“siddhu001” commited on
Commit
bbb5a3a
·
1 Parent(s): 1345d8b

Update model

Browse files
Files changed (22) hide show
  1. README.md +1380 -0
  2. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/RESULTS.md +65 -0
  3. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/config.yaml +1241 -0
  4. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/acc.png +0 -0
  5. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/backward_time.png +0 -0
  6. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/cer.png +0 -0
  7. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/cer_ctc.png +0 -0
  8. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/clip.png +0 -0
  9. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/forward_time.png +0 -0
  10. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/gpu_max_cached_mem_GB.png +0 -0
  11. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/grad_norm.png +0 -0
  12. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/iter_time.png +0 -0
  13. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss.png +0 -0
  14. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss_att.png +0 -0
  15. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss_ctc.png +0 -0
  16. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss_scale.png +0 -0
  17. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/optim0_lr0.png +0 -0
  18. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/optim_step_time.png +0 -0
  19. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/train_time.png +0 -0
  20. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/wer.png +0 -0
  21. exp/slu_train_asr_owsm_weighted_raw_en_word_sp/valid.acc.ave_10best.pth +3 -0
  22. meta.yaml +8 -0
README.md ADDED
@@ -0,0 +1,1380 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - espnet
4
+ - audio
5
+ - automatic-speech-recognition
6
+ language: en
7
+ datasets:
8
+ - slue-voxceleb
9
+ license: cc-by-4.0
10
+ ---
11
+
12
+ ## ESPnet2 ASR model
13
+
14
+ ### `espnet/sluevoxceleb_owsm_complex_slu`
15
+
16
+ This model was trained by “siddhu001” using slue-voxceleb recipe in [espnet](https://github.com/espnet/espnet/).
17
+
18
+ ### Demo: How to use in ESPnet2
19
+
20
+ Follow the [ESPnet installation instructions](https://espnet.github.io/espnet/installation.html)
21
+ if you haven't done that already.
22
+
23
+ ```bash
24
+ cd espnet
25
+ git checkout e23ef85f0b3116ad5c60d0833f186da0deec0734
26
+ pip install -e .
27
+ cd egs2/slue-voxceleb/slu1_correct
28
+ ./run.sh --skip_data_prep false --skip_train true --download_model espnet/sluevoxceleb_owsm_complex_slu
29
+ ```
30
+
31
+ <!-- Generated by scripts/utils/show_asr_result.sh -->
32
+ # RESULTS
33
+ ## Environments
34
+ - date: `Sat Feb 10 22:23:15 CST 2024`
35
+ - python version: `3.9.13 (main, Aug 25 2022, 23:26:10) [GCC 11.2.0]`
36
+ - espnet version: `espnet 202310`
37
+ - pytorch version: `pytorch 2.1.0+cu121`
38
+ - Git hash: `21d2105784e4da98397bf487b2550d4c6e16d40d`
39
+ - Commit date: `Wed Jan 31 13:40:37 2024 -0600`
40
+
41
+ ## exp/slu_train_asr_owsm_weighted_raw_en_word_sp
42
+ ### WER
43
+
44
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
45
+ |---|---|---|---|---|---|---|---|---|
46
+ |decode_asr_ctc0.3_beam10_slu_model_valid.acc.ave_10best/test|3530|144908|85.7|8.9|5.4|2.7|17.0|95.1|
47
+ |decode_asr_slu_model_valid.acc.ave_10best/devel|1450|58104|85.2|8.1|6.7|2.7|17.6|93.2|
48
+ |decode_asr_slu_model_valid.acc.ave_10best/test|3530|144908|82.9|9.9|7.2|3.3|20.4|95.7|
49
+
50
+ ### CER
51
+
52
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
53
+ |---|---|---|---|---|---|---|---|---|
54
+ |decode_asr_ctc0.3_beam10_slu_model_valid.acc.ave_10best/test|3530|647097|92.9|2.6|4.5|2.7|9.8|95.1|
55
+ |decode_asr_slu_model_valid.acc.ave_10best/devel|1450|256305|91.9|2.4|5.7|2.7|10.8|93.2|
56
+ |decode_asr_slu_model_valid.acc.ave_10best/test|3530|647097|90.7|3.0|6.3|3.1|12.4|95.7|
57
+
58
+ ### TER
59
+
60
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
61
+ |---|---|---|---|---|---|---|---|---|
62
+ ## exp/slu_train_asr_owsm_weighted_raw_en_word_sp/decode_asr_ctc0.3_beam10_slu_model_valid.acc.ave_10best
63
+ ### WER
64
+
65
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
66
+ |---|---|---|---|---|---|---|---|---|
67
+ |org/devel|1451|58267|87.8|7.3|4.8|2.2|14.3|92.7|
68
+
69
+ ### CER
70
+
71
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
72
+ |---|---|---|---|---|---|---|---|---|
73
+ |org/devel|1451|256942|94.0|2.2|3.9|2.2|8.2|92.7|
74
+
75
+ ### TER
76
+
77
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
78
+ |---|---|---|---|---|---|---|---|---|
79
+ ## exp/slu_train_asr_owsm_weighted_raw_en_word_sp/decode_asr_slu_model_valid.acc.ave_10best
80
+ ### WER
81
+
82
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
83
+ |---|---|---|---|---|---|---|---|---|
84
+ |org/devel|1451|58267|85.2|8.1|6.7|2.7|17.6|93.2|
85
+
86
+ ### CER
87
+
88
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
89
+ |---|---|---|---|---|---|---|---|---|
90
+ |org/devel|1451|256942|91.8|2.4|5.7|2.7|10.8|93.2|
91
+
92
+ ### TER
93
+
94
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
95
+ |---|---|---|---|---|---|---|---|---|
96
+
97
+ ## ASR config
98
+
99
+ <details><summary>expand</summary>
100
+
101
+ ```
102
+ config: conf/tuning/train_asr_owsm_weighted.yaml
103
+ print_config: false
104
+ log_level: INFO
105
+ drop_last_iter: false
106
+ dry_run: false
107
+ iterator_type: sequence
108
+ valid_iterator_type: null
109
+ output_dir: exp/slu_train_asr_owsm_weighted_raw_en_word_sp
110
+ ngpu: 1
111
+ seed: 2022
112
+ num_workers: 2
113
+ num_att_plot: 3
114
+ dist_backend: nccl
115
+ dist_init_method: env://
116
+ dist_world_size: 4
117
+ dist_rank: 0
118
+ local_rank: 0
119
+ dist_master_addr: localhost
120
+ dist_master_port: 52077
121
+ dist_launcher: null
122
+ multiprocessing_distributed: true
123
+ unused_parameters: false
124
+ sharded_ddp: false
125
+ cudnn_enabled: true
126
+ cudnn_benchmark: false
127
+ cudnn_deterministic: true
128
+ collect_stats: false
129
+ write_collected_feats: false
130
+ max_epoch: 70
131
+ patience: null
132
+ val_scheduler_criterion:
133
+ - valid
134
+ - loss
135
+ early_stopping_criterion:
136
+ - valid
137
+ - loss
138
+ - min
139
+ best_model_criterion:
140
+ - - valid
141
+ - acc
142
+ - max
143
+ keep_nbest_models: 10
144
+ nbest_averaging_interval: 0
145
+ grad_clip: 5.0
146
+ grad_clip_type: 2.0
147
+ grad_noise: false
148
+ accum_grad: 2
149
+ no_forward_run: false
150
+ resume: true
151
+ train_dtype: float32
152
+ use_amp: false
153
+ log_interval: null
154
+ use_matplotlib: true
155
+ use_tensorboard: true
156
+ create_graph_in_tensorboard: false
157
+ use_wandb: false
158
+ wandb_project: null
159
+ wandb_id: null
160
+ wandb_entity: null
161
+ wandb_name: null
162
+ wandb_model_log_interval: -1
163
+ detect_anomaly: false
164
+ use_lora: false
165
+ save_lora_only: true
166
+ lora_conf: {}
167
+ pretrain_path: null
168
+ init_param:
169
+ - /scratch/bbjs/arora1/new_download_espnet_egs2/harpervalley/slu1_superb_onlyda/owsm_v3.1_ebf/exp/s2t_train_s2t_ebf_conv2d_size1024_e18_d18_piecewise_lr2e-4_warmup60k_flashattn_raw_bpe50000/valid.total_count.ave_5best.till45epoch.pth:encoder:encoder
170
+ ignore_init_mismatch: false
171
+ freeze_param:
172
+ - encoder
173
+ num_iters_per_epoch: null
174
+ batch_size: 20
175
+ valid_batch_size: null
176
+ batch_bins: 6000000
177
+ valid_batch_bins: null
178
+ train_shape_file:
179
+ - exp/slu_stats_raw_en_word_sp/train/speech_shape
180
+ - exp/slu_stats_raw_en_word_sp/train/text_shape.word
181
+ valid_shape_file:
182
+ - exp/slu_stats_raw_en_word_sp/valid/speech_shape
183
+ - exp/slu_stats_raw_en_word_sp/valid/text_shape.word
184
+ batch_type: numel
185
+ valid_batch_type: null
186
+ fold_length:
187
+ - 80000
188
+ - 150
189
+ sort_in_batch: descending
190
+ shuffle_within_batch: false
191
+ sort_batch: descending
192
+ multiple_iterator: false
193
+ chunk_length: 500
194
+ chunk_shift_ratio: 0.5
195
+ num_cache_chunks: 1024
196
+ chunk_excluded_key_prefixes: []
197
+ chunk_default_fs: null
198
+ train_data_path_and_name_and_type:
199
+ - - dump/raw/train_sp/wav.scp
200
+ - speech
201
+ - sound
202
+ - - dump/raw/train_sp/text
203
+ - text
204
+ - text
205
+ valid_data_path_and_name_and_type:
206
+ - - dump/raw/devel/wav.scp
207
+ - speech
208
+ - sound
209
+ - - dump/raw/devel/text
210
+ - text
211
+ - text
212
+ allow_variable_data_keys: false
213
+ max_cache_size: 0.0
214
+ max_cache_fd: 32
215
+ allow_multi_rates: false
216
+ valid_max_cache_size: null
217
+ exclude_weight_decay: false
218
+ exclude_weight_decay_conf: {}
219
+ optim: adam
220
+ optim_conf:
221
+ lr: 0.002
222
+ weight_decay: 1.0e-06
223
+ scheduler: warmuplr
224
+ scheduler_conf:
225
+ warmup_steps: 5000
226
+ token_list:
227
+ - <blank>
228
+ - <unk>
229
+ - ▁i
230
+ - ▁and
231
+ - ''''
232
+ - s
233
+ - ▁the
234
+ - ▁a
235
+ - ▁it
236
+ - Neutral
237
+ - ▁to
238
+ - ▁you
239
+ - ▁that
240
+ - ▁of
241
+ - ▁in
242
+ - ▁was
243
+ - ▁uh
244
+ - ▁know
245
+ - t
246
+ - ▁so
247
+ - ▁we
248
+ - ▁he
249
+ - ing
250
+ - ▁um
251
+ - ed
252
+ - m
253
+ - ▁like
254
+ - ▁is
255
+ - ▁but
256
+ - Positive
257
+ - y
258
+ - ▁just
259
+ - ▁they
260
+ - re
261
+ - ▁this
262
+ - ▁for
263
+ - ▁be
264
+ - ▁my
265
+ - er
266
+ - ▁with
267
+ - ▁on
268
+ - ▁think
269
+ - ▁p
270
+ - ▁have
271
+ - ▁she
272
+ - e
273
+ - ▁me
274
+ - ▁really
275
+ - ▁there
276
+ - ▁what
277
+ - ▁m
278
+ - a
279
+ - ▁do
280
+ - ▁all
281
+ - i
282
+ - al
283
+ - ve
284
+ - c
285
+ - ▁as
286
+ - ▁about
287
+ - ▁not
288
+ - ▁t
289
+ - n
290
+ - ▁at
291
+ - l
292
+ - ▁had
293
+ - ▁b
294
+ - ▁when
295
+ - ▁c
296
+ - g
297
+ - ar
298
+ - ▁out
299
+ - en
300
+ - ▁s
301
+ - ▁an
302
+ - ▁people
303
+ - or
304
+ - an
305
+ - d
306
+ - o
307
+ - ll
308
+ - ▁are
309
+ - in
310
+ - ▁very
311
+ - p
312
+ - b
313
+ - u
314
+ - ▁because
315
+ - es
316
+ - ▁can
317
+ - ▁don
318
+ - ▁or
319
+ - ▁up
320
+ - it
321
+ - ▁one
322
+ - ly
323
+ - ▁if
324
+ - ▁f
325
+ - st
326
+ - ▁were
327
+ - ▁mean
328
+ - ▁d
329
+ - ▁who
330
+ - ▁then
331
+ - ic
332
+ - 'on'
333
+ - ▁no
334
+ - ▁go
335
+ - ▁her
336
+ - ▁g
337
+ - ent
338
+ - ▁st
339
+ - ▁kind
340
+ - ri
341
+ - ▁would
342
+ - ▁get
343
+ - ▁e
344
+ - le
345
+ - at
346
+ - r
347
+ - ▁time
348
+ - ▁w
349
+ - ▁re
350
+ - h
351
+ - ▁from
352
+ - ▁l
353
+ - ▁said
354
+ - ▁him
355
+ - ▁how
356
+ - v
357
+ - ▁well
358
+ - ▁h
359
+ - ▁gonna
360
+ - ▁lot
361
+ - ▁see
362
+ - f
363
+ - ▁his
364
+ - et
365
+ - ion
366
+ - ▁been
367
+ - ▁great
368
+ - ▁yeah
369
+ - ▁love
370
+ - ▁which
371
+ - ▁got
372
+ - k
373
+ - ▁them
374
+ - ▁way
375
+ - id
376
+ - ▁show
377
+ - w
378
+ - ▁some
379
+ - ▁your
380
+ - ▁did
381
+ - ▁sort
382
+ - ▁has
383
+ - ▁things
384
+ - ▁back
385
+ - ▁where
386
+ - ▁something
387
+ - ir
388
+ - ▁thing
389
+ - ad
390
+ - ▁su
391
+ - ▁ch
392
+ - ▁n
393
+ - il
394
+ - as
395
+ - ▁j
396
+ - ▁more
397
+ - se
398
+ - ▁say
399
+ - ▁co
400
+ - nd
401
+ - ▁much
402
+ - ▁always
403
+ - ine
404
+ - ▁r
405
+ - ation
406
+ - ur
407
+ - ▁other
408
+ - th
409
+ - ▁
410
+ - ▁se
411
+ - ▁now
412
+ - ate
413
+ - ▁doing
414
+ - ▁work
415
+ - ow
416
+ - ▁could
417
+ - ally
418
+ - ▁these
419
+ - Negative
420
+ - ▁good
421
+ - ▁any
422
+ - ers
423
+ - ce
424
+ - ▁cause
425
+ - ▁ex
426
+ - ▁pro
427
+ - ▁little
428
+ - ▁actually
429
+ - ▁into
430
+ - ▁make
431
+ - ▁first
432
+ - ▁being
433
+ - ra
434
+ - ▁our
435
+ - ▁al
436
+ - ▁by
437
+ - ▁film
438
+ - ▁didn
439
+ - ▁v
440
+ - ct
441
+ - ity
442
+ - ch
443
+ - un
444
+ - ▁part
445
+ - ▁de
446
+ - ▁come
447
+ - is
448
+ - ie
449
+ - ▁right
450
+ - ▁o
451
+ - ▁off
452
+ - ol
453
+ - ▁two
454
+ - ▁never
455
+ - ▁le
456
+ - ot
457
+ - ut
458
+ - ▁movie
459
+ - ▁play
460
+ - ge
461
+ - ies
462
+ - el
463
+ - ▁con
464
+ - am
465
+ - ▁going
466
+ - ke
467
+ - ▁want
468
+ - im
469
+ - ▁feel
470
+ - ive
471
+ - ro
472
+ - ▁mo
473
+ - ▁different
474
+ - ck
475
+ - ▁life
476
+ - ist
477
+ - ▁oh
478
+ - all
479
+ - ▁lo
480
+ - ard
481
+ - ▁went
482
+ - and
483
+ - ▁sh
484
+ - ▁even
485
+ - ry
486
+ - ▁years
487
+ - ▁look
488
+ - ▁us
489
+ - ant
490
+ - ▁te
491
+ - ▁k
492
+ - ▁li
493
+ - ▁happen
494
+ - ure
495
+ - ▁their
496
+ - ▁those
497
+ - ▁take
498
+ - ment
499
+ - ▁day
500
+ - ble
501
+ - ast
502
+ - ▁every
503
+ - um
504
+ - ill
505
+ - op
506
+ - ▁thought
507
+ - ou
508
+ - us
509
+ - ay
510
+ - ▁th
511
+ - ▁put
512
+ - ▁story
513
+ - ▁new
514
+ - ▁down
515
+ - ish
516
+ - ▁big
517
+ - ▁wanna
518
+ - ▁ro
519
+ - ▁also
520
+ - ▁read
521
+ - ▁around
522
+ - ous
523
+ - ▁through
524
+ - red
525
+ - ▁came
526
+ - ▁character
527
+ - ess
528
+ - te
529
+ - ver
530
+ - ▁will
531
+ - ag
532
+ - ss
533
+ - ▁fun
534
+ - ▁over
535
+ - ▁many
536
+ - ▁bl
537
+ - ▁cl
538
+ - ▁man
539
+ - ▁than
540
+ - ▁pre
541
+ - ▁world
542
+ - ▁person
543
+ - z
544
+ - ▁sp
545
+ - ven
546
+ - ▁wanted
547
+ - ▁bit
548
+ - ▁before
549
+ - ▁mar
550
+ - one
551
+ - ab
552
+ - ▁en
553
+ - ci
554
+ - ▁set
555
+ - ▁ha
556
+ - ▁find
557
+ - ul
558
+ - ▁fi
559
+ - ▁end
560
+ - ▁un
561
+ - ▁sc
562
+ - ▁after
563
+ - ind
564
+ - ter
565
+ - ▁working
566
+ - ▁why
567
+ - om
568
+ - me
569
+ - ▁such
570
+ - ▁whole
571
+ - ▁kinda
572
+ - ne
573
+ - ▁bo
574
+ - x
575
+ - ▁most
576
+ - ▁ad
577
+ - ▁guy
578
+ - ▁spe
579
+ - ars
580
+ - ▁am
581
+ - ful
582
+ - ▁together
583
+ - ▁let
584
+ - ▁quite
585
+ - ain
586
+ - ▁everything
587
+ - ▁made
588
+ - ig
589
+ - ▁old
590
+ - able
591
+ - ▁tr
592
+ - ak
593
+ - ▁fo
594
+ - ▁po
595
+ - ore
596
+ - ice
597
+ - ▁real
598
+ - ▁knew
599
+ - ▁hard
600
+ - pp
601
+ - age
602
+ - ated
603
+ - ▁same
604
+ - ▁start
605
+ - ▁ever
606
+ - ning
607
+ - ▁watch
608
+ - art
609
+ - ▁again
610
+ - ▁here
611
+ - are
612
+ - ght
613
+ - ong
614
+ - ▁done
615
+ - ▁only
616
+ - ▁live
617
+ - ▁wasn
618
+ - ▁ho
619
+ - ▁u
620
+ - ▁maybe
621
+ - ▁need
622
+ - ▁everybody
623
+ - ust
624
+ - ans
625
+ - ▁three
626
+ - ▁having
627
+ - ▁music
628
+ - ack
629
+ - ld
630
+ - ▁trying
631
+ - ▁guys
632
+ - rou
633
+ - ach
634
+ - ving
635
+ - ▁tell
636
+ - ▁should
637
+ - ff
638
+ - ide
639
+ - ▁four
640
+ - ▁started
641
+ - ▁com
642
+ - ass
643
+ - ▁long
644
+ - ▁fe
645
+ - ▁course
646
+ - ▁called
647
+ - ▁own
648
+ - ress
649
+ - ▁moment
650
+ - ▁pl
651
+ - ▁still
652
+ - ▁anything
653
+ - ▁family
654
+ - ▁fin
655
+ - ▁dan
656
+ - ▁bro
657
+ - 'no'
658
+ - ther
659
+ - ▁per
660
+ - ▁amazing
661
+ - ▁stuff
662
+ - per
663
+ - ▁jo
664
+ - ▁certain
665
+ - os
666
+ - ▁talk
667
+ - ater
668
+ - ▁help
669
+ - ▁too
670
+ - ▁year
671
+ - ight
672
+ - ▁fa
673
+ - self
674
+ - ces
675
+ - ▁br
676
+ - ▁bet
677
+ - ▁someone
678
+ - ▁di
679
+ - ▁sing
680
+ - nt
681
+ - ick
682
+ - ▁ph
683
+ - row
684
+ - ▁script
685
+ - ▁remember
686
+ - ▁try
687
+ - qu
688
+ - ite
689
+ - ▁young
690
+ - ▁wh
691
+ - ▁ser
692
+ - ▁ask
693
+ - ▁book
694
+ - ▁each
695
+ - ▁wr
696
+ - ▁best
697
+ - ▁ag
698
+ - ▁women
699
+ - ose
700
+ - ions
701
+ - ved
702
+ - j
703
+ - ue
704
+ - ▁does
705
+ - ▁five
706
+ - ▁both
707
+ - ▁friends
708
+ - ▁act
709
+ - iz
710
+ - cess
711
+ - pt
712
+ - ▁somebody
713
+ - ft
714
+ - ▁nice
715
+ - ▁myself
716
+ - een
717
+ - fe
718
+ - sp
719
+ - ict
720
+ - ty
721
+ - ▁child
722
+ - ud
723
+ - pe
724
+ - ▁hope
725
+ - ▁fact
726
+ - ▁saying
727
+ - ave
728
+ - icul
729
+ - au
730
+ - ale
731
+ - ris
732
+ - ▁twenty
733
+ - ▁school
734
+ - ▁doesn
735
+ - ▁able
736
+ - pect
737
+ - ▁last
738
+ - ber
739
+ - ▁song
740
+ - od
741
+ - ▁str
742
+ - ▁interesting
743
+ - lf
744
+ - ▁em
745
+ - ▁wor
746
+ - ap
747
+ - og
748
+ - ▁ra
749
+ - ▁dis
750
+ - ▁coming
751
+ - ▁ab
752
+ - ▁house
753
+ - ▁next
754
+ - ▁tra
755
+ - ▁okay
756
+ - ere
757
+ - ary
758
+ - ▁incredi
759
+ - ▁car
760
+ - ▁job
761
+ - ▁used
762
+ - ▁give
763
+ - ▁god
764
+ - ▁americ
765
+ - ▁characters
766
+ - ▁app
767
+ - ▁walk
768
+ - ▁yes
769
+ - rew
770
+ - ▁getting
771
+ - ▁six
772
+ - ▁chan
773
+ - ▁ne
774
+ - ▁pretty
775
+ - ang
776
+ - ▁creat
777
+ - ▁another
778
+ - ▁ter
779
+ - ▁kids
780
+ - ▁felt
781
+ - ▁sometimes
782
+ - ▁place
783
+ - out
784
+ - ▁funny
785
+ - ase
786
+ - ich
787
+ - act
788
+ - ▁days
789
+ - ▁hum
790
+ - ▁bring
791
+ - ts
792
+ - ▁making
793
+ - ▁comp
794
+ - ▁become
795
+ - ute
796
+ - ▁wonderful
797
+ - ron
798
+ - les
799
+ - ▁saw
800
+ - ▁point
801
+ - ia
802
+ - ▁realiz
803
+ - ▁int
804
+ - ▁away
805
+ - ays
806
+ - ▁home
807
+ - ace
808
+ - ▁relationship
809
+ - ▁woman
810
+ - ▁everyone
811
+ - ▁comes
812
+ - ▁high
813
+ - dd
814
+ - ▁night
815
+ - ath
816
+ - ▁else
817
+ - vent
818
+ - ▁shoot
819
+ - vers
820
+ - day
821
+ - ▁sure
822
+ - ried
823
+ - ned
824
+ - ▁obviously
825
+ - ▁dra
826
+ - ▁inter
827
+ - co
828
+ - ▁playing
829
+ - ▁important
830
+ - ort
831
+ - uck
832
+ - ision
833
+ - pport
834
+ - ▁seen
835
+ - pl
836
+ - ▁fl
837
+ - ound
838
+ - ▁bas
839
+ - ull
840
+ - est
841
+ - ▁actor
842
+ - ▁lear
843
+ - ▁worked
844
+ - ▁believe
845
+ - ▁gen
846
+ - ▁keep
847
+ - ▁friend
848
+ - ▁sw
849
+ - ▁des
850
+ - ▁times
851
+ - ▁im
852
+ - ▁sur
853
+ - ▁sit
854
+ - ▁probably
855
+ - ok
856
+ - ▁took
857
+ - ep
858
+ - ough
859
+ - ip
860
+ - ood
861
+ - ▁sa
862
+ - ▁season
863
+ - vel
864
+ - wn
865
+ - ▁dec
866
+ - ▁excited
867
+ - ian
868
+ - ire
869
+ - ph
870
+ - ▁month
871
+ - ner
872
+ - ▁min
873
+ - ▁rel
874
+ - ating
875
+ - body
876
+ - ition
877
+ - ▁loved
878
+ - ▁aw
879
+ - ▁hear
880
+ - ple
881
+ - ▁cool
882
+ - ▁y
883
+ - ord
884
+ - our
885
+ - ▁game
886
+ - ms
887
+ - ub
888
+ - ▁might
889
+ - ▁kid
890
+ - ▁movies
891
+ - ical
892
+ - ▁bad
893
+ - ▁scene
894
+ - iv
895
+ - ▁enough
896
+ - ▁sm
897
+ - bly
898
+ - ▁fift
899
+ - ▁eight
900
+ - ▁experience
901
+ - ▁actors
902
+ - ▁cou
903
+ - ▁understand
904
+ - ▁week
905
+ - ▁few
906
+ - gin
907
+ - ting
908
+ - ▁director
909
+ - ▁almost
910
+ - ▁open
911
+ - ren
912
+ - ▁star
913
+ - ▁room
914
+ - ▁call
915
+ - oy
916
+ - ▁goes
917
+ - ▁told
918
+ - ▁once
919
+ - ▁found
920
+ - arly
921
+ - ations
922
+ - ward
923
+ - ▁audience
924
+ - ird
925
+ - if
926
+ - ▁qu
927
+ - ▁ar
928
+ - ▁definitely
929
+ - ious
930
+ - iting
931
+ - ▁pol
932
+ - ▁huge
933
+ - ▁makes
934
+ - aking
935
+ - ream
936
+ - ance
937
+ - be
938
+ - ▁la
939
+ - ▁ac
940
+ - iter
941
+ - ▁run
942
+ - ▁gotta
943
+ - ▁gr
944
+ - ▁cam
945
+ - sh
946
+ - ▁gets
947
+ - ully
948
+ - ▁says
949
+ - ame
950
+ - side
951
+ - ▁bus
952
+ - ▁shows
953
+ - ▁dr
954
+ - ▁inv
955
+ - ▁idea
956
+ - ▁talking
957
+ - ▁wa
958
+ - way
959
+ - ▁art
960
+ - ▁whatever
961
+ - ▁write
962
+ - ash
963
+ - itt
964
+ - ▁met
965
+ - ▁wants
966
+ - ▁role
967
+ - ▁mu
968
+ - ▁boy
969
+ - ▁wrote
970
+ - ger
971
+ - ately
972
+ - ▁exc
973
+ - ▁mother
974
+ - ▁produ
975
+ - ▁cra
976
+ - ates
977
+ - ▁though
978
+ - av
979
+ - ▁episode
980
+ - ▁sl
981
+ - ▁change
982
+ - ▁voice
983
+ - ▁played
984
+ - ily
985
+ - ▁guess
986
+ - ves
987
+ - ▁hand
988
+ - ady
989
+ - ▁happy
990
+ - ith
991
+ - ▁name
992
+ - ny
993
+ - ▁gi
994
+ - ▁looking
995
+ - lev
996
+ - ▁acting
997
+ - aught
998
+ - iss
999
+ - ount
1000
+ - rom
1001
+ - ▁tw
1002
+ - ▁cont
1003
+ - ▁john
1004
+ - ▁far
1005
+ - ▁res
1006
+ - ▁sense
1007
+ - ake
1008
+ - ▁basically
1009
+ - ▁meet
1010
+ - ▁gu
1011
+ - ▁bre
1012
+ - ens
1013
+ - cept
1014
+ - ety
1015
+ - ▁girl
1016
+ - ▁york
1017
+ - ▁count
1018
+ - ▁shot
1019
+ - ise
1020
+ - ject
1021
+ - ▁tot
1022
+ - ▁stud
1023
+ - ▁feels
1024
+ - ▁thinking
1025
+ - ▁head
1026
+ - ▁cast
1027
+ - ▁writing
1028
+ - ▁rehe
1029
+ - ▁written
1030
+ - ▁perform
1031
+ - ▁fan
1032
+ - der
1033
+ - ect
1034
+ - ▁sk
1035
+ - ▁hour
1036
+ - ▁father
1037
+ - ered
1038
+ - ▁hundred
1039
+ - ▁ind
1040
+ - ▁norm
1041
+ - ▁acc
1042
+ - up
1043
+ - ▁while
1044
+ - fort
1045
+ - ▁nin
1046
+ - ▁true
1047
+ - itch
1048
+ - ▁inst
1049
+ - ▁second
1050
+ - ▁pick
1051
+ - ▁record
1052
+ - ross
1053
+ - ▁quest
1054
+ - ged
1055
+ - ▁career
1056
+ - ween
1057
+ - ▁bec
1058
+ - ▁reason
1059
+ - ▁since
1060
+ - ▁bra
1061
+ - ▁char
1062
+ - ▁imp
1063
+ - ree
1064
+ - ▁girls
1065
+ - ▁comple
1066
+ - ▁turn
1067
+ - ▁dad
1068
+ - ▁fant
1069
+ - ▁extra
1070
+ - ▁laugh
1071
+ - ▁stand
1072
+ - ▁honest
1073
+ - ▁comm
1074
+ - na
1075
+ - ▁listen
1076
+ - als
1077
+ - cial
1078
+ - spe
1079
+ - ▁ke
1080
+ - ory
1081
+ - view
1082
+ - ink
1083
+ - ▁direct
1084
+ - reat
1085
+ - round
1086
+ - ien
1087
+ - ▁under
1088
+ - ile
1089
+ - ▁diff
1090
+ - ually
1091
+ - ▁tur
1092
+ - thing
1093
+ - sic
1094
+ - ▁gon
1095
+ - ather
1096
+ - ▁aud
1097
+ - ▁scen
1098
+ - atch
1099
+ - ▁sho
1100
+ - ever
1101
+ - tra
1102
+ - ▁pe
1103
+ - mo
1104
+ - ild
1105
+ - ▁care
1106
+ - int
1107
+ - ▁fam
1108
+ - ▁ob
1109
+ - ▁ide
1110
+ - ade
1111
+ - right
1112
+ - ▁may
1113
+ - he
1114
+ - ody
1115
+ - ense
1116
+ - ▁interest
1117
+ - ah
1118
+ - form
1119
+ - ork
1120
+ - ▁episod
1121
+ - ▁rec
1122
+ - iew
1123
+ - ▁hop
1124
+ - ited
1125
+ - ▁exper
1126
+ - gh
1127
+ - ically
1128
+ - ▁bel
1129
+ - ▁el
1130
+ - enty
1131
+ - ▁gott
1132
+ - ▁stu
1133
+ - ▁id
1134
+ - rie
1135
+ - ▁nor
1136
+ - ▁inc
1137
+ - ertain
1138
+ - tain
1139
+ - ▁wo
1140
+ - ▁mon
1141
+ - az
1142
+ - xt
1143
+ - riend
1144
+ - now
1145
+ - ▁list
1146
+ - ime
1147
+ - ome
1148
+ - so
1149
+ - ause
1150
+ - iously
1151
+ - ▁sch
1152
+ - ▁vo
1153
+ - ▁op
1154
+ - ason
1155
+ - ▁mov
1156
+ - ▁hi
1157
+ - ▁pers
1158
+ - ▁ye
1159
+ - ▁def
1160
+ - orm
1161
+ - ▁belie
1162
+ - fore
1163
+ - ix
1164
+ - mber
1165
+ - very
1166
+ - ▁differe
1167
+ - ▁wonder
1168
+ - ek
1169
+ - nder
1170
+ - ▁obv
1171
+ - ▁ep
1172
+ - ship
1173
+ - ▁lau
1174
+ - ience
1175
+ - ool
1176
+ - ▁sin
1177
+ - rect
1178
+ - ▁happ
1179
+ - ▁gir
1180
+ - du
1181
+ - ng
1182
+ - ▁underst
1183
+ - most
1184
+ - eric
1185
+ - ouse
1186
+ - time
1187
+ - lm
1188
+ - ▁hel
1189
+ - redi
1190
+ - ▁cour
1191
+ - ▁relation
1192
+ - rough
1193
+ - q
1194
+ - ▁defin
1195
+ - ▁prob
1196
+ - ▁reme
1197
+ - ▁hu
1198
+ - ▁fir
1199
+ - anna
1200
+ - ways
1201
+ - itten
1202
+ - elt
1203
+ - ▁sometime
1204
+ - ':'
1205
+ - ▁kne
1206
+ - alk
1207
+ - ▁ok
1208
+ - ably
1209
+ - rote
1210
+ - gether
1211
+ - ▁definite
1212
+ - ▁import
1213
+ - '&'
1214
+ - fter
1215
+ - onest
1216
+ - erest
1217
+ - ▁amaz
1218
+ - ▁ano
1219
+ - <sos/eos>
1220
+ transcript_token_list: null
1221
+ two_pass: false
1222
+ pre_postencoder_norm: false
1223
+ init: null
1224
+ input_size: null
1225
+ ctc_conf:
1226
+ dropout_rate: 0.0
1227
+ ctc_type: builtin
1228
+ reduce: true
1229
+ ignore_nan_grad: null
1230
+ zero_infinity: true
1231
+ brctc_risk_strategy: exp
1232
+ brctc_group_strategy: end
1233
+ brctc_risk_factor: 0.0
1234
+ joint_net_conf: null
1235
+ use_preprocessor: true
1236
+ token_type: word
1237
+ bpemodel: null
1238
+ non_linguistic_symbols: null
1239
+ cleaner: null
1240
+ g2p: null
1241
+ speech_volume_normalize: null
1242
+ rir_scp: null
1243
+ rir_apply_prob: 1.0
1244
+ noise_scp: null
1245
+ noise_apply_prob: 1.0
1246
+ noise_db_range: '13_15'
1247
+ short_noise_thres: 0.5
1248
+ frontend: default
1249
+ frontend_conf:
1250
+ n_fft: 512
1251
+ win_length: 400
1252
+ hop_length: 160
1253
+ fs: 16k
1254
+ specaug: specaug
1255
+ specaug_conf:
1256
+ apply_time_warp: false
1257
+ time_warp_window: 5
1258
+ time_warp_mode: bicubic
1259
+ apply_freq_mask: true
1260
+ freq_mask_width_range:
1261
+ - 0
1262
+ - 27
1263
+ num_freq_mask: 2
1264
+ apply_time_mask: true
1265
+ time_mask_width_ratio_range:
1266
+ - 0.0
1267
+ - 0.05
1268
+ num_time_mask: 10
1269
+ normalize: global_mvn
1270
+ normalize_conf:
1271
+ stats_file: /scratch/bbjs/arora1/new_download_espnet_egs2/harpervalley/slu1_superb_onlyda/owsm_v3.1_ebf/exp/s2t_stats_raw_bpe50000/train/feats_stats.npz
1272
+ model: espnet
1273
+ model_conf:
1274
+ ctc_weight: 0.3
1275
+ lsm_weight: 0.1
1276
+ length_normalized_loss: false
1277
+ weighted_sum: true
1278
+ extract_feats_in_collect_stats: false
1279
+ preencoder: null
1280
+ preencoder_conf: {}
1281
+ encoder: e_branchformer
1282
+ encoder_conf:
1283
+ output_size: 1024
1284
+ attention_heads: 16
1285
+ attention_layer_type: selfattn
1286
+ pos_enc_layer_type: abs_pos
1287
+ rel_pos_type: latest
1288
+ cgmlp_linear_units: 4096
1289
+ cgmlp_conv_kernel: 31
1290
+ use_linear_after_conv: false
1291
+ gate_activation: identity
1292
+ num_blocks: 18
1293
+ dropout_rate: 0.1
1294
+ positional_dropout_rate: 0.1
1295
+ attention_dropout_rate: 0.1
1296
+ input_layer: conv2d
1297
+ layer_drop_rate: 0.0
1298
+ linear_units: 4096
1299
+ positionwise_layer_type: linear
1300
+ use_ffn: true
1301
+ macaron_ffn: true
1302
+ merge_conv_kernel: 31
1303
+ prepostencoder: linear
1304
+ prepostencoder_conf:
1305
+ input_size: 1024
1306
+ output_size: 80
1307
+ postencoder: conformer_full
1308
+ postencoder_conf:
1309
+ output_size: 256
1310
+ attention_heads: 4
1311
+ linear_units: 1024
1312
+ num_blocks: 12
1313
+ dropout_rate: 0.1
1314
+ positional_dropout_rate: 0.1
1315
+ attention_dropout_rate: 0.1
1316
+ input_layer: conv2d2
1317
+ normalize_before: true
1318
+ macaron_style: true
1319
+ rel_pos_type: latest
1320
+ pos_enc_layer_type: rel_pos
1321
+ selfattention_layer_type: rel_selfattn
1322
+ activation_type: swish
1323
+ use_cnn_module: true
1324
+ cnn_module_kernel: 31
1325
+ deliberationencoder: null
1326
+ deliberationencoder_conf: {}
1327
+ decoder: transformer
1328
+ decoder_conf:
1329
+ attention_heads: 4
1330
+ linear_units: 2048
1331
+ num_blocks: 6
1332
+ dropout_rate: 0.1
1333
+ positional_dropout_rate: 0.1
1334
+ self_attention_dropout_rate: 0.1
1335
+ src_attention_dropout_rate: 0.1
1336
+ postdecoder: null
1337
+ postdecoder_conf: {}
1338
+ required:
1339
+ - output_dir
1340
+ - token_list
1341
+ version: '202310'
1342
+ distributed: true
1343
+ ```
1344
+
1345
+ </details>
1346
+
1347
+
1348
+
1349
+ ### Citing ESPnet
1350
+
1351
+ ```BibTex
1352
+ @inproceedings{watanabe2018espnet,
1353
+ author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
1354
+ title={{ESPnet}: End-to-End Speech Processing Toolkit},
1355
+ year={2018},
1356
+ booktitle={Proceedings of Interspeech},
1357
+ pages={2207--2211},
1358
+ doi={10.21437/Interspeech.2018-1456},
1359
+ url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
1360
+ }
1361
+
1362
+
1363
+
1364
+
1365
+
1366
+
1367
+ ```
1368
+
1369
+ or arXiv:
1370
+
1371
+ ```bibtex
1372
+ @misc{watanabe2018espnet,
1373
+ title={ESPnet: End-to-End Speech Processing Toolkit},
1374
+ author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
1375
+ year={2018},
1376
+ eprint={1804.00015},
1377
+ archivePrefix={arXiv},
1378
+ primaryClass={cs.CL}
1379
+ }
1380
+ ```
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/RESULTS.md ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <!-- Generated by scripts/utils/show_asr_result.sh -->
2
+ # RESULTS
3
+ ## Environments
4
+ - date: `Sat Feb 10 22:23:15 CST 2024`
5
+ - python version: `3.9.13 (main, Aug 25 2022, 23:26:10) [GCC 11.2.0]`
6
+ - espnet version: `espnet 202310`
7
+ - pytorch version: `pytorch 2.1.0+cu121`
8
+ - Git hash: `21d2105784e4da98397bf487b2550d4c6e16d40d`
9
+ - Commit date: `Wed Jan 31 13:40:37 2024 -0600`
10
+
11
+ ## exp/slu_train_asr_owsm_weighted_raw_en_word_sp
12
+ ### WER
13
+
14
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
15
+ |---|---|---|---|---|---|---|---|---|
16
+ |decode_asr_ctc0.3_beam10_slu_model_valid.acc.ave_10best/test|3530|144908|85.7|8.9|5.4|2.7|17.0|95.1|
17
+ |decode_asr_slu_model_valid.acc.ave_10best/devel|1450|58104|85.2|8.1|6.7|2.7|17.6|93.2|
18
+ |decode_asr_slu_model_valid.acc.ave_10best/test|3530|144908|82.9|9.9|7.2|3.3|20.4|95.7|
19
+
20
+ ### CER
21
+
22
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
23
+ |---|---|---|---|---|---|---|---|---|
24
+ |decode_asr_ctc0.3_beam10_slu_model_valid.acc.ave_10best/test|3530|647097|92.9|2.6|4.5|2.7|9.8|95.1|
25
+ |decode_asr_slu_model_valid.acc.ave_10best/devel|1450|256305|91.9|2.4|5.7|2.7|10.8|93.2|
26
+ |decode_asr_slu_model_valid.acc.ave_10best/test|3530|647097|90.7|3.0|6.3|3.1|12.4|95.7|
27
+
28
+ ### TER
29
+
30
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
31
+ |---|---|---|---|---|---|---|---|---|
32
+ ## exp/slu_train_asr_owsm_weighted_raw_en_word_sp/decode_asr_ctc0.3_beam10_slu_model_valid.acc.ave_10best
33
+ ### WER
34
+
35
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
36
+ |---|---|---|---|---|---|---|---|---|
37
+ |org/devel|1451|58267|87.8|7.3|4.8|2.2|14.3|92.7|
38
+
39
+ ### CER
40
+
41
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
42
+ |---|---|---|---|---|---|---|---|---|
43
+ |org/devel|1451|256942|94.0|2.2|3.9|2.2|8.2|92.7|
44
+
45
+ ### TER
46
+
47
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
48
+ |---|---|---|---|---|---|---|---|---|
49
+ ## exp/slu_train_asr_owsm_weighted_raw_en_word_sp/decode_asr_slu_model_valid.acc.ave_10best
50
+ ### WER
51
+
52
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
53
+ |---|---|---|---|---|---|---|---|---|
54
+ |org/devel|1451|58267|85.2|8.1|6.7|2.7|17.6|93.2|
55
+
56
+ ### CER
57
+
58
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
59
+ |---|---|---|---|---|---|---|---|---|
60
+ |org/devel|1451|256942|91.8|2.4|5.7|2.7|10.8|93.2|
61
+
62
+ ### TER
63
+
64
+ |dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
65
+ |---|---|---|---|---|---|---|---|---|
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/config.yaml ADDED
@@ -0,0 +1,1241 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ config: conf/tuning/train_asr_owsm_weighted.yaml
2
+ print_config: false
3
+ log_level: INFO
4
+ drop_last_iter: false
5
+ dry_run: false
6
+ iterator_type: sequence
7
+ valid_iterator_type: null
8
+ output_dir: exp/slu_train_asr_owsm_weighted_raw_en_word_sp
9
+ ngpu: 1
10
+ seed: 2022
11
+ num_workers: 2
12
+ num_att_plot: 3
13
+ dist_backend: nccl
14
+ dist_init_method: env://
15
+ dist_world_size: 4
16
+ dist_rank: 0
17
+ local_rank: 0
18
+ dist_master_addr: localhost
19
+ dist_master_port: 52077
20
+ dist_launcher: null
21
+ multiprocessing_distributed: true
22
+ unused_parameters: false
23
+ sharded_ddp: false
24
+ cudnn_enabled: true
25
+ cudnn_benchmark: false
26
+ cudnn_deterministic: true
27
+ collect_stats: false
28
+ write_collected_feats: false
29
+ max_epoch: 70
30
+ patience: null
31
+ val_scheduler_criterion:
32
+ - valid
33
+ - loss
34
+ early_stopping_criterion:
35
+ - valid
36
+ - loss
37
+ - min
38
+ best_model_criterion:
39
+ - - valid
40
+ - acc
41
+ - max
42
+ keep_nbest_models: 10
43
+ nbest_averaging_interval: 0
44
+ grad_clip: 5.0
45
+ grad_clip_type: 2.0
46
+ grad_noise: false
47
+ accum_grad: 2
48
+ no_forward_run: false
49
+ resume: true
50
+ train_dtype: float32
51
+ use_amp: false
52
+ log_interval: null
53
+ use_matplotlib: true
54
+ use_tensorboard: true
55
+ create_graph_in_tensorboard: false
56
+ use_wandb: false
57
+ wandb_project: null
58
+ wandb_id: null
59
+ wandb_entity: null
60
+ wandb_name: null
61
+ wandb_model_log_interval: -1
62
+ detect_anomaly: false
63
+ use_lora: false
64
+ save_lora_only: true
65
+ lora_conf: {}
66
+ pretrain_path: null
67
+ init_param:
68
+ - /scratch/bbjs/arora1/new_download_espnet_egs2/harpervalley/slu1_superb_onlyda/owsm_v3.1_ebf/exp/s2t_train_s2t_ebf_conv2d_size1024_e18_d18_piecewise_lr2e-4_warmup60k_flashattn_raw_bpe50000/valid.total_count.ave_5best.till45epoch.pth:encoder:encoder
69
+ ignore_init_mismatch: false
70
+ freeze_param:
71
+ - encoder
72
+ num_iters_per_epoch: null
73
+ batch_size: 20
74
+ valid_batch_size: null
75
+ batch_bins: 6000000
76
+ valid_batch_bins: null
77
+ train_shape_file:
78
+ - exp/slu_stats_raw_en_word_sp/train/speech_shape
79
+ - exp/slu_stats_raw_en_word_sp/train/text_shape.word
80
+ valid_shape_file:
81
+ - exp/slu_stats_raw_en_word_sp/valid/speech_shape
82
+ - exp/slu_stats_raw_en_word_sp/valid/text_shape.word
83
+ batch_type: numel
84
+ valid_batch_type: null
85
+ fold_length:
86
+ - 80000
87
+ - 150
88
+ sort_in_batch: descending
89
+ shuffle_within_batch: false
90
+ sort_batch: descending
91
+ multiple_iterator: false
92
+ chunk_length: 500
93
+ chunk_shift_ratio: 0.5
94
+ num_cache_chunks: 1024
95
+ chunk_excluded_key_prefixes: []
96
+ chunk_default_fs: null
97
+ train_data_path_and_name_and_type:
98
+ - - dump/raw/train_sp/wav.scp
99
+ - speech
100
+ - sound
101
+ - - dump/raw/train_sp/text
102
+ - text
103
+ - text
104
+ valid_data_path_and_name_and_type:
105
+ - - dump/raw/devel/wav.scp
106
+ - speech
107
+ - sound
108
+ - - dump/raw/devel/text
109
+ - text
110
+ - text
111
+ allow_variable_data_keys: false
112
+ max_cache_size: 0.0
113
+ max_cache_fd: 32
114
+ allow_multi_rates: false
115
+ valid_max_cache_size: null
116
+ exclude_weight_decay: false
117
+ exclude_weight_decay_conf: {}
118
+ optim: adam
119
+ optim_conf:
120
+ lr: 0.002
121
+ weight_decay: 1.0e-06
122
+ scheduler: warmuplr
123
+ scheduler_conf:
124
+ warmup_steps: 5000
125
+ token_list:
126
+ - <blank>
127
+ - <unk>
128
+ - ▁i
129
+ - ▁and
130
+ - ''''
131
+ - s
132
+ - ▁the
133
+ - ▁a
134
+ - ▁it
135
+ - Neutral
136
+ - ▁to
137
+ - ▁you
138
+ - ▁that
139
+ - ▁of
140
+ - ▁in
141
+ - ▁was
142
+ - ▁uh
143
+ - ▁know
144
+ - t
145
+ - ▁so
146
+ - ▁we
147
+ - ▁he
148
+ - ing
149
+ - ▁um
150
+ - ed
151
+ - m
152
+ - ▁like
153
+ - ▁is
154
+ - ▁but
155
+ - Positive
156
+ - y
157
+ - ▁just
158
+ - ▁they
159
+ - re
160
+ - ▁this
161
+ - ▁for
162
+ - ▁be
163
+ - ▁my
164
+ - er
165
+ - ▁with
166
+ - ▁on
167
+ - ▁think
168
+ - ▁p
169
+ - ▁have
170
+ - ▁she
171
+ - e
172
+ - ▁me
173
+ - ▁really
174
+ - ▁there
175
+ - ▁what
176
+ - ▁m
177
+ - a
178
+ - ▁do
179
+ - ▁all
180
+ - i
181
+ - al
182
+ - ve
183
+ - c
184
+ - ▁as
185
+ - ▁about
186
+ - ▁not
187
+ - ▁t
188
+ - n
189
+ - ▁at
190
+ - l
191
+ - ▁had
192
+ - ▁b
193
+ - ▁when
194
+ - ▁c
195
+ - g
196
+ - ar
197
+ - ▁out
198
+ - en
199
+ - ▁s
200
+ - ▁an
201
+ - ▁people
202
+ - or
203
+ - an
204
+ - d
205
+ - o
206
+ - ll
207
+ - ▁are
208
+ - in
209
+ - ▁very
210
+ - p
211
+ - b
212
+ - u
213
+ - ▁because
214
+ - es
215
+ - ▁can
216
+ - ▁don
217
+ - ▁or
218
+ - ▁up
219
+ - it
220
+ - ▁one
221
+ - ly
222
+ - ▁if
223
+ - ▁f
224
+ - st
225
+ - ▁were
226
+ - ▁mean
227
+ - ▁d
228
+ - ▁who
229
+ - ▁then
230
+ - ic
231
+ - 'on'
232
+ - ▁no
233
+ - ▁go
234
+ - ▁her
235
+ - ▁g
236
+ - ent
237
+ - ▁st
238
+ - ▁kind
239
+ - ri
240
+ - ▁would
241
+ - ▁get
242
+ - ▁e
243
+ - le
244
+ - at
245
+ - r
246
+ - ▁time
247
+ - ▁w
248
+ - ▁re
249
+ - h
250
+ - ▁from
251
+ - ▁l
252
+ - ▁said
253
+ - ▁him
254
+ - ▁how
255
+ - v
256
+ - ▁well
257
+ - ▁h
258
+ - ▁gonna
259
+ - ▁lot
260
+ - ▁see
261
+ - f
262
+ - ▁his
263
+ - et
264
+ - ion
265
+ - ▁been
266
+ - ▁great
267
+ - ▁yeah
268
+ - ▁love
269
+ - ▁which
270
+ - ▁got
271
+ - k
272
+ - ▁them
273
+ - ▁way
274
+ - id
275
+ - ▁show
276
+ - w
277
+ - ▁some
278
+ - ▁your
279
+ - ▁did
280
+ - ▁sort
281
+ - ▁has
282
+ - ▁things
283
+ - ▁back
284
+ - ▁where
285
+ - ▁something
286
+ - ir
287
+ - ▁thing
288
+ - ad
289
+ - ▁su
290
+ - ▁ch
291
+ - ▁n
292
+ - il
293
+ - as
294
+ - ▁j
295
+ - ▁more
296
+ - se
297
+ - ▁say
298
+ - ▁co
299
+ - nd
300
+ - ▁much
301
+ - ▁always
302
+ - ine
303
+ - ▁r
304
+ - ation
305
+ - ur
306
+ - ▁other
307
+ - th
308
+ - ▁
309
+ - ▁se
310
+ - ▁now
311
+ - ate
312
+ - ▁doing
313
+ - ▁work
314
+ - ow
315
+ - ▁could
316
+ - ally
317
+ - ▁these
318
+ - Negative
319
+ - ▁good
320
+ - ▁any
321
+ - ers
322
+ - ce
323
+ - ▁cause
324
+ - ▁ex
325
+ - ▁pro
326
+ - ▁little
327
+ - ▁actually
328
+ - ▁into
329
+ - ▁make
330
+ - ▁first
331
+ - ▁being
332
+ - ra
333
+ - ▁our
334
+ - ▁al
335
+ - ▁by
336
+ - ▁film
337
+ - ▁didn
338
+ - ▁v
339
+ - ct
340
+ - ity
341
+ - ch
342
+ - un
343
+ - ▁part
344
+ - ▁de
345
+ - ▁come
346
+ - is
347
+ - ie
348
+ - ▁right
349
+ - ▁o
350
+ - ▁off
351
+ - ol
352
+ - ▁two
353
+ - ▁never
354
+ - ▁le
355
+ - ot
356
+ - ut
357
+ - ▁movie
358
+ - ▁play
359
+ - ge
360
+ - ies
361
+ - el
362
+ - ▁con
363
+ - am
364
+ - ▁going
365
+ - ke
366
+ - ��want
367
+ - im
368
+ - ▁feel
369
+ - ive
370
+ - ro
371
+ - ▁mo
372
+ - ▁different
373
+ - ck
374
+ - ▁life
375
+ - ist
376
+ - ▁oh
377
+ - all
378
+ - ▁lo
379
+ - ard
380
+ - ▁went
381
+ - and
382
+ - ▁sh
383
+ - ▁even
384
+ - ry
385
+ - ▁years
386
+ - ▁look
387
+ - ▁us
388
+ - ant
389
+ - ▁te
390
+ - ▁k
391
+ - ▁li
392
+ - ▁happen
393
+ - ure
394
+ - ▁their
395
+ - ▁those
396
+ - ▁take
397
+ - ment
398
+ - ▁day
399
+ - ble
400
+ - ast
401
+ - ▁every
402
+ - um
403
+ - ill
404
+ - op
405
+ - ▁thought
406
+ - ou
407
+ - us
408
+ - ay
409
+ - ▁th
410
+ - ▁put
411
+ - ▁story
412
+ - ▁new
413
+ - ▁down
414
+ - ish
415
+ - ▁big
416
+ - ▁wanna
417
+ - ▁ro
418
+ - ▁also
419
+ - ▁read
420
+ - ▁around
421
+ - ous
422
+ - ▁through
423
+ - red
424
+ - ▁came
425
+ - ▁character
426
+ - ess
427
+ - te
428
+ - ver
429
+ - ▁will
430
+ - ag
431
+ - ss
432
+ - ▁fun
433
+ - ▁over
434
+ - ▁many
435
+ - ▁bl
436
+ - ▁cl
437
+ - ▁man
438
+ - ▁than
439
+ - ▁pre
440
+ - ▁world
441
+ - ▁person
442
+ - z
443
+ - ▁sp
444
+ - ven
445
+ - ▁wanted
446
+ - ▁bit
447
+ - ▁before
448
+ - ▁mar
449
+ - one
450
+ - ab
451
+ - ▁en
452
+ - ci
453
+ - ▁set
454
+ - ▁ha
455
+ - ▁find
456
+ - ul
457
+ - ▁fi
458
+ - ▁end
459
+ - ▁un
460
+ - ▁sc
461
+ - ▁after
462
+ - ind
463
+ - ter
464
+ - ▁working
465
+ - ▁why
466
+ - om
467
+ - me
468
+ - ▁such
469
+ - ▁whole
470
+ - ▁kinda
471
+ - ne
472
+ - ▁bo
473
+ - x
474
+ - ▁most
475
+ - ▁ad
476
+ - ▁guy
477
+ - ▁spe
478
+ - ars
479
+ - ▁am
480
+ - ful
481
+ - ▁together
482
+ - ▁let
483
+ - ▁quite
484
+ - ain
485
+ - ▁everything
486
+ - ▁made
487
+ - ig
488
+ - ▁old
489
+ - able
490
+ - ▁tr
491
+ - ak
492
+ - ▁fo
493
+ - ▁po
494
+ - ore
495
+ - ice
496
+ - ▁real
497
+ - ▁knew
498
+ - ▁hard
499
+ - pp
500
+ - age
501
+ - ated
502
+ - ▁same
503
+ - ▁start
504
+ - ▁ever
505
+ - ning
506
+ - ▁watch
507
+ - art
508
+ - ▁again
509
+ - ▁here
510
+ - are
511
+ - ght
512
+ - ong
513
+ - ▁done
514
+ - ▁only
515
+ - ▁live
516
+ - ▁wasn
517
+ - ▁ho
518
+ - ▁u
519
+ - ▁maybe
520
+ - ▁need
521
+ - ▁everybody
522
+ - ust
523
+ - ans
524
+ - ▁three
525
+ - ▁having
526
+ - ▁music
527
+ - ack
528
+ - ld
529
+ - ▁trying
530
+ - ▁guys
531
+ - rou
532
+ - ach
533
+ - ving
534
+ - ▁tell
535
+ - ▁should
536
+ - ff
537
+ - ide
538
+ - ▁four
539
+ - ▁started
540
+ - ▁com
541
+ - ass
542
+ - ▁long
543
+ - ▁fe
544
+ - ▁course
545
+ - ▁called
546
+ - ▁own
547
+ - ress
548
+ - ▁moment
549
+ - ▁pl
550
+ - ▁still
551
+ - ▁anything
552
+ - ▁family
553
+ - ▁fin
554
+ - ▁dan
555
+ - ▁bro
556
+ - 'no'
557
+ - ther
558
+ - ▁per
559
+ - ▁amazing
560
+ - ▁stuff
561
+ - per
562
+ - ▁jo
563
+ - ▁certain
564
+ - os
565
+ - ▁talk
566
+ - ater
567
+ - ▁help
568
+ - ▁too
569
+ - ▁year
570
+ - ight
571
+ - ▁fa
572
+ - self
573
+ - ces
574
+ - ▁br
575
+ - ▁bet
576
+ - ▁someone
577
+ - ▁di
578
+ - ▁sing
579
+ - nt
580
+ - ick
581
+ - ▁ph
582
+ - row
583
+ - ▁script
584
+ - ▁remember
585
+ - ▁try
586
+ - qu
587
+ - ite
588
+ - ▁young
589
+ - ▁wh
590
+ - ▁ser
591
+ - ▁ask
592
+ - ▁book
593
+ - ▁each
594
+ - ▁wr
595
+ - ▁best
596
+ - ▁ag
597
+ - ▁women
598
+ - ose
599
+ - ions
600
+ - ved
601
+ - j
602
+ - ue
603
+ - ▁does
604
+ - ▁five
605
+ - ▁both
606
+ - ▁friends
607
+ - ▁act
608
+ - iz
609
+ - cess
610
+ - pt
611
+ - ▁somebody
612
+ - ft
613
+ - ▁nice
614
+ - ▁myself
615
+ - een
616
+ - fe
617
+ - sp
618
+ - ict
619
+ - ty
620
+ - ▁child
621
+ - ud
622
+ - pe
623
+ - ▁hope
624
+ - ▁fact
625
+ - ▁saying
626
+ - ave
627
+ - icul
628
+ - au
629
+ - ale
630
+ - ris
631
+ - ▁twenty
632
+ - ▁school
633
+ - ▁doesn
634
+ - ▁able
635
+ - pect
636
+ - ▁last
637
+ - ber
638
+ - ▁song
639
+ - od
640
+ - ▁str
641
+ - ▁interesting
642
+ - lf
643
+ - ▁em
644
+ - ▁wor
645
+ - ap
646
+ - og
647
+ - ▁ra
648
+ - ▁dis
649
+ - ▁coming
650
+ - ▁ab
651
+ - ▁house
652
+ - ▁next
653
+ - ▁tra
654
+ - ▁okay
655
+ - ere
656
+ - ary
657
+ - ▁incredi
658
+ - ▁car
659
+ - ▁job
660
+ - ▁used
661
+ - ▁give
662
+ - ▁god
663
+ - ▁americ
664
+ - ▁characters
665
+ - ▁app
666
+ - ▁walk
667
+ - ▁yes
668
+ - rew
669
+ - ▁getting
670
+ - ▁six
671
+ - ▁chan
672
+ - ▁ne
673
+ - ▁pretty
674
+ - ang
675
+ - ▁creat
676
+ - ▁another
677
+ - ▁ter
678
+ - ▁kids
679
+ - ▁felt
680
+ - ▁sometimes
681
+ - ▁place
682
+ - out
683
+ - ▁funny
684
+ - ase
685
+ - ich
686
+ - act
687
+ - ▁days
688
+ - ▁hum
689
+ - ▁bring
690
+ - ts
691
+ - ▁making
692
+ - ▁comp
693
+ - ▁become
694
+ - ute
695
+ - ▁wonderful
696
+ - ron
697
+ - les
698
+ - ▁saw
699
+ - ▁point
700
+ - ia
701
+ - ▁realiz
702
+ - ▁int
703
+ - ▁away
704
+ - ays
705
+ - ▁home
706
+ - ace
707
+ - ▁relationship
708
+ - ▁woman
709
+ - ▁everyone
710
+ - ▁comes
711
+ - ▁high
712
+ - dd
713
+ - ▁night
714
+ - ath
715
+ - ▁else
716
+ - vent
717
+ - ▁shoot
718
+ - vers
719
+ - day
720
+ - ▁sure
721
+ - ried
722
+ - ned
723
+ - ▁obviously
724
+ - ▁dra
725
+ - ▁inter
726
+ - co
727
+ - ▁playing
728
+ - ▁important
729
+ - ort
730
+ - uck
731
+ - ision
732
+ - pport
733
+ - ▁seen
734
+ - pl
735
+ - ▁fl
736
+ - ound
737
+ - ▁bas
738
+ - ull
739
+ - est
740
+ - ▁actor
741
+ - ▁lear
742
+ - ▁worked
743
+ - ▁believe
744
+ - ▁gen
745
+ - ▁keep
746
+ - ▁friend
747
+ - ▁sw
748
+ - ▁des
749
+ - ▁times
750
+ - ▁im
751
+ - ▁sur
752
+ - ▁sit
753
+ - ▁probably
754
+ - ok
755
+ - ▁took
756
+ - ep
757
+ - ough
758
+ - ip
759
+ - ood
760
+ - ▁sa
761
+ - ▁season
762
+ - vel
763
+ - wn
764
+ - ▁dec
765
+ - ▁excited
766
+ - ian
767
+ - ire
768
+ - ph
769
+ - ▁month
770
+ - ner
771
+ - ▁min
772
+ - ▁rel
773
+ - ating
774
+ - body
775
+ - ition
776
+ - ▁loved
777
+ - ▁aw
778
+ - ▁hear
779
+ - ple
780
+ - ▁cool
781
+ - ▁y
782
+ - ord
783
+ - our
784
+ - ▁game
785
+ - ms
786
+ - ub
787
+ - ▁might
788
+ - ▁kid
789
+ - ▁movies
790
+ - ical
791
+ - ▁bad
792
+ - ▁scene
793
+ - iv
794
+ - ▁enough
795
+ - ▁sm
796
+ - bly
797
+ - ▁fift
798
+ - ▁eight
799
+ - ▁experience
800
+ - ▁actors
801
+ - ▁cou
802
+ - ▁understand
803
+ - ▁week
804
+ - ▁few
805
+ - gin
806
+ - ting
807
+ - ▁director
808
+ - ▁almost
809
+ - ▁open
810
+ - ren
811
+ - ▁star
812
+ - ▁room
813
+ - ▁call
814
+ - oy
815
+ - ▁goes
816
+ - ▁told
817
+ - ▁once
818
+ - ▁found
819
+ - arly
820
+ - ations
821
+ - ward
822
+ - ▁audience
823
+ - ird
824
+ - if
825
+ - ▁qu
826
+ - ▁ar
827
+ - ▁definitely
828
+ - ious
829
+ - iting
830
+ - ▁pol
831
+ - ▁huge
832
+ - ▁makes
833
+ - aking
834
+ - ream
835
+ - ance
836
+ - be
837
+ - ▁la
838
+ - ▁ac
839
+ - iter
840
+ - ▁run
841
+ - ▁gotta
842
+ - ▁gr
843
+ - ▁cam
844
+ - sh
845
+ - ▁gets
846
+ - ully
847
+ - ▁says
848
+ - ame
849
+ - side
850
+ - ▁bus
851
+ - ▁shows
852
+ - ▁dr
853
+ - ▁inv
854
+ - ▁idea
855
+ - ▁talking
856
+ - ▁wa
857
+ - way
858
+ - ▁art
859
+ - ▁whatever
860
+ - ▁write
861
+ - ash
862
+ - itt
863
+ - ▁met
864
+ - ▁wants
865
+ - ▁role
866
+ - ▁mu
867
+ - ▁boy
868
+ - ▁wrote
869
+ - ger
870
+ - ately
871
+ - ▁exc
872
+ - ▁mother
873
+ - ▁produ
874
+ - ▁cra
875
+ - ates
876
+ - ▁though
877
+ - av
878
+ - ▁episode
879
+ - ▁sl
880
+ - ▁change
881
+ - ▁voice
882
+ - ▁played
883
+ - ily
884
+ - ▁guess
885
+ - ves
886
+ - ▁hand
887
+ - ady
888
+ - ▁happy
889
+ - ith
890
+ - ▁name
891
+ - ny
892
+ - ▁gi
893
+ - ▁looking
894
+ - lev
895
+ - ▁acting
896
+ - aught
897
+ - iss
898
+ - ount
899
+ - rom
900
+ - ▁tw
901
+ - ▁cont
902
+ - ▁john
903
+ - ▁far
904
+ - ▁res
905
+ - ▁sense
906
+ - ake
907
+ - ▁basically
908
+ - ▁meet
909
+ - ▁gu
910
+ - ▁bre
911
+ - ens
912
+ - cept
913
+ - ety
914
+ - ▁girl
915
+ - ▁york
916
+ - ▁count
917
+ - ▁shot
918
+ - ise
919
+ - ject
920
+ - ▁tot
921
+ - ▁stud
922
+ - ▁feels
923
+ - ▁thinking
924
+ - ▁head
925
+ - ▁cast
926
+ - ▁writing
927
+ - ▁rehe
928
+ - ▁written
929
+ - ▁perform
930
+ - ▁fan
931
+ - der
932
+ - ect
933
+ - ▁sk
934
+ - ▁hour
935
+ - ▁father
936
+ - ered
937
+ - ▁hundred
938
+ - ▁ind
939
+ - ▁norm
940
+ - ▁acc
941
+ - up
942
+ - ▁while
943
+ - fort
944
+ - ▁nin
945
+ - ▁true
946
+ - itch
947
+ - ▁inst
948
+ - ▁second
949
+ - ▁pick
950
+ - ▁record
951
+ - ross
952
+ - ▁quest
953
+ - ged
954
+ - ▁career
955
+ - ween
956
+ - ▁bec
957
+ - ▁reason
958
+ - ▁since
959
+ - ▁bra
960
+ - ▁char
961
+ - ▁imp
962
+ - ree
963
+ - ▁girls
964
+ - ▁comple
965
+ - ▁turn
966
+ - ▁dad
967
+ - ▁fant
968
+ - ▁extra
969
+ - ▁laugh
970
+ - ▁stand
971
+ - ▁honest
972
+ - ▁comm
973
+ - na
974
+ - ▁listen
975
+ - als
976
+ - cial
977
+ - spe
978
+ - ▁ke
979
+ - ory
980
+ - view
981
+ - ink
982
+ - ▁direct
983
+ - reat
984
+ - round
985
+ - ien
986
+ - ▁under
987
+ - ile
988
+ - ▁diff
989
+ - ually
990
+ - ▁tur
991
+ - thing
992
+ - sic
993
+ - ▁gon
994
+ - ather
995
+ - ▁aud
996
+ - ▁scen
997
+ - atch
998
+ - ▁sho
999
+ - ever
1000
+ - tra
1001
+ - ▁pe
1002
+ - mo
1003
+ - ild
1004
+ - ▁care
1005
+ - int
1006
+ - ▁fam
1007
+ - ▁ob
1008
+ - ▁ide
1009
+ - ade
1010
+ - right
1011
+ - ▁may
1012
+ - he
1013
+ - ody
1014
+ - ense
1015
+ - ▁interest
1016
+ - ah
1017
+ - form
1018
+ - ork
1019
+ - ▁episod
1020
+ - ▁rec
1021
+ - iew
1022
+ - ▁hop
1023
+ - ited
1024
+ - ▁exper
1025
+ - gh
1026
+ - ically
1027
+ - ▁bel
1028
+ - ▁el
1029
+ - enty
1030
+ - ▁gott
1031
+ - ▁stu
1032
+ - ▁id
1033
+ - rie
1034
+ - ▁nor
1035
+ - ▁inc
1036
+ - ertain
1037
+ - tain
1038
+ - ▁wo
1039
+ - ▁mon
1040
+ - az
1041
+ - xt
1042
+ - riend
1043
+ - now
1044
+ - ▁list
1045
+ - ime
1046
+ - ome
1047
+ - so
1048
+ - ause
1049
+ - iously
1050
+ - ▁sch
1051
+ - ▁vo
1052
+ - ▁op
1053
+ - ason
1054
+ - ▁mov
1055
+ - ▁hi
1056
+ - ▁pers
1057
+ - ▁ye
1058
+ - ▁def
1059
+ - orm
1060
+ - ▁belie
1061
+ - fore
1062
+ - ix
1063
+ - mber
1064
+ - very
1065
+ - ▁differe
1066
+ - ▁wonder
1067
+ - ek
1068
+ - nder
1069
+ - ▁obv
1070
+ - ▁ep
1071
+ - ship
1072
+ - ▁lau
1073
+ - ience
1074
+ - ool
1075
+ - ▁sin
1076
+ - rect
1077
+ - ▁happ
1078
+ - ▁gir
1079
+ - du
1080
+ - ng
1081
+ - ▁underst
1082
+ - most
1083
+ - eric
1084
+ - ouse
1085
+ - time
1086
+ - lm
1087
+ - ▁hel
1088
+ - redi
1089
+ - ▁cour
1090
+ - ▁relation
1091
+ - rough
1092
+ - q
1093
+ - ▁defin
1094
+ - ▁prob
1095
+ - ▁reme
1096
+ - ▁hu
1097
+ - ▁fir
1098
+ - anna
1099
+ - ways
1100
+ - itten
1101
+ - elt
1102
+ - ▁sometime
1103
+ - ':'
1104
+ - ▁kne
1105
+ - alk
1106
+ - ▁ok
1107
+ - ably
1108
+ - rote
1109
+ - gether
1110
+ - ▁definite
1111
+ - ▁import
1112
+ - '&'
1113
+ - fter
1114
+ - onest
1115
+ - erest
1116
+ - ▁amaz
1117
+ - ▁ano
1118
+ - <sos/eos>
1119
+ transcript_token_list: null
1120
+ two_pass: false
1121
+ pre_postencoder_norm: false
1122
+ init: null
1123
+ input_size: null
1124
+ ctc_conf:
1125
+ dropout_rate: 0.0
1126
+ ctc_type: builtin
1127
+ reduce: true
1128
+ ignore_nan_grad: null
1129
+ zero_infinity: true
1130
+ brctc_risk_strategy: exp
1131
+ brctc_group_strategy: end
1132
+ brctc_risk_factor: 0.0
1133
+ joint_net_conf: null
1134
+ use_preprocessor: true
1135
+ token_type: word
1136
+ bpemodel: null
1137
+ non_linguistic_symbols: null
1138
+ cleaner: null
1139
+ g2p: null
1140
+ speech_volume_normalize: null
1141
+ rir_scp: null
1142
+ rir_apply_prob: 1.0
1143
+ noise_scp: null
1144
+ noise_apply_prob: 1.0
1145
+ noise_db_range: '13_15'
1146
+ short_noise_thres: 0.5
1147
+ frontend: default
1148
+ frontend_conf:
1149
+ n_fft: 512
1150
+ win_length: 400
1151
+ hop_length: 160
1152
+ fs: 16k
1153
+ specaug: specaug
1154
+ specaug_conf:
1155
+ apply_time_warp: false
1156
+ time_warp_window: 5
1157
+ time_warp_mode: bicubic
1158
+ apply_freq_mask: true
1159
+ freq_mask_width_range:
1160
+ - 0
1161
+ - 27
1162
+ num_freq_mask: 2
1163
+ apply_time_mask: true
1164
+ time_mask_width_ratio_range:
1165
+ - 0.0
1166
+ - 0.05
1167
+ num_time_mask: 10
1168
+ normalize: global_mvn
1169
+ normalize_conf:
1170
+ stats_file: /scratch/bbjs/arora1/new_download_espnet_egs2/harpervalley/slu1_superb_onlyda/owsm_v3.1_ebf/exp/s2t_stats_raw_bpe50000/train/feats_stats.npz
1171
+ model: espnet
1172
+ model_conf:
1173
+ ctc_weight: 0.3
1174
+ lsm_weight: 0.1
1175
+ length_normalized_loss: false
1176
+ weighted_sum: true
1177
+ extract_feats_in_collect_stats: false
1178
+ preencoder: null
1179
+ preencoder_conf: {}
1180
+ encoder: e_branchformer
1181
+ encoder_conf:
1182
+ output_size: 1024
1183
+ attention_heads: 16
1184
+ attention_layer_type: selfattn
1185
+ pos_enc_layer_type: abs_pos
1186
+ rel_pos_type: latest
1187
+ cgmlp_linear_units: 4096
1188
+ cgmlp_conv_kernel: 31
1189
+ use_linear_after_conv: false
1190
+ gate_activation: identity
1191
+ num_blocks: 18
1192
+ dropout_rate: 0.1
1193
+ positional_dropout_rate: 0.1
1194
+ attention_dropout_rate: 0.1
1195
+ input_layer: conv2d
1196
+ layer_drop_rate: 0.0
1197
+ linear_units: 4096
1198
+ positionwise_layer_type: linear
1199
+ use_ffn: true
1200
+ macaron_ffn: true
1201
+ merge_conv_kernel: 31
1202
+ prepostencoder: linear
1203
+ prepostencoder_conf:
1204
+ input_size: 1024
1205
+ output_size: 80
1206
+ postencoder: conformer_full
1207
+ postencoder_conf:
1208
+ output_size: 256
1209
+ attention_heads: 4
1210
+ linear_units: 1024
1211
+ num_blocks: 12
1212
+ dropout_rate: 0.1
1213
+ positional_dropout_rate: 0.1
1214
+ attention_dropout_rate: 0.1
1215
+ input_layer: conv2d2
1216
+ normalize_before: true
1217
+ macaron_style: true
1218
+ rel_pos_type: latest
1219
+ pos_enc_layer_type: rel_pos
1220
+ selfattention_layer_type: rel_selfattn
1221
+ activation_type: swish
1222
+ use_cnn_module: true
1223
+ cnn_module_kernel: 31
1224
+ deliberationencoder: null
1225
+ deliberationencoder_conf: {}
1226
+ decoder: transformer
1227
+ decoder_conf:
1228
+ attention_heads: 4
1229
+ linear_units: 2048
1230
+ num_blocks: 6
1231
+ dropout_rate: 0.1
1232
+ positional_dropout_rate: 0.1
1233
+ self_attention_dropout_rate: 0.1
1234
+ src_attention_dropout_rate: 0.1
1235
+ postdecoder: null
1236
+ postdecoder_conf: {}
1237
+ required:
1238
+ - output_dir
1239
+ - token_list
1240
+ version: '202310'
1241
+ distributed: true
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/acc.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/backward_time.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/cer.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/cer_ctc.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/clip.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/forward_time.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/gpu_max_cached_mem_GB.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/grad_norm.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/iter_time.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss_att.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss_ctc.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/loss_scale.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/optim0_lr0.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/optim_step_time.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/train_time.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/images/wer.png ADDED
exp/slu_train_asr_owsm_weighted_raw_en_word_sp/valid.acc.ave_10best.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf6adea0364615ed6d0c7f0e48ef3bb56a1b6e41f17bf32d410d4ede0592ec28
3
+ size 2373620026
meta.yaml ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ espnet: '202310'
2
+ files:
3
+ slu_model_file: exp/slu_train_asr_owsm_weighted_raw_en_word_sp/valid.acc.ave_10best.pth
4
+ python: "3.9.13 (main, Aug 25 2022, 23:26:10) \n[GCC 11.2.0]"
5
+ timestamp: 1715356647.863976
6
+ torch: 2.1.0+cu121
7
+ yaml_files:
8
+ slu_train_config: exp/slu_train_asr_owsm_weighted_raw_en_word_sp/config.yaml