Upload 2 files
Browse files- khummuang/config.yaml +242 -0
- khummuang/valid.acc.ave_10best.pth +3 -0
khummuang/config.yaml
ADDED
@@ -0,0 +1,242 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
config: conf/train_asr_hubert_transfer.yaml
|
2 |
+
print_config: false
|
3 |
+
log_level: INFO
|
4 |
+
dry_run: false
|
5 |
+
iterator_type: sequence
|
6 |
+
output_dir: exp/asr_hubert_official_korat_khummuang_pattani_transfer_seed100
|
7 |
+
ngpu: 1
|
8 |
+
seed: 100
|
9 |
+
num_workers: 1
|
10 |
+
num_att_plot: 0
|
11 |
+
dist_backend: nccl
|
12 |
+
dist_init_method: env://
|
13 |
+
dist_world_size: 2
|
14 |
+
dist_rank: 0
|
15 |
+
local_rank: 0
|
16 |
+
dist_master_addr: localhost
|
17 |
+
dist_master_port: null
|
18 |
+
dist_launcher: null
|
19 |
+
multiprocessing_distributed: true
|
20 |
+
unused_parameters: true
|
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: 134
|
28 |
+
patience: 20
|
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: 5.0
|
43 |
+
grad_clip_type: 2.0
|
44 |
+
grad_noise: false
|
45 |
+
accum_grad: 2
|
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 |
+
create_graph_in_tensorboard: false
|
54 |
+
use_wandb: false
|
55 |
+
wandb_project: null
|
56 |
+
wandb_id: null
|
57 |
+
wandb_entity: null
|
58 |
+
wandb_name: null
|
59 |
+
wandb_model_log_interval: -1
|
60 |
+
detect_anomaly: false
|
61 |
+
pretrain_path: null
|
62 |
+
init_param:
|
63 |
+
- /mnt/espnet_models/asr1/official_korat_khummuang_pattani/exp/asr_hubert_scratch_seed100/20epoch.pth:encoder:encoder
|
64 |
+
ignore_init_mismatch: true
|
65 |
+
freeze_param: []
|
66 |
+
num_iters_per_epoch: 1500
|
67 |
+
batch_size: 20
|
68 |
+
valid_batch_size: null
|
69 |
+
batch_bins: 9000000
|
70 |
+
valid_batch_bins: null
|
71 |
+
train_shape_file:
|
72 |
+
- exp/asr_stats_raw_char_sp/train/speech_shape
|
73 |
+
- exp/asr_stats_raw_char_sp/train/text_shape.char
|
74 |
+
valid_shape_file:
|
75 |
+
- exp/asr_stats_raw_char_sp/valid/speech_shape
|
76 |
+
- exp/asr_stats_raw_char_sp/valid/text_shape.char
|
77 |
+
batch_type: numel
|
78 |
+
valid_batch_type: null
|
79 |
+
fold_length:
|
80 |
+
- 80000
|
81 |
+
- 150
|
82 |
+
sort_in_batch: descending
|
83 |
+
sort_batch: descending
|
84 |
+
multiple_iterator: false
|
85 |
+
chunk_length: 500
|
86 |
+
chunk_shift_ratio: 0.5
|
87 |
+
num_cache_chunks: 1024
|
88 |
+
chunk_excluded_key_prefixes: []
|
89 |
+
train_data_path_and_name_and_type:
|
90 |
+
- - dump/raw/cheat_sp/wav.scp
|
91 |
+
- speech
|
92 |
+
- sound
|
93 |
+
- - dump/raw/cheat_sp/text
|
94 |
+
- text
|
95 |
+
- text
|
96 |
+
valid_data_path_and_name_and_type:
|
97 |
+
- - dump/raw/test/wav.scp
|
98 |
+
- speech
|
99 |
+
- sound
|
100 |
+
- - dump/raw/test/text
|
101 |
+
- text
|
102 |
+
- text
|
103 |
+
allow_variable_data_keys: false
|
104 |
+
max_cache_size: 0.0
|
105 |
+
max_cache_fd: 32
|
106 |
+
valid_max_cache_size: null
|
107 |
+
exclude_weight_decay: false
|
108 |
+
exclude_weight_decay_conf: {}
|
109 |
+
optim: adam
|
110 |
+
optim_conf:
|
111 |
+
lr: 0.0002
|
112 |
+
weight_decay: 1.0e-06
|
113 |
+
scheduler: warmuplr
|
114 |
+
scheduler_conf:
|
115 |
+
warmup_steps: 3000
|
116 |
+
token_list:
|
117 |
+
- <blank>
|
118 |
+
- <unk>
|
119 |
+
- <space>
|
120 |
+
- อ
|
121 |
+
- า
|
122 |
+
- น
|
123 |
+
- ม
|
124 |
+
- ่
|
125 |
+
- ร
|
126 |
+
- ะ
|
127 |
+
- ง
|
128 |
+
- ด
|
129 |
+
- เ
|
130 |
+
- ห
|
131 |
+
- ี
|
132 |
+
- ้
|
133 |
+
- ก
|
134 |
+
- ั
|
135 |
+
- ย
|
136 |
+
- บ
|
137 |
+
- ล
|
138 |
+
- แ
|
139 |
+
- ิ
|
140 |
+
- ค
|
141 |
+
- ว
|
142 |
+
- ส
|
143 |
+
- ต
|
144 |
+
- ข
|
145 |
+
- ป
|
146 |
+
- ์
|
147 |
+
- ู
|
148 |
+
- ท
|
149 |
+
- ไ
|
150 |
+
- จ
|
151 |
+
- ื
|
152 |
+
- ุ
|
153 |
+
- พ
|
154 |
+
- ำ
|
155 |
+
- ช
|
156 |
+
- โ
|
157 |
+
- ๋
|
158 |
+
- ซ
|
159 |
+
- ึ
|
160 |
+
- ็
|
161 |
+
- ฮ
|
162 |
+
- ใ
|
163 |
+
- ๊
|
164 |
+
- ญ
|
165 |
+
- ผ
|
166 |
+
- ษ
|
167 |
+
- ศ
|
168 |
+
- ฆ
|
169 |
+
- ณ
|
170 |
+
- ธ
|
171 |
+
- ถ
|
172 |
+
- ฟ
|
173 |
+
- ภ
|
174 |
+
- ฤ
|
175 |
+
- ฝ
|
176 |
+
- ฉ
|
177 |
+
- ฐ
|
178 |
+
- ฑ
|
179 |
+
- ฬ
|
180 |
+
- ฎ
|
181 |
+
- <sos/eos>
|
182 |
+
init: null
|
183 |
+
input_size: 1
|
184 |
+
ctc_conf:
|
185 |
+
dropout_rate: 0.0
|
186 |
+
ctc_type: builtin
|
187 |
+
reduce: true
|
188 |
+
ignore_nan_grad: null
|
189 |
+
zero_infinity: true
|
190 |
+
joint_net_conf: null
|
191 |
+
use_preprocessor: true
|
192 |
+
token_type: char
|
193 |
+
bpemodel: null
|
194 |
+
non_linguistic_symbols: null
|
195 |
+
cleaner: null
|
196 |
+
g2p: null
|
197 |
+
speech_volume_normalize: null
|
198 |
+
rir_scp: null
|
199 |
+
rir_apply_prob: 1.0
|
200 |
+
noise_scp: null
|
201 |
+
noise_apply_prob: 1.0
|
202 |
+
noise_db_range: '13_15'
|
203 |
+
short_noise_thres: 0.5
|
204 |
+
aux_ctc_tasks: []
|
205 |
+
frontend: null
|
206 |
+
frontend_conf: {}
|
207 |
+
specaug: null
|
208 |
+
specaug_conf: {}
|
209 |
+
normalize: utterance_mvn
|
210 |
+
normalize_conf: {}
|
211 |
+
model: espnet
|
212 |
+
model_conf:
|
213 |
+
ctc_weight: 0.3
|
214 |
+
lsm_weight: 0.1
|
215 |
+
length_normalized_loss: false
|
216 |
+
preencoder: null
|
217 |
+
preencoder_conf: {}
|
218 |
+
encoder: hubert
|
219 |
+
encoder_conf:
|
220 |
+
output_size: 256
|
221 |
+
normalize_before: false
|
222 |
+
hubert_url: https://dl.fbaipublicfiles.com/hubert/hubert_base_ls960.pt
|
223 |
+
hubert_dir_path: ./downloads/hubert_pretrained_models/hubert_base_ls960.pt
|
224 |
+
freeze_finetune_updates: 3000
|
225 |
+
postencoder: null
|
226 |
+
postencoder_conf: {}
|
227 |
+
decoder: transformer
|
228 |
+
decoder_conf:
|
229 |
+
attention_heads: 4
|
230 |
+
linear_units: 2048
|
231 |
+
num_blocks: 6
|
232 |
+
dropout_rate: 0.1
|
233 |
+
positional_dropout_rate: 0.1
|
234 |
+
self_attention_dropout_rate: 0.1
|
235 |
+
src_attention_dropout_rate: 0.1
|
236 |
+
preprocessor: default
|
237 |
+
preprocessor_conf: {}
|
238 |
+
required:
|
239 |
+
- output_dir
|
240 |
+
- token_list
|
241 |
+
version: '202301'
|
242 |
+
distributed: true
|
khummuang/valid.acc.ave_10best.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:10ab38525443f14058430a6b81e2c0bd52cd62b1286c8864ed97a3c0b924e617
|
3 |
+
size 417820655
|