Yurii Paniv commited on
Commit
1743317
1 Parent(s): d0e54b9

Initial model

Browse files
config.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-xls-r-300m",
3
+ "activation_dropout": 0.0,
4
+ "apply_spec_augment": true,
5
+ "architectures": [
6
+ "Wav2Vec2ForCTC"
7
+ ],
8
+ "attention_dropout": 0.2,
9
+ "bos_token_id": 1,
10
+ "classifier_proj_size": 256,
11
+ "codevector_dim": 768,
12
+ "contrastive_logits_temperature": 0.1,
13
+ "conv_bias": true,
14
+ "conv_dim": [
15
+ 512,
16
+ 512,
17
+ 512,
18
+ 512,
19
+ 512,
20
+ 512,
21
+ 512
22
+ ],
23
+ "conv_kernel": [
24
+ 10,
25
+ 3,
26
+ 3,
27
+ 3,
28
+ 3,
29
+ 2,
30
+ 2
31
+ ],
32
+ "conv_stride": [
33
+ 5,
34
+ 2,
35
+ 2,
36
+ 2,
37
+ 2,
38
+ 2,
39
+ 2
40
+ ],
41
+ "ctc_loss_reduction": "mean",
42
+ "ctc_zero_infinity": false,
43
+ "diversity_loss_weight": 0.1,
44
+ "do_stable_layer_norm": true,
45
+ "eos_token_id": 2,
46
+ "feat_extract_activation": "gelu",
47
+ "feat_extract_dropout": 0.0,
48
+ "feat_extract_norm": "layer",
49
+ "feat_proj_dropout": 0.1,
50
+ "feat_quantizer_dropout": 0.0,
51
+ "final_dropout": 0.0,
52
+ "gradient_checkpointing": false,
53
+ "hidden_act": "gelu",
54
+ "hidden_dropout": 0.2,
55
+ "hidden_size": 1024,
56
+ "initializer_range": 0.02,
57
+ "intermediate_size": 4096,
58
+ "layer_norm_eps": 1e-05,
59
+ "layerdrop": 0.0,
60
+ "mask_feature_length": 10,
61
+ "mask_feature_prob": 0.0,
62
+ "mask_time_length": 10,
63
+ "mask_time_prob": 0.05,
64
+ "model_type": "wav2vec2",
65
+ "num_attention_heads": 16,
66
+ "num_codevector_groups": 2,
67
+ "num_codevectors_per_group": 320,
68
+ "num_conv_pos_embedding_groups": 16,
69
+ "num_conv_pos_embeddings": 128,
70
+ "num_feat_extract_layers": 7,
71
+ "num_hidden_layers": 24,
72
+ "num_negatives": 100,
73
+ "pad_token_id": 36,
74
+ "proj_codevector_dim": 768,
75
+ "torch_dtype": "float32",
76
+ "transformers_version": "4.11.0",
77
+ "use_weighted_layer_sum": false,
78
+ "vocab_size": 39
79
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0fbd68c9108bdd7f2b74d2d6ab60ec454155756c1066837141fb986be18d13b0
3
+ size 2490378769
preprocessor_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0.0,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44e381719215ff2cd65a2fbf5588b8deeaf4f712e2800f38410d404e2527695f
3
+ size 1262083569
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:76eba952d902f592e5dc53b1ace73547fece6bddedfb82abd6d368838cc3983f
3
+ size 14567
scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12ab3ab57bdd4de9271b2ed77c5b527f019219b09da8c0a67492f35863f058ca
3
+ size 559
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8cd458f9d8a9355ed82afe3dc4a8877431efaf85b6481a523d12fee5200e931c
3
+ size 623
trainer_state.json ADDED
@@ -0,0 +1,1501 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 35.73957158962796,
5
+ "global_step": 39600,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.36,
12
+ "learning_rate": 0.0002388,
13
+ "loss": 5.1061,
14
+ "step": 400
15
+ },
16
+ {
17
+ "epoch": 0.36,
18
+ "eval_loss": 3.2615509033203125,
19
+ "eval_runtime": 209.4571,
20
+ "eval_samples_per_second": 18.892,
21
+ "eval_steps_per_second": 2.363,
22
+ "eval_wer": 1.0,
23
+ "step": 400
24
+ },
25
+ {
26
+ "epoch": 0.72,
27
+ "learning_rate": 0.0002990989719814553,
28
+ "loss": 1.8298,
29
+ "step": 800
30
+ },
31
+ {
32
+ "epoch": 0.72,
33
+ "eval_loss": 0.5897451639175415,
34
+ "eval_runtime": 208.9676,
35
+ "eval_samples_per_second": 18.936,
36
+ "eval_steps_per_second": 2.369,
37
+ "eval_wer": 0.7382658232272045,
38
+ "step": 800
39
+ },
40
+ {
41
+ "epoch": 1.08,
42
+ "learning_rate": 0.0002978895383995162,
43
+ "loss": 0.7312,
44
+ "step": 1200
45
+ },
46
+ {
47
+ "epoch": 1.08,
48
+ "eval_loss": 0.38931864500045776,
49
+ "eval_runtime": 205.7748,
50
+ "eval_samples_per_second": 19.23,
51
+ "eval_steps_per_second": 2.406,
52
+ "eval_wer": 0.5565651976374375,
53
+ "step": 1200
54
+ },
55
+ {
56
+ "epoch": 1.44,
57
+ "learning_rate": 0.0002966801048175771,
58
+ "loss": 0.5859,
59
+ "step": 1600
60
+ },
61
+ {
62
+ "epoch": 1.44,
63
+ "eval_loss": 0.3028641939163208,
64
+ "eval_runtime": 210.4962,
65
+ "eval_samples_per_second": 18.798,
66
+ "eval_steps_per_second": 2.352,
67
+ "eval_wer": 0.43102785447174363,
68
+ "step": 1600
69
+ },
70
+ {
71
+ "epoch": 1.8,
72
+ "learning_rate": 0.00029547067123563797,
73
+ "loss": 0.5272,
74
+ "step": 2000
75
+ },
76
+ {
77
+ "epoch": 1.8,
78
+ "eval_loss": 0.2646712362766266,
79
+ "eval_runtime": 208.1951,
80
+ "eval_samples_per_second": 19.006,
81
+ "eval_steps_per_second": 2.378,
82
+ "eval_wer": 0.3874462656834306,
83
+ "step": 2000
84
+ },
85
+ {
86
+ "epoch": 2.17,
87
+ "learning_rate": 0.00029426123765369884,
88
+ "loss": 0.4691,
89
+ "step": 2400
90
+ },
91
+ {
92
+ "epoch": 2.17,
93
+ "eval_loss": 0.25143691897392273,
94
+ "eval_runtime": 208.3951,
95
+ "eval_samples_per_second": 18.988,
96
+ "eval_steps_per_second": 2.375,
97
+ "eval_wer": 0.3728025722573655,
98
+ "step": 2400
99
+ },
100
+ {
101
+ "epoch": 2.53,
102
+ "learning_rate": 0.0002930518040717597,
103
+ "loss": 0.4313,
104
+ "step": 2800
105
+ },
106
+ {
107
+ "epoch": 2.53,
108
+ "eval_loss": 0.23935846984386444,
109
+ "eval_runtime": 208.5439,
110
+ "eval_samples_per_second": 18.974,
111
+ "eval_steps_per_second": 2.374,
112
+ "eval_wer": 0.33872715199384895,
113
+ "step": 2800
114
+ },
115
+ {
116
+ "epoch": 2.89,
117
+ "learning_rate": 0.0002918423704898206,
118
+ "loss": 0.4157,
119
+ "step": 3200
120
+ },
121
+ {
122
+ "epoch": 2.89,
123
+ "eval_loss": 0.22154289484024048,
124
+ "eval_runtime": 207.5705,
125
+ "eval_samples_per_second": 19.063,
126
+ "eval_steps_per_second": 2.385,
127
+ "eval_wer": 0.3165344423863279,
128
+ "step": 3200
129
+ },
130
+ {
131
+ "epoch": 3.25,
132
+ "learning_rate": 0.00029063293690788144,
133
+ "loss": 0.382,
134
+ "step": 3600
135
+ },
136
+ {
137
+ "epoch": 3.25,
138
+ "eval_loss": 0.22716625034809113,
139
+ "eval_runtime": 208.4179,
140
+ "eval_samples_per_second": 18.986,
141
+ "eval_steps_per_second": 2.375,
142
+ "eval_wer": 0.310103798972495,
143
+ "step": 3600
144
+ },
145
+ {
146
+ "epoch": 3.61,
147
+ "learning_rate": 0.0002894235033259423,
148
+ "loss": 0.3636,
149
+ "step": 4000
150
+ },
151
+ {
152
+ "epoch": 3.61,
153
+ "eval_loss": 0.21131443977355957,
154
+ "eval_runtime": 206.1846,
155
+ "eval_samples_per_second": 19.192,
156
+ "eval_steps_per_second": 2.401,
157
+ "eval_wer": 0.30538566385908505,
158
+ "step": 4000
159
+ },
160
+ {
161
+ "epoch": 3.97,
162
+ "learning_rate": 0.00028821406974400323,
163
+ "loss": 0.3665,
164
+ "step": 4400
165
+ },
166
+ {
167
+ "epoch": 3.97,
168
+ "eval_loss": 0.21285590529441833,
169
+ "eval_runtime": 205.9986,
170
+ "eval_samples_per_second": 19.209,
171
+ "eval_steps_per_second": 2.403,
172
+ "eval_wer": 0.2907769195820082,
173
+ "step": 4400
174
+ },
175
+ {
176
+ "epoch": 4.33,
177
+ "learning_rate": 0.00028700463616206405,
178
+ "loss": 0.333,
179
+ "step": 4800
180
+ },
181
+ {
182
+ "epoch": 4.33,
183
+ "eval_loss": 0.2077477127313614,
184
+ "eval_runtime": 208.1768,
185
+ "eval_samples_per_second": 19.008,
186
+ "eval_steps_per_second": 2.378,
187
+ "eval_wer": 0.2788243106280362,
188
+ "step": 4800
189
+ },
190
+ {
191
+ "epoch": 4.69,
192
+ "learning_rate": 0.00028579520258012497,
193
+ "loss": 0.3313,
194
+ "step": 5200
195
+ },
196
+ {
197
+ "epoch": 4.69,
198
+ "eval_loss": 0.18993255496025085,
199
+ "eval_runtime": 205.0681,
200
+ "eval_samples_per_second": 19.296,
201
+ "eval_steps_per_second": 2.414,
202
+ "eval_wer": 0.2659979729493587,
203
+ "step": 5200
204
+ },
205
+ {
206
+ "epoch": 5.05,
207
+ "learning_rate": 0.00028458576899818584,
208
+ "loss": 0.3222,
209
+ "step": 5600
210
+ },
211
+ {
212
+ "epoch": 5.05,
213
+ "eval_loss": 0.19064170122146606,
214
+ "eval_runtime": 205.7167,
215
+ "eval_samples_per_second": 19.235,
216
+ "eval_steps_per_second": 2.406,
217
+ "eval_wer": 0.2542201097403278,
218
+ "step": 5600
219
+ },
220
+ {
221
+ "epoch": 5.41,
222
+ "learning_rate": 0.0002833763354162467,
223
+ "loss": 0.3003,
224
+ "step": 6000
225
+ },
226
+ {
227
+ "epoch": 5.41,
228
+ "eval_loss": 0.1921728551387787,
229
+ "eval_runtime": 204.1327,
230
+ "eval_samples_per_second": 19.384,
231
+ "eval_steps_per_second": 2.425,
232
+ "eval_wer": 0.2612099395379722,
233
+ "step": 6000
234
+ },
235
+ {
236
+ "epoch": 5.78,
237
+ "learning_rate": 0.00028216690183430757,
238
+ "loss": 0.3057,
239
+ "step": 6400
240
+ },
241
+ {
242
+ "epoch": 5.78,
243
+ "eval_loss": 0.19207137823104858,
244
+ "eval_runtime": 205.293,
245
+ "eval_samples_per_second": 19.275,
246
+ "eval_steps_per_second": 2.411,
247
+ "eval_wer": 0.2560724146367036,
248
+ "step": 6400
249
+ },
250
+ {
251
+ "epoch": 6.14,
252
+ "learning_rate": 0.0002809604918363233,
253
+ "loss": 0.297,
254
+ "step": 6800
255
+ },
256
+ {
257
+ "epoch": 6.14,
258
+ "eval_loss": 0.1768147200345993,
259
+ "eval_runtime": 204.3464,
260
+ "eval_samples_per_second": 19.364,
261
+ "eval_steps_per_second": 2.422,
262
+ "eval_wer": 0.24146367035962674,
263
+ "step": 6800
264
+ },
265
+ {
266
+ "epoch": 6.5,
267
+ "learning_rate": 0.0002797510582543842,
268
+ "loss": 0.2834,
269
+ "step": 7200
270
+ },
271
+ {
272
+ "epoch": 6.5,
273
+ "eval_loss": 0.18651635944843292,
274
+ "eval_runtime": 205.3683,
275
+ "eval_samples_per_second": 19.268,
276
+ "eval_steps_per_second": 2.41,
277
+ "eval_wer": 0.24894278824310628,
278
+ "step": 7200
279
+ },
280
+ {
281
+ "epoch": 6.86,
282
+ "learning_rate": 0.00027854162467244505,
283
+ "loss": 0.2787,
284
+ "step": 7600
285
+ },
286
+ {
287
+ "epoch": 6.86,
288
+ "eval_loss": 0.18522889912128448,
289
+ "eval_runtime": 204.2237,
290
+ "eval_samples_per_second": 19.376,
291
+ "eval_steps_per_second": 2.424,
292
+ "eval_wer": 0.24726522909167162,
293
+ "step": 7600
294
+ },
295
+ {
296
+ "epoch": 7.22,
297
+ "learning_rate": 0.0002773321910905059,
298
+ "loss": 0.2678,
299
+ "step": 8000
300
+ },
301
+ {
302
+ "epoch": 7.22,
303
+ "eval_loss": 0.1824478805065155,
304
+ "eval_runtime": 203.4819,
305
+ "eval_samples_per_second": 19.446,
306
+ "eval_steps_per_second": 2.433,
307
+ "eval_wer": 0.23087407821619543,
308
+ "step": 8000
309
+ },
310
+ {
311
+ "epoch": 7.58,
312
+ "learning_rate": 0.0002761227575085668,
313
+ "loss": 0.2665,
314
+ "step": 8400
315
+ },
316
+ {
317
+ "epoch": 7.58,
318
+ "eval_loss": 0.17768050730228424,
319
+ "eval_runtime": 203.7918,
320
+ "eval_samples_per_second": 19.417,
321
+ "eval_steps_per_second": 2.429,
322
+ "eval_wer": 0.2309789256631601,
323
+ "step": 8400
324
+ },
325
+ {
326
+ "epoch": 7.94,
327
+ "learning_rate": 0.00027491332392662766,
328
+ "loss": 0.2722,
329
+ "step": 8800
330
+ },
331
+ {
332
+ "epoch": 7.94,
333
+ "eval_loss": 0.17655304074287415,
334
+ "eval_runtime": 203.4054,
335
+ "eval_samples_per_second": 19.454,
336
+ "eval_steps_per_second": 2.434,
337
+ "eval_wer": 0.23216719672875966,
338
+ "step": 8800
339
+ },
340
+ {
341
+ "epoch": 8.3,
342
+ "learning_rate": 0.0002737038903446885,
343
+ "loss": 0.2549,
344
+ "step": 9200
345
+ },
346
+ {
347
+ "epoch": 8.3,
348
+ "eval_loss": 0.18208913505077362,
349
+ "eval_runtime": 203.5406,
350
+ "eval_samples_per_second": 19.441,
351
+ "eval_steps_per_second": 2.432,
352
+ "eval_wer": 0.23590675567049943,
353
+ "step": 9200
354
+ },
355
+ {
356
+ "epoch": 8.66,
357
+ "learning_rate": 0.00027249748034670427,
358
+ "loss": 0.2535,
359
+ "step": 9600
360
+ },
361
+ {
362
+ "epoch": 8.66,
363
+ "eval_loss": 0.17272169888019562,
364
+ "eval_runtime": 204.5912,
365
+ "eval_samples_per_second": 19.341,
366
+ "eval_steps_per_second": 2.419,
367
+ "eval_wer": 0.22619089225177366,
368
+ "step": 9600
369
+ },
370
+ {
371
+ "epoch": 9.03,
372
+ "learning_rate": 0.00027128804676476514,
373
+ "loss": 0.2508,
374
+ "step": 10000
375
+ },
376
+ {
377
+ "epoch": 9.03,
378
+ "eval_loss": 0.17887552082538605,
379
+ "eval_runtime": 202.6986,
380
+ "eval_samples_per_second": 19.522,
381
+ "eval_steps_per_second": 2.442,
382
+ "eval_wer": 0.22301051969384544,
383
+ "step": 10000
384
+ },
385
+ {
386
+ "epoch": 9.39,
387
+ "learning_rate": 0.000270078613182826,
388
+ "loss": 0.2354,
389
+ "step": 10400
390
+ },
391
+ {
392
+ "epoch": 9.39,
393
+ "eval_loss": 0.17052198946475983,
394
+ "eval_runtime": 202.2989,
395
+ "eval_samples_per_second": 19.56,
396
+ "eval_steps_per_second": 2.447,
397
+ "eval_wer": 0.2227658756509279,
398
+ "step": 10400
399
+ },
400
+ {
401
+ "epoch": 9.75,
402
+ "learning_rate": 0.0002688691796008869,
403
+ "loss": 0.2449,
404
+ "step": 10800
405
+ },
406
+ {
407
+ "epoch": 9.75,
408
+ "eval_loss": 0.17250895500183105,
409
+ "eval_runtime": 202.4538,
410
+ "eval_samples_per_second": 19.545,
411
+ "eval_steps_per_second": 2.445,
412
+ "eval_wer": 0.2203893335197288,
413
+ "step": 10800
414
+ },
415
+ {
416
+ "epoch": 10.11,
417
+ "learning_rate": 0.0002676597460189478,
418
+ "loss": 0.2461,
419
+ "step": 11200
420
+ },
421
+ {
422
+ "epoch": 10.11,
423
+ "eval_loss": 0.18136142194271088,
424
+ "eval_runtime": 202.9746,
425
+ "eval_samples_per_second": 19.495,
426
+ "eval_steps_per_second": 2.439,
427
+ "eval_wer": 0.23083912906720722,
428
+ "step": 11200
429
+ },
430
+ {
431
+ "epoch": 10.47,
432
+ "learning_rate": 0.0002664533360209635,
433
+ "loss": 0.2354,
434
+ "step": 11600
435
+ },
436
+ {
437
+ "epoch": 10.47,
438
+ "eval_loss": 0.17660640180110931,
439
+ "eval_runtime": 202.5754,
440
+ "eval_samples_per_second": 19.533,
441
+ "eval_steps_per_second": 2.444,
442
+ "eval_wer": 0.21902631670918812,
443
+ "step": 11600
444
+ },
445
+ {
446
+ "epoch": 10.83,
447
+ "learning_rate": 0.00026524390243902435,
448
+ "loss": 0.2271,
449
+ "step": 12000
450
+ },
451
+ {
452
+ "epoch": 10.83,
453
+ "eval_loss": 0.16712866723537445,
454
+ "eval_runtime": 204.6252,
455
+ "eval_samples_per_second": 19.338,
456
+ "eval_steps_per_second": 2.419,
457
+ "eval_wer": 0.22643553629469124,
458
+ "step": 12000
459
+ },
460
+ {
461
+ "epoch": 11.19,
462
+ "learning_rate": 0.0002640344688570853,
463
+ "loss": 0.2262,
464
+ "step": 12400
465
+ },
466
+ {
467
+ "epoch": 11.19,
468
+ "eval_loss": 0.16640537977218628,
469
+ "eval_runtime": 205.9457,
470
+ "eval_samples_per_second": 19.214,
471
+ "eval_steps_per_second": 2.404,
472
+ "eval_wer": 0.2237444518225981,
473
+ "step": 12400
474
+ },
475
+ {
476
+ "epoch": 11.55,
477
+ "learning_rate": 0.0002628250352751461,
478
+ "loss": 0.222,
479
+ "step": 12800
480
+ },
481
+ {
482
+ "epoch": 11.55,
483
+ "eval_loss": 0.16319511830806732,
484
+ "eval_runtime": 203.9378,
485
+ "eval_samples_per_second": 19.403,
486
+ "eval_steps_per_second": 2.427,
487
+ "eval_wer": 0.20773774158599237,
488
+ "step": 12800
489
+ },
490
+ {
491
+ "epoch": 11.91,
492
+ "learning_rate": 0.000261615601693207,
493
+ "loss": 0.2254,
494
+ "step": 13200
495
+ },
496
+ {
497
+ "epoch": 11.91,
498
+ "eval_loss": 0.16412602365016937,
499
+ "eval_runtime": 203.9337,
500
+ "eval_samples_per_second": 19.403,
501
+ "eval_steps_per_second": 2.427,
502
+ "eval_wer": 0.21906126585817634,
503
+ "step": 13200
504
+ },
505
+ {
506
+ "epoch": 12.27,
507
+ "learning_rate": 0.0002604061681112679,
508
+ "loss": 0.2198,
509
+ "step": 13600
510
+ },
511
+ {
512
+ "epoch": 12.27,
513
+ "eval_loss": 0.16197219491004944,
514
+ "eval_runtime": 202.2546,
515
+ "eval_samples_per_second": 19.564,
516
+ "eval_steps_per_second": 2.447,
517
+ "eval_wer": 0.2122112326564848,
518
+ "step": 13600
519
+ },
520
+ {
521
+ "epoch": 12.63,
522
+ "learning_rate": 0.00025919673452932875,
523
+ "loss": 0.2167,
524
+ "step": 14000
525
+ },
526
+ {
527
+ "epoch": 12.63,
528
+ "eval_loss": 0.15757694840431213,
529
+ "eval_runtime": 202.2278,
530
+ "eval_samples_per_second": 19.567,
531
+ "eval_steps_per_second": 2.448,
532
+ "eval_wer": 0.2082270296718275,
533
+ "step": 14000
534
+ },
535
+ {
536
+ "epoch": 13.0,
537
+ "learning_rate": 0.0002579903245313445,
538
+ "loss": 0.2138,
539
+ "step": 14400
540
+ },
541
+ {
542
+ "epoch": 13.0,
543
+ "eval_loss": 0.16401419043540955,
544
+ "eval_runtime": 202.7059,
545
+ "eval_samples_per_second": 19.521,
546
+ "eval_steps_per_second": 2.442,
547
+ "eval_wer": 0.2156711984063188,
548
+ "step": 14400
549
+ },
550
+ {
551
+ "epoch": 13.36,
552
+ "learning_rate": 0.0002567808909494053,
553
+ "loss": 0.2112,
554
+ "step": 14800
555
+ },
556
+ {
557
+ "epoch": 13.36,
558
+ "eval_loss": 0.17054887115955353,
559
+ "eval_runtime": 203.4699,
560
+ "eval_samples_per_second": 19.448,
561
+ "eval_steps_per_second": 2.433,
562
+ "eval_wer": 0.21696431691888302,
563
+ "step": 14800
564
+ },
565
+ {
566
+ "epoch": 13.72,
567
+ "learning_rate": 0.00025557145736746623,
568
+ "loss": 0.2038,
569
+ "step": 15200
570
+ },
571
+ {
572
+ "epoch": 13.72,
573
+ "eval_loss": 0.1620979905128479,
574
+ "eval_runtime": 203.3439,
575
+ "eval_samples_per_second": 19.46,
576
+ "eval_steps_per_second": 2.434,
577
+ "eval_wer": 0.2075979449900395,
578
+ "step": 15200
579
+ },
580
+ {
581
+ "epoch": 14.08,
582
+ "learning_rate": 0.0002543620237855271,
583
+ "loss": 0.2101,
584
+ "step": 15600
585
+ },
586
+ {
587
+ "epoch": 14.08,
588
+ "eval_loss": 0.16726231575012207,
589
+ "eval_runtime": 203.5727,
590
+ "eval_samples_per_second": 19.438,
591
+ "eval_steps_per_second": 2.432,
592
+ "eval_wer": 0.19962953902072483,
593
+ "step": 15600
594
+ },
595
+ {
596
+ "epoch": 14.44,
597
+ "learning_rate": 0.00025315259020358797,
598
+ "loss": 0.2013,
599
+ "step": 16000
600
+ },
601
+ {
602
+ "epoch": 14.44,
603
+ "eval_loss": 0.16415779292583466,
604
+ "eval_runtime": 202.8281,
605
+ "eval_samples_per_second": 19.509,
606
+ "eval_steps_per_second": 2.44,
607
+ "eval_wer": 0.20885611435361548,
608
+ "step": 16000
609
+ },
610
+ {
611
+ "epoch": 14.8,
612
+ "learning_rate": 0.00025194315662164883,
613
+ "loss": 0.21,
614
+ "step": 16400
615
+ },
616
+ {
617
+ "epoch": 14.8,
618
+ "eval_loss": 0.1616578996181488,
619
+ "eval_runtime": 203.1464,
620
+ "eval_samples_per_second": 19.479,
621
+ "eval_steps_per_second": 2.437,
622
+ "eval_wer": 0.2186768252193059,
623
+ "step": 16400
624
+ },
625
+ {
626
+ "epoch": 15.16,
627
+ "learning_rate": 0.0002507337230397097,
628
+ "loss": 0.2002,
629
+ "step": 16800
630
+ },
631
+ {
632
+ "epoch": 15.16,
633
+ "eval_loss": 0.1627209484577179,
634
+ "eval_runtime": 202.7336,
635
+ "eval_samples_per_second": 19.518,
636
+ "eval_steps_per_second": 2.442,
637
+ "eval_wer": 0.21839723202740013,
638
+ "step": 16800
639
+ },
640
+ {
641
+ "epoch": 15.52,
642
+ "learning_rate": 0.00024952428945777057,
643
+ "loss": 0.2007,
644
+ "step": 17200
645
+ },
646
+ {
647
+ "epoch": 15.52,
648
+ "eval_loss": 0.1601659506559372,
649
+ "eval_runtime": 202.199,
650
+ "eval_samples_per_second": 19.57,
651
+ "eval_steps_per_second": 2.448,
652
+ "eval_wer": 0.20924055499248592,
653
+ "step": 17200
654
+ },
655
+ {
656
+ "epoch": 15.88,
657
+ "learning_rate": 0.0002483178794597863,
658
+ "loss": 0.2042,
659
+ "step": 17600
660
+ },
661
+ {
662
+ "epoch": 15.88,
663
+ "eval_loss": 0.15520912408828735,
664
+ "eval_runtime": 202.5397,
665
+ "eval_samples_per_second": 19.537,
666
+ "eval_steps_per_second": 2.444,
667
+ "eval_wer": 0.1952958445461853,
668
+ "step": 17600
669
+ },
670
+ {
671
+ "epoch": 16.25,
672
+ "learning_rate": 0.0002471084458778472,
673
+ "loss": 0.187,
674
+ "step": 18000
675
+ },
676
+ {
677
+ "epoch": 16.25,
678
+ "eval_loss": 0.16360661387443542,
679
+ "eval_runtime": 202.7459,
680
+ "eval_samples_per_second": 19.517,
681
+ "eval_steps_per_second": 2.441,
682
+ "eval_wer": 0.19442211582147975,
683
+ "step": 18000
684
+ },
685
+ {
686
+ "epoch": 16.61,
687
+ "learning_rate": 0.00024589901229590805,
688
+ "loss": 0.1931,
689
+ "step": 18400
690
+ },
691
+ {
692
+ "epoch": 16.61,
693
+ "eval_loss": 0.1533084213733673,
694
+ "eval_runtime": 203.0495,
695
+ "eval_samples_per_second": 19.488,
696
+ "eval_steps_per_second": 2.438,
697
+ "eval_wer": 0.1916960822003984,
698
+ "step": 18400
699
+ },
700
+ {
701
+ "epoch": 16.97,
702
+ "learning_rate": 0.000244689578713969,
703
+ "loss": 0.1891,
704
+ "step": 18800
705
+ },
706
+ {
707
+ "epoch": 16.97,
708
+ "eval_loss": 0.16208086907863617,
709
+ "eval_runtime": 202.8946,
710
+ "eval_samples_per_second": 19.503,
711
+ "eval_steps_per_second": 2.44,
712
+ "eval_wer": 0.1907175060287282,
713
+ "step": 18800
714
+ },
715
+ {
716
+ "epoch": 17.33,
717
+ "learning_rate": 0.00024348014513202981,
718
+ "loss": 0.1863,
719
+ "step": 19200
720
+ },
721
+ {
722
+ "epoch": 17.33,
723
+ "eval_loss": 0.15993733704090118,
724
+ "eval_runtime": 202.8467,
725
+ "eval_samples_per_second": 19.507,
726
+ "eval_steps_per_second": 2.44,
727
+ "eval_wer": 0.1905777094327753,
728
+ "step": 19200
729
+ },
730
+ {
731
+ "epoch": 17.69,
732
+ "learning_rate": 0.00024227071155009068,
733
+ "loss": 0.1863,
734
+ "step": 19600
735
+ },
736
+ {
737
+ "epoch": 17.69,
738
+ "eval_loss": 0.15354220569133759,
739
+ "eval_runtime": 202.9873,
740
+ "eval_samples_per_second": 19.494,
741
+ "eval_steps_per_second": 2.439,
742
+ "eval_wer": 0.1911718449655751,
743
+ "step": 19600
744
+ },
745
+ {
746
+ "epoch": 18.05,
747
+ "learning_rate": 0.00024106127796815155,
748
+ "loss": 0.1827,
749
+ "step": 20000
750
+ },
751
+ {
752
+ "epoch": 18.05,
753
+ "eval_loss": 0.1616511195898056,
754
+ "eval_runtime": 204.4003,
755
+ "eval_samples_per_second": 19.359,
756
+ "eval_steps_per_second": 2.422,
757
+ "eval_wer": 0.1880264215566351,
758
+ "step": 20000
759
+ },
760
+ {
761
+ "epoch": 18.41,
762
+ "learning_rate": 0.00023985184438621244,
763
+ "loss": 0.1822,
764
+ "step": 20400
765
+ },
766
+ {
767
+ "epoch": 18.41,
768
+ "eval_loss": 0.1649104207754135,
769
+ "eval_runtime": 206.8059,
770
+ "eval_samples_per_second": 19.134,
771
+ "eval_steps_per_second": 2.394,
772
+ "eval_wer": 0.18806137070562332,
773
+ "step": 20400
774
+ },
775
+ {
776
+ "epoch": 18.77,
777
+ "learning_rate": 0.0002386424108042733,
778
+ "loss": 0.1841,
779
+ "step": 20800
780
+ },
781
+ {
782
+ "epoch": 18.77,
783
+ "eval_loss": 0.16039612889289856,
784
+ "eval_runtime": 204.0742,
785
+ "eval_samples_per_second": 19.39,
786
+ "eval_steps_per_second": 2.426,
787
+ "eval_wer": 0.18942438751616397,
788
+ "step": 20800
789
+ },
790
+ {
791
+ "epoch": 19.13,
792
+ "learning_rate": 0.00023743297722233418,
793
+ "loss": 0.1817,
794
+ "step": 21200
795
+ },
796
+ {
797
+ "epoch": 19.13,
798
+ "eval_loss": 0.157293900847435,
799
+ "eval_runtime": 202.3143,
800
+ "eval_samples_per_second": 19.559,
801
+ "eval_steps_per_second": 2.447,
802
+ "eval_wer": 0.1810715409079789,
803
+ "step": 21200
804
+ },
805
+ {
806
+ "epoch": 19.49,
807
+ "learning_rate": 0.00023622959080830475,
808
+ "loss": 0.1756,
809
+ "step": 21600
810
+ },
811
+ {
812
+ "epoch": 19.49,
813
+ "eval_loss": 0.16328036785125732,
814
+ "eval_runtime": 204.2348,
815
+ "eval_samples_per_second": 19.375,
816
+ "eval_steps_per_second": 2.424,
817
+ "eval_wer": 0.18809631985461153,
818
+ "step": 21600
819
+ },
820
+ {
821
+ "epoch": 19.86,
822
+ "learning_rate": 0.00023502015722636564,
823
+ "loss": 0.1804,
824
+ "step": 22000
825
+ },
826
+ {
827
+ "epoch": 19.86,
828
+ "eval_loss": 0.1585751324892044,
829
+ "eval_runtime": 204.9615,
830
+ "eval_samples_per_second": 19.306,
831
+ "eval_steps_per_second": 2.415,
832
+ "eval_wer": 0.18823611645056443,
833
+ "step": 22000
834
+ },
835
+ {
836
+ "epoch": 20.22,
837
+ "learning_rate": 0.0002338107236444265,
838
+ "loss": 0.1736,
839
+ "step": 22400
840
+ },
841
+ {
842
+ "epoch": 20.22,
843
+ "eval_loss": 0.15415678918361664,
844
+ "eval_runtime": 202.4017,
845
+ "eval_samples_per_second": 19.55,
846
+ "eval_steps_per_second": 2.446,
847
+ "eval_wer": 0.19033306538985775,
848
+ "step": 22400
849
+ },
850
+ {
851
+ "epoch": 20.58,
852
+ "learning_rate": 0.00023260129006248738,
853
+ "loss": 0.1704,
854
+ "step": 22800
855
+ },
856
+ {
857
+ "epoch": 20.58,
858
+ "eval_loss": 0.1622617095708847,
859
+ "eval_runtime": 206.2682,
860
+ "eval_samples_per_second": 19.184,
861
+ "eval_steps_per_second": 2.4,
862
+ "eval_wer": 0.1983014713591724,
863
+ "step": 22800
864
+ },
865
+ {
866
+ "epoch": 20.94,
867
+ "learning_rate": 0.00023139185648054828,
868
+ "loss": 0.1741,
869
+ "step": 23200
870
+ },
871
+ {
872
+ "epoch": 20.94,
873
+ "eval_loss": 0.15422095358371735,
874
+ "eval_runtime": 208.5273,
875
+ "eval_samples_per_second": 18.976,
876
+ "eval_steps_per_second": 2.374,
877
+ "eval_wer": 0.19568028518505573,
878
+ "step": 23200
879
+ },
880
+ {
881
+ "epoch": 21.3,
882
+ "learning_rate": 0.00023018242289860912,
883
+ "loss": 0.1642,
884
+ "step": 23600
885
+ },
886
+ {
887
+ "epoch": 21.3,
888
+ "eval_loss": 0.155488982796669,
889
+ "eval_runtime": 206.3858,
890
+ "eval_samples_per_second": 19.173,
891
+ "eval_steps_per_second": 2.398,
892
+ "eval_wer": 0.19704330199559641,
893
+ "step": 23600
894
+ },
895
+ {
896
+ "epoch": 21.66,
897
+ "learning_rate": 0.00022897601290062486,
898
+ "loss": 0.1707,
899
+ "step": 24000
900
+ },
901
+ {
902
+ "epoch": 21.66,
903
+ "eval_loss": 0.15275663137435913,
904
+ "eval_runtime": 206.6413,
905
+ "eval_samples_per_second": 19.149,
906
+ "eval_steps_per_second": 2.395,
907
+ "eval_wer": 0.18442665921084822,
908
+ "step": 24000
909
+ },
910
+ {
911
+ "epoch": 22.02,
912
+ "learning_rate": 0.00022776657931868573,
913
+ "loss": 0.1673,
914
+ "step": 24400
915
+ },
916
+ {
917
+ "epoch": 22.02,
918
+ "eval_loss": 0.15888585150241852,
919
+ "eval_runtime": 208.9307,
920
+ "eval_samples_per_second": 18.939,
921
+ "eval_steps_per_second": 2.369,
922
+ "eval_wer": 0.1858246251703771,
923
+ "step": 24400
924
+ },
925
+ {
926
+ "epoch": 22.38,
927
+ "learning_rate": 0.0002265571457367466,
928
+ "loss": 0.1593,
929
+ "step": 24800
930
+ },
931
+ {
932
+ "epoch": 22.38,
933
+ "eval_loss": 0.15319041907787323,
934
+ "eval_runtime": 206.6808,
935
+ "eval_samples_per_second": 19.145,
936
+ "eval_steps_per_second": 2.395,
937
+ "eval_wer": 0.17750672771118023,
938
+ "step": 24800
939
+ },
940
+ {
941
+ "epoch": 22.74,
942
+ "learning_rate": 0.0002253477121548075,
943
+ "loss": 0.1595,
944
+ "step": 25200
945
+ },
946
+ {
947
+ "epoch": 22.74,
948
+ "eval_loss": 0.15495522320270538,
949
+ "eval_runtime": 211.5152,
950
+ "eval_samples_per_second": 18.708,
951
+ "eval_steps_per_second": 2.34,
952
+ "eval_wer": 0.1755495753678398,
953
+ "step": 25200
954
+ },
955
+ {
956
+ "epoch": 23.1,
957
+ "learning_rate": 0.00022413827857286836,
958
+ "loss": 0.1598,
959
+ "step": 25600
960
+ },
961
+ {
962
+ "epoch": 23.1,
963
+ "eval_loss": 0.16047397255897522,
964
+ "eval_runtime": 213.237,
965
+ "eval_samples_per_second": 18.557,
966
+ "eval_steps_per_second": 2.321,
967
+ "eval_wer": 0.18624401495823575,
968
+ "step": 25600
969
+ },
970
+ {
971
+ "epoch": 23.47,
972
+ "learning_rate": 0.00022293186857488405,
973
+ "loss": 0.1573,
974
+ "step": 26000
975
+ },
976
+ {
977
+ "epoch": 23.47,
978
+ "eval_loss": 0.15952068567276,
979
+ "eval_runtime": 205.0554,
980
+ "eval_samples_per_second": 19.297,
981
+ "eval_steps_per_second": 2.414,
982
+ "eval_wer": 0.17796106664802713,
983
+ "step": 26000
984
+ },
985
+ {
986
+ "epoch": 23.83,
987
+ "learning_rate": 0.00022172243499294495,
988
+ "loss": 0.1596,
989
+ "step": 26400
990
+ },
991
+ {
992
+ "epoch": 23.83,
993
+ "eval_loss": 0.15038904547691345,
994
+ "eval_runtime": 203.7621,
995
+ "eval_samples_per_second": 19.42,
996
+ "eval_steps_per_second": 2.429,
997
+ "eval_wer": 0.17879984622374445,
998
+ "step": 26400
999
+ },
1000
+ {
1001
+ "epoch": 24.19,
1002
+ "learning_rate": 0.00022051300141100584,
1003
+ "loss": 0.1589,
1004
+ "step": 26800
1005
+ },
1006
+ {
1007
+ "epoch": 24.19,
1008
+ "eval_loss": 0.15989159047603607,
1009
+ "eval_runtime": 202.8293,
1010
+ "eval_samples_per_second": 19.509,
1011
+ "eval_steps_per_second": 2.44,
1012
+ "eval_wer": 0.17677279558242756,
1013
+ "step": 26800
1014
+ },
1015
+ {
1016
+ "epoch": 24.55,
1017
+ "learning_rate": 0.00021930356782906668,
1018
+ "loss": 0.1551,
1019
+ "step": 27200
1020
+ },
1021
+ {
1022
+ "epoch": 24.55,
1023
+ "eval_loss": 0.15309767425060272,
1024
+ "eval_runtime": 203.5071,
1025
+ "eval_samples_per_second": 19.444,
1026
+ "eval_steps_per_second": 2.432,
1027
+ "eval_wer": 0.1685946947191836,
1028
+ "step": 27200
1029
+ },
1030
+ {
1031
+ "epoch": 24.91,
1032
+ "learning_rate": 0.00021809413424712758,
1033
+ "loss": 0.1593,
1034
+ "step": 27600
1035
+ },
1036
+ {
1037
+ "epoch": 24.91,
1038
+ "eval_loss": 0.1611892282962799,
1039
+ "eval_runtime": 203.404,
1040
+ "eval_samples_per_second": 19.454,
1041
+ "eval_steps_per_second": 2.434,
1042
+ "eval_wer": 0.181560828993814,
1043
+ "step": 27600
1044
+ },
1045
+ {
1046
+ "epoch": 25.27,
1047
+ "learning_rate": 0.00021688470066518845,
1048
+ "loss": 0.1519,
1049
+ "step": 28000
1050
+ },
1051
+ {
1052
+ "epoch": 25.27,
1053
+ "eval_loss": 0.16373491287231445,
1054
+ "eval_runtime": 203.251,
1055
+ "eval_samples_per_second": 19.469,
1056
+ "eval_steps_per_second": 2.435,
1057
+ "eval_wer": 0.17638835494355712,
1058
+ "step": 28000
1059
+ },
1060
+ {
1061
+ "epoch": 25.63,
1062
+ "learning_rate": 0.0002156752670832493,
1063
+ "loss": 0.1509,
1064
+ "step": 28400
1065
+ },
1066
+ {
1067
+ "epoch": 25.63,
1068
+ "eval_loss": 0.15399527549743652,
1069
+ "eval_runtime": 203.0998,
1070
+ "eval_samples_per_second": 19.483,
1071
+ "eval_steps_per_second": 2.437,
1072
+ "eval_wer": 0.17093628770139446,
1073
+ "step": 28400
1074
+ },
1075
+ {
1076
+ "epoch": 25.99,
1077
+ "learning_rate": 0.0002144658335013102,
1078
+ "loss": 0.1524,
1079
+ "step": 28800
1080
+ },
1081
+ {
1082
+ "epoch": 25.99,
1083
+ "eval_loss": 0.1519121676683426,
1084
+ "eval_runtime": 203.5041,
1085
+ "eval_samples_per_second": 19.444,
1086
+ "eval_steps_per_second": 2.432,
1087
+ "eval_wer": 0.18072204941809666,
1088
+ "step": 28800
1089
+ },
1090
+ {
1091
+ "epoch": 26.35,
1092
+ "learning_rate": 0.00021325639991937108,
1093
+ "loss": 0.1452,
1094
+ "step": 29200
1095
+ },
1096
+ {
1097
+ "epoch": 26.35,
1098
+ "eval_loss": 0.15314562618732452,
1099
+ "eval_runtime": 203.4098,
1100
+ "eval_samples_per_second": 19.453,
1101
+ "eval_steps_per_second": 2.434,
1102
+ "eval_wer": 0.17310313493866425,
1103
+ "step": 29200
1104
+ },
1105
+ {
1106
+ "epoch": 26.71,
1107
+ "learning_rate": 0.0002120499899213868,
1108
+ "loss": 0.1485,
1109
+ "step": 29600
1110
+ },
1111
+ {
1112
+ "epoch": 26.71,
1113
+ "eval_loss": 0.14827923476696014,
1114
+ "eval_runtime": 203.2718,
1115
+ "eval_samples_per_second": 19.467,
1116
+ "eval_steps_per_second": 2.435,
1117
+ "eval_wer": 0.16136022087862162,
1118
+ "step": 29600
1119
+ },
1120
+ {
1121
+ "epoch": 27.08,
1122
+ "learning_rate": 0.0002108405563394477,
1123
+ "loss": 0.1513,
1124
+ "step": 30000
1125
+ },
1126
+ {
1127
+ "epoch": 27.08,
1128
+ "eval_loss": 0.15502476692199707,
1129
+ "eval_runtime": 203.5744,
1130
+ "eval_samples_per_second": 19.438,
1131
+ "eval_steps_per_second": 2.432,
1132
+ "eval_wer": 0.16579876280012582,
1133
+ "step": 30000
1134
+ },
1135
+ {
1136
+ "epoch": 27.44,
1137
+ "learning_rate": 0.00020963112275750853,
1138
+ "loss": 0.1437,
1139
+ "step": 30400
1140
+ },
1141
+ {
1142
+ "epoch": 27.44,
1143
+ "eval_loss": 0.15631796419620514,
1144
+ "eval_runtime": 204.0627,
1145
+ "eval_samples_per_second": 19.391,
1146
+ "eval_steps_per_second": 2.426,
1147
+ "eval_wer": 0.16614825429000804,
1148
+ "step": 30400
1149
+ },
1150
+ {
1151
+ "epoch": 27.8,
1152
+ "learning_rate": 0.00020842168917556943,
1153
+ "loss": 0.1445,
1154
+ "step": 30800
1155
+ },
1156
+ {
1157
+ "epoch": 27.8,
1158
+ "eval_loss": 0.15840862691402435,
1159
+ "eval_runtime": 204.5713,
1160
+ "eval_samples_per_second": 19.343,
1161
+ "eval_steps_per_second": 2.42,
1162
+ "eval_wer": 0.17086638940341803,
1163
+ "step": 30800
1164
+ },
1165
+ {
1166
+ "epoch": 28.16,
1167
+ "learning_rate": 0.0002072122555936303,
1168
+ "loss": 0.1438,
1169
+ "step": 31200
1170
+ },
1171
+ {
1172
+ "epoch": 28.16,
1173
+ "eval_loss": 0.15680429339408875,
1174
+ "eval_runtime": 204.0022,
1175
+ "eval_samples_per_second": 19.397,
1176
+ "eval_steps_per_second": 2.426,
1177
+ "eval_wer": 0.1630727291790445,
1178
+ "step": 31200
1179
+ },
1180
+ {
1181
+ "epoch": 28.52,
1182
+ "learning_rate": 0.00020600282201169116,
1183
+ "loss": 0.1423,
1184
+ "step": 31600
1185
+ },
1186
+ {
1187
+ "epoch": 28.52,
1188
+ "eval_loss": 0.15523846447467804,
1189
+ "eval_runtime": 203.5692,
1190
+ "eval_samples_per_second": 19.438,
1191
+ "eval_steps_per_second": 2.432,
1192
+ "eval_wer": 0.16090588194177471,
1193
+ "step": 31600
1194
+ },
1195
+ {
1196
+ "epoch": 28.88,
1197
+ "learning_rate": 0.00020479338842975206,
1198
+ "loss": 0.1451,
1199
+ "step": 32000
1200
+ },
1201
+ {
1202
+ "epoch": 28.88,
1203
+ "eval_loss": 0.1542510688304901,
1204
+ "eval_runtime": 203.7413,
1205
+ "eval_samples_per_second": 19.422,
1206
+ "eval_steps_per_second": 2.43,
1207
+ "eval_wer": 0.16314262747702094,
1208
+ "step": 32000
1209
+ },
1210
+ {
1211
+ "epoch": 29.24,
1212
+ "learning_rate": 0.00020358395484781292,
1213
+ "loss": 0.1396,
1214
+ "step": 32400
1215
+ },
1216
+ {
1217
+ "epoch": 29.24,
1218
+ "eval_loss": 0.15091465413570404,
1219
+ "eval_runtime": 203.6781,
1220
+ "eval_samples_per_second": 19.428,
1221
+ "eval_steps_per_second": 2.43,
1222
+ "eval_wer": 0.16317757662600915,
1223
+ "step": 32400
1224
+ },
1225
+ {
1226
+ "epoch": 29.6,
1227
+ "learning_rate": 0.0002023745212658738,
1228
+ "loss": 0.142,
1229
+ "step": 32800
1230
+ },
1231
+ {
1232
+ "epoch": 29.6,
1233
+ "eval_loss": 0.15581683814525604,
1234
+ "eval_runtime": 203.4725,
1235
+ "eval_samples_per_second": 19.447,
1236
+ "eval_steps_per_second": 2.433,
1237
+ "eval_wer": 0.1635620172648796,
1238
+ "step": 32800
1239
+ },
1240
+ {
1241
+ "epoch": 29.96,
1242
+ "learning_rate": 0.0002011650876839347,
1243
+ "loss": 0.1385,
1244
+ "step": 33200
1245
+ },
1246
+ {
1247
+ "epoch": 29.96,
1248
+ "eval_loss": 0.145892933011055,
1249
+ "eval_runtime": 203.8769,
1250
+ "eval_samples_per_second": 19.409,
1251
+ "eval_steps_per_second": 2.428,
1252
+ "eval_wer": 0.17338272813057,
1253
+ "step": 33200
1254
+ },
1255
+ {
1256
+ "epoch": 30.32,
1257
+ "learning_rate": 0.00019995565410199556,
1258
+ "loss": 0.1376,
1259
+ "step": 33600
1260
+ },
1261
+ {
1262
+ "epoch": 30.32,
1263
+ "eval_loss": 0.15049244463443756,
1264
+ "eval_runtime": 203.9232,
1265
+ "eval_samples_per_second": 19.404,
1266
+ "eval_steps_per_second": 2.427,
1267
+ "eval_wer": 0.17236920280991158,
1268
+ "step": 33600
1269
+ },
1270
+ {
1271
+ "epoch": 30.69,
1272
+ "learning_rate": 0.00019874924410401127,
1273
+ "loss": 0.1399,
1274
+ "step": 34000
1275
+ },
1276
+ {
1277
+ "epoch": 30.69,
1278
+ "eval_loss": 0.14067970216274261,
1279
+ "eval_runtime": 203.8686,
1280
+ "eval_samples_per_second": 19.41,
1281
+ "eval_steps_per_second": 2.428,
1282
+ "eval_wer": 0.1610107293887394,
1283
+ "step": 34000
1284
+ },
1285
+ {
1286
+ "epoch": 31.05,
1287
+ "learning_rate": 0.00019753981052207217,
1288
+ "loss": 0.1377,
1289
+ "step": 34400
1290
+ },
1291
+ {
1292
+ "epoch": 31.05,
1293
+ "eval_loss": 0.1523168534040451,
1294
+ "eval_runtime": 203.6263,
1295
+ "eval_samples_per_second": 19.433,
1296
+ "eval_steps_per_second": 2.431,
1297
+ "eval_wer": 0.17292838919372314,
1298
+ "step": 34400
1299
+ },
1300
+ {
1301
+ "epoch": 31.41,
1302
+ "learning_rate": 0.000196330376940133,
1303
+ "loss": 0.1331,
1304
+ "step": 34800
1305
+ },
1306
+ {
1307
+ "epoch": 31.41,
1308
+ "eval_loss": 0.14309507608413696,
1309
+ "eval_runtime": 204.8423,
1310
+ "eval_samples_per_second": 19.317,
1311
+ "eval_steps_per_second": 2.416,
1312
+ "eval_wer": 0.1706566945094887,
1313
+ "step": 34800
1314
+ },
1315
+ {
1316
+ "epoch": 31.77,
1317
+ "learning_rate": 0.0001951209433581939,
1318
+ "loss": 0.1356,
1319
+ "step": 35200
1320
+ },
1321
+ {
1322
+ "epoch": 31.77,
1323
+ "eval_loss": 0.1558520793914795,
1324
+ "eval_runtime": 204.495,
1325
+ "eval_samples_per_second": 19.35,
1326
+ "eval_steps_per_second": 2.421,
1327
+ "eval_wer": 0.16391150875476182,
1328
+ "step": 35200
1329
+ },
1330
+ {
1331
+ "epoch": 32.13,
1332
+ "learning_rate": 0.00019391150977625477,
1333
+ "loss": 0.1381,
1334
+ "step": 35600
1335
+ },
1336
+ {
1337
+ "epoch": 32.13,
1338
+ "eval_loss": 0.14920052886009216,
1339
+ "eval_runtime": 204.3124,
1340
+ "eval_samples_per_second": 19.367,
1341
+ "eval_steps_per_second": 2.423,
1342
+ "eval_wer": 0.1660783559920316,
1343
+ "step": 35600
1344
+ },
1345
+ {
1346
+ "epoch": 32.49,
1347
+ "learning_rate": 0.00019270207619431564,
1348
+ "loss": 0.1332,
1349
+ "step": 36000
1350
+ },
1351
+ {
1352
+ "epoch": 32.49,
1353
+ "eval_loss": 0.15177668631076813,
1354
+ "eval_runtime": 204.8291,
1355
+ "eval_samples_per_second": 19.319,
1356
+ "eval_steps_per_second": 2.417,
1357
+ "eval_wer": 0.16139517002760984,
1358
+ "step": 36000
1359
+ },
1360
+ {
1361
+ "epoch": 32.85,
1362
+ "learning_rate": 0.00019149264261237654,
1363
+ "loss": 0.1345,
1364
+ "step": 36400
1365
+ },
1366
+ {
1367
+ "epoch": 32.85,
1368
+ "eval_loss": 0.14764094352722168,
1369
+ "eval_runtime": 204.5031,
1370
+ "eval_samples_per_second": 19.349,
1371
+ "eval_steps_per_second": 2.421,
1372
+ "eval_wer": 0.1562925942753294,
1373
+ "step": 36400
1374
+ },
1375
+ {
1376
+ "epoch": 33.21,
1377
+ "learning_rate": 0.00019028320903043738,
1378
+ "loss": 0.1307,
1379
+ "step": 36800
1380
+ },
1381
+ {
1382
+ "epoch": 33.21,
1383
+ "eval_loss": 0.1433647871017456,
1384
+ "eval_runtime": 205.0597,
1385
+ "eval_samples_per_second": 19.297,
1386
+ "eval_steps_per_second": 2.414,
1387
+ "eval_wer": 0.15922832279034005,
1388
+ "step": 36800
1389
+ },
1390
+ {
1391
+ "epoch": 33.57,
1392
+ "learning_rate": 0.00018907679903245312,
1393
+ "loss": 0.1309,
1394
+ "step": 37200
1395
+ },
1396
+ {
1397
+ "epoch": 33.57,
1398
+ "eval_loss": 0.1497233361005783,
1399
+ "eval_runtime": 204.4523,
1400
+ "eval_samples_per_second": 19.354,
1401
+ "eval_steps_per_second": 2.421,
1402
+ "eval_wer": 0.17090133855240625,
1403
+ "step": 37200
1404
+ },
1405
+ {
1406
+ "epoch": 33.93,
1407
+ "learning_rate": 0.00018786736545051402,
1408
+ "loss": 0.1319,
1409
+ "step": 37600
1410
+ },
1411
+ {
1412
+ "epoch": 33.93,
1413
+ "eval_loss": 0.15207722783088684,
1414
+ "eval_runtime": 204.6556,
1415
+ "eval_samples_per_second": 19.335,
1416
+ "eval_steps_per_second": 2.419,
1417
+ "eval_wer": 0.1708314402544298,
1418
+ "step": 37600
1419
+ },
1420
+ {
1421
+ "epoch": 34.3,
1422
+ "learning_rate": 0.00018665793186857486,
1423
+ "loss": 0.1279,
1424
+ "step": 38000
1425
+ },
1426
+ {
1427
+ "epoch": 34.3,
1428
+ "eval_loss": 0.14533580839633942,
1429
+ "eval_runtime": 205.2587,
1430
+ "eval_samples_per_second": 19.278,
1431
+ "eval_steps_per_second": 2.412,
1432
+ "eval_wer": 0.1569566281061056,
1433
+ "step": 38000
1434
+ },
1435
+ {
1436
+ "epoch": 34.66,
1437
+ "learning_rate": 0.00018544849828663575,
1438
+ "loss": 0.1268,
1439
+ "step": 38400
1440
+ },
1441
+ {
1442
+ "epoch": 34.66,
1443
+ "eval_loss": 0.14409175515174866,
1444
+ "eval_runtime": 204.8962,
1445
+ "eval_samples_per_second": 19.312,
1446
+ "eval_steps_per_second": 2.416,
1447
+ "eval_wer": 0.1608709327927865,
1448
+ "step": 38400
1449
+ },
1450
+ {
1451
+ "epoch": 35.02,
1452
+ "learning_rate": 0.0001842390647046966,
1453
+ "loss": 0.1301,
1454
+ "step": 38800
1455
+ },
1456
+ {
1457
+ "epoch": 35.02,
1458
+ "eval_loss": 0.1421143114566803,
1459
+ "eval_runtime": 205.2147,
1460
+ "eval_samples_per_second": 19.282,
1461
+ "eval_steps_per_second": 2.412,
1462
+ "eval_wer": 0.1567818823611645,
1463
+ "step": 38800
1464
+ },
1465
+ {
1466
+ "epoch": 35.38,
1467
+ "learning_rate": 0.0001830296311227575,
1468
+ "loss": 0.1248,
1469
+ "step": 39200
1470
+ },
1471
+ {
1472
+ "epoch": 35.38,
1473
+ "eval_loss": 0.14348700642585754,
1474
+ "eval_runtime": 209.2746,
1475
+ "eval_samples_per_second": 18.908,
1476
+ "eval_steps_per_second": 2.365,
1477
+ "eval_wer": 0.15727117044699962,
1478
+ "step": 39200
1479
+ },
1480
+ {
1481
+ "epoch": 35.74,
1482
+ "learning_rate": 0.00018182322112477323,
1483
+ "loss": 0.1256,
1484
+ "step": 39600
1485
+ },
1486
+ {
1487
+ "epoch": 35.74,
1488
+ "eval_loss": 0.14100748300552368,
1489
+ "eval_runtime": 214.9156,
1490
+ "eval_samples_per_second": 18.412,
1491
+ "eval_steps_per_second": 2.303,
1492
+ "eval_wer": 0.1528326285254954,
1493
+ "step": 39600
1494
+ }
1495
+ ],
1496
+ "max_steps": 99720,
1497
+ "num_train_epochs": 90,
1498
+ "total_flos": 1.728945548938538e+20,
1499
+ "trial_name": null,
1500
+ "trial_params": null
1501
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a14684982016a59a1ac5f3e95d14e445c0af8431dab5efc04d3bde3ec2b1ae6e
3
+ size 2799
vocab.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"a": 1, "b": 2, "c": 3, "d": 4, "e": 5, "f": 6, "g": 7, "h": 8, "i": 9, "j": 10, "k": 11, "l": 12, "m": 13, "n": 14, "o": 15, "p": 16, "q": 17, "r": 18, "s": 19, "t": 20, "u": 21, "v": 22, "w": 23, "x": 24, "y": 25, "z": 26, "\u00e7": 27, "\u00eb": 28, "\u00f6": 29, "\u00fc": 30, "\u011f": 31, "\u0131": 32, "\u015f": 33, "\u0307": 34, "|": 0, "[UNK]": 35, "[PAD]": 36}