marinone94 commited on
Commit
8ac322f
1 Parent(s): 166c885

End of training

Browse files
all_results.json CHANGED
@@ -1,32 +1,33 @@
1
  {
2
- "epoch": 2.32,
3
  "eval_loss": 1.6191972494125366,
4
- "eval_pretrained_loss": 1.710707187652588,
5
- "eval_pretrained_runtime": 125.9444,
6
- "eval_pretrained_samples_per_second": 2.62,
7
- "eval_pretrained_steps_per_second": 0.087,
8
- "eval_pretrained_wer": 258.2308797700783,
9
  "eval_runtime": 56.3363,
10
  "eval_samples_per_second": 0.071,
11
  "eval_steps_per_second": 0.036,
12
  "eval_wer": 153.2258064516129,
13
- "test_finetuned_loss": 1.0840972661972046,
14
- "test_finetuned_runtime": 169.6782,
15
- "test_finetuned_samples_per_second": 4.473,
16
- "test_finetuned_steps_per_second": 0.141,
17
- "test_finetuned_wer": 172.4357697669787,
18
  "test_loss": 1.7568330764770508,
19
- "test_pretrained_loss": 1.723103642463684,
20
- "test_pretrained_runtime": 196.5873,
21
- "test_pretrained_samples_per_second": 3.861,
22
- "test_pretrained_steps_per_second": 0.122,
23
- "test_pretrained_wer": 261.9265750514506,
24
  "test_runtime": 37.8582,
25
  "test_samples_per_second": 0.106,
26
  "test_steps_per_second": 0.053,
27
  "test_wer": 138.5964912280702,
28
- "train_loss": 1.0522634956453527,
29
- "train_runtime": 1735.2585,
30
- "train_samples_per_second": 4.131,
31
- "train_steps_per_second": 0.065
 
32
  }
 
1
  {
2
+ "epoch": 1.0,
3
  "eval_loss": 1.6191972494125366,
4
+ "eval_pretrained_loss": 1.8532216548919678,
5
+ "eval_pretrained_runtime": 33.9735,
6
+ "eval_pretrained_samples_per_second": 0.118,
7
+ "eval_pretrained_steps_per_second": 0.059,
8
+ "eval_pretrained_wer": 123.40425531914893,
9
  "eval_runtime": 56.3363,
10
  "eval_samples_per_second": 0.071,
11
  "eval_steps_per_second": 0.036,
12
  "eval_wer": 153.2258064516129,
13
+ "test_finetuned_loss": 1.7792198657989502,
14
+ "test_finetuned_runtime": 13.3999,
15
+ "test_finetuned_samples_per_second": 0.299,
16
+ "test_finetuned_steps_per_second": 0.149,
17
+ "test_finetuned_wer": 127.90697674418605,
18
  "test_loss": 1.7568330764770508,
19
+ "test_pretrained_loss": 1.7792198657989502,
20
+ "test_pretrained_runtime": 33.5108,
21
+ "test_pretrained_samples_per_second": 0.119,
22
+ "test_pretrained_steps_per_second": 0.06,
23
+ "test_pretrained_wer": 127.90697674418605,
24
  "test_runtime": 37.8582,
25
  "test_samples_per_second": 0.106,
26
  "test_steps_per_second": 0.053,
27
  "test_wer": 138.5964912280702,
28
+ "total_flos": 196951080960000.0,
29
+ "train_loss": 1.7391430735588074,
30
+ "train_runtime": 61.5449,
31
+ "train_samples_per_second": 0.13,
32
+ "train_steps_per_second": 0.032
33
  }
config.json CHANGED
@@ -2,6 +2,7 @@
2
  "_name_or_path": "openai/whisper-tiny",
3
  "activation_dropout": 0.0,
4
  "activation_function": "gelu",
 
5
  "architectures": [
6
  "WhisperForConditionalGeneration"
7
  ],
@@ -11,6 +12,7 @@
11
  50257
12
  ],
13
  "bos_token_id": 50257,
 
14
  "d_model": 384,
15
  "decoder_attention_heads": 6,
16
  "decoder_ffn_dim": 1536,
@@ -39,6 +41,12 @@
39
  ],
40
  "init_std": 0.02,
41
  "is_encoder_decoder": true,
 
 
 
 
 
 
42
  "max_length": 448,
43
  "max_source_positions": 1500,
44
  "max_target_positions": 448,
@@ -136,7 +144,8 @@
136
  50362
137
  ],
138
  "torch_dtype": "float32",
139
- "transformers_version": "4.26.0.dev0",
140
  "use_cache": true,
 
141
  "vocab_size": 51865
142
  }
 
2
  "_name_or_path": "openai/whisper-tiny",
3
  "activation_dropout": 0.0,
4
  "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
  "architectures": [
7
  "WhisperForConditionalGeneration"
8
  ],
 
12
  50257
13
  ],
14
  "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
  "d_model": 384,
17
  "decoder_attention_heads": 6,
18
  "decoder_ffn_dim": 1536,
 
41
  ],
42
  "init_std": 0.02,
43
  "is_encoder_decoder": true,
44
+ "mask_feature_length": 10,
45
+ "mask_feature_min_masks": 0,
46
+ "mask_feature_prob": 0.0,
47
+ "mask_time_length": 10,
48
+ "mask_time_min_masks": 2,
49
+ "mask_time_prob": 0.05,
50
  "max_length": 448,
51
  "max_source_positions": 1500,
52
  "max_target_positions": 448,
 
144
  50362
145
  ],
146
  "torch_dtype": "float32",
147
+ "transformers_version": "4.27.3",
148
  "use_cache": true,
149
+ "use_weighted_layer_sum": false,
150
  "vocab_size": 51865
151
  }
eval_pretrained_results.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "eval_pretrained_loss": 1.710707187652588,
3
- "eval_pretrained_runtime": 125.9444,
4
- "eval_pretrained_samples_per_second": 2.62,
5
- "eval_pretrained_steps_per_second": 0.087,
6
- "eval_pretrained_wer": 258.2308797700783
7
  }
 
1
  {
2
+ "eval_pretrained_loss": 1.8532216548919678,
3
+ "eval_pretrained_runtime": 33.9735,
4
+ "eval_pretrained_samples_per_second": 0.118,
5
+ "eval_pretrained_steps_per_second": 0.059,
6
+ "eval_pretrained_wer": 123.40425531914893
7
  }
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
+ 50259
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": "4.27.3"
223
+ }
preprocessor_config.json CHANGED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2027faf2910d36a40b249965d707ab19532ef0effc8d243d184de4439275106e
3
  size 151098921
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3843686519777a4550909e8bd4961dcf7425e7183295f03d09a433a271f0887
3
  size 151098921
test_finetuned_results.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
- "epoch": 2.32,
3
- "test_finetuned_loss": 1.0840972661972046,
4
- "test_finetuned_runtime": 169.6782,
5
- "test_finetuned_samples_per_second": 4.473,
6
- "test_finetuned_steps_per_second": 0.141,
7
- "test_finetuned_wer": 172.4357697669787
8
  }
 
1
  {
2
+ "epoch": 1.0,
3
+ "test_finetuned_loss": 1.7792198657989502,
4
+ "test_finetuned_runtime": 13.3999,
5
+ "test_finetuned_samples_per_second": 0.299,
6
+ "test_finetuned_steps_per_second": 0.149,
7
+ "test_finetuned_wer": 127.90697674418605
8
  }
test_pretrained_results.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "test_pretrained_loss": 1.723103642463684,
3
- "test_pretrained_runtime": 196.5873,
4
- "test_pretrained_samples_per_second": 3.861,
5
- "test_pretrained_steps_per_second": 0.122,
6
- "test_pretrained_wer": 261.9265750514506
7
  }
 
1
  {
2
+ "test_pretrained_loss": 1.7792198657989502,
3
+ "test_pretrained_runtime": 33.5108,
4
+ "test_pretrained_samples_per_second": 0.119,
5
+ "test_pretrained_steps_per_second": 0.06,
6
+ "test_pretrained_wer": 127.90697674418605
7
  }
train_results.json CHANGED
@@ -1,7 +1,8 @@
1
  {
2
- "epoch": 2.32,
3
- "train_loss": 1.0522634956453527,
4
- "train_runtime": 1735.2585,
5
- "train_samples_per_second": 4.131,
6
- "train_steps_per_second": 0.065
 
7
  }
 
1
  {
2
+ "epoch": 1.0,
3
+ "total_flos": 196951080960000.0,
4
+ "train_loss": 1.7391430735588074,
5
+ "train_runtime": 61.5449,
6
+ "train_samples_per_second": 0.13,
7
+ "train_steps_per_second": 0.032
8
  }
trainer_state.json CHANGED
@@ -1,787 +1,55 @@
1
  {
2
- "best_metric": 161.45617116397892,
3
- "best_model_checkpoint": "./checkpoint-88",
4
- "epoch": 2.3214285714285716,
5
- "global_step": 112,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
- "epoch": 0.01,
12
  "learning_rate": 0.0,
13
- "loss": 1.6569,
14
  "step": 1
15
  },
16
  {
17
- "epoch": 0.02,
18
- "learning_rate": 2.2058823529411765e-07,
19
- "loss": 1.6564,
20
- "step": 2
21
- },
22
- {
23
- "epoch": 0.03,
24
- "learning_rate": 4.411764705882353e-07,
25
- "loss": 1.6518,
26
- "step": 3
27
- },
28
- {
29
- "epoch": 0.04,
30
- "learning_rate": 6.61764705882353e-07,
31
- "loss": 1.6237,
32
- "step": 4
33
- },
34
- {
35
- "epoch": 0.04,
36
- "learning_rate": 8.823529411764706e-07,
37
- "loss": 1.7141,
38
- "step": 5
39
- },
40
- {
41
- "epoch": 0.05,
42
- "learning_rate": 1.1029411764705884e-06,
43
- "loss": 1.655,
44
- "step": 6
45
- },
46
- {
47
- "epoch": 0.06,
48
- "learning_rate": 1.323529411764706e-06,
49
- "loss": 1.5904,
50
- "step": 7
51
- },
52
- {
53
- "epoch": 0.07,
54
- "learning_rate": 1.5441176470588234e-06,
55
- "loss": 1.6232,
56
- "step": 8
57
- },
58
- {
59
- "epoch": 0.08,
60
- "learning_rate": 1.7647058823529412e-06,
61
- "loss": 1.6403,
62
- "step": 9
63
- },
64
- {
65
- "epoch": 0.09,
66
- "learning_rate": 1.985294117647059e-06,
67
- "loss": 1.6065,
68
- "step": 10
69
- },
70
- {
71
- "epoch": 0.1,
72
- "learning_rate": 2.2058823529411767e-06,
73
- "loss": 1.5947,
74
- "step": 11
75
- },
76
- {
77
- "epoch": 0.1,
78
- "eval_loss": 1.629066824913025,
79
- "eval_runtime": 125.637,
80
- "eval_samples_per_second": 2.627,
81
- "eval_steps_per_second": 0.088,
82
- "eval_wer": 227.94188088775348,
83
- "step": 11
84
- },
85
- {
86
- "epoch": 0.11,
87
- "learning_rate": 2.4264705882352943e-06,
88
- "loss": 1.52,
89
- "step": 12
90
- },
91
- {
92
- "epoch": 0.12,
93
- "learning_rate": 2.647058823529412e-06,
94
- "loss": 1.4456,
95
- "step": 13
96
- },
97
- {
98
- "epoch": 0.12,
99
- "learning_rate": 2.867647058823529e-06,
100
- "loss": 1.58,
101
- "step": 14
102
- },
103
- {
104
- "epoch": 0.13,
105
- "learning_rate": 3.088235294117647e-06,
106
- "loss": 1.5958,
107
- "step": 15
108
- },
109
- {
110
- "epoch": 0.14,
111
- "learning_rate": 3.308823529411765e-06,
112
- "loss": 1.4695,
113
- "step": 16
114
- },
115
- {
116
- "epoch": 0.15,
117
- "learning_rate": 3.5294117647058825e-06,
118
- "loss": 1.4718,
119
- "step": 17
120
- },
121
- {
122
- "epoch": 0.16,
123
- "learning_rate": 3.75e-06,
124
- "loss": 1.4406,
125
- "step": 18
126
- },
127
- {
128
- "epoch": 0.17,
129
- "learning_rate": 3.970588235294118e-06,
130
- "loss": 1.4339,
131
- "step": 19
132
- },
133
- {
134
- "epoch": 0.18,
135
- "learning_rate": 4.191176470588236e-06,
136
- "loss": 1.313,
137
- "step": 20
138
- },
139
- {
140
- "epoch": 0.19,
141
- "learning_rate": 4.411764705882353e-06,
142
- "loss": 1.2825,
143
- "step": 21
144
- },
145
- {
146
- "epoch": 0.2,
147
- "learning_rate": 4.632352941176471e-06,
148
- "loss": 1.3194,
149
- "step": 22
150
- },
151
- {
152
- "epoch": 0.2,
153
- "eval_loss": 1.460469126701355,
154
- "eval_runtime": 118.4642,
155
- "eval_samples_per_second": 2.786,
156
- "eval_steps_per_second": 0.093,
157
- "eval_wer": 235.94124221619032,
158
- "step": 22
159
- },
160
- {
161
- "epoch": 0.21,
162
- "learning_rate": 4.852941176470589e-06,
163
- "loss": 1.2301,
164
- "step": 23
165
- },
166
- {
167
- "epoch": 0.21,
168
- "learning_rate": 5.073529411764706e-06,
169
- "loss": 1.3231,
170
- "step": 24
171
- },
172
- {
173
- "epoch": 0.22,
174
- "learning_rate": 5.294117647058824e-06,
175
- "loss": 1.2689,
176
- "step": 25
177
- },
178
- {
179
- "epoch": 0.23,
180
- "learning_rate": 5.5147058823529415e-06,
181
- "loss": 1.2768,
182
- "step": 26
183
- },
184
- {
185
- "epoch": 0.24,
186
- "learning_rate": 5.735294117647058e-06,
187
- "loss": 1.2927,
188
- "step": 27
189
- },
190
- {
191
- "epoch": 0.25,
192
- "learning_rate": 5.955882352941176e-06,
193
- "loss": 1.3419,
194
- "step": 28
195
- },
196
- {
197
- "epoch": 0.26,
198
- "learning_rate": 6.176470588235294e-06,
199
- "loss": 1.141,
200
- "step": 29
201
- },
202
- {
203
- "epoch": 0.27,
204
- "learning_rate": 6.397058823529412e-06,
205
- "loss": 1.2907,
206
- "step": 30
207
- },
208
- {
209
- "epoch": 0.28,
210
- "learning_rate": 6.61764705882353e-06,
211
- "loss": 1.2248,
212
- "step": 31
213
- },
214
- {
215
- "epoch": 0.29,
216
- "learning_rate": 6.838235294117647e-06,
217
- "loss": 1.149,
218
- "step": 32
219
- },
220
- {
221
- "epoch": 0.29,
222
- "learning_rate": 7.058823529411765e-06,
223
- "loss": 1.2636,
224
- "step": 33
225
- },
226
- {
227
- "epoch": 0.29,
228
- "eval_loss": 1.3149287700653076,
229
- "eval_runtime": 115.745,
230
- "eval_samples_per_second": 2.851,
231
- "eval_steps_per_second": 0.095,
232
- "eval_wer": 177.19942519559316,
233
- "step": 33
234
- },
235
- {
236
- "epoch": 0.3,
237
- "learning_rate": 7.2794117647058826e-06,
238
- "loss": 1.2351,
239
- "step": 34
240
  },
241
  {
242
- "epoch": 0.31,
243
  "learning_rate": 7.5e-06,
244
- "loss": 1.2055,
245
- "step": 35
246
- },
247
- {
248
- "epoch": 0.32,
249
- "learning_rate": 7.403846153846155e-06,
250
- "loss": 1.1864,
251
- "step": 36
252
- },
253
- {
254
- "epoch": 0.33,
255
- "learning_rate": 7.307692307692308e-06,
256
- "loss": 1.148,
257
- "step": 37
258
- },
259
- {
260
- "epoch": 0.34,
261
- "learning_rate": 7.211538461538462e-06,
262
- "loss": 1.1505,
263
- "step": 38
264
- },
265
- {
266
- "epoch": 1.01,
267
- "learning_rate": 7.115384615384615e-06,
268
- "loss": 1.0859,
269
- "step": 39
270
- },
271
- {
272
- "epoch": 1.02,
273
- "learning_rate": 7.01923076923077e-06,
274
- "loss": 1.119,
275
- "step": 40
276
- },
277
- {
278
- "epoch": 1.03,
279
- "learning_rate": 6.923076923076923e-06,
280
- "loss": 1.0238,
281
- "step": 41
282
- },
283
- {
284
- "epoch": 1.04,
285
- "learning_rate": 6.826923076923077e-06,
286
- "loss": 1.0468,
287
- "step": 42
288
- },
289
- {
290
- "epoch": 1.04,
291
- "learning_rate": 6.730769230769231e-06,
292
- "loss": 1.061,
293
- "step": 43
294
- },
295
- {
296
- "epoch": 1.05,
297
- "learning_rate": 6.6346153846153846e-06,
298
- "loss": 1.0238,
299
- "step": 44
300
- },
301
- {
302
- "epoch": 1.05,
303
- "eval_loss": 1.1836069822311401,
304
- "eval_runtime": 109.1507,
305
- "eval_samples_per_second": 3.023,
306
- "eval_steps_per_second": 0.101,
307
- "eval_wer": 199.80839853105542,
308
- "step": 44
309
- },
310
- {
311
- "epoch": 1.06,
312
- "learning_rate": 6.538461538461539e-06,
313
- "loss": 1.0016,
314
- "step": 45
315
- },
316
- {
317
- "epoch": 1.07,
318
- "learning_rate": 6.442307692307692e-06,
319
- "loss": 0.9635,
320
- "step": 46
321
- },
322
- {
323
- "epoch": 1.08,
324
- "learning_rate": 6.3461538461538466e-06,
325
- "loss": 0.9953,
326
- "step": 47
327
- },
328
- {
329
- "epoch": 1.09,
330
- "learning_rate": 6.25e-06,
331
- "loss": 1.0061,
332
- "step": 48
333
- },
334
- {
335
- "epoch": 1.1,
336
- "learning_rate": 6.153846153846154e-06,
337
- "loss": 0.9933,
338
- "step": 49
339
- },
340
- {
341
- "epoch": 1.11,
342
- "learning_rate": 6.057692307692308e-06,
343
- "loss": 0.9479,
344
- "step": 50
345
- },
346
- {
347
- "epoch": 1.12,
348
- "learning_rate": 5.9615384615384615e-06,
349
- "loss": 0.9132,
350
- "step": 51
351
- },
352
- {
353
- "epoch": 1.12,
354
- "learning_rate": 5.865384615384615e-06,
355
- "loss": 1.0232,
356
- "step": 52
357
- },
358
- {
359
- "epoch": 1.13,
360
- "learning_rate": 5.76923076923077e-06,
361
- "loss": 1.0153,
362
- "step": 53
363
- },
364
- {
365
- "epoch": 1.14,
366
- "learning_rate": 5.673076923076923e-06,
367
- "loss": 0.9548,
368
- "step": 54
369
- },
370
- {
371
- "epoch": 1.15,
372
- "learning_rate": 5.576923076923077e-06,
373
- "loss": 0.9306,
374
- "step": 55
375
- },
376
- {
377
- "epoch": 1.15,
378
- "eval_loss": 1.1363615989685059,
379
- "eval_runtime": 115.8861,
380
- "eval_samples_per_second": 2.848,
381
- "eval_steps_per_second": 0.095,
382
- "eval_wer": 210.21874501037843,
383
- "step": 55
384
- },
385
- {
386
- "epoch": 1.16,
387
- "learning_rate": 5.480769230769231e-06,
388
- "loss": 0.9088,
389
- "step": 56
390
- },
391
- {
392
- "epoch": 1.17,
393
- "learning_rate": 5.384615384615385e-06,
394
- "loss": 0.8902,
395
- "step": 57
396
- },
397
- {
398
- "epoch": 1.18,
399
- "learning_rate": 5.288461538461539e-06,
400
- "loss": 0.861,
401
- "step": 58
402
- },
403
- {
404
- "epoch": 1.19,
405
- "learning_rate": 5.192307692307692e-06,
406
- "loss": 0.8047,
407
- "step": 59
408
- },
409
- {
410
- "epoch": 1.2,
411
- "learning_rate": 5.096153846153847e-06,
412
- "loss": 0.8459,
413
- "step": 60
414
- },
415
- {
416
- "epoch": 1.21,
417
- "learning_rate": 4.9999999999999996e-06,
418
- "loss": 0.8103,
419
- "step": 61
420
- },
421
- {
422
- "epoch": 1.21,
423
- "learning_rate": 4.903846153846154e-06,
424
- "loss": 0.8752,
425
- "step": 62
426
- },
427
- {
428
- "epoch": 1.22,
429
- "learning_rate": 4.807692307692308e-06,
430
- "loss": 0.8398,
431
- "step": 63
432
- },
433
- {
434
- "epoch": 1.23,
435
- "learning_rate": 4.711538461538462e-06,
436
- "loss": 0.8571,
437
- "step": 64
438
- },
439
- {
440
- "epoch": 1.24,
441
- "learning_rate": 4.615384615384616e-06,
442
- "loss": 0.8998,
443
- "step": 65
444
- },
445
- {
446
- "epoch": 1.25,
447
- "learning_rate": 4.519230769230769e-06,
448
- "loss": 0.9233,
449
- "step": 66
450
- },
451
- {
452
- "epoch": 1.25,
453
- "eval_loss": 1.1090655326843262,
454
- "eval_runtime": 104.9772,
455
- "eval_samples_per_second": 3.144,
456
- "eval_steps_per_second": 0.105,
457
- "eval_wer": 175.20357656075362,
458
- "step": 66
459
- },
460
- {
461
- "epoch": 1.26,
462
- "learning_rate": 4.423076923076924e-06,
463
- "loss": 0.7878,
464
- "step": 67
465
- },
466
- {
467
- "epoch": 1.27,
468
- "learning_rate": 4.3269230769230765e-06,
469
- "loss": 0.8948,
470
- "step": 68
471
- },
472
- {
473
- "epoch": 1.28,
474
- "learning_rate": 4.230769230769231e-06,
475
- "loss": 0.8711,
476
- "step": 69
477
- },
478
- {
479
- "epoch": 1.29,
480
- "learning_rate": 4.134615384615385e-06,
481
- "loss": 0.8292,
482
- "step": 70
483
- },
484
- {
485
- "epoch": 1.29,
486
- "learning_rate": 4.0384615384615385e-06,
487
- "loss": 0.9318,
488
- "step": 71
489
- },
490
- {
491
- "epoch": 1.3,
492
- "learning_rate": 3.942307692307692e-06,
493
- "loss": 0.9272,
494
- "step": 72
495
- },
496
- {
497
- "epoch": 1.31,
498
- "learning_rate": 3.846153846153846e-06,
499
- "loss": 0.8923,
500
- "step": 73
501
- },
502
- {
503
- "epoch": 1.32,
504
- "learning_rate": 3.75e-06,
505
- "loss": 0.9013,
506
- "step": 74
507
- },
508
- {
509
- "epoch": 1.33,
510
- "learning_rate": 3.653846153846154e-06,
511
- "loss": 0.8659,
512
- "step": 75
513
- },
514
- {
515
- "epoch": 1.34,
516
- "learning_rate": 3.5576923076923075e-06,
517
- "loss": 0.8422,
518
- "step": 76
519
- },
520
- {
521
- "epoch": 2.01,
522
- "learning_rate": 3.4615384615384617e-06,
523
- "loss": 0.8482,
524
- "step": 77
525
- },
526
- {
527
- "epoch": 2.01,
528
- "eval_loss": 1.0900639295578003,
529
- "eval_runtime": 99.705,
530
- "eval_samples_per_second": 3.31,
531
- "eval_steps_per_second": 0.11,
532
- "eval_wer": 161.85534089094685,
533
- "step": 77
534
- },
535
- {
536
- "epoch": 2.02,
537
- "learning_rate": 3.3653846153846154e-06,
538
- "loss": 0.8925,
539
- "step": 78
540
- },
541
- {
542
- "epoch": 2.03,
543
- "learning_rate": 3.2692307692307696e-06,
544
- "loss": 0.8202,
545
- "step": 79
546
- },
547
- {
548
- "epoch": 2.04,
549
- "learning_rate": 3.1730769230769233e-06,
550
- "loss": 0.8384,
551
- "step": 80
552
- },
553
- {
554
- "epoch": 2.04,
555
- "learning_rate": 3.076923076923077e-06,
556
- "loss": 0.8469,
557
- "step": 81
558
- },
559
- {
560
- "epoch": 2.05,
561
- "learning_rate": 2.9807692307692307e-06,
562
- "loss": 0.8172,
563
- "step": 82
564
- },
565
- {
566
- "epoch": 2.06,
567
- "learning_rate": 2.884615384615385e-06,
568
- "loss": 0.8119,
569
- "step": 83
570
- },
571
- {
572
- "epoch": 2.07,
573
- "learning_rate": 2.7884615384615386e-06,
574
- "loss": 0.7811,
575
- "step": 84
576
- },
577
- {
578
- "epoch": 2.08,
579
- "learning_rate": 2.6923076923076923e-06,
580
- "loss": 0.819,
581
- "step": 85
582
- },
583
- {
584
- "epoch": 2.09,
585
- "learning_rate": 2.596153846153846e-06,
586
- "loss": 0.8324,
587
- "step": 86
588
- },
589
- {
590
- "epoch": 2.1,
591
- "learning_rate": 2.4999999999999998e-06,
592
- "loss": 0.8336,
593
- "step": 87
594
- },
595
- {
596
- "epoch": 2.11,
597
- "learning_rate": 2.403846153846154e-06,
598
- "loss": 0.7929,
599
- "step": 88
600
- },
601
- {
602
- "epoch": 2.11,
603
- "eval_loss": 1.0782241821289062,
604
- "eval_runtime": 111.1236,
605
- "eval_samples_per_second": 2.97,
606
- "eval_steps_per_second": 0.099,
607
- "eval_wer": 161.45617116397892,
608
- "step": 88
609
- },
610
- {
611
- "epoch": 2.12,
612
- "learning_rate": 2.307692307692308e-06,
613
- "loss": 0.774,
614
- "step": 89
615
- },
616
- {
617
- "epoch": 2.12,
618
- "learning_rate": 2.211538461538462e-06,
619
- "loss": 0.8693,
620
- "step": 90
621
- },
622
- {
623
- "epoch": 2.13,
624
- "learning_rate": 2.1153846153846155e-06,
625
- "loss": 0.8628,
626
- "step": 91
627
- },
628
- {
629
- "epoch": 2.14,
630
- "learning_rate": 2.0192307692307692e-06,
631
- "loss": 0.8199,
632
- "step": 92
633
- },
634
- {
635
- "epoch": 2.15,
636
- "learning_rate": 1.923076923076923e-06,
637
- "loss": 0.7995,
638
- "step": 93
639
- },
640
- {
641
- "epoch": 2.16,
642
- "learning_rate": 1.826923076923077e-06,
643
- "loss": 0.7824,
644
- "step": 94
645
- },
646
- {
647
- "epoch": 2.17,
648
- "learning_rate": 1.7307692307692308e-06,
649
- "loss": 0.7694,
650
- "step": 95
651
- },
652
- {
653
- "epoch": 2.18,
654
- "learning_rate": 1.6346153846153848e-06,
655
- "loss": 0.7526,
656
- "step": 96
657
- },
658
- {
659
- "epoch": 2.19,
660
- "learning_rate": 1.5384615384615385e-06,
661
- "loss": 0.7037,
662
- "step": 97
663
- },
664
- {
665
- "epoch": 2.2,
666
- "learning_rate": 1.4423076923076924e-06,
667
- "loss": 0.7366,
668
- "step": 98
669
- },
670
- {
671
- "epoch": 2.21,
672
- "learning_rate": 1.3461538461538462e-06,
673
- "loss": 0.7134,
674
- "step": 99
675
- },
676
- {
677
- "epoch": 2.21,
678
- "eval_loss": 1.0737253427505493,
679
- "eval_runtime": 96.9018,
680
- "eval_samples_per_second": 3.406,
681
- "eval_steps_per_second": 0.114,
682
- "eval_wer": 177.1195912501996,
683
- "step": 99
684
- },
685
- {
686
- "epoch": 2.21,
687
- "learning_rate": 1.2499999999999999e-06,
688
- "loss": 0.7738,
689
- "step": 100
690
- },
691
- {
692
- "epoch": 2.22,
693
- "learning_rate": 1.153846153846154e-06,
694
- "loss": 0.7373,
695
- "step": 101
696
- },
697
- {
698
- "epoch": 2.23,
699
- "learning_rate": 1.0576923076923078e-06,
700
- "loss": 0.7623,
701
- "step": 102
702
- },
703
- {
704
- "epoch": 2.24,
705
- "learning_rate": 9.615384615384615e-07,
706
- "loss": 0.815,
707
- "step": 103
708
- },
709
- {
710
- "epoch": 2.25,
711
- "learning_rate": 8.653846153846154e-07,
712
- "loss": 0.829,
713
- "step": 104
714
- },
715
- {
716
- "epoch": 2.26,
717
- "learning_rate": 7.692307692307693e-07,
718
- "loss": 0.7113,
719
- "step": 105
720
- },
721
- {
722
- "epoch": 2.27,
723
- "learning_rate": 6.730769230769231e-07,
724
- "loss": 0.8106,
725
- "step": 106
726
- },
727
- {
728
- "epoch": 2.28,
729
- "learning_rate": 5.76923076923077e-07,
730
- "loss": 0.7909,
731
- "step": 107
732
- },
733
- {
734
- "epoch": 2.29,
735
- "learning_rate": 4.807692307692307e-07,
736
- "loss": 0.7564,
737
- "step": 108
738
- },
739
- {
740
- "epoch": 2.29,
741
- "learning_rate": 3.8461538461538463e-07,
742
- "loss": 0.8574,
743
- "step": 109
744
- },
745
- {
746
- "epoch": 2.3,
747
- "learning_rate": 2.884615384615385e-07,
748
- "loss": 0.8543,
749
- "step": 110
750
- },
751
- {
752
- "epoch": 2.3,
753
- "eval_loss": 1.0717276334762573,
754
- "eval_runtime": 95.923,
755
- "eval_samples_per_second": 3.44,
756
- "eval_steps_per_second": 0.115,
757
- "eval_wer": 187.5459045186013,
758
- "step": 110
759
- },
760
- {
761
- "epoch": 2.31,
762
- "learning_rate": 1.9230769230769231e-07,
763
- "loss": 0.824,
764
- "step": 111
765
  },
766
  {
767
- "epoch": 2.32,
768
- "learning_rate": 9.615384615384616e-08,
769
- "loss": 0.8352,
770
- "step": 112
 
 
 
771
  },
772
  {
773
- "epoch": 2.32,
774
- "step": 112,
775
- "total_flos": 1.7415399333888e+17,
776
- "train_loss": 1.0522634956453527,
777
- "train_runtime": 1735.2585,
778
- "train_samples_per_second": 4.131,
779
- "train_steps_per_second": 0.065
780
  }
781
  ],
782
- "max_steps": 112,
783
  "num_train_epochs": 9223372036854775807,
784
- "total_flos": 1.7415399333888e+17,
785
  "trial_name": null,
786
  "trial_params": null
787
  }
 
1
  {
2
+ "best_metric": 123.40425531914893,
3
+ "best_model_checkpoint": "./whisper-training-blog/checkpoint-1",
4
+ "epoch": 1.0,
5
+ "global_step": 2,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
+ "epoch": 0.5,
12
  "learning_rate": 0.0,
13
+ "loss": 1.8019,
14
  "step": 1
15
  },
16
  {
17
+ "epoch": 0.5,
18
+ "eval_loss": 1.8532216548919678,
19
+ "eval_runtime": 6.8084,
20
+ "eval_samples_per_second": 0.588,
21
+ "eval_steps_per_second": 0.294,
22
+ "eval_wer": 123.40425531914893,
23
+ "step": 1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24
  },
25
  {
26
+ "epoch": 1.0,
27
  "learning_rate": 7.5e-06,
28
+ "loss": 1.6763,
29
+ "step": 2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30
  },
31
  {
32
+ "epoch": 1.0,
33
+ "eval_loss": 1.8532216548919678,
34
+ "eval_runtime": 5.8615,
35
+ "eval_samples_per_second": 0.682,
36
+ "eval_steps_per_second": 0.341,
37
+ "eval_wer": 123.40425531914893,
38
+ "step": 2
39
  },
40
  {
41
+ "epoch": 1.0,
42
+ "step": 2,
43
+ "total_flos": 196951080960000.0,
44
+ "train_loss": 1.7391430735588074,
45
+ "train_runtime": 61.5449,
46
+ "train_samples_per_second": 0.13,
47
+ "train_steps_per_second": 0.032
48
  }
49
  ],
50
+ "max_steps": 2,
51
  "num_train_epochs": 9223372036854775807,
52
+ "total_flos": 196951080960000.0,
53
  "trial_name": null,
54
  "trial_params": null
55
  }
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4578ee205fe35fdbb3b98224d5f855f806bd9eedb29095f0f731a29261db2d94
3
- size 3579
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3f657957109096465f8089361114a5bcf715d17db0a314d2a9784c7deda67c16
3
+ size 3707