itachi176 commited on
Commit
f695c0f
1 Parent(s): 3ec87e9

Upload 3 files

Browse files
generation_config.json ADDED
@@ -0,0 +1,223 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ null
13
+ ],
14
+ [
15
+ 2,
16
+ 50359
17
+ ],
18
+ [
19
+ 3,
20
+ 50363
21
+ ]
22
+ ],
23
+ "is_multilingual": true,
24
+ "lang_to_id": {
25
+ "<|af|>": 50327,
26
+ "<|am|>": 50334,
27
+ "<|ar|>": 50272,
28
+ "<|as|>": 50350,
29
+ "<|az|>": 50304,
30
+ "<|ba|>": 50355,
31
+ "<|be|>": 50330,
32
+ "<|bg|>": 50292,
33
+ "<|bn|>": 50302,
34
+ "<|bo|>": 50347,
35
+ "<|br|>": 50309,
36
+ "<|bs|>": 50315,
37
+ "<|ca|>": 50270,
38
+ "<|cs|>": 50283,
39
+ "<|cy|>": 50297,
40
+ "<|da|>": 50285,
41
+ "<|de|>": 50261,
42
+ "<|el|>": 50281,
43
+ "<|en|>": 50259,
44
+ "<|es|>": 50262,
45
+ "<|et|>": 50307,
46
+ "<|eu|>": 50310,
47
+ "<|fa|>": 50300,
48
+ "<|fi|>": 50277,
49
+ "<|fo|>": 50338,
50
+ "<|fr|>": 50265,
51
+ "<|gl|>": 50319,
52
+ "<|gu|>": 50333,
53
+ "<|haw|>": 50352,
54
+ "<|ha|>": 50354,
55
+ "<|he|>": 50279,
56
+ "<|hi|>": 50276,
57
+ "<|hr|>": 50291,
58
+ "<|ht|>": 50339,
59
+ "<|hu|>": 50286,
60
+ "<|hy|>": 50312,
61
+ "<|id|>": 50275,
62
+ "<|is|>": 50311,
63
+ "<|it|>": 50274,
64
+ "<|ja|>": 50266,
65
+ "<|jw|>": 50356,
66
+ "<|ka|>": 50329,
67
+ "<|kk|>": 50316,
68
+ "<|km|>": 50323,
69
+ "<|kn|>": 50306,
70
+ "<|ko|>": 50264,
71
+ "<|la|>": 50294,
72
+ "<|lb|>": 50345,
73
+ "<|ln|>": 50353,
74
+ "<|lo|>": 50336,
75
+ "<|lt|>": 50293,
76
+ "<|lv|>": 50301,
77
+ "<|mg|>": 50349,
78
+ "<|mi|>": 50295,
79
+ "<|mk|>": 50308,
80
+ "<|ml|>": 50296,
81
+ "<|mn|>": 50314,
82
+ "<|mr|>": 50320,
83
+ "<|ms|>": 50282,
84
+ "<|mt|>": 50343,
85
+ "<|my|>": 50346,
86
+ "<|ne|>": 50313,
87
+ "<|nl|>": 50271,
88
+ "<|nn|>": 50342,
89
+ "<|no|>": 50288,
90
+ "<|oc|>": 50328,
91
+ "<|pa|>": 50321,
92
+ "<|pl|>": 50269,
93
+ "<|ps|>": 50340,
94
+ "<|pt|>": 50267,
95
+ "<|ro|>": 50284,
96
+ "<|ru|>": 50263,
97
+ "<|sa|>": 50344,
98
+ "<|sd|>": 50332,
99
+ "<|si|>": 50322,
100
+ "<|sk|>": 50298,
101
+ "<|sl|>": 50305,
102
+ "<|sn|>": 50324,
103
+ "<|so|>": 50326,
104
+ "<|sq|>": 50317,
105
+ "<|sr|>": 50303,
106
+ "<|su|>": 50357,
107
+ "<|sv|>": 50273,
108
+ "<|sw|>": 50318,
109
+ "<|ta|>": 50287,
110
+ "<|te|>": 50299,
111
+ "<|tg|>": 50331,
112
+ "<|th|>": 50289,
113
+ "<|tk|>": 50341,
114
+ "<|tl|>": 50348,
115
+ "<|tr|>": 50268,
116
+ "<|tt|>": 50351,
117
+ "<|uk|>": 50280,
118
+ "<|ur|>": 50290,
119
+ "<|uz|>": 50337,
120
+ "<|vi|>": 50278,
121
+ "<|yi|>": 50335,
122
+ "<|yo|>": 50325,
123
+ "<|zh|>": 50260
124
+ },
125
+ "max_initial_timestamp_index": 1,
126
+ "max_length": 448,
127
+ "no_timestamps_token_id": 50363,
128
+ "pad_token_id": 50257,
129
+ "return_timestamps": false,
130
+ "suppress_tokens": [
131
+ 1,
132
+ 2,
133
+ 7,
134
+ 8,
135
+ 9,
136
+ 10,
137
+ 14,
138
+ 25,
139
+ 26,
140
+ 27,
141
+ 28,
142
+ 29,
143
+ 31,
144
+ 58,
145
+ 59,
146
+ 60,
147
+ 61,
148
+ 62,
149
+ 63,
150
+ 90,
151
+ 91,
152
+ 92,
153
+ 93,
154
+ 359,
155
+ 503,
156
+ 522,
157
+ 542,
158
+ 873,
159
+ 893,
160
+ 902,
161
+ 918,
162
+ 922,
163
+ 931,
164
+ 1350,
165
+ 1853,
166
+ 1982,
167
+ 2460,
168
+ 2627,
169
+ 3246,
170
+ 3253,
171
+ 3268,
172
+ 3536,
173
+ 3846,
174
+ 3961,
175
+ 4183,
176
+ 4667,
177
+ 6585,
178
+ 6647,
179
+ 7273,
180
+ 9061,
181
+ 9383,
182
+ 10428,
183
+ 10929,
184
+ 11938,
185
+ 12033,
186
+ 12331,
187
+ 12562,
188
+ 13793,
189
+ 14157,
190
+ 14635,
191
+ 15265,
192
+ 15618,
193
+ 16553,
194
+ 16604,
195
+ 18362,
196
+ 18956,
197
+ 20075,
198
+ 21675,
199
+ 22520,
200
+ 26130,
201
+ 26161,
202
+ 26435,
203
+ 28279,
204
+ 29464,
205
+ 31650,
206
+ 32302,
207
+ 32470,
208
+ 36865,
209
+ 42863,
210
+ 47425,
211
+ 49870,
212
+ 50254,
213
+ 50258,
214
+ 50360,
215
+ 50361,
216
+ 50362
217
+ ],
218
+ "task_to_id": {
219
+ "transcribe": 50359,
220
+ "translate": 50358
221
+ },
222
+ "transformers_version": "4.27.0.dev0"
223
+ }
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
+ }
trainer_state.json ADDED
@@ -0,0 +1,763 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 814.6945937054502,
3
+ "best_model_checkpoint": "./whisper-small-vi/checkpoint-2000",
4
+ "epoch": 4.322766570605188,
5
+ "global_step": 3000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.04,
12
+ "learning_rate": 4.0000000000000003e-07,
13
+ "loss": 5.7877,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.07,
18
+ "learning_rate": 9.000000000000001e-07,
19
+ "loss": 4.8451,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.11,
24
+ "learning_rate": 1.4000000000000001e-06,
25
+ "loss": 3.5507,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.14,
30
+ "learning_rate": 1.9000000000000002e-06,
31
+ "loss": 2.6175,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.18,
36
+ "learning_rate": 2.4000000000000003e-06,
37
+ "loss": 1.6457,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.22,
42
+ "learning_rate": 2.9e-06,
43
+ "loss": 1.3034,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.25,
48
+ "learning_rate": 3.4000000000000005e-06,
49
+ "loss": 1.1651,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.29,
54
+ "learning_rate": 3.900000000000001e-06,
55
+ "loss": 1.109,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.32,
60
+ "learning_rate": 4.38e-06,
61
+ "loss": 0.9525,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.36,
66
+ "learning_rate": 4.880000000000001e-06,
67
+ "loss": 0.7372,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.4,
72
+ "learning_rate": 5.380000000000001e-06,
73
+ "loss": 0.3308,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.43,
78
+ "learning_rate": 5.8800000000000005e-06,
79
+ "loss": 0.288,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.47,
84
+ "learning_rate": 6.380000000000001e-06,
85
+ "loss": 0.3499,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.5,
90
+ "learning_rate": 6.88e-06,
91
+ "loss": 0.3132,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.54,
96
+ "learning_rate": 7.3800000000000005e-06,
97
+ "loss": 0.2645,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.58,
102
+ "learning_rate": 7.88e-06,
103
+ "loss": 0.2469,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.61,
108
+ "learning_rate": 8.380000000000001e-06,
109
+ "loss": 0.2709,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.65,
114
+ "learning_rate": 8.880000000000001e-06,
115
+ "loss": 0.2689,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.68,
120
+ "learning_rate": 9.38e-06,
121
+ "loss": 0.2883,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.72,
126
+ "learning_rate": 9.88e-06,
127
+ "loss": 0.2295,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.76,
132
+ "learning_rate": 9.945714285714286e-06,
133
+ "loss": 0.2475,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.79,
138
+ "learning_rate": 9.874285714285715e-06,
139
+ "loss": 0.2683,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.83,
144
+ "learning_rate": 9.802857142857144e-06,
145
+ "loss": 0.274,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.86,
150
+ "learning_rate": 9.731428571428573e-06,
151
+ "loss": 0.2694,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.9,
156
+ "learning_rate": 9.66e-06,
157
+ "loss": 0.2621,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.94,
162
+ "learning_rate": 9.58857142857143e-06,
163
+ "loss": 0.2185,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.97,
168
+ "learning_rate": 9.517142857142859e-06,
169
+ "loss": 0.2433,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 1.01,
174
+ "learning_rate": 9.445714285714288e-06,
175
+ "loss": 0.2303,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 1.04,
180
+ "learning_rate": 9.374285714285715e-06,
181
+ "loss": 0.148,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 1.08,
186
+ "learning_rate": 9.302857142857144e-06,
187
+ "loss": 0.1834,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 1.12,
192
+ "learning_rate": 9.231428571428573e-06,
193
+ "loss": 0.1685,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 1.15,
198
+ "learning_rate": 9.16e-06,
199
+ "loss": 0.1677,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 1.19,
204
+ "learning_rate": 9.08857142857143e-06,
205
+ "loss": 0.1705,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 1.22,
210
+ "learning_rate": 9.017142857142858e-06,
211
+ "loss": 0.1427,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 1.26,
216
+ "learning_rate": 8.945714285714286e-06,
217
+ "loss": 0.1716,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 1.3,
222
+ "learning_rate": 8.874285714285715e-06,
223
+ "loss": 0.1328,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 1.33,
228
+ "learning_rate": 8.802857142857144e-06,
229
+ "loss": 0.1332,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 1.37,
234
+ "learning_rate": 8.731428571428571e-06,
235
+ "loss": 0.1705,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 1.4,
240
+ "learning_rate": 8.66e-06,
241
+ "loss": 0.1306,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 1.44,
246
+ "learning_rate": 8.588571428571429e-06,
247
+ "loss": 0.1627,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 1.44,
252
+ "eval_loss": 0.5460556149482727,
253
+ "eval_runtime": 1644.0015,
254
+ "eval_samples_per_second": 0.752,
255
+ "eval_steps_per_second": 0.094,
256
+ "eval_wer": 1004.5290053733962,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 1.48,
261
+ "learning_rate": 8.517142857142858e-06,
262
+ "loss": 0.1499,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 1.51,
267
+ "learning_rate": 8.445714285714285e-06,
268
+ "loss": 0.1401,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 1.55,
273
+ "learning_rate": 8.374285714285714e-06,
274
+ "loss": 0.1588,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 1.59,
279
+ "learning_rate": 8.302857142857143e-06,
280
+ "loss": 0.1326,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 1.62,
285
+ "learning_rate": 8.231428571428572e-06,
286
+ "loss": 0.1421,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 1.66,
291
+ "learning_rate": 8.16e-06,
292
+ "loss": 0.1479,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 1.69,
297
+ "learning_rate": 8.088571428571429e-06,
298
+ "loss": 0.1357,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 1.73,
303
+ "learning_rate": 8.017142857142858e-06,
304
+ "loss": 0.1458,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 1.77,
309
+ "learning_rate": 7.945714285714287e-06,
310
+ "loss": 0.1338,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 1.8,
315
+ "learning_rate": 7.874285714285716e-06,
316
+ "loss": 0.1448,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 1.84,
321
+ "learning_rate": 7.802857142857143e-06,
322
+ "loss": 0.1507,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 1.87,
327
+ "learning_rate": 7.731428571428572e-06,
328
+ "loss": 0.124,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 1.91,
333
+ "learning_rate": 7.660000000000001e-06,
334
+ "loss": 0.1538,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 1.95,
339
+ "learning_rate": 7.588571428571429e-06,
340
+ "loss": 0.1425,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 1.98,
345
+ "learning_rate": 7.5171428571428575e-06,
346
+ "loss": 0.1787,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 2.02,
351
+ "learning_rate": 7.445714285714286e-06,
352
+ "loss": 0.111,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 2.05,
357
+ "learning_rate": 7.374285714285715e-06,
358
+ "loss": 0.062,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 2.09,
363
+ "learning_rate": 7.302857142857144e-06,
364
+ "loss": 0.0726,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 2.13,
369
+ "learning_rate": 7.231428571428573e-06,
370
+ "loss": 0.0712,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 2.16,
375
+ "learning_rate": 7.16e-06,
376
+ "loss": 0.07,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 2.2,
381
+ "learning_rate": 7.088571428571429e-06,
382
+ "loss": 0.0758,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 2.23,
387
+ "learning_rate": 7.017142857142858e-06,
388
+ "loss": 0.0817,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 2.27,
393
+ "learning_rate": 6.945714285714287e-06,
394
+ "loss": 0.068,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 2.31,
399
+ "learning_rate": 6.874285714285714e-06,
400
+ "loss": 0.0598,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 2.34,
405
+ "learning_rate": 6.8028571428571434e-06,
406
+ "loss": 0.0582,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 2.38,
411
+ "learning_rate": 6.7314285714285724e-06,
412
+ "loss": 0.0921,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 2.41,
417
+ "learning_rate": 6.660000000000001e-06,
418
+ "loss": 0.0647,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 2.45,
423
+ "learning_rate": 6.588571428571429e-06,
424
+ "loss": 0.0769,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 2.49,
429
+ "learning_rate": 6.517142857142858e-06,
430
+ "loss": 0.0833,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 2.52,
435
+ "learning_rate": 6.445714285714286e-06,
436
+ "loss": 0.0806,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 2.56,
441
+ "learning_rate": 6.374285714285715e-06,
442
+ "loss": 0.073,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 2.59,
447
+ "learning_rate": 6.302857142857144e-06,
448
+ "loss": 0.0691,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 2.63,
453
+ "learning_rate": 6.231428571428571e-06,
454
+ "loss": 0.0553,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 2.67,
459
+ "learning_rate": 6.16e-06,
460
+ "loss": 0.0937,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 2.7,
465
+ "learning_rate": 6.088571428571429e-06,
466
+ "loss": 0.0746,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 2.74,
471
+ "learning_rate": 6.017142857142858e-06,
472
+ "loss": 0.0642,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 2.77,
477
+ "learning_rate": 5.945714285714286e-06,
478
+ "loss": 0.0809,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 2.81,
483
+ "learning_rate": 5.874285714285715e-06,
484
+ "loss": 0.062,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 2.85,
489
+ "learning_rate": 5.802857142857144e-06,
490
+ "loss": 0.0652,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 2.88,
495
+ "learning_rate": 5.731428571428572e-06,
496
+ "loss": 0.07,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 2.88,
501
+ "eval_loss": 0.5394195318222046,
502
+ "eval_runtime": 1644.6856,
503
+ "eval_samples_per_second": 0.752,
504
+ "eval_steps_per_second": 0.094,
505
+ "eval_wer": 814.6945937054502,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 2.92,
510
+ "learning_rate": 5.66e-06,
511
+ "loss": 0.0761,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 2.95,
516
+ "learning_rate": 5.588571428571429e-06,
517
+ "loss": 0.0641,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 2.99,
522
+ "learning_rate": 5.517142857142857e-06,
523
+ "loss": 0.0699,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 3.03,
528
+ "learning_rate": 5.445714285714286e-06,
529
+ "loss": 0.0506,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 3.06,
534
+ "learning_rate": 5.374285714285715e-06,
535
+ "loss": 0.0347,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 3.1,
540
+ "learning_rate": 5.3028571428571425e-06,
541
+ "loss": 0.0264,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 3.13,
546
+ "learning_rate": 5.2314285714285716e-06,
547
+ "loss": 0.0262,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 3.17,
552
+ "learning_rate": 5.1600000000000006e-06,
553
+ "loss": 0.0321,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 3.21,
558
+ "learning_rate": 5.08857142857143e-06,
559
+ "loss": 0.0363,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 3.24,
564
+ "learning_rate": 5.017142857142857e-06,
565
+ "loss": 0.0313,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 3.28,
570
+ "learning_rate": 4.945714285714286e-06,
571
+ "loss": 0.0365,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 3.31,
576
+ "learning_rate": 4.874285714285715e-06,
577
+ "loss": 0.0285,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 3.35,
582
+ "learning_rate": 4.802857142857143e-06,
583
+ "loss": 0.0355,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 3.39,
588
+ "learning_rate": 4.731428571428572e-06,
589
+ "loss": 0.0346,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 3.42,
594
+ "learning_rate": 4.66e-06,
595
+ "loss": 0.0268,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 3.46,
600
+ "learning_rate": 4.588571428571429e-06,
601
+ "loss": 0.0304,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 3.49,
606
+ "learning_rate": 4.5171428571428575e-06,
607
+ "loss": 0.0297,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 3.53,
612
+ "learning_rate": 4.445714285714286e-06,
613
+ "loss": 0.0359,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 3.57,
618
+ "learning_rate": 4.374285714285715e-06,
619
+ "loss": 0.0349,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 3.6,
624
+ "learning_rate": 4.302857142857143e-06,
625
+ "loss": 0.0278,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 3.64,
630
+ "learning_rate": 4.231428571428572e-06,
631
+ "loss": 0.0433,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 3.67,
636
+ "learning_rate": 4.16e-06,
637
+ "loss": 0.0322,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 3.71,
642
+ "learning_rate": 4.088571428571429e-06,
643
+ "loss": 0.0398,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 3.75,
648
+ "learning_rate": 4.017142857142857e-06,
649
+ "loss": 0.0316,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 3.78,
654
+ "learning_rate": 3.945714285714286e-06,
655
+ "loss": 0.0397,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 3.82,
660
+ "learning_rate": 3.874285714285715e-06,
661
+ "loss": 0.0318,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 3.85,
666
+ "learning_rate": 3.802857142857143e-06,
667
+ "loss": 0.0247,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 3.89,
672
+ "learning_rate": 3.731428571428572e-06,
673
+ "loss": 0.0285,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 3.93,
678
+ "learning_rate": 3.66e-06,
679
+ "loss": 0.0321,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 3.96,
684
+ "learning_rate": 3.588571428571429e-06,
685
+ "loss": 0.028,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 4.0,
690
+ "learning_rate": 3.5171428571428573e-06,
691
+ "loss": 0.0251,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 4.03,
696
+ "learning_rate": 3.4457142857142863e-06,
697
+ "loss": 0.0125,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 4.07,
702
+ "learning_rate": 3.3742857142857145e-06,
703
+ "loss": 0.0152,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 4.11,
708
+ "learning_rate": 3.302857142857143e-06,
709
+ "loss": 0.0131,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 4.14,
714
+ "learning_rate": 3.2314285714285716e-06,
715
+ "loss": 0.0117,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 4.18,
720
+ "learning_rate": 3.1600000000000002e-06,
721
+ "loss": 0.0156,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 4.21,
726
+ "learning_rate": 3.0885714285714284e-06,
727
+ "loss": 0.009,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 4.25,
732
+ "learning_rate": 3.0171428571428574e-06,
733
+ "loss": 0.0101,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 4.29,
738
+ "learning_rate": 2.945714285714286e-06,
739
+ "loss": 0.0143,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 4.32,
744
+ "learning_rate": 2.8742857142857146e-06,
745
+ "loss": 0.0151,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 4.32,
750
+ "eval_loss": 0.5925257205963135,
751
+ "eval_runtime": 1647.3609,
752
+ "eval_samples_per_second": 0.751,
753
+ "eval_steps_per_second": 0.094,
754
+ "eval_wer": 1116.8110538436233,
755
+ "step": 3000
756
+ }
757
+ ],
758
+ "max_steps": 4000,
759
+ "num_train_epochs": 6,
760
+ "total_flos": 3.45956177903616e+18,
761
+ "trial_name": null,
762
+ "trial_params": null
763
+ }