518
+ "epoch": 10.34,
519
+ "grad_norm": 0.04002631828188896,
520
+ "learning_rate": 6.297142857142857e-06,
521
+ "loss": 0.0007,
522
+ "step": 1800
523
+ },
524
+ {
525
+ "epoch": 10.49,
526
+ "grad_norm": 0.04201885685324669,
527
+ "learning_rate": 6.225714285714286e-06,
528
+ "loss": 0.0006,
529
+ "step": 1825
530
+ },
531
+ {
532
+ "epoch": 10.63,
533
+ "grad_norm": 0.042038705199956894,
534
+ "learning_rate": 6.1542857142857145e-06,
535
+ "loss": 0.0007,
536
+ "step": 1850
537
+ },
538
+ {
539
+ "epoch": 10.78,
540
+ "grad_norm": 0.037008922547101974,
541
+ "learning_rate": 6.0828571428571435e-06,
542
+ "loss": 0.0006,
543
+ "step": 1875
544
+ },
545
+ {
546
+ "epoch": 10.92,
547
+ "grad_norm": 0.03480172157287598,
548
+ "learning_rate": 6.011428571428572e-06,
549
+ "loss": 0.0006,
550
+ "step": 1900
551
+ },
552
+ {
553
+ "epoch": 11.06,
554
+ "grad_norm": 0.0292174331843853,
555
+ "learning_rate": 5.94e-06,
556
+ "loss": 0.0006,
557
+ "step": 1925
558
+ },
559
+ {
560
+ "epoch": 11.21,
561
+ "grad_norm": 0.03173697739839554,
562
+ "learning_rate": 5.868571428571429e-06,
563
+ "loss": 0.0005,
564
+ "step": 1950
565
+ },
566
+ {
567
+ "epoch": 11.35,
568
+ "grad_norm": 0.026889758184552193,
569
+ "learning_rate": 5.797142857142858e-06,
570
+ "loss": 0.0005,
571
+ "step": 1975
572
+ },
573
+ {
574
+ "epoch": 11.49,
575
+ "grad_norm": 0.02639606036245823,
576
+ "learning_rate": 5.725714285714287e-06,
577
+ "loss": 0.0005,
578
+ "step": 2000
579
+ },
580
+ {
581
+ "epoch": 11.49,
582
+ "eval_loss": 0.6665691137313843,
583
+ "eval_runtime": 480.6597,
584
+ "eval_samples_per_second": 2.574,
585
+ "eval_steps_per_second": 0.322,
586
+ "eval_wer": 27.557846255071826,
587
+ "step": 2000
588
+ },
589
+ {
590
+ "epoch": 11.64,
591
+ "grad_norm": 0.029212338849902153,
592
+ "learning_rate": 5.654285714285714e-06,
593
+ "loss": 0.0005,
594
+ "step": 2025
595
+ },
596
+ {
597
+ "epoch": 11.78,
598
+ "grad_norm": 0.025965554639697075,
599
+ "learning_rate": 5.582857142857143e-06,
600
+ "loss": 0.0005,
601
+ "step": 2050
602
+ },
603
+ {
604
+ "epoch": 11.93,
605
+ "grad_norm": 0.0334208607673645,
606
+ "learning_rate": 5.511428571428572e-06,
607
+ "loss": 0.0005,
608
+ "step": 2075
609
+ },
610
+ {
611
+ "epoch": 12.07,
612
+ "grad_norm": 0.022900965064764023,
613
+ "learning_rate": 5.4400000000000004e-06,
614
+ "loss": 0.0005,
615
+ "step": 2100
616
+ },
617
+ {
618
+ "epoch": 12.21,
619
+ "grad_norm": 0.024590345099568367,
620
+ "learning_rate": 5.368571428571429e-06,
621
+ "loss": 0.0004,
622
+ "step": 2125
623
+ },
624
+ {
625
+ "epoch": 12.36,
626
+ "grad_norm": 0.02628406509757042,
627
+ "learning_rate": 5.297142857142858e-06,
628
+ "loss": 0.0005,
629
+ "step": 2150
630
+ },
631
+ {
632
+ "epoch": 12.5,
633
+ "grad_norm": 0.037630677223205566,
634
+ "learning_rate": 5.225714285714286e-06,
635
+ "loss": 0.0004,
636
+ "step": 2175
637
+ },
638
+ {
639
+ "epoch": 12.64,
640
+ "grad_norm": 0.020544828847050667,
641
+ "learning_rate": 5.154285714285715e-06,
642
+ "loss": 0.0004,
643
+ "step": 2200
644
+ },
645
+ {
646
+ "epoch": 12.79,
647
+ "grad_norm": 0.02015620656311512,
648
+ "learning_rate": 5.082857142857144e-06,
649
+ "loss": 0.0004,
650
+ "step": 2225
651
+ },
652
+ {
653
+ "epoch": 12.93,
654
+ "grad_norm": 0.027246667072176933,
655
+ "learning_rate": 5.011428571428571e-06,
656
+ "loss": 0.0005,
657
+ "step": 2250
658
+ },
659
+ {
660
+ "epoch": 13.07,
661
+ "grad_norm": 0.023458700627088547,
662
+ "learning_rate": 4.94e-06,
663
+ "loss": 0.0004,
664
+ "step": 2275
665
+ },
666
+ {
667
+ "epoch": 13.22,
668
+ "grad_norm": 0.020197845995426178,
669
+ "learning_rate": 4.868571428571429e-06,
670
+ "loss": 0.0004,
671
+ "step": 2300
672
+ },
673
+ {
674
+ "epoch": 13.36,
675
+ "grad_norm": 0.020962951704859734,
676
+ "learning_rate": 4.797142857142857e-06,
677
+ "loss": 0.0004,
678
+ "step": 2325
679
+ },
680
+ {
681
+ "epoch": 13.51,
682
+ "grad_norm": 0.024561889469623566,
683
+ "learning_rate": 4.725714285714286e-06,
684
+ "loss": 0.0004,
685
+ "step": 2350
686
+ },
687
+ {
688
+ "epoch": 13.65,
689
+ "grad_norm": 0.021130310371518135,
690
+ "learning_rate": 4.6542857142857145e-06,
691
+ "loss": 0.0004,
692
+ "step": 2375
693
+ },
694
+ {
695
+ "epoch": 13.79,
696
+ "grad_norm": 0.01879502646625042,
697
+ "learning_rate": 4.5828571428571435e-06,
698
+ "loss": 0.0004,
699
+ "step": 2400
700
+ },
701
+ {
702
+ "epoch": 13.94,
703
+ "grad_norm": 0.020792394876480103,
704
+ "learning_rate": 4.511428571428572e-06,
705
+ "loss": 0.0004,
706
+ "step": 2425
707
+ },
708
+ {
709
+ "epoch": 14.08,
710
+ "grad_norm": 0.015595002099871635,
711
+ "learning_rate": 4.440000000000001e-06,
712
+ "loss": 0.0004,
713
+ "step": 2450
714
+ },
715
+ {
716
+ "epoch": 14.22,
717
+ "grad_norm": 0.017891034483909607,
718
+ "learning_rate": 4.368571428571429e-06,
719
+ "loss": 0.0003,
720
+ "step": 2475
721
+ },
722
+ {
723
+ "epoch": 14.37,
724
+ "grad_norm": 0.02275862917304039,
725
+ "learning_rate": 4.297142857142858e-06,
726
+ "loss": 0.0004,
727
+ "step": 2500
728
+ },
729
+ {
730
+ "epoch": 14.51,
731
+ "grad_norm": 0.019310912117362022,
732
+ "learning_rate": 4.225714285714286e-06,
733
+ "loss": 0.0004,
734
+ "step": 2525
735
+ },
736
+ {
737
+ "epoch": 14.66,
738
+ "grad_norm": 0.018501998856663704,
739
+ "learning_rate": 4.154285714285714e-06,
740
+ "loss": 0.0003,
741
+ "step": 2550
742
+ },
743
+ {
744
+ "epoch": 14.8,
745
+ "grad_norm": 0.024563327431678772,
746
+ "learning_rate": 4.082857142857143e-06,
747
+ "loss": 0.0004,
748
+ "step": 2575
749
+ },
750
+ {
751
+ "epoch": 14.94,
752
+ "grad_norm": 0.015636250376701355,
753
+ "learning_rate": 4.011428571428571e-06,
754
+ "loss": 0.0003,
755
+ "step": 2600
756
+ },
757
+ {
758
+ "epoch": 15.09,
759
+ "grad_norm": 0.017548007890582085,
760
+ "learning_rate": 3.94e-06,
761
+ "loss": 0.0003,
762
+ "step": 2625
763
+ },
764
+ {
765
+ "epoch": 15.23,
766
+ "grad_norm": 0.016372185200452805,
767
+ "learning_rate": 3.8685714285714286e-06,
768
+ "loss": 0.0003,
769
+ "step": 2650
770
+ },
771
+ {
772
+ "epoch": 15.37,
773
+ "grad_norm": 0.0156166460365057,
774
+ "learning_rate": 3.7971428571428576e-06,
775
+ "loss": 0.0003,
776
+ "step": 2675
777
+ },
778
+ {
779
+ "epoch": 15.52,
780
+ "grad_norm": 0.017448540776968002,
781
+ "learning_rate": 3.7257142857142857e-06,
782
+ "loss": 0.0003,
783
+ "step": 2700
784
+ },
785
+ {
786
+ "epoch": 15.66,
787
+ "grad_norm": 0.01800905540585518,
788
+ "learning_rate": 3.6542857142857148e-06,
789
+ "loss": 0.0003,
790
+ "step": 2725
791
+ },
792
+ {
793
+ "epoch": 15.8,
794
+ "grad_norm": 0.01949344575405121,
795
+ "learning_rate": 3.582857142857143e-06,
796
+ "loss": 0.0003,
797
+ "step": 2750
798
+ },
799
+ {
800
+ "epoch": 15.95,
801
+ "grad_norm": 0.01464366540312767,
802
+ "learning_rate": 3.511428571428572e-06,
803
+ "loss": 0.0003,
804
+ "step": 2775
805
+ },
806
+ {
807
+ "epoch": 16.09,
808
+ "grad_norm": 0.017990361899137497,
809
+ "learning_rate": 3.44e-06,
810
+ "loss": 0.0003,
811
+ "step": 2800
812
+ },
813
+ {
814
+ "epoch": 16.24,
815
+ "grad_norm": 0.013816111721098423,
816
+ "learning_rate": 3.3685714285714287e-06,
817
+ "loss": 0.0003,
818
+ "step": 2825
819
+ },
820
+ {
821
+ "epoch": 16.38,
822
+ "grad_norm": 0.014646317809820175,
823
+ "learning_rate": 3.2971428571428577e-06,
824
+ "loss": 0.0003,
825
+ "step": 2850
826
+ },
827
+ {
828
+ "epoch": 16.52,
829
+ "grad_norm": 0.014701823703944683,
830
+ "learning_rate": 3.225714285714286e-06,
831
+ "loss": 0.0003,
832
+ "step": 2875
833
+ },
834
+ {
835
+ "epoch": 16.67,
836
+ "grad_norm": 0.016701674088835716,
837
+ "learning_rate": 3.154285714285715e-06,
838
+ "loss": 0.0003,
839
+ "step": 2900
840
+ },
841
+ {
842
+ "epoch": 16.81,
843
+ "grad_norm": 0.015475871041417122,
844
+ "learning_rate": 3.082857142857143e-06,
845
+ "loss": 0.0003,
846
+ "step": 2925
847
+ },
848
+ {
849
+ "epoch": 16.95,
850
+ "grad_norm": 0.013779123313724995,
851
+ "learning_rate": 3.0114285714285716e-06,
852
+ "loss": 0.0003,
853
+ "step": 2950
854
+ },
855
+ {
856
+ "epoch": 17.1,
857
+ "grad_norm": 0.01210096850991249,
858
+ "learning_rate": 2.9400000000000002e-06,
859
+ "loss": 0.0003,
860
+ "step": 2975
861
+ },
862
+ {
863
+ "epoch": 17.24,
864
+ "grad_norm": 0.01389345247298479,
865
+ "learning_rate": 2.868571428571429e-06,
866
+ "loss": 0.0003,
867
+ "step": 3000
868
+ },
869
+ {
870
+ "epoch": 17.24,
871
+ "eval_loss": 0.6930021047592163,
872
+ "eval_runtime": 480.9102,
873
+ "eval_samples_per_second": 2.572,
874
+ "eval_steps_per_second": 0.322,
875
+ "eval_wer": 27.601710713894068,
876
+ "step": 3000
877
+ }
878
+ ],
879
+ "logging_steps": 25,
880
+ "max_steps": 4000,
881
+ "num_input_tokens_seen": 0,
882
+ "num_train_epochs": 23,
883
+ "save_steps": 1000,
884
+ "total_flos": 1.379813374550016e+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 sha256:3d3010299f45b67cddc7e6d6f4bc607816f3110b6d0b4165041ce3a03b7b1d04
3
+ size 5112
checkpoint-4000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "median_filter_width": 7,
41
+ "model_type": "whisper",
42
+ "num_hidden_layers": 12,
43
+ "num_mel_bins": 80,
44
+ "pad_token_id": 50257,
45
+ "scale_embedding": false,
46
+ "suppress_tokens": [],
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.39.1",
49
+ "use_cache": true,
50
+ "use_weighted_layer_sum": false,
51
+ "vocab_size": 51865
52
+ }
checkpoint-4000/generation_config.json ADDED
@@ -0,0 +1,265 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 5,
5
+ 3
6
+ ],
7
+ [
8
+ 5,
9
+ 9
10
+ ],
11
+ [
12
+ 8,
13
+ 0
14
+ ],
15
+ [
16
+ 8,
17
+ 4
18
+ ],
19
+ [
20
+ 8,
21
+ 7
22
+ ],
23
+ [
24
+ 8,
25
+ 8
26
+ ],
27
+ [
28
+ 9,
29
+ 0
30
+ ],
31
+ [
32
+ 9,
33
+ 7
34
+ ],
35
+ [
36
+ 9,
37
+ 9
38
+ ],
39
+ [
40
+ 10,
41
+ 5
42
+ ]
43
+ ],
44
+ "begin_suppress_tokens": [
45
+ 220,
46
+ 50257
47
+ ],
48
+ "bos_token_id": 50257,
49
+ "decoder_start_token_id": 50258,
50
+ "eos_token_id": 50257,
51
+ "forced_decoder_ids": [
52
+ [
53
+ 1,
54
+ null
55
+ ],
56
+ [
57
+ 2,
58
+ 50359
59
+ ]
60
+ ],
61
+ "is_multilingual": true,
62
+ "lang_to_id": {
63
+ "<|af|>": 50327,
64
+ "<|am|>": 50334,
65
+ "<|ar|>": 50272,
66
+ "<|as|>": 50350,
67
+ "<|az|>": 50304,
68
+ "<|ba|>": 50355,
69
+ "<|be|>": 50330,
70
+ "<|bg|>": 50292,
71
+ "<|bn|>": 50302,
72
+ "<|bo|>": 50347,
73
+ "<|br|>": 50309,
74
+ "<|bs|>": 50315,
75
+ "<|ca|>": 50270,
76
+ "<|cs|>": 50283,
77
+ "<|cy|>": 50297,
78
+ "<|da|>": 50285,
79
+ "<|de|>": 50261,
80
+ "<|el|>": 50281,
81
+ "<|en|>": 50259,
82
+ "<|es|>": 50262,
83
+ "<|et|>": 50307,
84
+ "<|eu|>": 50310,
85
+ "<|fa|>": 50300,
86
+ "<|fi|>": 50277,
87
+ "<|fo|>": 50338,
88
+ "<|fr|>": 50265,
89
+ "<|gl|>": 50319,
90
+ "<|gu|>": 50333,
91
+ "<|haw|>": 50352,
92
+ "<|ha|>": 50354,
93
+ "<|he|>": 50279,
94
+ "<|hi|>": 50276,
95
+ "<|hr|>": 50291,
96
+ "<|ht|>": 50339,
97
+ "<|hu|>": 50286,
98
+ "<|hy|>": 50312,
99
+ "<|id|>": 50275,
100
+ "<|is|>": 50311,
101
+ "<|it|>": 50274,
102
+ "<|ja|>": 50266,
103
+ "<|jw|>": 50356,
104
+ "<|ka|>": 50329,
105
+ "<|kk|>": 50316,
106
+ "<|km|>": 50323,
107
+ "<|kn|>": 50306,
108
+ "<|ko|>": 50264,
109
+ "<|la|>": 50294,
110
+ "<|lb|>": 50345,
111
+ "<|ln|>": 50353,
112
+ "<|lo|>": 50336,
113
+ "<|lt|>": 50293,
114
+ "<|lv|>": 50301,
115
+ "<|mg|>": 50349,
116
+ "<|mi|>": 50295,
117
+ "<|mk|>": 50308,
118
+ "<|ml|>": 50296,
119
+ "<|mn|>": 50314,
120
+ "<|mr|>": 50320,
121
+ "<|ms|>": 50282,
122
+ "<|mt|>": 50343,
123
+ "<|my|>": 50346,
124
+ "<|ne|>": 50313,
125
+ "<|nl|>": 50271,
126
+ "<|nn|>": 50342,
127
+ "<|no|>": 50288,
128
+ "<|oc|>": 50328,
129
+ "<|pa|>": 50321,
130
+ "<|pl|>": 50269,
131
+ "<|ps|>": 50340,
132
+ "<|pt|>": 50267,
133
+ "<|ro|>": 50284,
134
+ "<|ru|>": 50263,
135
+ "<|sa|>": 50344,
136
+ "<|sd|>": 50332,
137
+ "<|si|>": 50322,
138
+ "<|sk|>": 50298,
139
+ "<|sl|>": 50305,
140
+ "<|sn|>": 50324,
141
+ "<|so|>": 50326,
142
+ "<|sq|>": 50317,
143
+ "<|sr|>": 50303,
144
+ "<|su|>": 50357,
145
+ "<|sv|>": 50273,
146
+ "<|sw|>": 50318,
147
+ "<|ta|>": 50287,
148
+ "<|te|>": 50299,
149
+ "<|tg|>": 50331,
150
+ "<|th|>": 50289,
151
+ "<|tk|>": 50341,
152
+ "<|tl|>": 50348,
153
+ "<|tr|>": 50268,
154
+ "<|tt|>": 50351,
155
+ "<|uk|>": 50280,
156
+ "<|ur|>": 50290,
157
+ "<|uz|>": 50337,
158
+ "<|vi|>": 50278,
159
+ "<|yi|>": 50335,
160
+ "<|yo|>": 50325,
161
+ "<|zh|>": 50260
162
+ },
163
+ "language": "vi",
164
+ "max_initial_timestamp_index": 50,
165
+ "max_length": 448,
166
+ "no_timestamps_token_id": 50363,
167
+ "pad_token_id": 50257,
168
+ "prev_sot_token_id": 50361,
169
+ "return_timestamps": false,
170
+ "suppress_tokens": [
171
+ 1,
172
+ 2,
173
+ 7,
174
+ 8,
175
+ 9,
176
+ 10,
177
+ 14,
178
+ 25,
179
+ 26,
180
+ 27,
181
+ 28,
182
+ 29,
183
+ 31,
184
+ 58,
185
+ 59,
186
+ 60,
187
+ 61,
188
+ 62,
189
+ 63,
190
+ 90,
191
+ 91,
192
+ 92,
193
+ 93,
194
+ 359,
195
+ 503,
196
+ 522,
197
+ 542,
198
+ 873,
199
+ 893,
200
+ 902,
201
+ 918,
202
+ 922,
203
+ 931,
204
+ 1350,
205
+ 1853,
206
+ 1982,
207
+ 2460,
208
+ 2627,
209
+ 3246,
210
+ 3253,
211
+ 3268,
212
+ 3536,
213
+ 3846,
214
+ 3961,
215
+ 4183,
216
+ 4667,
217
+ 6585,
218
+ 6647,
219
+ 7273,
220
+ 9061,
221
+ 9383,
222
+ 10428,
223
+ 10929,
224
+ 11938,
225
+ 12033,
226
+ 12331,
227
+ 12562,
228
+ 13793,
229
+ 14157,
230
+ 14635,
231
+ 15265,
232
+ 15618,
233
+ 16553,
234
+ 16604,
235
+ 18362,
236
+ 18956,
237
+ 20075,
238
+ 21675,
239
+ 22520,
240
+ 26130,
241
+ 26161,
242
+ 26435,
243
+ 28279,
244
+ 29464,
245
+ 31650,
246
+ 32302,
247
+ 32470,
248
+ 36865,
249
+ 42863,
250
+ 47425,
251
+ 49870,
252
+ 50254,
253
+ 50258,
254
+ 50358,
255
+ 50359,
256
+ 50360,
257
+ 50361,
258
+ 50362
259
+ ],
260
+ "task_to_id": {
261
+ "transcribe": 50359,
262
+ "translate": 50358
263
+ },
264
+ "transformers_version": "4.39.1"
265
+ }
checkpoint-4000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f82590a114717789303a557310a9ea9444af68be7c65d2312c39302d3fe2d97c
3
+ size 966995080
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:f1728fa1022bd2e8c511070754bdda2175962566d4e426353015e278b113a162
3
+ size 14244
checkpoint-4000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a1b605b6a726fe4508975d3c1001bdb069d7f7d2da956e03a21fdf298a91ad4
3
+ size 1064
checkpoint-4000/trainer_state.json ADDED
@@ -0,0 +1,1177 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 27.557846255071826,
3
+ "best_model_checkpoint": "./whisper-small-vietnamese\\checkpoint-2000",
4
+ "epoch": 22.988505747126435,
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": 0.14,
13
+ "grad_norm": 118.77095794677734,
14
+ "learning_rate": 4.2000000000000006e-07,
15
+ "loss": 5.0409,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.29,
20
+ "grad_norm": 29.974971771240234,
21
+ "learning_rate": 9.200000000000001e-07,
22
+ "loss": 4.0132,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.43,
27
+ "grad_norm": 25.654932022094727,
28
+ "learning_rate": 1.42e-06,
29
+ "loss": 2.9199,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.57,
34
+ "grad_norm": 16.862260818481445,
35
+ "learning_rate": 1.9200000000000003e-06,
36
+ "loss": 1.7066,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.72,
41
+ "grad_norm": 13.715407371520996,
42
+ "learning_rate": 2.42e-06,
43
+ "loss": 1.2867,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.86,
48
+ "grad_norm": 16.711305618286133,
49
+ "learning_rate": 2.92e-06,
50
+ "loss": 1.1578,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 1.01,
55
+ "grad_norm": 14.38102912902832,
56
+ "learning_rate": 3.4200000000000007e-06,
57
+ "loss": 1.0035,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 1.15,
62
+ "grad_norm": 15.285762786865234,
63
+ "learning_rate": 3.920000000000001e-06,
64
+ "loss": 0.8052,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 1.29,
69
+ "grad_norm": 8.91514778137207,
70
+ "learning_rate": 4.42e-06,
71
+ "loss": 0.286,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 1.44,
76
+ "grad_norm": 7.789880275726318,
77
+ "learning_rate": 4.92e-06,
78
+ "loss": 0.2305,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 1.58,
83
+ "grad_norm": 8.093725204467773,
84
+ "learning_rate": 5.420000000000001e-06,
85
+ "loss": 0.2184,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 1.72,
90
+ "grad_norm": 4.7622785568237305,
91
+ "learning_rate": 5.92e-06,
92
+ "loss": 0.2219,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 1.87,
97
+ "grad_norm": 7.469940662384033,
98
+ "learning_rate": 6.42e-06,
99
+ "loss": 0.2067,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 2.01,
104
+ "grad_norm": 4.024804592132568,
105
+ "learning_rate": 6.92e-06,
106
+ "loss": 0.2128,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 2.16,
111
+ "grad_norm": 4.569382667541504,
112
+ "learning_rate": 7.420000000000001e-06,
113
+ "loss": 0.1294,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 2.3,
118
+ "grad_norm": 5.233477592468262,
119
+ "learning_rate": 7.92e-06,
120
+ "loss": 0.1217,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 2.44,
125
+ "grad_norm": 6.401954174041748,
126
+ "learning_rate": 8.42e-06,
127
+ "loss": 0.1272,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 2.59,
132
+ "grad_norm": 4.058222770690918,
133
+ "learning_rate": 8.920000000000001e-06,
134
+ "loss": 0.1188,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 2.73,
139
+ "grad_norm": 4.228504657745361,
140
+ "learning_rate": 9.42e-06,
141
+ "loss": 0.1294,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 2.87,
146
+ "grad_norm": 5.160878658294678,
147
+ "learning_rate": 9.920000000000002e-06,
148
+ "loss": 0.1241,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 3.02,
153
+ "grad_norm": 3.9591662883758545,
154
+ "learning_rate": 9.940000000000001e-06,
155
+ "loss": 0.1264,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 3.16,
160
+ "grad_norm": 4.170319080352783,
161
+ "learning_rate": 9.86857142857143e-06,
162
+ "loss": 0.0793,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 3.3,
167
+ "grad_norm": 3.4333674907684326,
168
+ "learning_rate": 9.797142857142858e-06,
169
+ "loss": 0.0643,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 3.45,
174
+ "grad_norm": 4.089048862457275,
175
+ "learning_rate": 9.725714285714287e-06,
176
+ "loss": 0.0668,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 3.59,
181
+ "grad_norm": 3.6274781227111816,
182
+ "learning_rate": 9.654285714285716e-06,
183
+ "loss": 0.069,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 3.74,
188
+ "grad_norm": 3.2134461402893066,
189
+ "learning_rate": 9.582857142857143e-06,
190
+ "loss": 0.0759,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 3.88,
195
+ "grad_norm": 3.4892189502716064,
196
+ "learning_rate": 9.511428571428572e-06,
197
+ "loss": 0.0698,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 4.02,
202
+ "grad_norm": 2.3172402381896973,
203
+ "learning_rate": 9.440000000000001e-06,
204
+ "loss": 0.0651,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 4.17,
209
+ "grad_norm": 2.6511449813842773,
210
+ "learning_rate": 9.368571428571428e-06,
211
+ "loss": 0.0373,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 4.31,
216
+ "grad_norm": 3.1017539501190186,
217
+ "learning_rate": 9.297142857142857e-06,
218
+ "loss": 0.0415,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 4.45,
223
+ "grad_norm": 3.3337509632110596,
224
+ "learning_rate": 9.225714285714286e-06,
225
+ "loss": 0.0388,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 4.6,
230
+ "grad_norm": 2.3944547176361084,
231
+ "learning_rate": 9.154285714285715e-06,
232
+ "loss": 0.0411,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 4.74,
237
+ "grad_norm": 3.1757237911224365,
238
+ "learning_rate": 9.082857142857143e-06,
239
+ "loss": 0.0381,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 4.89,
244
+ "grad_norm": 4.307478427886963,
245
+ "learning_rate": 9.011428571428572e-06,
246
+ "loss": 0.0434,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 5.03,
251
+ "grad_norm": 1.4969714879989624,
252
+ "learning_rate": 8.94e-06,
253
+ "loss": 0.0384,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 5.17,
258
+ "grad_norm": 1.416281819343567,
259
+ "learning_rate": 8.86857142857143e-06,
260
+ "loss": 0.0207,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 5.32,
265
+ "grad_norm": 4.1161322593688965,
266
+ "learning_rate": 8.797142857142857e-06,
267
+ "loss": 0.0222,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 5.46,
272
+ "grad_norm": 0.9343141317367554,
273
+ "learning_rate": 8.725714285714286e-06,
274
+ "loss": 0.0213,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 5.6,
279
+ "grad_norm": 3.1798384189605713,
280
+ "learning_rate": 8.654285714285715e-06,
281
+ "loss": 0.0239,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 5.75,
286
+ "grad_norm": 1.7415159940719604,
287
+ "learning_rate": 8.582857142857144e-06,
288
+ "loss": 0.0243,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 5.75,
293
+ "eval_loss": 0.6310169696807861,
294
+ "eval_runtime": 482.8872,
295
+ "eval_samples_per_second": 2.562,
296
+ "eval_steps_per_second": 0.321,
297
+ "eval_wer": 29.213729575611364,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 5.89,
302
+ "grad_norm": 2.3477253913879395,
303
+ "learning_rate": 8.511428571428571e-06,
304
+ "loss": 0.0215,
305
+ "step": 1025
306
+ },
307
+ {
308
+ "epoch": 6.03,
309
+ "grad_norm": 0.7934111952781677,
310
+ "learning_rate": 8.44e-06,
311
+ "loss": 0.019,
312
+ "step": 1050
313
+ },
314
+ {
315
+ "epoch": 6.18,
316
+ "grad_norm": 1.1651448011398315,
317
+ "learning_rate": 8.36857142857143e-06,
318
+ "loss": 0.0096,
319
+ "step": 1075
320
+ },
321
+ {
322
+ "epoch": 6.32,
323
+ "grad_norm": 1.3079229593276978,
324
+ "learning_rate": 8.297142857142859e-06,
325
+ "loss": 0.0117,
326
+ "step": 1100
327
+ },
328
+ {
329
+ "epoch": 6.47,
330
+ "grad_norm": 2.6112120151519775,
331
+ "learning_rate": 8.225714285714288e-06,
332
+ "loss": 0.0097,
333
+ "step": 1125
334
+ },
335
+ {
336
+ "epoch": 6.61,
337
+ "grad_norm": 0.8934247493743896,
338
+ "learning_rate": 8.154285714285715e-06,
339
+ "loss": 0.0085,
340
+ "step": 1150
341
+ },
342
+ {
343
+ "epoch": 6.75,
344
+ "grad_norm": 1.6696081161499023,
345
+ "learning_rate": 8.082857142857144e-06,
346
+ "loss": 0.011,
347
+ "step": 1175
348
+ },
349
+ {
350
+ "epoch": 6.9,
351
+ "grad_norm": 1.9043184518814087,
352
+ "learning_rate": 8.011428571428573e-06,
353
+ "loss": 0.0159,
354
+ "step": 1200
355
+ },
356
+ {
357
+ "epoch": 7.04,
358
+ "grad_norm": 0.7788882255554199,
359
+ "learning_rate": 7.94e-06,
360
+ "loss": 0.0081,
361
+ "step": 1225
362
+ },
363
+ {
364
+ "epoch": 7.18,
365
+ "grad_norm": 0.44921088218688965,
366
+ "learning_rate": 7.86857142857143e-06,
367
+ "loss": 0.0075,
368
+ "step": 1250
369
+ },
370
+ {
371
+ "epoch": 7.33,
372
+ "grad_norm": 1.6392821073532104,
373
+ "learning_rate": 7.797142857142858e-06,
374
+ "loss": 0.0055,
375
+ "step": 1275
376
+ },
377
+ {
378
+ "epoch": 7.47,
379
+ "grad_norm": 2.472712278366089,
380
+ "learning_rate": 7.725714285714286e-06,
381
+ "loss": 0.0059,
382
+ "step": 1300
383
+ },
384
+ {
385
+ "epoch": 7.61,
386
+ "grad_norm": 0.16271165013313293,
387
+ "learning_rate": 7.654285714285715e-06,
388
+ "loss": 0.0046,
389
+ "step": 1325
390
+ },
391
+ {
392
+ "epoch": 7.76,
393
+ "grad_norm": 0.6483273506164551,
394
+ "learning_rate": 7.5828571428571444e-06,
395
+ "loss": 0.0044,
396
+ "step": 1350
397
+ },
398
+ {
399
+ "epoch": 7.9,
400
+ "grad_norm": 1.0499032735824585,
401
+ "learning_rate": 7.511428571428572e-06,
402
+ "loss": 0.0046,
403
+ "step": 1375
404
+ },
405
+ {
406
+ "epoch": 8.05,
407
+ "grad_norm": 0.6188054084777832,
408
+ "learning_rate": 7.440000000000001e-06,
409
+ "loss": 0.0039,
410
+ "step": 1400
411
+ },
412
+ {
413
+ "epoch": 8.19,
414
+ "grad_norm": 0.11255456507205963,
415
+ "learning_rate": 7.36857142857143e-06,
416
+ "loss": 0.0035,
417
+ "step": 1425
418
+ },
419
+ {
420
+ "epoch": 8.33,
421
+ "grad_norm": 0.21570441126823425,
422
+ "learning_rate": 7.297142857142858e-06,
423
+ "loss": 0.0044,
424
+ "step": 1450
425
+ },
426
+ {
427
+ "epoch": 8.48,
428
+ "grad_norm": 0.2393558919429779,
429
+ "learning_rate": 7.225714285714286e-06,
430
+ "loss": 0.0025,
431
+ "step": 1475
432
+ },
433
+ {
434
+ "epoch": 8.62,
435
+ "grad_norm": 0.8370733857154846,
436
+ "learning_rate": 7.154285714285715e-06,
437
+ "loss": 0.0022,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 8.76,
442
+ "grad_norm": 0.25669318437576294,
443
+ "learning_rate": 7.082857142857143e-06,
444
+ "loss": 0.002,
445
+ "step": 1525
446
+ },
447
+ {
448
+ "epoch": 8.91,
449
+ "grad_norm": 0.2188798040151596,
450
+ "learning_rate": 7.011428571428572e-06,
451
+ "loss": 0.002,
452
+ "step": 1550
453
+ },
454
+ {
455
+ "epoch": 9.05,
456
+ "grad_norm": 0.06543828547000885,
457
+ "learning_rate": 6.9400000000000005e-06,
458
+ "loss": 0.0035,
459
+ "step": 1575
460
+ },
461
+ {
462
+ "epoch": 9.2,
463
+ "grad_norm": 0.05953472480177879,
464
+ "learning_rate": 6.868571428571429e-06,
465
+ "loss": 0.0012,
466
+ "step": 1600
467
+ },
468
+ {
469
+ "epoch": 9.34,
470
+ "grad_norm": 0.05562310665845871,
471
+ "learning_rate": 6.797142857142858e-06,
472
+ "loss": 0.0012,
473
+ "step": 1625
474
+ },
475
+ {
476
+ "epoch": 9.48,
477
+ "grad_norm": 0.06505300104618073,
478
+ "learning_rate": 6.725714285714287e-06,
479
+ "loss": 0.0011,
480
+ "step": 1650
481
+ },
482
+ {
483
+ "epoch": 9.63,
484
+ "grad_norm": 0.05420365184545517,
485
+ "learning_rate": 6.654285714285716e-06,
486
+ "loss": 0.001,
487
+ "step": 1675
488
+ },
489
+ {
490
+ "epoch": 9.77,
491
+ "grad_norm": 0.0459221787750721,
492
+ "learning_rate": 6.582857142857143e-06,
493
+ "loss": 0.0009,
494
+ "step": 1700
495
+ },
496
+ {
497
+ "epoch": 9.91,
498
+ "grad_norm": 0.05463138595223427,
499
+ "learning_rate": 6.511428571428572e-06,
500
+ "loss": 0.0009,
501
+ "step": 1725
502
+ },
503
+ {
504
+ "epoch": 10.06,
505
+ "grad_norm": 0.2885764539241791,
506
+ "learning_rate": 6.440000000000001e-06,
507
+ "loss": 0.0009,
508
+ "step": 1750
509
+ },
510
+ {
511
+ "epoch": 10.2,
512
+ "grad_norm": 0.03790697455406189,
513
+ "learning_rate": 6.368571428571429e-06,
514
+ "loss": 0.0007,
515
+ "step": 1775
516
+ },
517
+ {
518
+ "epoch": 10.34,
519
+ "grad_norm": 0.04002631828188896,
520
+ "learning_rate": 6.297142857142857e-06,
521
+ "loss": 0.0007,
522
+ "step": 1800
523
+ },
524
+ {
525
+ "epoch": 10.49,
526
+ "grad_norm": 0.04201885685324669,
527
+ "learning_rate": 6.225714285714286e-06,
528
+ "loss": 0.0006,
529
+ "step": 1825
530
+ },
531
+ {
532
+ "epoch": 10.63,
533
+ "grad_norm": 0.042038705199956894,
534
+ "learning_rate": 6.1542857142857145e-06,
535
+ "loss": 0.0007,
536
+ "step": 1850
537
+ },
538
+ {
539
+ "epoch": 10.78,
540
+ "grad_norm": 0.037008922547101974,
541
+ "learning_rate": 6.0828571428571435e-06,
542
+ "loss": 0.0006,
543
+ "step": 1875
544
+ },
545
+ {
546
+ "epoch": 10.92,
547
+ "grad_norm": 0.03480172157287598,
548
+ "learning_rate": 6.011428571428572e-06,
549
+ "loss": 0.0006,
550
+ "step": 1900
551
+ },
552
+ {
553
+ "epoch": 11.06,
554
+ "grad_norm": 0.0292174331843853,
555
+ "learning_rate": 5.94e-06,
556
+ "loss": 0.0006,
557
+ "step": 1925
558
+ },
559
+ {
560
+ "epoch": 11.21,
561
+ "grad_norm": 0.03173697739839554,
562
+ "learning_rate": 5.868571428571429e-06,
563
+ "loss": 0.0005,
564
+ "step": 1950
565
+ },
566
+ {
567
+ "epoch": 11.35,
568
+ "grad_norm": 0.026889758184552193,
569
+ "learning_rate": 5.797142857142858e-06,
570
+ "loss": 0.0005,
571
+ "step": 1975
572
+ },
573
+ {
574
+ "epoch": 11.49,
575
+ "grad_norm": 0.02639606036245823,
576
+ "learning_rate": 5.725714285714287e-06,
577
+ "loss": 0.0005,
578
+ "step": 2000
579
+ },
580
+ {
581
+ "epoch": 11.49,
582
+ "eval_loss": 0.6665691137313843,
583
+ "eval_runtime": 480.6597,
584
+ "eval_samples_per_second": 2.574,
585
+ "eval_steps_per_second": 0.322,
586
+ "eval_wer": 27.557846255071826,
587
+ "step": 2000
588
+ },
589
+ {
590
+ "epoch": 11.64,
591
+ "grad_norm": 0.029212338849902153,
592
+ "learning_rate": 5.654285714285714e-06,
593
+ "loss": 0.0005,
594
+ "step": 2025
595
+ },
596
+ {
597
+ "epoch": 11.78,
598
+ "grad_norm": 0.025965554639697075,
599
+ "learning_rate": 5.582857142857143e-06,
600
+ "loss": 0.0005,
601
+ "step": 2050
602
+ },
603
+ {
604
+ "epoch": 11.93,
605
+ "grad_norm": 0.0334208607673645,
606
+ "learning_rate": 5.511428571428572e-06,
607
+ "loss": 0.0005,
608
+ "step": 2075
609
+ },
610
+ {
611
+ "epoch": 12.07,
612
+ "grad_norm": 0.022900965064764023,
613
+ "learning_rate": 5.4400000000000004e-06,
614
+ "loss": 0.0005,
615
+ "step": 2100
616
+ },
617
+ {
618
+ "epoch": 12.21,
619
+ "grad_norm": 0.024590345099568367,
620
+ "learning_rate": 5.368571428571429e-06,
621
+ "loss": 0.0004,
622
+ "step": 2125
623
+ },
624
+ {
625
+ "epoch": 12.36,
626
+ "grad_norm": 0.02628406509757042,
627
+ "learning_rate": 5.297142857142858e-06,
628
+ "loss": 0.0005,
629
+ "step": 2150
630
+ },
631
+ {
632
+ "epoch": 12.5,
633
+ "grad_norm": 0.037630677223205566,
634
+ "learning_rate": 5.225714285714286e-06,
635
+ "loss": 0.0004,
636
+ "step": 2175
637
+ },
638
+ {
639
+ "epoch": 12.64,
640
+ "grad_norm": 0.020544828847050667,
641
+ "learning_rate": 5.154285714285715e-06,
642
+ "loss": 0.0004,
643
+ "step": 2200
644
+ },
645
+ {
646
+ "epoch": 12.79,
647
+ "grad_norm": 0.02015620656311512,
648
+ "learning_rate": 5.082857142857144e-06,
649
+ "loss": 0.0004,
650
+ "step": 2225
651
+ },
652
+ {
653
+ "epoch": 12.93,
654
+ "grad_norm": 0.027246667072176933,
655
+ "learning_rate": 5.011428571428571e-06,
656
+ "loss": 0.0005,
657
+ "step": 2250
658
+ },
659
+ {
660
+ "epoch": 13.07,
661
+ "grad_norm": 0.023458700627088547,
662
+ "learning_rate": 4.94e-06,
663
+ "loss": 0.0004,
664
+ "step": 2275
665
+ },
666
+ {
667
+ "epoch": 13.22,
668
+ "grad_norm": 0.020197845995426178,
669
+ "learning_rate": 4.868571428571429e-06,
670
+ "loss": 0.0004,
671
+ "step": 2300
672
+ },
673
+ {
674
+ "epoch": 13.36,
675
+ "grad_norm": 0.020962951704859734,
676
+ "learning_rate": 4.797142857142857e-06,
677
+ "loss": 0.0004,
678
+ "step": 2325
679
+ },
680
+ {
681
+ "epoch": 13.51,
682
+ "grad_norm": 0.024561889469623566,
683
+ "learning_rate": 4.725714285714286e-06,
684
+ "loss": 0.0004,
685
+ "step": 2350
686
+ },
687
+ {
688
+ "epoch": 13.65,
689
+ "grad_norm": 0.021130310371518135,
690
+ "learning_rate": 4.6542857142857145e-06,
691
+ "loss": 0.0004,
692
+ "step": 2375
693
+ },
694
+ {
695
+ "epoch": 13.79,
696
+ "grad_norm": 0.01879502646625042,
697
+ "learning_rate": 4.5828571428571435e-06,
698
+ "loss": 0.0004,
699
+ "step": 2400
700
+ },
701
+ {
702
+ "epoch": 13.94,
703
+ "grad_norm": 0.020792394876480103,
704
+ "learning_rate": 4.511428571428572e-06,
705
+ "loss": 0.0004,
706
+ "step": 2425
707
+ },
708
+ {
709
+ "epoch": 14.08,
710
+ "grad_norm": 0.015595002099871635,
711
+ "learning_rate": 4.440000000000001e-06,
712
+ "loss": 0.0004,
713
+ "step": 2450
714
+ },
715
+ {
716
+ "epoch": 14.22,
717
+ "grad_norm": 0.017891034483909607,
718
+ "learning_rate": 4.368571428571429e-06,
719
+ "loss": 0.0003,
720
+ "step": 2475
721
+ },
722
+ {
723
+ "epoch": 14.37,
724
+ "grad_norm": 0.02275862917304039,
725
+ "learning_rate": 4.297142857142858e-06,
726
+ "loss": 0.0004,
727
+ "step": 2500
728
+ },
729
+ {
730
+ "epoch": 14.51,
731
+ "grad_norm": 0.019310912117362022,
732
+ "learning_rate": 4.225714285714286e-06,
733
+ "loss": 0.0004,
734
+ "step": 2525
735
+ },
736
+ {
737
+ "epoch": 14.66,
738
+ "grad_norm": 0.018501998856663704,
739
+ "learning_rate": 4.154285714285714e-06,
740
+ "loss": 0.0003,
741
+ "step": 2550
742
+ },
743
+ {
744
+ "epoch": 14.8,
745
+ "grad_norm": 0.024563327431678772,
746
+ "learning_rate": 4.082857142857143e-06,
747
+ "loss": 0.0004,
748
+ "step": 2575
749
+ },
750
+ {
751
+ "epoch": 14.94,
752
+ "grad_norm": 0.015636250376701355,
753
+ "learning_rate": 4.011428571428571e-06,
754
+ "loss": 0.0003,
755
+ "step": 2600
756
+ },
757
+ {
758
+ "epoch": 15.09,
759
+ "grad_norm": 0.017548007890582085,
760
+ "learning_rate": 3.94e-06,
761
+ "loss": 0.0003,
762
+ "step": 2625
763
+ },
764
+ {
765
+ "epoch": 15.23,
766
+ "grad_norm": 0.016372185200452805,
767
+ "learning_rate": 3.8685714285714286e-06,
768
+ "loss": 0.0003,
769
+ "step": 2650
770
+ },
771
+ {
772
+ "epoch": 15.37,
773
+ "grad_norm": 0.0156166460365057,
774
+ "learning_rate": 3.7971428571428576e-06,
775
+ "loss": 0.0003,
776
+ "step": 2675
777
+ },
778
+ {
779
+ "epoch": 15.52,
780
+ "grad_norm": 0.017448540776968002,
781
+ "learning_rate": 3.7257142857142857e-06,
782
+ "loss": 0.0003,
783
+ "step": 2700
784
+ },
785
+ {
786
+ "epoch": 15.66,
787
+ "grad_norm": 0.01800905540585518,
788
+ "learning_rate": 3.6542857142857148e-06,
789
+ "loss": 0.0003,
790
+ "step": 2725
791
+ },
792
+ {
793
+ "epoch": 15.8,
794
+ "grad_norm": 0.01949344575405121,
795
+ "learning_rate": 3.582857142857143e-06,
796
+ "loss": 0.0003,
797
+ "step": 2750
798
+ },
799
+ {
800
+ "epoch": 15.95,
801
+ "grad_norm": 0.01464366540312767,
802
+ "learning_rate": 3.511428571428572e-06,
803
+ "loss": 0.0003,
804
+ "step": 2775
805
+ },
806
+ {
807
+ "epoch": 16.09,
808
+ "grad_norm": 0.017990361899137497,
809
+ "learning_rate": 3.44e-06,
810
+ "loss": 0.0003,
811
+ "step": 2800
812
+ },
813
+ {
814
+ "epoch": 16.24,
815
+ "grad_norm": 0.013816111721098423,
816
+ "learning_rate": 3.3685714285714287e-06,
817
+ "loss": 0.0003,
818
+ "step": 2825
819
+ },
820
+ {
821
+ "epoch": 16.38,
822
+ "grad_norm": 0.014646317809820175,
823
+ "learning_rate": 3.2971428571428577e-06,
824
+ "loss": 0.0003,
825
+ "step": 2850
826
+ },
827
+ {
828
+ "epoch": 16.52,
829
+ "grad_norm": 0.014701823703944683,
830
+ "learning_rate": 3.225714285714286e-06,
831
+ "loss": 0.0003,
832
+ "step": 2875
833
+ },
834
+ {
835
+ "epoch": 16.67,
836
+ "grad_norm": 0.016701674088835716,
837
+ "learning_rate": 3.154285714285715e-06,
838
+ "loss": 0.0003,
839
+ "step": 2900
840
+ },
841
+ {
842
+ "epoch": 16.81,
843
+ "grad_norm": 0.015475871041417122,
844
+ "learning_rate": 3.082857142857143e-06,
845
+ "loss": 0.0003,
846
+ "step": 2925
847
+ },
848
+ {
849
+ "epoch": 16.95,
850
+ "grad_norm": 0.013779123313724995,
851
+ "learning_rate": 3.0114285714285716e-06,
852
+ "loss": 0.0003,
853
+ "step": 2950
854
+ },
855
+ {
856
+ "epoch": 17.1,
857
+ "grad_norm": 0.01210096850991249,
858
+ "learning_rate": 2.9400000000000002e-06,
859
+ "loss": 0.0003,
860
+ "step": 2975
861
+ },
862
+ {
863
+ "epoch": 17.24,
864
+ "grad_norm": 0.01389345247298479,
865
+ "learning_rate": 2.868571428571429e-06,
866
+ "loss": 0.0003,
867
+ "step": 3000
868
+ },
869
+ {
870
+ "epoch": 17.24,
871
+ "eval_loss": 0.6930021047592163,
872
+ "eval_runtime": 480.9102,
873
+ "eval_samples_per_second": 2.572,
874
+ "eval_steps_per_second": 0.322,
875
+ "eval_wer": 27.601710713894068,
876
+ "step": 3000
877
+ },
878
+ {
879
+ "epoch": 17.39,
880
+ "grad_norm": 0.013713260181248188,
881
+ "learning_rate": 2.797142857142857e-06,
882
+ "loss": 0.0003,
883
+ "step": 3025
884
+ },
885
+ {
886
+ "epoch": 17.53,
887
+ "grad_norm": 0.012472675181925297,
888
+ "learning_rate": 2.725714285714286e-06,
889
+ "loss": 0.0003,
890
+ "step": 3050
891
+ },
892
+ {
893
+ "epoch": 17.67,
894
+ "grad_norm": 0.013508176431059837,
895
+ "learning_rate": 2.654285714285714e-06,
896
+ "loss": 0.0003,
897
+ "step": 3075
898
+ },
899
+ {
900
+ "epoch": 17.82,
901
+ "grad_norm": 0.01630580611526966,
902
+ "learning_rate": 2.582857142857143e-06,
903
+ "loss": 0.0003,
904
+ "step": 3100
905
+ },
906
+ {
907
+ "epoch": 17.96,
908
+ "grad_norm": 0.014032123610377312,
909
+ "learning_rate": 2.5114285714285718e-06,
910
+ "loss": 0.0003,
911
+ "step": 3125
912
+ },
913
+ {
914
+ "epoch": 18.1,
915
+ "grad_norm": 0.014417425729334354,
916
+ "learning_rate": 2.4400000000000004e-06,
917
+ "loss": 0.0003,
918
+ "step": 3150
919
+ },
920
+ {
921
+ "epoch": 18.25,
922
+ "grad_norm": 0.013058402575552464,
923
+ "learning_rate": 2.3685714285714285e-06,
924
+ "loss": 0.0003,
925
+ "step": 3175
926
+ },
927
+ {
928
+ "epoch": 18.39,
929
+ "grad_norm": 0.011255180463194847,
930
+ "learning_rate": 2.297142857142857e-06,
931
+ "loss": 0.0003,
932
+ "step": 3200
933
+ },
934
+ {
935
+ "epoch": 18.53,
936
+ "grad_norm": 0.011813076213002205,
937
+ "learning_rate": 2.2257142857142857e-06,
938
+ "loss": 0.0003,
939
+ "step": 3225
940
+ },
941
+ {
942
+ "epoch": 18.68,
943
+ "grad_norm": 0.016077186912298203,
944
+ "learning_rate": 2.1542857142857147e-06,
945
+ "loss": 0.0003,
946
+ "step": 3250
947
+ },
948
+ {
949
+ "epoch": 18.82,
950
+ "grad_norm": 0.012286370620131493,
951
+ "learning_rate": 2.0828571428571433e-06,
952
+ "loss": 0.0002,
953
+ "step": 3275
954
+ },
955
+ {
956
+ "epoch": 18.97,
957
+ "grad_norm": 0.010851682163774967,
958
+ "learning_rate": 2.0114285714285715e-06,
959
+ "loss": 0.0003,
960
+ "step": 3300
961
+ },
962
+ {
963
+ "epoch": 19.11,
964
+ "grad_norm": 0.012027677148580551,
965
+ "learning_rate": 1.94e-06,
966
+ "loss": 0.0002,
967
+ "step": 3325
968
+ },
969
+ {
970
+ "epoch": 19.25,
971
+ "grad_norm": 0.012901864014565945,
972
+ "learning_rate": 1.8685714285714289e-06,
973
+ "loss": 0.0002,
974
+ "step": 3350
975
+ },
976
+ {
977
+ "epoch": 19.4,
978
+ "grad_norm": 0.010309391655027866,
979
+ "learning_rate": 1.7971428571428572e-06,
980
+ "loss": 0.0002,
981
+ "step": 3375
982
+ },
983
+ {
984
+ "epoch": 19.54,
985
+ "grad_norm": 0.012053144164383411,
986
+ "learning_rate": 1.7257142857142858e-06,
987
+ "loss": 0.0002,
988
+ "step": 3400
989
+ },
990
+ {
991
+ "epoch": 19.68,
992
+ "grad_norm": 0.01089125033468008,
993
+ "learning_rate": 1.6542857142857144e-06,
994
+ "loss": 0.0002,
995
+ "step": 3425
996
+ },
997
+ {
998
+ "epoch": 19.83,
999
+ "grad_norm": 0.011995966546237469,
1000
+ "learning_rate": 1.582857142857143e-06,
1001
+ "loss": 0.0002,
1002
+ "step": 3450
1003
+ },
1004
+ {
1005
+ "epoch": 19.97,
1006
+ "grad_norm": 0.01478640642017126,
1007
+ "learning_rate": 1.5114285714285714e-06,
1008
+ "loss": 0.0003,
1009
+ "step": 3475
1010
+ },
1011
+ {
1012
+ "epoch": 20.11,
1013
+ "grad_norm": 0.010389993898570538,
1014
+ "learning_rate": 1.44e-06,
1015
+ "loss": 0.0003,
1016
+ "step": 3500
1017
+ },
1018
+ {
1019
+ "epoch": 20.26,
1020
+ "grad_norm": 0.01525922492146492,
1021
+ "learning_rate": 1.3685714285714286e-06,
1022
+ "loss": 0.0002,
1023
+ "step": 3525
1024
+ },
1025
+ {
1026
+ "epoch": 20.4,
1027
+ "grad_norm": 0.01428759004920721,
1028
+ "learning_rate": 1.2971428571428574e-06,
1029
+ "loss": 0.0002,
1030
+ "step": 3550
1031
+ },
1032
+ {
1033
+ "epoch": 20.55,
1034
+ "grad_norm": 0.012515067122876644,
1035
+ "learning_rate": 1.2257142857142857e-06,
1036
+ "loss": 0.0002,
1037
+ "step": 3575
1038
+ },
1039
+ {
1040
+ "epoch": 20.69,
1041
+ "grad_norm": 0.01102293748408556,
1042
+ "learning_rate": 1.1542857142857143e-06,
1043
+ "loss": 0.0002,
1044
+ "step": 3600
1045
+ },
1046
+ {
1047
+ "epoch": 20.83,
1048
+ "grad_norm": 0.012940412387251854,
1049
+ "learning_rate": 1.082857142857143e-06,
1050
+ "loss": 0.0002,
1051
+ "step": 3625
1052
+ },
1053
+ {
1054
+ "epoch": 20.98,
1055
+ "grad_norm": 0.009889173321425915,
1056
+ "learning_rate": 1.0114285714285715e-06,
1057
+ "loss": 0.0002,
1058
+ "step": 3650
1059
+ },
1060
+ {
1061
+ "epoch": 21.12,
1062
+ "grad_norm": 0.011683104559779167,
1063
+ "learning_rate": 9.400000000000001e-07,
1064
+ "loss": 0.0002,
1065
+ "step": 3675
1066
+ },
1067
+ {
1068
+ "epoch": 21.26,
1069
+ "grad_norm": 0.012568894773721695,
1070
+ "learning_rate": 8.685714285714286e-07,
1071
+ "loss": 0.0002,
1072
+ "step": 3700
1073
+ },
1074
+ {
1075
+ "epoch": 21.41,
1076
+ "grad_norm": 0.01185997761785984,
1077
+ "learning_rate": 7.971428571428572e-07,
1078
+ "loss": 0.0002,
1079
+ "step": 3725
1080
+ },
1081
+ {
1082
+ "epoch": 21.55,
1083
+ "grad_norm": 0.01229412853717804,
1084
+ "learning_rate": 7.257142857142857e-07,
1085
+ "loss": 0.0002,
1086
+ "step": 3750
1087
+ },
1088
+ {
1089
+ "epoch": 21.7,
1090
+ "grad_norm": 0.013192495331168175,
1091
+ "learning_rate": 6.542857142857144e-07,
1092
+ "loss": 0.0002,
1093
+ "step": 3775
1094
+ },
1095
+ {
1096
+ "epoch": 21.84,
1097
+ "grad_norm": 0.010738629847764969,
1098
+ "learning_rate": 5.82857142857143e-07,
1099
+ "loss": 0.0002,
1100
+ "step": 3800
1101
+ },
1102
+ {
1103
+ "epoch": 21.98,
1104
+ "grad_norm": 0.010789092630147934,
1105
+ "learning_rate": 5.114285714285714e-07,
1106
+ "loss": 0.0002,
1107
+ "step": 3825
1108
+ },
1109
+ {
1110
+ "epoch": 22.13,
1111
+ "grad_norm": 0.011501124128699303,
1112
+ "learning_rate": 4.4e-07,
1113
+ "loss": 0.0002,
1114
+ "step": 3850
1115
+ },
1116
+ {
1117
+ "epoch": 22.27,
1118
+ "grad_norm": 0.01108501572161913,
1119
+ "learning_rate": 3.685714285714286e-07,
1120
+ "loss": 0.0002,
1121
+ "step": 3875
1122
+ },
1123
+ {
1124
+ "epoch": 22.41,
1125
+ "grad_norm": 0.0128969457000494,
1126
+ "learning_rate": 2.9714285714285715e-07,
1127
+ "loss": 0.0002,
1128
+ "step": 3900
1129
+ },
1130
+ {
1131
+ "epoch": 22.56,
1132
+ "grad_norm": 0.014468998648226261,
1133
+ "learning_rate": 2.2571428571428574e-07,
1134
+ "loss": 0.0002,
1135
+ "step": 3925
1136
+ },
1137
+ {
1138
+ "epoch": 22.7,
1139
+ "grad_norm": 0.013158345595002174,
1140
+ "learning_rate": 1.542857142857143e-07,
1141
+ "loss": 0.0002,
1142
+ "step": 3950
1143
+ },
1144
+ {
1145
+ "epoch": 22.84,
1146
+ "grad_norm": 0.010379764251410961,
1147
+ "learning_rate": 8.285714285714285e-08,
1148
+ "loss": 0.0002,
1149
+ "step": 3975
1150
+ },
1151
+ {
1152
+ "epoch": 22.99,
1153
+ "grad_norm": 0.008165850304067135,
1154
+ "learning_rate": 1.142857142857143e-08,
1155
+ "loss": 0.0002,
1156
+ "step": 4000
1157
+ },
1158
+ {
1159
+ "epoch": 22.99,
1160
+ "eval_loss": 0.7029648423194885,
1161
+ "eval_runtime": 485.2218,
1162
+ "eval_samples_per_second": 2.549,
1163
+ "eval_steps_per_second": 0.319,
1164
+ "eval_wer": 27.579778484482947,
1165
+ "step": 4000
1166
+ }
1167
+ ],
1168
+ "logging_steps": 25,
1169
+ "max_steps": 4000,
1170
+ "num_input_tokens_seen": 0,
1171
+ "num_train_epochs": 23,
1172
+ "save_steps": 1000,
1173
+ "total_flos": 1.839962795360256e+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 sha256:3d3010299f45b67cddc7e6d6f4bc607816f3110b6d0b4165041ce3a03b7b1d04
3
+ size 5112
runs/Apr09_18-47-55_DESKTOP-477V1NF/events.out.tfevents.1712659686.DESKTOP-477V1NF.13644.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf1ef070ebd2a453d4b17a6f1afec97d1535abe1c27af862c602e265bb3c6fd1
3
+ size 40826