“siddhu001”
commited on
Commit
·
fd66565
1
Parent(s):
46e1bd0
Update model
Browse files- README.md +823 -0
- data/en_token_list/bpe_unigram500/bpe.model +3 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/RESULTS.md +1 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/config.yaml +748 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/acc.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/backward_time.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/cer.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/clip.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/forward_time.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/gpu_max_cached_mem_GB.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/grad_norm.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/iter_time.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/loss.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/loss_att.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/loss_scale.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/optim0_lr0.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/optim_step_time.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/train_time.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/wer.png +0 -0
- exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/valid.acc.ave_10best.pth +3 -0
- meta.yaml +8 -0
README.md
ADDED
@@ -0,0 +1,823 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- espnet
|
4 |
+
- audio
|
5 |
+
- automatic-speech-recognition
|
6 |
+
language: en
|
7 |
+
datasets:
|
8 |
+
- slue-ted
|
9 |
+
license: cc-by-4.0
|
10 |
+
---
|
11 |
+
|
12 |
+
## ESPnet2 ASR model
|
13 |
+
|
14 |
+
### `espnet/slueted_owsm_summ`
|
15 |
+
|
16 |
+
This model was trained by “siddhu001” using slue-ted 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-ted/slu1
|
28 |
+
./run.sh --skip_data_prep false --skip_train true --download_model espnet/slueted_owsm_summ
|
29 |
+
```
|
30 |
+
|
31 |
+
RESULT 23.72001537034864 5.0727273644425015 16.49858244502415 22.432552212134674 83.55819195438502
|
32 |
+
|
33 |
+
## ASR config
|
34 |
+
|
35 |
+
<details><summary>expand</summary>
|
36 |
+
|
37 |
+
```
|
38 |
+
config: conf//train_asr_owsm_weighted_dropout.yaml
|
39 |
+
print_config: false
|
40 |
+
log_level: INFO
|
41 |
+
drop_last_iter: false
|
42 |
+
dry_run: false
|
43 |
+
iterator_type: sequence
|
44 |
+
valid_iterator_type: null
|
45 |
+
output_dir: exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp
|
46 |
+
ngpu: 1
|
47 |
+
seed: 2022
|
48 |
+
num_workers: 2
|
49 |
+
num_att_plot: 3
|
50 |
+
dist_backend: nccl
|
51 |
+
dist_init_method: env://
|
52 |
+
dist_world_size: 4
|
53 |
+
dist_rank: 0
|
54 |
+
local_rank: 0
|
55 |
+
dist_master_addr: localhost
|
56 |
+
dist_master_port: 55179
|
57 |
+
dist_launcher: null
|
58 |
+
multiprocessing_distributed: true
|
59 |
+
unused_parameters: false
|
60 |
+
sharded_ddp: false
|
61 |
+
cudnn_enabled: true
|
62 |
+
cudnn_benchmark: false
|
63 |
+
cudnn_deterministic: true
|
64 |
+
collect_stats: false
|
65 |
+
write_collected_feats: false
|
66 |
+
max_epoch: 25
|
67 |
+
patience: null
|
68 |
+
val_scheduler_criterion:
|
69 |
+
- valid
|
70 |
+
- loss
|
71 |
+
early_stopping_criterion:
|
72 |
+
- valid
|
73 |
+
- loss
|
74 |
+
- min
|
75 |
+
best_model_criterion:
|
76 |
+
- - valid
|
77 |
+
- acc
|
78 |
+
- max
|
79 |
+
keep_nbest_models: 10
|
80 |
+
nbest_averaging_interval: 0
|
81 |
+
grad_clip: 5.0
|
82 |
+
grad_clip_type: 2.0
|
83 |
+
grad_noise: false
|
84 |
+
accum_grad: 1
|
85 |
+
no_forward_run: false
|
86 |
+
resume: true
|
87 |
+
train_dtype: float32
|
88 |
+
use_amp: false
|
89 |
+
log_interval: 100
|
90 |
+
use_matplotlib: true
|
91 |
+
use_tensorboard: true
|
92 |
+
create_graph_in_tensorboard: false
|
93 |
+
use_wandb: false
|
94 |
+
wandb_project: null
|
95 |
+
wandb_id: null
|
96 |
+
wandb_entity: null
|
97 |
+
wandb_name: null
|
98 |
+
wandb_model_log_interval: -1
|
99 |
+
detect_anomaly: false
|
100 |
+
use_lora: false
|
101 |
+
save_lora_only: true
|
102 |
+
lora_conf: {}
|
103 |
+
pretrain_path: null
|
104 |
+
init_param:
|
105 |
+
- /scratch/bbjs/arora1/espnet_slue_PR/espnet/egs2/tedlium3/asr1/exp/asr_train_asr_owsm_weighted_dropout_raw_en_bpe500/valid.acc.ave_10best.pth:::ctc
|
106 |
+
ignore_init_mismatch: false
|
107 |
+
freeze_param:
|
108 |
+
- encoder
|
109 |
+
num_iters_per_epoch: null
|
110 |
+
batch_size: 20
|
111 |
+
valid_batch_size: null
|
112 |
+
batch_bins: 12000000
|
113 |
+
valid_batch_bins: null
|
114 |
+
train_shape_file:
|
115 |
+
- exp/slu_stats_raw_en_bpe500_sp/train/speech_shape
|
116 |
+
- exp/slu_stats_raw_en_bpe500_sp/train/text_shape.bpe
|
117 |
+
valid_shape_file:
|
118 |
+
- exp/slu_stats_raw_en_bpe500_sp/valid/speech_shape
|
119 |
+
- exp/slu_stats_raw_en_bpe500_sp/valid/text_shape.bpe
|
120 |
+
batch_type: numel
|
121 |
+
valid_batch_type: null
|
122 |
+
fold_length:
|
123 |
+
- 80000
|
124 |
+
- 150
|
125 |
+
sort_in_batch: descending
|
126 |
+
shuffle_within_batch: false
|
127 |
+
sort_batch: descending
|
128 |
+
multiple_iterator: false
|
129 |
+
chunk_length: 500
|
130 |
+
chunk_shift_ratio: 0.5
|
131 |
+
num_cache_chunks: 1024
|
132 |
+
chunk_excluded_key_prefixes: []
|
133 |
+
chunk_default_fs: null
|
134 |
+
train_data_path_and_name_and_type:
|
135 |
+
- - dump/raw/train_sp/wav.scp
|
136 |
+
- speech
|
137 |
+
- kaldi_ark
|
138 |
+
- - dump/raw/train_sp/text
|
139 |
+
- text
|
140 |
+
- text
|
141 |
+
valid_data_path_and_name_and_type:
|
142 |
+
- - dump/raw/devel/wav.scp
|
143 |
+
- speech
|
144 |
+
- kaldi_ark
|
145 |
+
- - dump/raw/devel/text
|
146 |
+
- text
|
147 |
+
- text
|
148 |
+
allow_variable_data_keys: false
|
149 |
+
max_cache_size: 0.0
|
150 |
+
max_cache_fd: 32
|
151 |
+
allow_multi_rates: false
|
152 |
+
valid_max_cache_size: null
|
153 |
+
exclude_weight_decay: false
|
154 |
+
exclude_weight_decay_conf: {}
|
155 |
+
optim: adam
|
156 |
+
optim_conf:
|
157 |
+
lr: 0.002
|
158 |
+
weight_decay: 1.0e-06
|
159 |
+
scheduler: warmuplr
|
160 |
+
scheduler_conf:
|
161 |
+
warmup_steps: 5000
|
162 |
+
token_list:
|
163 |
+
- <blank>
|
164 |
+
- <unk>
|
165 |
+
- '[sep]'
|
166 |
+
- '"'
|
167 |
+
- s
|
168 |
+
- ▁
|
169 |
+
- ▁the
|
170 |
+
- ','
|
171 |
+
- t
|
172 |
+
- d
|
173 |
+
- ▁a
|
174 |
+
- .
|
175 |
+
- ing
|
176 |
+
- o
|
177 |
+
- e
|
178 |
+
- ▁to
|
179 |
+
- a
|
180 |
+
- ▁and
|
181 |
+
- y
|
182 |
+
- n
|
183 |
+
- ▁of
|
184 |
+
- r
|
185 |
+
- ▁in
|
186 |
+
- u
|
187 |
+
- i
|
188 |
+
- m
|
189 |
+
- p
|
190 |
+
- c
|
191 |
+
- er
|
192 |
+
- g
|
193 |
+
- l
|
194 |
+
- al
|
195 |
+
- re
|
196 |
+
- ed
|
197 |
+
- b
|
198 |
+
- ''''
|
199 |
+
- ar
|
200 |
+
- k
|
201 |
+
- in
|
202 |
+
- f
|
203 |
+
- ▁"
|
204 |
+
- le
|
205 |
+
- 'on'
|
206 |
+
- v
|
207 |
+
- or
|
208 |
+
- th
|
209 |
+
- '-'
|
210 |
+
- ▁c
|
211 |
+
- en
|
212 |
+
- ▁f
|
213 |
+
- ▁--
|
214 |
+
- ▁we
|
215 |
+
- ▁for
|
216 |
+
- ▁how
|
217 |
+
- ly
|
218 |
+
- ▁re
|
219 |
+
- se
|
220 |
+
- ▁that
|
221 |
+
- es
|
222 |
+
- w
|
223 |
+
- ic
|
224 |
+
- st
|
225 |
+
- ▁w
|
226 |
+
- ▁be
|
227 |
+
- ri
|
228 |
+
- an
|
229 |
+
- ra
|
230 |
+
- ve
|
231 |
+
- ce
|
232 |
+
- ur
|
233 |
+
- ▁by
|
234 |
+
- ▁it
|
235 |
+
- li
|
236 |
+
- ▁de
|
237 |
+
- '?'
|
238 |
+
- it
|
239 |
+
- ch
|
240 |
+
- ent
|
241 |
+
- ▁is
|
242 |
+
- ter
|
243 |
+
- el
|
244 |
+
- ▁on
|
245 |
+
- ▁e
|
246 |
+
- ▁he
|
247 |
+
- ▁co
|
248 |
+
- ▁an
|
249 |
+
- ▁ma
|
250 |
+
- ▁st
|
251 |
+
- ll
|
252 |
+
- ▁with
|
253 |
+
- ▁can
|
254 |
+
- il
|
255 |
+
- ▁you
|
256 |
+
- ▁us
|
257 |
+
- ation
|
258 |
+
- te
|
259 |
+
- ▁this
|
260 |
+
- ▁b
|
261 |
+
- ▁do
|
262 |
+
- ▁g
|
263 |
+
- me
|
264 |
+
- ▁what
|
265 |
+
- ck
|
266 |
+
- ▁from
|
267 |
+
- ate
|
268 |
+
- ▁p
|
269 |
+
- z
|
270 |
+
- la
|
271 |
+
- ▁mo
|
272 |
+
- ▁di
|
273 |
+
- ive
|
274 |
+
- mp
|
275 |
+
- ▁talk
|
276 |
+
- ity
|
277 |
+
- vi
|
278 |
+
- ta
|
279 |
+
- at
|
280 |
+
- ge
|
281 |
+
- ▁tr
|
282 |
+
- ▁she
|
283 |
+
- ▁our
|
284 |
+
- ▁pa
|
285 |
+
- ci
|
286 |
+
- et
|
287 |
+
- h
|
288 |
+
- ▁su
|
289 |
+
- ver
|
290 |
+
- ▁world
|
291 |
+
- pe
|
292 |
+
- ▁about
|
293 |
+
- ▁me
|
294 |
+
- ▁so
|
295 |
+
- and
|
296 |
+
- ▁con
|
297 |
+
- tion
|
298 |
+
- de
|
299 |
+
- ir
|
300 |
+
- ▁her
|
301 |
+
- im
|
302 |
+
- ':'
|
303 |
+
- ▁his
|
304 |
+
- ies
|
305 |
+
- ▁po
|
306 |
+
- ▁are
|
307 |
+
- ect
|
308 |
+
- lo
|
309 |
+
- ▁your
|
310 |
+
- un
|
311 |
+
- ist
|
312 |
+
- hi
|
313 |
+
- ▁mi
|
314 |
+
- x
|
315 |
+
- id
|
316 |
+
- ment
|
317 |
+
- ol
|
318 |
+
- ul
|
319 |
+
- ti
|
320 |
+
- ne
|
321 |
+
- qu
|
322 |
+
- ▁but
|
323 |
+
- ▁ca
|
324 |
+
- ▁fa
|
325 |
+
- ▁as
|
326 |
+
- ▁un
|
327 |
+
- ers
|
328 |
+
- ight
|
329 |
+
- ▁says
|
330 |
+
- '0'
|
331 |
+
- ng
|
332 |
+
- op
|
333 |
+
- '1'
|
334 |
+
- ▁k
|
335 |
+
- ad
|
336 |
+
- j
|
337 |
+
- ma
|
338 |
+
- ▁pro
|
339 |
+
- ▁work
|
340 |
+
- ▁ba
|
341 |
+
- ▁share
|
342 |
+
- ▁new
|
343 |
+
- ▁more
|
344 |
+
- ▁vi
|
345 |
+
- ▁sa
|
346 |
+
- ▁at
|
347 |
+
- ▁la
|
348 |
+
- ut
|
349 |
+
- bi
|
350 |
+
- sion
|
351 |
+
- ▁ho
|
352 |
+
- na
|
353 |
+
- act
|
354 |
+
- age
|
355 |
+
- ke
|
356 |
+
- if
|
357 |
+
- ▁bo
|
358 |
+
- ▁br
|
359 |
+
- ▁ha
|
360 |
+
- ▁no
|
361 |
+
- co
|
362 |
+
- ▁lo
|
363 |
+
- mi
|
364 |
+
- ▁make
|
365 |
+
- ▁people
|
366 |
+
- ▁why
|
367 |
+
- ant
|
368 |
+
- ▁their
|
369 |
+
- ▁i
|
370 |
+
- ▁life
|
371 |
+
- ▁all
|
372 |
+
- ting
|
373 |
+
- ▁human
|
374 |
+
- ▁have
|
375 |
+
- om
|
376 |
+
- )
|
377 |
+
- ▁(
|
378 |
+
- ▁help
|
379 |
+
- ▁ted
|
380 |
+
- wa
|
381 |
+
- sh
|
382 |
+
- ▁da
|
383 |
+
- ▁le
|
384 |
+
- ▁out
|
385 |
+
- ph
|
386 |
+
- ical
|
387 |
+
- ▁way
|
388 |
+
- ff
|
389 |
+
- ▁ro
|
390 |
+
- able
|
391 |
+
- ▁some
|
392 |
+
- est
|
393 |
+
- ure
|
394 |
+
- em
|
395 |
+
- ho
|
396 |
+
- ▁ex
|
397 |
+
- gen
|
398 |
+
- ha
|
399 |
+
- ia
|
400 |
+
- ine
|
401 |
+
- ▁into
|
402 |
+
- ca
|
403 |
+
- ▁was
|
404 |
+
- ▁who
|
405 |
+
- ther
|
406 |
+
- ▁they
|
407 |
+
- ow
|
408 |
+
- he
|
409 |
+
- ▁one
|
410 |
+
- ▁when
|
411 |
+
- form
|
412 |
+
- ▁pre
|
413 |
+
- ni
|
414 |
+
- ▁could
|
415 |
+
- ▁like
|
416 |
+
- ▁per
|
417 |
+
- ▁up
|
418 |
+
- ance
|
419 |
+
- com
|
420 |
+
- ▁go
|
421 |
+
- ion
|
422 |
+
- tor
|
423 |
+
- ▁fe
|
424 |
+
- ▁ra
|
425 |
+
- ▁or
|
426 |
+
- ▁en
|
427 |
+
- ▁change
|
428 |
+
- tic
|
429 |
+
- ▁every
|
430 |
+
- ▁jo
|
431 |
+
- ence
|
432 |
+
- ▁not
|
433 |
+
- ▁art
|
434 |
+
- one
|
435 |
+
- use
|
436 |
+
- ous
|
437 |
+
- ▁plan
|
438 |
+
- ▁music
|
439 |
+
- ▁exp
|
440 |
+
- und
|
441 |
+
- ▁ne
|
442 |
+
- um
|
443 |
+
- ative
|
444 |
+
- pp
|
445 |
+
- ▁need
|
446 |
+
- tro
|
447 |
+
- directed
|
448 |
+
- ▁learn
|
449 |
+
- ▁narrate
|
450 |
+
- ▁has
|
451 |
+
- lar
|
452 |
+
- '].'
|
453 |
+
- man
|
454 |
+
- ▁car
|
455 |
+
- ▁future
|
456 |
+
- ▁real
|
457 |
+
- ▁time
|
458 |
+
- ize
|
459 |
+
- ▁live
|
460 |
+
- ber
|
461 |
+
- ▁mar
|
462 |
+
- ▁ga
|
463 |
+
- ▁take
|
464 |
+
- ▁dr
|
465 |
+
- ful
|
466 |
+
- ▁get
|
467 |
+
- ▁shows
|
468 |
+
- day
|
469 |
+
- ▁cha
|
470 |
+
- ▁than
|
471 |
+
- ▁know
|
472 |
+
- ian
|
473 |
+
- ▁see
|
474 |
+
- ▁just
|
475 |
+
- '2'
|
476 |
+
- ▁other
|
477 |
+
- old
|
478 |
+
- ▁design
|
479 |
+
- ▁chi
|
480 |
+
- ▁build
|
481 |
+
- ious
|
482 |
+
- ▁most
|
483 |
+
- ▁si
|
484 |
+
- ▁will
|
485 |
+
- ▁power
|
486 |
+
- ▁think
|
487 |
+
- port
|
488 |
+
- ▁over
|
489 |
+
- ▁ja
|
490 |
+
- ish
|
491 |
+
- ▁climate
|
492 |
+
- ▁sha
|
493 |
+
- ▁through
|
494 |
+
- less
|
495 |
+
- '3'
|
496 |
+
- ▁my
|
497 |
+
- ▁where
|
498 |
+
- ▁global
|
499 |
+
- ▁health
|
500 |
+
- ▁pri
|
501 |
+
- ▁20
|
502 |
+
- ▁story
|
503 |
+
- gu
|
504 |
+
- ugh
|
505 |
+
- ▁create
|
506 |
+
- ▁look
|
507 |
+
- ▁trans
|
508 |
+
- ▁har
|
509 |
+
- ▁even
|
510 |
+
- ▁part
|
511 |
+
- ▁years
|
512 |
+
- ▁lead
|
513 |
+
- side
|
514 |
+
- low
|
515 |
+
- long
|
516 |
+
- ▁technolog
|
517 |
+
- ness
|
518 |
+
- '5'
|
519 |
+
- ▁call
|
520 |
+
- ▁sc
|
521 |
+
- ▁system
|
522 |
+
- '9'
|
523 |
+
- line
|
524 |
+
- ▁brain
|
525 |
+
- ▁data
|
526 |
+
- ▁own
|
527 |
+
- ition
|
528 |
+
- ▁explains
|
529 |
+
- ▁tell
|
530 |
+
- ▁explore
|
531 |
+
- ▁start
|
532 |
+
- ▁ru
|
533 |
+
- ▁which
|
534 |
+
- ▁anderson
|
535 |
+
- ▁find
|
536 |
+
- ▁hu
|
537 |
+
- ▁women
|
538 |
+
- ▁better
|
539 |
+
- ▁idea
|
540 |
+
- ▁history
|
541 |
+
- ▁research
|
542 |
+
- ▁science
|
543 |
+
- ism
|
544 |
+
- ▁first
|
545 |
+
- ▁grow
|
546 |
+
- ▁right
|
547 |
+
- clu
|
548 |
+
- ▁space
|
549 |
+
- ▁develop
|
550 |
+
- ▁problem
|
551 |
+
- ▁two
|
552 |
+
- ▁earth
|
553 |
+
- ologist
|
554 |
+
- ▁many
|
555 |
+
- ▁should
|
556 |
+
- ▁three
|
557 |
+
- ▁fellow
|
558 |
+
- ▁social
|
559 |
+
- ▁africa
|
560 |
+
- ▁...
|
561 |
+
- '4'
|
562 |
+
- ▁addis
|
563 |
+
- ▁powerful
|
564 |
+
- ▁found
|
565 |
+
- ▁under
|
566 |
+
- ▁understand
|
567 |
+
- ▁after
|
568 |
+
- ▁stories
|
569 |
+
- ▁around
|
570 |
+
- ▁personal
|
571 |
+
- ▁project
|
572 |
+
- ▁between
|
573 |
+
- ▁question
|
574 |
+
- ▁play
|
575 |
+
- ▁scientist
|
576 |
+
- ▁happen
|
577 |
+
- ▁good
|
578 |
+
- ▁produc
|
579 |
+
- ▁experience
|
580 |
+
- ▁step
|
581 |
+
- ▁america
|
582 |
+
- '8'
|
583 |
+
- ▁great
|
584 |
+
- ▁down
|
585 |
+
- ▁high
|
586 |
+
- ▁would
|
587 |
+
- ▁turn
|
588 |
+
- ▁surprising
|
589 |
+
- ▁imagin
|
590 |
+
- ▁teach
|
591 |
+
- cross
|
592 |
+
- ▁place
|
593 |
+
- ▁medic
|
594 |
+
- ▁million
|
595 |
+
- ▁things
|
596 |
+
- '7'
|
597 |
+
- ▁reveal
|
598 |
+
- ▁without
|
599 |
+
- ▁challenge
|
600 |
+
- ▁next
|
601 |
+
- ▁each
|
602 |
+
- ▁studio
|
603 |
+
- organ
|
604 |
+
- '6'
|
605 |
+
- ▁business
|
606 |
+
- ▁much
|
607 |
+
- ▁show
|
608 |
+
- ▁conversation
|
609 |
+
- ▁energy
|
610 |
+
- ▁school
|
611 |
+
- ▁ocean
|
612 |
+
- ▁while
|
613 |
+
- source
|
614 |
+
- ization
|
615 |
+
- ▁break
|
616 |
+
- ▁robot
|
617 |
+
- ▁disease
|
618 |
+
- ▁behind
|
619 |
+
- ability
|
620 |
+
- ▁team
|
621 |
+
- ▁chris
|
622 |
+
- ▁become
|
623 |
+
- ▁solution
|
624 |
+
- ▁protect
|
625 |
+
- ▁collect
|
626 |
+
- ▁different
|
627 |
+
- ▁those
|
628 |
+
- ▁connect
|
629 |
+
- ▁architect
|
630 |
+
- ▁language
|
631 |
+
- ▁simple
|
632 |
+
- ▁solve
|
633 |
+
- ▁before
|
634 |
+
- ▁community
|
635 |
+
- ▁country
|
636 |
+
- ▁secret
|
637 |
+
- ▁keep
|
638 |
+
- ▁food
|
639 |
+
- ▁thought
|
640 |
+
- ▁discover
|
641 |
+
- ▁environment
|
642 |
+
- ▁government
|
643 |
+
- ▁public
|
644 |
+
- ;
|
645 |
+
- '!'
|
646 |
+
- /
|
647 |
+
- q
|
648 |
+
- '%'
|
649 |
+
- '@'
|
650 |
+
- ']'
|
651 |
+
- +
|
652 |
+
- '&'
|
653 |
+
- '|'
|
654 |
+
- _
|
655 |
+
- (
|
656 |
+
- '"'
|
657 |
+
- $
|
658 |
+
- '*'
|
659 |
+
- '='
|
660 |
+
- '['
|
661 |
+
- '`'
|
662 |
+
- <sos/eos>
|
663 |
+
transcript_token_list: null
|
664 |
+
two_pass: false
|
665 |
+
pre_postencoder_norm: false
|
666 |
+
init: null
|
667 |
+
input_size: null
|
668 |
+
ctc_conf:
|
669 |
+
dropout_rate: 0.0
|
670 |
+
ctc_type: builtin
|
671 |
+
reduce: true
|
672 |
+
ignore_nan_grad: null
|
673 |
+
zero_infinity: true
|
674 |
+
brctc_risk_strategy: exp
|
675 |
+
brctc_group_strategy: end
|
676 |
+
brctc_risk_factor: 0.0
|
677 |
+
joint_net_conf: null
|
678 |
+
use_preprocessor: true
|
679 |
+
token_type: bpe
|
680 |
+
bpemodel: data/en_token_list/bpe_unigram500/bpe.model
|
681 |
+
non_linguistic_symbols: null
|
682 |
+
cleaner: null
|
683 |
+
g2p: null
|
684 |
+
speech_volume_normalize: null
|
685 |
+
rir_scp: null
|
686 |
+
rir_apply_prob: 1.0
|
687 |
+
noise_scp: null
|
688 |
+
noise_apply_prob: 1.0
|
689 |
+
noise_db_range: '13_15'
|
690 |
+
short_noise_thres: 0.5
|
691 |
+
frontend: default
|
692 |
+
frontend_conf:
|
693 |
+
n_fft: 512
|
694 |
+
win_length: 400
|
695 |
+
hop_length: 160
|
696 |
+
fs: 16k
|
697 |
+
specaug: specaug
|
698 |
+
specaug_conf:
|
699 |
+
apply_time_warp: false
|
700 |
+
time_warp_window: 5
|
701 |
+
time_warp_mode: bicubic
|
702 |
+
apply_freq_mask: true
|
703 |
+
freq_mask_width_range:
|
704 |
+
- 0
|
705 |
+
- 27
|
706 |
+
num_freq_mask: 2
|
707 |
+
apply_time_mask: true
|
708 |
+
time_mask_width_ratio_range:
|
709 |
+
- 0.0
|
710 |
+
- 0.05
|
711 |
+
num_time_mask: 10
|
712 |
+
normalize: global_mvn
|
713 |
+
normalize_conf:
|
714 |
+
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
|
715 |
+
model: espnet
|
716 |
+
model_conf:
|
717 |
+
ctc_weight: 0.0
|
718 |
+
lsm_weight: 0.1
|
719 |
+
length_normalized_loss: false
|
720 |
+
weighted_sum: true
|
721 |
+
extract_feats_in_collect_stats: false
|
722 |
+
preencoder: null
|
723 |
+
preencoder_conf: {}
|
724 |
+
encoder: e_branchformer
|
725 |
+
encoder_conf:
|
726 |
+
output_size: 1024
|
727 |
+
attention_heads: 16
|
728 |
+
attention_layer_type: selfattn
|
729 |
+
pos_enc_layer_type: abs_pos
|
730 |
+
rel_pos_type: latest
|
731 |
+
cgmlp_linear_units: 4096
|
732 |
+
cgmlp_conv_kernel: 31
|
733 |
+
use_linear_after_conv: false
|
734 |
+
gate_activation: identity
|
735 |
+
num_blocks: 18
|
736 |
+
dropout_rate: 0.2
|
737 |
+
positional_dropout_rate: 0.2
|
738 |
+
attention_dropout_rate: 0.2
|
739 |
+
input_layer: conv2d
|
740 |
+
layer_drop_rate: 0.0
|
741 |
+
linear_units: 4096
|
742 |
+
positionwise_layer_type: linear
|
743 |
+
use_ffn: true
|
744 |
+
macaron_ffn: true
|
745 |
+
merge_conv_kernel: 31
|
746 |
+
prepostencoder: linear
|
747 |
+
prepostencoder_conf:
|
748 |
+
input_size: 1024
|
749 |
+
output_size: 80
|
750 |
+
postencoder: conformer_full
|
751 |
+
postencoder_conf:
|
752 |
+
output_size: 256
|
753 |
+
attention_heads: 4
|
754 |
+
linear_units: 1024
|
755 |
+
num_blocks: 12
|
756 |
+
dropout_rate: 0.1
|
757 |
+
positional_dropout_rate: 0.2
|
758 |
+
attention_dropout_rate: 0.1
|
759 |
+
input_layer: conv2d1
|
760 |
+
normalize_before: true
|
761 |
+
macaron_style: true
|
762 |
+
rel_pos_type: latest
|
763 |
+
pos_enc_layer_type: rel_pos
|
764 |
+
selfattention_layer_type: rel_selfattn
|
765 |
+
activation_type: swish
|
766 |
+
use_cnn_module: true
|
767 |
+
cnn_module_kernel: 31
|
768 |
+
deliberationencoder: null
|
769 |
+
deliberationencoder_conf: {}
|
770 |
+
decoder: transformer
|
771 |
+
decoder_conf:
|
772 |
+
attention_heads: 4
|
773 |
+
linear_units: 2048
|
774 |
+
num_blocks: 6
|
775 |
+
dropout_rate: 0.1
|
776 |
+
positional_dropout_rate: 0.1
|
777 |
+
self_attention_dropout_rate: 0.1
|
778 |
+
src_attention_dropout_rate: 0.1
|
779 |
+
postdecoder: null
|
780 |
+
postdecoder_conf: {}
|
781 |
+
required:
|
782 |
+
- output_dir
|
783 |
+
- token_list
|
784 |
+
version: '202310'
|
785 |
+
distributed: true
|
786 |
+
```
|
787 |
+
|
788 |
+
</details>
|
789 |
+
|
790 |
+
|
791 |
+
|
792 |
+
### Citing ESPnet
|
793 |
+
|
794 |
+
```BibTex
|
795 |
+
@inproceedings{watanabe2018espnet,
|
796 |
+
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},
|
797 |
+
title={{ESPnet}: End-to-End Speech Processing Toolkit},
|
798 |
+
year={2018},
|
799 |
+
booktitle={Proceedings of Interspeech},
|
800 |
+
pages={2207--2211},
|
801 |
+
doi={10.21437/Interspeech.2018-1456},
|
802 |
+
url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
|
803 |
+
}
|
804 |
+
|
805 |
+
|
806 |
+
|
807 |
+
|
808 |
+
|
809 |
+
|
810 |
+
```
|
811 |
+
|
812 |
+
or arXiv:
|
813 |
+
|
814 |
+
```bibtex
|
815 |
+
@misc{watanabe2018espnet,
|
816 |
+
title={ESPnet: End-to-End Speech Processing Toolkit},
|
817 |
+
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},
|
818 |
+
year={2018},
|
819 |
+
eprint={1804.00015},
|
820 |
+
archivePrefix={arXiv},
|
821 |
+
primaryClass={cs.CL}
|
822 |
+
}
|
823 |
+
```
|
data/en_token_list/bpe_unigram500/bpe.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9070b5fe2b321a13162f2aa8159c6f637402b46fb032510dbb1dabfcf0afa24f
|
3 |
+
size 244766
|
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/RESULTS.md
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
RESULT 23.72001537034864 5.0727273644425015 16.49858244502415 22.432552212134674 83.55819195438502
|
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/config.yaml
ADDED
@@ -0,0 +1,748 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
config: conf//train_asr_owsm_weighted_dropout.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_dropout_raw_en_bpe500_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: 55179
|
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: 25
|
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: 1
|
48 |
+
no_forward_run: false
|
49 |
+
resume: true
|
50 |
+
train_dtype: float32
|
51 |
+
use_amp: false
|
52 |
+
log_interval: 100
|
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/espnet_slue_PR/espnet/egs2/tedlium3/asr1/exp/asr_train_asr_owsm_weighted_dropout_raw_en_bpe500/valid.acc.ave_10best.pth:::ctc
|
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: 12000000
|
76 |
+
valid_batch_bins: null
|
77 |
+
train_shape_file:
|
78 |
+
- exp/slu_stats_raw_en_bpe500_sp/train/speech_shape
|
79 |
+
- exp/slu_stats_raw_en_bpe500_sp/train/text_shape.bpe
|
80 |
+
valid_shape_file:
|
81 |
+
- exp/slu_stats_raw_en_bpe500_sp/valid/speech_shape
|
82 |
+
- exp/slu_stats_raw_en_bpe500_sp/valid/text_shape.bpe
|
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 |
+
- kaldi_ark
|
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 |
+
- kaldi_ark
|
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 |
+
- '[sep]'
|
129 |
+
- '"'
|
130 |
+
- s
|
131 |
+
- ▁
|
132 |
+
- ▁the
|
133 |
+
- ','
|
134 |
+
- t
|
135 |
+
- d
|
136 |
+
- ▁a
|
137 |
+
- .
|
138 |
+
- ing
|
139 |
+
- o
|
140 |
+
- e
|
141 |
+
- ▁to
|
142 |
+
- a
|
143 |
+
- ▁and
|
144 |
+
- y
|
145 |
+
- n
|
146 |
+
- ▁of
|
147 |
+
- r
|
148 |
+
- ▁in
|
149 |
+
- u
|
150 |
+
- i
|
151 |
+
- m
|
152 |
+
- p
|
153 |
+
- c
|
154 |
+
- er
|
155 |
+
- g
|
156 |
+
- l
|
157 |
+
- al
|
158 |
+
- re
|
159 |
+
- ed
|
160 |
+
- b
|
161 |
+
- ''''
|
162 |
+
- ar
|
163 |
+
- k
|
164 |
+
- in
|
165 |
+
- f
|
166 |
+
- ▁"
|
167 |
+
- le
|
168 |
+
- 'on'
|
169 |
+
- v
|
170 |
+
- or
|
171 |
+
- th
|
172 |
+
- '-'
|
173 |
+
- ▁c
|
174 |
+
- en
|
175 |
+
- ▁f
|
176 |
+
- ▁--
|
177 |
+
- ▁we
|
178 |
+
- ▁for
|
179 |
+
- ▁how
|
180 |
+
- ly
|
181 |
+
- ▁re
|
182 |
+
- se
|
183 |
+
- ▁that
|
184 |
+
- es
|
185 |
+
- w
|
186 |
+
- ic
|
187 |
+
- st
|
188 |
+
- ▁w
|
189 |
+
- ▁be
|
190 |
+
- ri
|
191 |
+
- an
|
192 |
+
- ra
|
193 |
+
- ve
|
194 |
+
- ce
|
195 |
+
- ur
|
196 |
+
- ▁by
|
197 |
+
- ▁it
|
198 |
+
- li
|
199 |
+
- ▁de
|
200 |
+
- '?'
|
201 |
+
- it
|
202 |
+
- ch
|
203 |
+
- ent
|
204 |
+
- ▁is
|
205 |
+
- ter
|
206 |
+
- el
|
207 |
+
- ▁on
|
208 |
+
- ▁e
|
209 |
+
- ▁he
|
210 |
+
- ▁co
|
211 |
+
- ▁an
|
212 |
+
- ▁ma
|
213 |
+
- ▁st
|
214 |
+
- ll
|
215 |
+
- ▁with
|
216 |
+
- ▁can
|
217 |
+
- il
|
218 |
+
- ▁you
|
219 |
+
- ▁us
|
220 |
+
- ation
|
221 |
+
- te
|
222 |
+
- ▁this
|
223 |
+
- ▁b
|
224 |
+
- ▁do
|
225 |
+
- ▁g
|
226 |
+
- me
|
227 |
+
- ▁what
|
228 |
+
- ck
|
229 |
+
- ▁from
|
230 |
+
- ate
|
231 |
+
- ▁p
|
232 |
+
- z
|
233 |
+
- la
|
234 |
+
- ▁mo
|
235 |
+
- ▁di
|
236 |
+
- ive
|
237 |
+
- mp
|
238 |
+
- ▁talk
|
239 |
+
- ity
|
240 |
+
- vi
|
241 |
+
- ta
|
242 |
+
- at
|
243 |
+
- ge
|
244 |
+
- ▁tr
|
245 |
+
- ▁she
|
246 |
+
- ▁our
|
247 |
+
- ▁pa
|
248 |
+
- ci
|
249 |
+
- et
|
250 |
+
- h
|
251 |
+
- ▁su
|
252 |
+
- ver
|
253 |
+
- ▁world
|
254 |
+
- pe
|
255 |
+
- ▁about
|
256 |
+
- ▁me
|
257 |
+
- ▁so
|
258 |
+
- and
|
259 |
+
- ▁con
|
260 |
+
- tion
|
261 |
+
- de
|
262 |
+
- ir
|
263 |
+
- ▁her
|
264 |
+
- im
|
265 |
+
- ':'
|
266 |
+
- ▁his
|
267 |
+
- ies
|
268 |
+
- ▁po
|
269 |
+
- ▁are
|
270 |
+
- ect
|
271 |
+
- lo
|
272 |
+
- ▁your
|
273 |
+
- un
|
274 |
+
- ist
|
275 |
+
- hi
|
276 |
+
- ▁mi
|
277 |
+
- x
|
278 |
+
- id
|
279 |
+
- ment
|
280 |
+
- ol
|
281 |
+
- ul
|
282 |
+
- ti
|
283 |
+
- ne
|
284 |
+
- qu
|
285 |
+
- ▁but
|
286 |
+
- ▁ca
|
287 |
+
- ▁fa
|
288 |
+
- ▁as
|
289 |
+
- ▁un
|
290 |
+
- ers
|
291 |
+
- ight
|
292 |
+
- ▁says
|
293 |
+
- '0'
|
294 |
+
- ng
|
295 |
+
- op
|
296 |
+
- '1'
|
297 |
+
- ▁k
|
298 |
+
- ad
|
299 |
+
- j
|
300 |
+
- ma
|
301 |
+
- ▁pro
|
302 |
+
- ▁work
|
303 |
+
- ▁ba
|
304 |
+
- ▁share
|
305 |
+
- ▁new
|
306 |
+
- ▁more
|
307 |
+
- ▁vi
|
308 |
+
- ▁sa
|
309 |
+
- ▁at
|
310 |
+
- ▁la
|
311 |
+
- ut
|
312 |
+
- bi
|
313 |
+
- sion
|
314 |
+
- ▁ho
|
315 |
+
- na
|
316 |
+
- act
|
317 |
+
- age
|
318 |
+
- ke
|
319 |
+
- if
|
320 |
+
- ▁bo
|
321 |
+
- ▁br
|
322 |
+
- ▁ha
|
323 |
+
- ▁no
|
324 |
+
- co
|
325 |
+
- ▁lo
|
326 |
+
- mi
|
327 |
+
- ▁make
|
328 |
+
- ▁people
|
329 |
+
- ▁why
|
330 |
+
- ant
|
331 |
+
- ▁their
|
332 |
+
- ▁i
|
333 |
+
- ▁life
|
334 |
+
- ▁all
|
335 |
+
- ting
|
336 |
+
- ▁human
|
337 |
+
- ▁have
|
338 |
+
- om
|
339 |
+
- )
|
340 |
+
- ▁(
|
341 |
+
- ▁help
|
342 |
+
- ▁ted
|
343 |
+
- wa
|
344 |
+
- sh
|
345 |
+
- ▁da
|
346 |
+
- ▁le
|
347 |
+
- ▁out
|
348 |
+
- ph
|
349 |
+
- ical
|
350 |
+
- ▁way
|
351 |
+
- ff
|
352 |
+
- ▁ro
|
353 |
+
- able
|
354 |
+
- ▁some
|
355 |
+
- est
|
356 |
+
- ure
|
357 |
+
- em
|
358 |
+
- ho
|
359 |
+
- ▁ex
|
360 |
+
- gen
|
361 |
+
- ha
|
362 |
+
- ia
|
363 |
+
- ine
|
364 |
+
- ▁into
|
365 |
+
- ca
|
366 |
+
- ▁was
|
367 |
+
- ▁who
|
368 |
+
- ther
|
369 |
+
- ▁they
|
370 |
+
- ow
|
371 |
+
- he
|
372 |
+
- ▁one
|
373 |
+
- ▁when
|
374 |
+
- form
|
375 |
+
- ▁pre
|
376 |
+
- ni
|
377 |
+
- ▁could
|
378 |
+
- ▁like
|
379 |
+
- ▁per
|
380 |
+
- ▁up
|
381 |
+
- ance
|
382 |
+
- com
|
383 |
+
- ▁go
|
384 |
+
- ion
|
385 |
+
- tor
|
386 |
+
- ▁fe
|
387 |
+
- ▁ra
|
388 |
+
- ▁or
|
389 |
+
- ▁en
|
390 |
+
- ▁change
|
391 |
+
- tic
|
392 |
+
- ▁every
|
393 |
+
- ▁jo
|
394 |
+
- ence
|
395 |
+
- ▁not
|
396 |
+
- ▁art
|
397 |
+
- one
|
398 |
+
- use
|
399 |
+
- ous
|
400 |
+
- ▁plan
|
401 |
+
- ▁music
|
402 |
+
- ▁exp
|
403 |
+
- und
|
404 |
+
- ▁ne
|
405 |
+
- um
|
406 |
+
- ative
|
407 |
+
- pp
|
408 |
+
- ▁need
|
409 |
+
- tro
|
410 |
+
- directed
|
411 |
+
- ▁learn
|
412 |
+
- ▁narrate
|
413 |
+
- ▁has
|
414 |
+
- lar
|
415 |
+
- '].'
|
416 |
+
- man
|
417 |
+
- ▁car
|
418 |
+
- ▁future
|
419 |
+
- ▁real
|
420 |
+
- ▁time
|
421 |
+
- ize
|
422 |
+
- ▁live
|
423 |
+
- ber
|
424 |
+
- ▁mar
|
425 |
+
- ▁ga
|
426 |
+
- ▁take
|
427 |
+
- ▁dr
|
428 |
+
- ful
|
429 |
+
- ▁get
|
430 |
+
- ▁shows
|
431 |
+
- day
|
432 |
+
- ▁cha
|
433 |
+
- ▁than
|
434 |
+
- ▁know
|
435 |
+
- ian
|
436 |
+
- ▁see
|
437 |
+
- ▁just
|
438 |
+
- '2'
|
439 |
+
- ▁other
|
440 |
+
- old
|
441 |
+
- ▁design
|
442 |
+
- ▁chi
|
443 |
+
- ▁build
|
444 |
+
- ious
|
445 |
+
- ▁most
|
446 |
+
- ▁si
|
447 |
+
- ▁will
|
448 |
+
- ▁power
|
449 |
+
- ▁think
|
450 |
+
- port
|
451 |
+
- ▁over
|
452 |
+
- ▁ja
|
453 |
+
- ish
|
454 |
+
- ▁climate
|
455 |
+
- ▁sha
|
456 |
+
- ▁through
|
457 |
+
- less
|
458 |
+
- '3'
|
459 |
+
- ▁my
|
460 |
+
- ▁where
|
461 |
+
- ▁global
|
462 |
+
- ▁health
|
463 |
+
- ▁pri
|
464 |
+
- ▁20
|
465 |
+
- ▁story
|
466 |
+
- gu
|
467 |
+
- ugh
|
468 |
+
- ▁create
|
469 |
+
- ▁look
|
470 |
+
- ▁trans
|
471 |
+
- ▁har
|
472 |
+
- ▁even
|
473 |
+
- ▁part
|
474 |
+
- ▁years
|
475 |
+
- ▁lead
|
476 |
+
- side
|
477 |
+
- low
|
478 |
+
- long
|
479 |
+
- ▁technolog
|
480 |
+
- ness
|
481 |
+
- '5'
|
482 |
+
- ▁call
|
483 |
+
- ▁sc
|
484 |
+
- ▁system
|
485 |
+
- '9'
|
486 |
+
- line
|
487 |
+
- ▁brain
|
488 |
+
- ▁data
|
489 |
+
- ▁own
|
490 |
+
- ition
|
491 |
+
- ▁explains
|
492 |
+
- ▁tell
|
493 |
+
- ▁explore
|
494 |
+
- ▁start
|
495 |
+
- ▁ru
|
496 |
+
- ▁which
|
497 |
+
- ▁anderson
|
498 |
+
- ▁find
|
499 |
+
- ▁hu
|
500 |
+
- ▁women
|
501 |
+
- ▁better
|
502 |
+
- ▁idea
|
503 |
+
- ▁history
|
504 |
+
- ▁research
|
505 |
+
- ▁science
|
506 |
+
- ism
|
507 |
+
- ▁first
|
508 |
+
- ▁grow
|
509 |
+
- ▁right
|
510 |
+
- clu
|
511 |
+
- ▁space
|
512 |
+
- ▁develop
|
513 |
+
- ▁problem
|
514 |
+
- ▁two
|
515 |
+
- ▁earth
|
516 |
+
- ologist
|
517 |
+
- ▁many
|
518 |
+
- ▁should
|
519 |
+
- ▁three
|
520 |
+
- ▁fellow
|
521 |
+
- ▁social
|
522 |
+
- ▁africa
|
523 |
+
- ▁...
|
524 |
+
- '4'
|
525 |
+
- ▁addis
|
526 |
+
- ▁powerful
|
527 |
+
- ▁found
|
528 |
+
- ▁under
|
529 |
+
- ▁understand
|
530 |
+
- ▁after
|
531 |
+
- ▁stories
|
532 |
+
- ▁around
|
533 |
+
- ▁personal
|
534 |
+
- ▁project
|
535 |
+
- ▁between
|
536 |
+
- ▁question
|
537 |
+
- ▁play
|
538 |
+
- ▁scientist
|
539 |
+
- ▁happen
|
540 |
+
- ▁good
|
541 |
+
- ▁produc
|
542 |
+
- ▁experience
|
543 |
+
- ▁step
|
544 |
+
- ▁america
|
545 |
+
- '8'
|
546 |
+
- ▁great
|
547 |
+
- ▁down
|
548 |
+
- ▁high
|
549 |
+
- ▁would
|
550 |
+
- ▁turn
|
551 |
+
- ▁surprising
|
552 |
+
- ▁imagin
|
553 |
+
- ▁teach
|
554 |
+
- cross
|
555 |
+
- ▁place
|
556 |
+
- ▁medic
|
557 |
+
- ▁million
|
558 |
+
- ▁things
|
559 |
+
- '7'
|
560 |
+
- ▁reveal
|
561 |
+
- ▁without
|
562 |
+
- ▁challenge
|
563 |
+
- ▁next
|
564 |
+
- ▁each
|
565 |
+
- ▁studio
|
566 |
+
- organ
|
567 |
+
- '6'
|
568 |
+
- ▁business
|
569 |
+
- ▁much
|
570 |
+
- ▁show
|
571 |
+
- ▁conversation
|
572 |
+
- ▁energy
|
573 |
+
- ▁school
|
574 |
+
- ▁ocean
|
575 |
+
- ▁while
|
576 |
+
- source
|
577 |
+
- ization
|
578 |
+
- ▁break
|
579 |
+
- ▁robot
|
580 |
+
- ▁disease
|
581 |
+
- ▁behind
|
582 |
+
- ability
|
583 |
+
- ▁team
|
584 |
+
- ▁chris
|
585 |
+
- ▁become
|
586 |
+
- ▁solution
|
587 |
+
- ▁protect
|
588 |
+
- ▁collect
|
589 |
+
- ▁different
|
590 |
+
- ▁those
|
591 |
+
- ▁connect
|
592 |
+
- ▁architect
|
593 |
+
- ▁language
|
594 |
+
- ▁simple
|
595 |
+
- ▁solve
|
596 |
+
- ▁before
|
597 |
+
- ▁community
|
598 |
+
- ▁country
|
599 |
+
- ▁secret
|
600 |
+
- ▁keep
|
601 |
+
- ▁food
|
602 |
+
- ▁thought
|
603 |
+
- ▁discover
|
604 |
+
- ▁environment
|
605 |
+
- ▁government
|
606 |
+
- ▁public
|
607 |
+
- ;
|
608 |
+
- '!'
|
609 |
+
- /
|
610 |
+
- q
|
611 |
+
- '%'
|
612 |
+
- '@'
|
613 |
+
- ']'
|
614 |
+
- +
|
615 |
+
- '&'
|
616 |
+
- '|'
|
617 |
+
- _
|
618 |
+
- (
|
619 |
+
- '"'
|
620 |
+
- $
|
621 |
+
- '*'
|
622 |
+
- '='
|
623 |
+
- '['
|
624 |
+
- '`'
|
625 |
+
- <sos/eos>
|
626 |
+
transcript_token_list: null
|
627 |
+
two_pass: false
|
628 |
+
pre_postencoder_norm: false
|
629 |
+
init: null
|
630 |
+
input_size: null
|
631 |
+
ctc_conf:
|
632 |
+
dropout_rate: 0.0
|
633 |
+
ctc_type: builtin
|
634 |
+
reduce: true
|
635 |
+
ignore_nan_grad: null
|
636 |
+
zero_infinity: true
|
637 |
+
brctc_risk_strategy: exp
|
638 |
+
brctc_group_strategy: end
|
639 |
+
brctc_risk_factor: 0.0
|
640 |
+
joint_net_conf: null
|
641 |
+
use_preprocessor: true
|
642 |
+
token_type: bpe
|
643 |
+
bpemodel: data/en_token_list/bpe_unigram500/bpe.model
|
644 |
+
non_linguistic_symbols: null
|
645 |
+
cleaner: null
|
646 |
+
g2p: null
|
647 |
+
speech_volume_normalize: null
|
648 |
+
rir_scp: null
|
649 |
+
rir_apply_prob: 1.0
|
650 |
+
noise_scp: null
|
651 |
+
noise_apply_prob: 1.0
|
652 |
+
noise_db_range: '13_15'
|
653 |
+
short_noise_thres: 0.5
|
654 |
+
frontend: default
|
655 |
+
frontend_conf:
|
656 |
+
n_fft: 512
|
657 |
+
win_length: 400
|
658 |
+
hop_length: 160
|
659 |
+
fs: 16k
|
660 |
+
specaug: specaug
|
661 |
+
specaug_conf:
|
662 |
+
apply_time_warp: false
|
663 |
+
time_warp_window: 5
|
664 |
+
time_warp_mode: bicubic
|
665 |
+
apply_freq_mask: true
|
666 |
+
freq_mask_width_range:
|
667 |
+
- 0
|
668 |
+
- 27
|
669 |
+
num_freq_mask: 2
|
670 |
+
apply_time_mask: true
|
671 |
+
time_mask_width_ratio_range:
|
672 |
+
- 0.0
|
673 |
+
- 0.05
|
674 |
+
num_time_mask: 10
|
675 |
+
normalize: global_mvn
|
676 |
+
normalize_conf:
|
677 |
+
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
|
678 |
+
model: espnet
|
679 |
+
model_conf:
|
680 |
+
ctc_weight: 0.0
|
681 |
+
lsm_weight: 0.1
|
682 |
+
length_normalized_loss: false
|
683 |
+
weighted_sum: true
|
684 |
+
extract_feats_in_collect_stats: false
|
685 |
+
preencoder: null
|
686 |
+
preencoder_conf: {}
|
687 |
+
encoder: e_branchformer
|
688 |
+
encoder_conf:
|
689 |
+
output_size: 1024
|
690 |
+
attention_heads: 16
|
691 |
+
attention_layer_type: selfattn
|
692 |
+
pos_enc_layer_type: abs_pos
|
693 |
+
rel_pos_type: latest
|
694 |
+
cgmlp_linear_units: 4096
|
695 |
+
cgmlp_conv_kernel: 31
|
696 |
+
use_linear_after_conv: false
|
697 |
+
gate_activation: identity
|
698 |
+
num_blocks: 18
|
699 |
+
dropout_rate: 0.2
|
700 |
+
positional_dropout_rate: 0.2
|
701 |
+
attention_dropout_rate: 0.2
|
702 |
+
input_layer: conv2d
|
703 |
+
layer_drop_rate: 0.0
|
704 |
+
linear_units: 4096
|
705 |
+
positionwise_layer_type: linear
|
706 |
+
use_ffn: true
|
707 |
+
macaron_ffn: true
|
708 |
+
merge_conv_kernel: 31
|
709 |
+
prepostencoder: linear
|
710 |
+
prepostencoder_conf:
|
711 |
+
input_size: 1024
|
712 |
+
output_size: 80
|
713 |
+
postencoder: conformer_full
|
714 |
+
postencoder_conf:
|
715 |
+
output_size: 256
|
716 |
+
attention_heads: 4
|
717 |
+
linear_units: 1024
|
718 |
+
num_blocks: 12
|
719 |
+
dropout_rate: 0.1
|
720 |
+
positional_dropout_rate: 0.2
|
721 |
+
attention_dropout_rate: 0.1
|
722 |
+
input_layer: conv2d1
|
723 |
+
normalize_before: true
|
724 |
+
macaron_style: true
|
725 |
+
rel_pos_type: latest
|
726 |
+
pos_enc_layer_type: rel_pos
|
727 |
+
selfattention_layer_type: rel_selfattn
|
728 |
+
activation_type: swish
|
729 |
+
use_cnn_module: true
|
730 |
+
cnn_module_kernel: 31
|
731 |
+
deliberationencoder: null
|
732 |
+
deliberationencoder_conf: {}
|
733 |
+
decoder: transformer
|
734 |
+
decoder_conf:
|
735 |
+
attention_heads: 4
|
736 |
+
linear_units: 2048
|
737 |
+
num_blocks: 6
|
738 |
+
dropout_rate: 0.1
|
739 |
+
positional_dropout_rate: 0.1
|
740 |
+
self_attention_dropout_rate: 0.1
|
741 |
+
src_attention_dropout_rate: 0.1
|
742 |
+
postdecoder: null
|
743 |
+
postdecoder_conf: {}
|
744 |
+
required:
|
745 |
+
- output_dir
|
746 |
+
- token_list
|
747 |
+
version: '202310'
|
748 |
+
distributed: true
|
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/acc.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/backward_time.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/cer.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/clip.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/forward_time.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/gpu_max_cached_mem_GB.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/grad_norm.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/iter_time.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/loss.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/loss_att.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/loss_scale.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/optim0_lr0.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/optim_step_time.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/train_time.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/images/wer.png
ADDED
exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/valid.acc.ave_10best.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0aefdf647fd9bef1d3d957e27d8a6e561671e081106dd4a45fdd69e01342948b
|
3 |
+
size 2381810530
|
meta.yaml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
espnet: '202310'
|
2 |
+
files:
|
3 |
+
slu_model_file: exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_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: 1715364701.23388
|
6 |
+
torch: 2.1.0+cu121
|
7 |
+
yaml_files:
|
8 |
+
slu_train_config: exp/slu_train_asr_owsm_weighted_dropout_raw_en_bpe500_sp/config.yaml
|