Commit
•
2833cba
1
Parent(s):
6bb1687
Add model files
Browse files- exp/asr_stats_raw_ml_bpe150_sp/train/feats_lengths_stats.npz +0 -0
- exp/asr_stats_raw_ml_bpe150_sp/train/feats_stats.npz +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/RESULTS.md +32 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/config.yaml +348 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/acc.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/backward_time.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/cer.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/cer_ctc.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/forward_time.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/gpu_max_cached_mem_GB.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/iter_time.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/loss.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/loss_att.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/loss_ctc.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/optim0_lr0.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/optim_step_time.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/train_time.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/wer.png +0 -0
- exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/valid.acc.ave_10best.pth +3 -0
- exp/lm_train_lm_ml_bpe150/config.yaml +267 -0
- exp/lm_train_lm_ml_bpe150/images/backward_time.png +0 -0
- exp/lm_train_lm_ml_bpe150/images/forward_time.png +0 -0
- exp/lm_train_lm_ml_bpe150/images/gpu_max_cached_mem_GB.png +0 -0
- exp/lm_train_lm_ml_bpe150/images/iter_time.png +0 -0
- exp/lm_train_lm_ml_bpe150/images/loss.png +0 -0
- exp/lm_train_lm_ml_bpe150/images/optim0_lr0.png +0 -0
- exp/lm_train_lm_ml_bpe150/images/optim_step_time.png +0 -0
- exp/lm_train_lm_ml_bpe150/images/train_time.png +0 -0
- exp/lm_train_lm_ml_bpe150/valid.loss.ave_1best.pth +1 -0
exp/asr_stats_raw_ml_bpe150_sp/train/feats_lengths_stats.npz
ADDED
Binary file (778 Bytes). View file
|
|
exp/asr_stats_raw_ml_bpe150_sp/train/feats_stats.npz
ADDED
Binary file (1.4 kB). View file
|
|
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/RESULTS.md
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<!-- Generated by scripts/utils/show_asr_result.sh -->
|
2 |
+
# RESULTS
|
3 |
+
## Environments
|
4 |
+
- date: `Sat Mar 19 20:34:49 UTC 2022`
|
5 |
+
- python version: `3.9.10 | packaged by conda-forge | (main, Feb 1 2022, 21:24:11) [GCC 9.4.0]`
|
6 |
+
- espnet version: `espnet 0.10.7a1`
|
7 |
+
- pytorch version: `pytorch 1.10.1`
|
8 |
+
- Git hash: `d2410457152872f63c51ee76ed746a6ea3153f09`
|
9 |
+
- Commit date: `Sat Mar 19 09:04:54 2022 +0000`
|
10 |
+
|
11 |
+
## asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp
|
12 |
+
### WER
|
13 |
+
|
14 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
15 |
+
|---|---|---|---|---|---|---|---|---|
|
16 |
+
|decode_asr_lm_lm_train_lm_ml_bpe150_valid.loss.ave_asr_model_valid.acc.ave/dev_ml|369|2345|75.2|21.8|3.0|2.4|27.2|71.5|
|
17 |
+
|decode_asr_lm_lm_train_lm_ml_bpe150_valid.loss.ave_asr_model_valid.acc.ave/test_ml|1062|6136|67.0|28.7|4.3|2.6|35.6|71.8|
|
18 |
+
|
19 |
+
### CER
|
20 |
+
|
21 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
22 |
+
|---|---|---|---|---|---|---|---|---|
|
23 |
+
|decode_asr_lm_lm_train_lm_ml_bpe150_valid.loss.ave_asr_model_valid.acc.ave/dev_ml|369|21321|96.1|2.2|1.7|0.9|4.7|71.5|
|
24 |
+
|decode_asr_lm_lm_train_lm_ml_bpe150_valid.loss.ave_asr_model_valid.acc.ave/test_ml|1062|57065|93.5|3.2|3.3|1.3|7.7|71.8|
|
25 |
+
|
26 |
+
### TER
|
27 |
+
|
28 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
29 |
+
|---|---|---|---|---|---|---|---|---|
|
30 |
+
|decode_asr_lm_lm_train_lm_ml_bpe150_valid.loss.ave_asr_model_valid.acc.ave/dev_ml|369|13402|93.5|4.4|2.1|0.9|7.4|71.3|
|
31 |
+
|decode_asr_lm_lm_train_lm_ml_bpe150_valid.loss.ave_asr_model_valid.acc.ave/test_ml|1062|35911|89.9|6.3|3.8|1.3|11.4|70.4|
|
32 |
+
|
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/config.yaml
ADDED
@@ -0,0 +1,348 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
config: conf/tuning/train_asr_conformer_s3prlfrontend_hubert_fused.yaml
|
2 |
+
print_config: false
|
3 |
+
log_level: INFO
|
4 |
+
dry_run: false
|
5 |
+
iterator_type: sequence
|
6 |
+
output_dir: exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp
|
7 |
+
ngpu: 1
|
8 |
+
seed: 0
|
9 |
+
num_workers: 1
|
10 |
+
num_att_plot: 3
|
11 |
+
dist_backend: nccl
|
12 |
+
dist_init_method: env://
|
13 |
+
dist_world_size: null
|
14 |
+
dist_rank: null
|
15 |
+
local_rank: 0
|
16 |
+
dist_master_addr: null
|
17 |
+
dist_master_port: null
|
18 |
+
dist_launcher: null
|
19 |
+
multiprocessing_distributed: false
|
20 |
+
unused_parameters: false
|
21 |
+
sharded_ddp: false
|
22 |
+
cudnn_enabled: true
|
23 |
+
cudnn_benchmark: false
|
24 |
+
cudnn_deterministic: false
|
25 |
+
collect_stats: false
|
26 |
+
write_collected_feats: false
|
27 |
+
max_epoch: 50
|
28 |
+
patience: null
|
29 |
+
val_scheduler_criterion:
|
30 |
+
- valid
|
31 |
+
- loss
|
32 |
+
early_stopping_criterion:
|
33 |
+
- valid
|
34 |
+
- loss
|
35 |
+
- min
|
36 |
+
best_model_criterion:
|
37 |
+
- - valid
|
38 |
+
- acc
|
39 |
+
- max
|
40 |
+
keep_nbest_models: 10
|
41 |
+
nbest_averaging_interval: 0
|
42 |
+
grad_clip: 3
|
43 |
+
grad_clip_type: 2.0
|
44 |
+
grad_noise: false
|
45 |
+
accum_grad: 1
|
46 |
+
no_forward_run: false
|
47 |
+
resume: true
|
48 |
+
train_dtype: float32
|
49 |
+
use_amp: true
|
50 |
+
log_interval: null
|
51 |
+
use_matplotlib: true
|
52 |
+
use_tensorboard: true
|
53 |
+
use_wandb: false
|
54 |
+
wandb_project: null
|
55 |
+
wandb_id: null
|
56 |
+
wandb_entity: null
|
57 |
+
wandb_name: null
|
58 |
+
wandb_model_log_interval: -1
|
59 |
+
detect_anomaly: false
|
60 |
+
pretrain_path: null
|
61 |
+
init_param: []
|
62 |
+
ignore_init_mismatch: false
|
63 |
+
freeze_param: []
|
64 |
+
num_iters_per_epoch: null
|
65 |
+
batch_size: 20
|
66 |
+
valid_batch_size: null
|
67 |
+
batch_bins: 2000000
|
68 |
+
valid_batch_bins: null
|
69 |
+
train_shape_file:
|
70 |
+
- exp/asr_stats_raw_ml_bpe150_sp/train/speech_shape
|
71 |
+
- exp/asr_stats_raw_ml_bpe150_sp/train/text_shape.bpe
|
72 |
+
valid_shape_file:
|
73 |
+
- exp/asr_stats_raw_ml_bpe150_sp/valid/speech_shape
|
74 |
+
- exp/asr_stats_raw_ml_bpe150_sp/valid/text_shape.bpe
|
75 |
+
batch_type: numel
|
76 |
+
valid_batch_type: null
|
77 |
+
fold_length:
|
78 |
+
- 80000
|
79 |
+
- 150
|
80 |
+
sort_in_batch: descending
|
81 |
+
sort_batch: descending
|
82 |
+
multiple_iterator: false
|
83 |
+
chunk_length: 500
|
84 |
+
chunk_shift_ratio: 0.5
|
85 |
+
num_cache_chunks: 1024
|
86 |
+
train_data_path_and_name_and_type:
|
87 |
+
- - dump/raw/train_ml_sp/wav.scp
|
88 |
+
- speech
|
89 |
+
- sound
|
90 |
+
- - dump/raw/train_ml_sp/text
|
91 |
+
- text
|
92 |
+
- text
|
93 |
+
valid_data_path_and_name_and_type:
|
94 |
+
- - dump/raw/dev_ml/wav.scp
|
95 |
+
- speech
|
96 |
+
- sound
|
97 |
+
- - dump/raw/dev_ml/text
|
98 |
+
- text
|
99 |
+
- text
|
100 |
+
allow_variable_data_keys: false
|
101 |
+
max_cache_size: 0.0
|
102 |
+
max_cache_fd: 32
|
103 |
+
valid_max_cache_size: null
|
104 |
+
optim: adam
|
105 |
+
optim_conf:
|
106 |
+
lr: 0.5
|
107 |
+
scheduler: noamlr
|
108 |
+
scheduler_conf:
|
109 |
+
warmup_steps: 2500
|
110 |
+
token_list:
|
111 |
+
- <blank>
|
112 |
+
- <unk>
|
113 |
+
- ്
|
114 |
+
- ാ
|
115 |
+
- ി
|
116 |
+
- ു
|
117 |
+
- ര
|
118 |
+
- ▁
|
119 |
+
- ം
|
120 |
+
- യ
|
121 |
+
- ല
|
122 |
+
- െ
|
123 |
+
- വ
|
124 |
+
- ക
|
125 |
+
- ത
|
126 |
+
- ക്ക
|
127 |
+
- ▁അ
|
128 |
+
- ന
|
129 |
+
- ട
|
130 |
+
- മ
|
131 |
+
- േ
|
132 |
+
- ോ
|
133 |
+
- റ
|
134 |
+
- ▁പ
|
135 |
+
- ർ
|
136 |
+
- ണ
|
137 |
+
- ▁ക
|
138 |
+
- ▁സ
|
139 |
+
- ▁മ
|
140 |
+
- ത്ത
|
141 |
+
- ള
|
142 |
+
- ദ
|
143 |
+
- ▁വ
|
144 |
+
- ും
|
145 |
+
- ▁ആ
|
146 |
+
- ന്ന
|
147 |
+
- സ
|
148 |
+
- ത്
|
149 |
+
- ീ
|
150 |
+
- ▁ത
|
151 |
+
- പ
|
152 |
+
- ▁ന
|
153 |
+
- ന്
|
154 |
+
- ൂ
|
155 |
+
- ശ
|
156 |
+
- ട്ട
|
157 |
+
- ിയ
|
158 |
+
- ൻ
|
159 |
+
- ഗ
|
160 |
+
-
|
161 |
+
- ▁എന്ന
|
162 |
+
- പ്പ
|
163 |
+
- തി
|
164 |
+
- സ്
|
165 |
+
- ്യ
|
166 |
+
- ▁ഇ
|
167 |
+
- ൽ
|
168 |
+
- ങ്ങ
|
169 |
+
- റ്റ
|
170 |
+
- ുന്ന
|
171 |
+
- ഭ
|
172 |
+
- ാണ്
|
173 |
+
- രി
|
174 |
+
- ഹ
|
175 |
+
- ണ്ട
|
176 |
+
- ത്തി
|
177 |
+
- ▁അവ
|
178 |
+
- ധ
|
179 |
+
- ഷ
|
180 |
+
- ▁എ
|
181 |
+
- ാന
|
182 |
+
- ▁ച
|
183 |
+
- ത്ര
|
184 |
+
- ▁പ്ര
|
185 |
+
- ായി
|
186 |
+
- ▁നി
|
187 |
+
- മായ
|
188 |
+
- ിൽ
|
189 |
+
- ▁ഒരു
|
190 |
+
- ഥ
|
191 |
+
- കള
|
192 |
+
- ▁വി
|
193 |
+
- ജ
|
194 |
+
- ിന്റെ
|
195 |
+
- ബ
|
196 |
+
- ക്കുന്ന
|
197 |
+
- ിച്ച
|
198 |
+
- ങ്ങൾ
|
199 |
+
- ▁കാ
|
200 |
+
- ുണ്ട
|
201 |
+
- ുള്ള
|
202 |
+
- ായിരുന്നു
|
203 |
+
- ച
|
204 |
+
- ക്ഷ
|
205 |
+
- കൾ
|
206 |
+
- ിരുന്ന
|
207 |
+
- ല്ല
|
208 |
+
- ൾ
|
209 |
+
- യും
|
210 |
+
- ▁ജ
|
211 |
+
- ▁ബ
|
212 |
+
- ഡ
|
213 |
+
- ച്ച
|
214 |
+
- ▁അതിന
|
215 |
+
- ുടെ
|
216 |
+
- ിച്ചു
|
217 |
+
- ൈ
|
218 |
+
- ില്ല
|
219 |
+
- ▁ശ
|
220 |
+
- യിൽ
|
221 |
+
- മാണ്
|
222 |
+
- യുടെ
|
223 |
+
- ത്തിൽ
|
224 |
+
- ൃ
|
225 |
+
- ഫ
|
226 |
+
- ഖ
|
227 |
+
- ഠ
|
228 |
+
- e
|
229 |
+
- ൺ
|
230 |
+
- ഘ
|
231 |
+
- _
|
232 |
+
- '-'
|
233 |
+
- ഐ
|
234 |
+
- എ
|
235 |
+
- n
|
236 |
+
- ൌ
|
237 |
+
- ഃ
|
238 |
+
- അ
|
239 |
+
- ഔ
|
240 |
+
- ഊ
|
241 |
+
- ഛ
|
242 |
+
- r
|
243 |
+
- l
|
244 |
+
- t
|
245 |
+
- ഓ
|
246 |
+
- ൗ
|
247 |
+
- ഏ
|
248 |
+
- ഈ
|
249 |
+
- ഉ
|
250 |
+
- ൊ
|
251 |
+
- ഴ
|
252 |
+
- ഞ
|
253 |
+
- ഒ
|
254 |
+
- ഇ
|
255 |
+
- ആ
|
256 |
+
- ഋ
|
257 |
+
- ഝ
|
258 |
+
- ങ
|
259 |
+
- ഢ
|
260 |
+
- <sos/eos>
|
261 |
+
init: null
|
262 |
+
input_size: null
|
263 |
+
ctc_conf:
|
264 |
+
dropout_rate: 0.0
|
265 |
+
ctc_type: builtin
|
266 |
+
reduce: true
|
267 |
+
ignore_nan_grad: true
|
268 |
+
joint_net_conf: null
|
269 |
+
model_conf:
|
270 |
+
ctc_weight: 0.3
|
271 |
+
lsm_weight: 0.1
|
272 |
+
length_normalized_loss: false
|
273 |
+
use_preprocessor: true
|
274 |
+
token_type: bpe
|
275 |
+
bpemodel: data/ml_token_list/bpe_unigram150/bpe.model
|
276 |
+
non_linguistic_symbols: null
|
277 |
+
cleaner: null
|
278 |
+
g2p: null
|
279 |
+
speech_volume_normalize: null
|
280 |
+
rir_scp: null
|
281 |
+
rir_apply_prob: 1.0
|
282 |
+
noise_scp: null
|
283 |
+
noise_apply_prob: 1.0
|
284 |
+
noise_db_range: '13_15'
|
285 |
+
frontend: fused
|
286 |
+
frontend_conf:
|
287 |
+
frontends:
|
288 |
+
- frontend_type: s3prl
|
289 |
+
frontend_conf:
|
290 |
+
upstream: hubert_large_ll60k
|
291 |
+
download_dir: ./hub
|
292 |
+
multilayer_feature: true
|
293 |
+
- frontend_type: default
|
294 |
+
n_fft: 512
|
295 |
+
win_length: 400
|
296 |
+
hop_length: 160
|
297 |
+
align_method: linear_projection
|
298 |
+
proj_dim: 100
|
299 |
+
fs: 16k
|
300 |
+
specaug: specaug
|
301 |
+
specaug_conf:
|
302 |
+
apply_time_warp: true
|
303 |
+
time_warp_window: 5
|
304 |
+
time_warp_mode: bicubic
|
305 |
+
apply_freq_mask: true
|
306 |
+
freq_mask_width_range:
|
307 |
+
- 0
|
308 |
+
- 30
|
309 |
+
num_freq_mask: 2
|
310 |
+
apply_time_mask: true
|
311 |
+
time_mask_width_range:
|
312 |
+
- 0
|
313 |
+
- 40
|
314 |
+
num_time_mask: 2
|
315 |
+
normalize: utterance_mvn
|
316 |
+
normalize_conf: {}
|
317 |
+
preencoder: linear
|
318 |
+
preencoder_conf:
|
319 |
+
input_size: 200
|
320 |
+
output_size: 80
|
321 |
+
encoder: conformer
|
322 |
+
encoder_conf:
|
323 |
+
input_layer: conv2d
|
324 |
+
num_blocks: 12
|
325 |
+
linear_units: 2048
|
326 |
+
dropout_rate: 0.1
|
327 |
+
output_size: 256
|
328 |
+
attention_heads: 4
|
329 |
+
attention_dropout_rate: 0.0
|
330 |
+
pos_enc_layer_type: rel_pos
|
331 |
+
selfattention_layer_type: rel_selfattn
|
332 |
+
activation_type: swish
|
333 |
+
macaron_style: true
|
334 |
+
use_cnn_module: true
|
335 |
+
cnn_module_kernel: 15
|
336 |
+
postencoder: null
|
337 |
+
postencoder_conf: {}
|
338 |
+
decoder: transformer
|
339 |
+
decoder_conf:
|
340 |
+
input_layer: embed
|
341 |
+
num_blocks: 6
|
342 |
+
linear_units: 2048
|
343 |
+
dropout_rate: 0.1
|
344 |
+
required:
|
345 |
+
- output_dir
|
346 |
+
- token_list
|
347 |
+
version: 0.10.7a1
|
348 |
+
distributed: false
|
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/acc.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/backward_time.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/cer.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/cer_ctc.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/forward_time.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/gpu_max_cached_mem_GB.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/iter_time.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/loss.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/loss_att.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/loss_ctc.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/optim0_lr0.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/optim_step_time.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/train_time.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/images/wer.png
ADDED
![]() |
exp/asr_train_asr_conformer_s3prlfrontend_hubert_fused_raw_ml_bpe150_sp/valid.acc.ave_10best.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c446bbc1dbcf517f45d633657575dd052e1735e83fabe8f61e28c989d9c9b092
|
3 |
+
size 1440097508
|
exp/lm_train_lm_ml_bpe150/config.yaml
ADDED
@@ -0,0 +1,267 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
config: conf/train_lm.yaml
|
2 |
+
print_config: false
|
3 |
+
log_level: INFO
|
4 |
+
dry_run: false
|
5 |
+
iterator_type: sequence
|
6 |
+
output_dir: exp/lm_train_lm_ml_bpe150
|
7 |
+
ngpu: 1
|
8 |
+
seed: 0
|
9 |
+
num_workers: 1
|
10 |
+
num_att_plot: 3
|
11 |
+
dist_backend: nccl
|
12 |
+
dist_init_method: env://
|
13 |
+
dist_world_size: null
|
14 |
+
dist_rank: null
|
15 |
+
local_rank: 0
|
16 |
+
dist_master_addr: null
|
17 |
+
dist_master_port: null
|
18 |
+
dist_launcher: null
|
19 |
+
multiprocessing_distributed: false
|
20 |
+
unused_parameters: false
|
21 |
+
sharded_ddp: false
|
22 |
+
cudnn_enabled: true
|
23 |
+
cudnn_benchmark: false
|
24 |
+
cudnn_deterministic: true
|
25 |
+
collect_stats: false
|
26 |
+
write_collected_feats: false
|
27 |
+
max_epoch: 30
|
28 |
+
patience: 3
|
29 |
+
val_scheduler_criterion:
|
30 |
+
- valid
|
31 |
+
- loss
|
32 |
+
early_stopping_criterion:
|
33 |
+
- valid
|
34 |
+
- loss
|
35 |
+
- min
|
36 |
+
best_model_criterion:
|
37 |
+
- - valid
|
38 |
+
- loss
|
39 |
+
- min
|
40 |
+
keep_nbest_models: 1
|
41 |
+
nbest_averaging_interval: 0
|
42 |
+
grad_clip: 5.0
|
43 |
+
grad_clip_type: 2.0
|
44 |
+
grad_noise: false
|
45 |
+
accum_grad: 1
|
46 |
+
no_forward_run: false
|
47 |
+
resume: true
|
48 |
+
train_dtype: float32
|
49 |
+
use_amp: false
|
50 |
+
log_interval: null
|
51 |
+
use_matplotlib: true
|
52 |
+
use_tensorboard: true
|
53 |
+
use_wandb: false
|
54 |
+
wandb_project: null
|
55 |
+
wandb_id: null
|
56 |
+
wandb_entity: null
|
57 |
+
wandb_name: null
|
58 |
+
wandb_model_log_interval: -1
|
59 |
+
detect_anomaly: false
|
60 |
+
pretrain_path: null
|
61 |
+
init_param: []
|
62 |
+
ignore_init_mismatch: false
|
63 |
+
freeze_param: []
|
64 |
+
num_iters_per_epoch: null
|
65 |
+
batch_size: 64
|
66 |
+
valid_batch_size: null
|
67 |
+
batch_bins: 1000000
|
68 |
+
valid_batch_bins: null
|
69 |
+
train_shape_file:
|
70 |
+
- exp/lm_stats_ml_bpe150/train/text_shape.bpe
|
71 |
+
valid_shape_file:
|
72 |
+
- exp/lm_stats_ml_bpe150/valid/text_shape.bpe
|
73 |
+
batch_type: folded
|
74 |
+
valid_batch_type: null
|
75 |
+
fold_length:
|
76 |
+
- 150
|
77 |
+
sort_in_batch: descending
|
78 |
+
sort_batch: descending
|
79 |
+
multiple_iterator: false
|
80 |
+
chunk_length: 500
|
81 |
+
chunk_shift_ratio: 0.5
|
82 |
+
num_cache_chunks: 1024
|
83 |
+
train_data_path_and_name_and_type:
|
84 |
+
- - dump/raw/lm_train.txt
|
85 |
+
- text
|
86 |
+
- text
|
87 |
+
valid_data_path_and_name_and_type:
|
88 |
+
- - dump/raw/dev_ml/text
|
89 |
+
- text
|
90 |
+
- text
|
91 |
+
allow_variable_data_keys: false
|
92 |
+
max_cache_size: 0.0
|
93 |
+
max_cache_fd: 32
|
94 |
+
valid_max_cache_size: null
|
95 |
+
optim: sgd
|
96 |
+
optim_conf: {}
|
97 |
+
scheduler: null
|
98 |
+
scheduler_conf: {}
|
99 |
+
token_list:
|
100 |
+
- <blank>
|
101 |
+
- <unk>
|
102 |
+
- ്
|
103 |
+
- ാ
|
104 |
+
- ി
|
105 |
+
- ു
|
106 |
+
- ര
|
107 |
+
- ▁
|
108 |
+
- ം
|
109 |
+
- യ
|
110 |
+
- ല
|
111 |
+
- െ
|
112 |
+
- വ
|
113 |
+
- ക
|
114 |
+
- ത
|
115 |
+
- ക്ക
|
116 |
+
- ▁അ
|
117 |
+
- ന
|
118 |
+
- ട
|
119 |
+
- മ
|
120 |
+
- േ
|
121 |
+
- ോ
|
122 |
+
- റ
|
123 |
+
- ▁പ
|
124 |
+
- ർ
|
125 |
+
- ണ
|
126 |
+
- ▁ക
|
127 |
+
- ▁സ
|
128 |
+
- ▁മ
|
129 |
+
- ത്ത
|
130 |
+
- ള
|
131 |
+
- ദ
|
132 |
+
- ▁വ
|
133 |
+
- ും
|
134 |
+
- ▁ആ
|
135 |
+
- ന്ന
|
136 |
+
- സ
|
137 |
+
- ത്
|
138 |
+
- ീ
|
139 |
+
- ▁ത
|
140 |
+
- പ
|
141 |
+
- ▁ന
|
142 |
+
- ന്
|
143 |
+
- ൂ
|
144 |
+
- ശ
|
145 |
+
- ട്ട
|
146 |
+
- ിയ
|
147 |
+
- ൻ
|
148 |
+
- ഗ
|
149 |
+
-
|
150 |
+
- ▁എന്ന
|
151 |
+
- പ്പ
|
152 |
+
- തി
|
153 |
+
- സ്
|
154 |
+
- ്യ
|
155 |
+
- ▁ഇ
|
156 |
+
- ൽ
|
157 |
+
- ങ്ങ
|
158 |
+
- റ്റ
|
159 |
+
- ുന്ന
|
160 |
+
- ഭ
|
161 |
+
- ാണ്
|
162 |
+
- രി
|
163 |
+
- ഹ
|
164 |
+
- ണ്ട
|
165 |
+
- ത്തി
|
166 |
+
- ▁അവ
|
167 |
+
- ധ
|
168 |
+
- ഷ
|
169 |
+
- ▁എ
|
170 |
+
- ാന
|
171 |
+
- ▁ച
|
172 |
+
- ത്ര
|
173 |
+
- ▁പ്ര
|
174 |
+
- ായി
|
175 |
+
- ▁നി
|
176 |
+
- മായ
|
177 |
+
- ിൽ
|
178 |
+
- ▁ഒരു
|
179 |
+
- ഥ
|
180 |
+
- കള
|
181 |
+
- ▁വി
|
182 |
+
- ജ
|
183 |
+
- ിന്റെ
|
184 |
+
- ബ
|
185 |
+
- ക്കുന്ന
|
186 |
+
- ിച്ച
|
187 |
+
- ങ്ങൾ
|
188 |
+
- ▁കാ
|
189 |
+
- ുണ്ട
|
190 |
+
- ുള്ള
|
191 |
+
- ായിരുന്നു
|
192 |
+
- ച
|
193 |
+
- ക്ഷ
|
194 |
+
- കൾ
|
195 |
+
- ിരുന്ന
|
196 |
+
- ല്ല
|
197 |
+
- ൾ
|
198 |
+
- യും
|
199 |
+
- ▁ജ
|
200 |
+
- ▁ബ
|
201 |
+
- ഡ
|
202 |
+
- ച്ച
|
203 |
+
- ▁അതിന
|
204 |
+
- ുടെ
|
205 |
+
- ിച്ചു
|
206 |
+
- ൈ
|
207 |
+
- ില്ല
|
208 |
+
- ▁ശ
|
209 |
+
- യിൽ
|
210 |
+
- മാണ്
|
211 |
+
- യുടെ
|
212 |
+
- ത്തിൽ
|
213 |
+
- ൃ
|
214 |
+
- ഫ
|
215 |
+
- ഖ
|
216 |
+
- ഠ
|
217 |
+
- e
|
218 |
+
- ൺ
|
219 |
+
- ഘ
|
220 |
+
- _
|
221 |
+
- '-'
|
222 |
+
- ഐ
|
223 |
+
- എ
|
224 |
+
- n
|
225 |
+
- ൌ
|
226 |
+
- ഃ
|
227 |
+
- അ
|
228 |
+
- ഔ
|
229 |
+
- ഊ
|
230 |
+
- ഛ
|
231 |
+
- r
|
232 |
+
- l
|
233 |
+
- t
|
234 |
+
- ഓ
|
235 |
+
- ൗ
|
236 |
+
- ഏ
|
237 |
+
- ഈ
|
238 |
+
- ഉ
|
239 |
+
- ൊ
|
240 |
+
- ഴ
|
241 |
+
- ഞ
|
242 |
+
- ഒ
|
243 |
+
- ഇ
|
244 |
+
- ആ
|
245 |
+
- ഋ
|
246 |
+
- ഝ
|
247 |
+
- ങ
|
248 |
+
- ഢ
|
249 |
+
- <sos/eos>
|
250 |
+
init: null
|
251 |
+
model_conf:
|
252 |
+
ignore_id: 0
|
253 |
+
use_preprocessor: true
|
254 |
+
token_type: bpe
|
255 |
+
bpemodel: data/ml_token_list/bpe_unigram150/bpe.model
|
256 |
+
non_linguistic_symbols: null
|
257 |
+
cleaner: null
|
258 |
+
g2p: null
|
259 |
+
lm: seq_rnn
|
260 |
+
lm_conf:
|
261 |
+
nlayers: 2
|
262 |
+
unit: 650
|
263 |
+
required:
|
264 |
+
- output_dir
|
265 |
+
- token_list
|
266 |
+
version: 0.10.7a1
|
267 |
+
distributed: false
|
exp/lm_train_lm_ml_bpe150/images/backward_time.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/images/forward_time.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/images/gpu_max_cached_mem_GB.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/images/iter_time.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/images/loss.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/images/optim0_lr0.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/images/optim_step_time.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/images/train_time.png
ADDED
![]() |
exp/lm_train_lm_ml_bpe150/valid.loss.ave_1best.pth
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
30epoch.pth
|