Skip to content
项目
群组
代码片段
帮助
当前项目
正在载入...
登录 / 注册
切换导航面板
F
Fairseq-S2T
概览
Overview
Details
Activity
Cycle Analytics
版本库
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Charts
问题
0
Issues
0
列表
Board
标记
里程碑
合并请求
0
Merge Requests
0
CI / CD
CI / CD
流水线
作业
日程表
图表
维基
Wiki
代码片段
Snippets
成员
Collapse sidebar
Close sidebar
活动
图像
聊天
创建新问题
作业
提交
Issue Boards
Open sidebar
xuchen
Fairseq-S2T
Commits
99763132
Commit
99763132
authored
Oct 23, 2021
by
xuchen
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
optimize the shell scripts
parent
5160a9f5
隐藏空白字符变更
内嵌
并排
正在显示
34 个修改的文件
包含
356 行增加
和
300 行删除
+356
-300
egs/librispeech/asr/conf/base.yaml
+1
-1
egs/librispeech/asr/conf/conformer.yaml
+1
-1
egs/librispeech/asr/conf/ctc.yaml
+0
-44
egs/librispeech/asr/conf/debug.yaml
+7
-11
egs/librispeech/asr/conf/large.yaml
+11
-15
egs/librispeech/asr/conf/pyramid3_base.yaml
+7
-11
egs/librispeech/asr/conf/pyramid4_base.yaml
+18
-18
egs/librispeech/asr/conf/pyramid4_base_sr8.yaml
+56
-0
egs/librispeech/asr/conf/pyramid4_deep18.yaml
+56
-0
egs/librispeech/asr/conf/pyramid4_deep30.yaml
+14
-14
egs/librispeech/asr/conf/pyramid4_large.yaml
+9
-14
egs/librispeech/asr/conf/rpr.yaml
+3
-3
egs/librispeech/asr/decode.sh
+1
-1
egs/librispeech/asr/run.sh
+9
-7
egs/librispeech/asr/train.sh
+9
-7
egs/mustc/asr/conf/base.yaml
+2
-2
egs/mustc/asr/conf/ctc.yaml
+0
-44
egs/mustc/asr/conf/pyramid4_base.yaml
+11
-9
egs/mustc/asr/conf/pyramid4_base_sr8.yaml
+16
-13
egs/mustc/asr/conf/rpr.yaml
+3
-3
egs/mustc/asr/conf/tmp.yaml
+0
-2
egs/mustc/asr/local/monitor.sh
+1
-1
egs/mustc/asr/run.sh
+2
-2
egs/mustc/asr/train.sh
+8
-2
egs/mustc/mt/decode.sh
+4
-4
egs/mustc/mt/run.sh
+13
-12
egs/mustc/mt/train.sh
+2
-2
egs/mustc/st/conf/base.yaml
+2
-2
egs/mustc/st/conf/ctc.yaml
+2
-45
egs/mustc/st/conf/pyramid4_base_sr8.yaml
+58
-0
egs/mustc/st/conf/sate_ctc.yaml
+20
-6
egs/mustc/st/conf/tmp.yaml
+0
-2
egs/mustc/st/decode.sh
+1
-1
egs/mustc/st/train.sh
+9
-1
没有找到文件。
egs/librispeech/asr/conf/base.yaml
查看文件 @
99763132
...
@@ -24,7 +24,7 @@ warmup-updates: 10000
...
@@ -24,7 +24,7 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy
criterion
:
label_smoothed_cross_entropy
_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-kernel-sizes
:
5,5
...
...
egs/librispeech/asr/conf/conformer.yaml
查看文件 @
99763132
...
@@ -2,4 +2,4 @@ arch: s2t_conformer_s
...
@@ -2,4 +2,4 @@ arch: s2t_conformer_s
macaron-style
:
True
macaron-style
:
True
use-cnn-module
:
True
use-cnn-module
:
True
cnn-module-kernel
:
15
cnn-module-kernel
:
31
egs/librispeech/asr/conf/ctc.yaml
查看文件 @
99763132
train-subset
:
train-clean-100,train-clean-360,train-other-500
valid-subset
:
dev-clean
max-epoch
:
100
max-update
:
300000
num-workers
:
8
patience
:
10
no-progress-bar
:
True
log-interval
:
100
seed
:
1
report-accuracy
:
True
#load-pretrained-encoder-from:
#load-pretrained-decoder-from:
arch
:
s2t_transformer_s
share-decoder-input-output-embed
:
True
optimizer
:
adam
clip-norm
:
10.0
lr-scheduler
:
inverse_sqrt
warmup-init-lr
:
1e-7
warmup-updates
:
10000
lr
:
2e-3
#adam_betas: (0.9,0.98)
ctc-weight
:
0.3
ctc-weight
:
0.3
criterion
:
label_smoothed_cross_entropy_with_ctc
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-channels
:
1024
dropout
:
0.1
activation-fn
:
relu
encoder-embed-dim
:
256
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
decoder-layers
:
6
encoder-attention-heads
:
4
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
attention-dropout
:
0.1
activation-dropout
:
0.1
egs/librispeech/asr/conf/
ctc_
debug.yaml
→
egs/librispeech/asr/conf/debug.yaml
查看文件 @
99763132
#train-subset: train-clean-100,train-clean-360,train-other-500
train-subset
:
train-clean-100
train-subset
:
train-clean-100
valid-subset
:
dev-clean
valid-subset
:
dev-clean
max-epoch
:
100
max-epoch
:
100
max-update
:
300000
max-update
:
300000
num-workers
:
0
num-workers
:
8
patience
:
10
patience
:
10
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
report-accuracy
:
True
report-accuracy
:
True
#load-pretrained-encoder-from:
#load-pretrained-decoder-from:
arch
:
s2t_transformer_s
arch
:
s2t_transformer_s
share-decoder-input-output-embed
:
True
share-decoder-input-output-embed
:
True
optimizer
:
adam
optimizer
:
adam
...
@@ -22,26 +24,20 @@ warmup-updates: 10000
...
@@ -22,26 +24,20 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
ctc-weight
:
0.3
criterion
:
label_smoothed_cross_entropy_with_ctc
criterion
:
label_smoothed_cross_entropy_with_ctc
ctc-weight
:
0.3
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-kernel-sizes
:
5,5
conv-channels
:
1024
conv-channels
:
1024
dropout
:
0.1
dropout
:
0.1
activation-fn
:
relu
activation-fn
:
relu
#
encoder-embed-dim: 256
encoder-embed-dim
:
256
encoder-ffn-embed-dim
:
2048
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
encoder-layers
:
12
decoder-layers
:
3
decoder-layers
:
6
encoder-attention-heads
:
4
encoder-attention-heads
:
4
macaron-style
:
True
use-cnn-module
:
True
cnn-module-kernel
:
31
adpater
:
subsample
decoder-embed-dim
:
256
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
decoder-attention-heads
:
4
...
...
egs/
mustc/st/conf/pyramid
.yaml
→
egs/
librispeech/asr/conf/large
.yaml
查看文件 @
99763132
arch
:
pys2t_transformer_s
train-subset
:
train-clean-100,train-clean-360,train-other-500
#conv-kernel-sizes: 5
valid-subset
:
dev-clean
encoder-embed-dim
:
512
pyramid-layers
:
3_6_9
ctc-layer
:
7
train-subset
:
train_st
valid-subset
:
dev_st
max-epoch
:
5
0
max-epoch
:
10
0
max-update
:
1
00000
max-update
:
3
00000
num-workers
:
8
num-workers
:
8
patience
:
1
0
patience
:
2
0
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
...
@@ -20,6 +14,7 @@ report-accuracy: True
...
@@ -20,6 +14,7 @@ report-accuracy: True
#load-pretrained-encoder-from:
#load-pretrained-encoder-from:
#load-pretrained-decoder-from:
#load-pretrained-decoder-from:
arch
:
s2t_transformer_m
share-decoder-input-output-embed
:
True
share-decoder-input-output-embed
:
True
optimizer
:
adam
optimizer
:
adam
clip-norm
:
10.0
clip-norm
:
10.0
...
@@ -29,20 +24,21 @@ warmup-updates: 10000
...
@@ -29,20 +24,21 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
ctc-weight
:
0.3
criterion
:
label_smoothed_cross_entropy_with_ctc
criterion
:
label_smoothed_cross_entropy_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-channels
:
1024
conv-channels
:
1024
dropout
:
0.1
dropout
:
0.1
activation-fn
:
relu
activation-fn
:
relu
encoder-embed-dim
:
512
encoder-ffn-embed-dim
:
2048
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
encoder-layers
:
12
decoder-layers
:
6
decoder-layers
:
6
encoder-attention-heads
:
4
encoder-attention-heads
:
8
decoder-embed-dim
:
256
decoder-embed-dim
:
512
decoder-ffn-embed-dim
:
2048
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
decoder-attention-heads
:
8
attention-dropout
:
0.1
attention-dropout
:
0.1
activation-dropout
:
0.1
activation-dropout
:
0.1
egs/librispeech/asr/conf/pyramid
_stage3
.yaml
→
egs/librispeech/asr/conf/pyramid
3_base
.yaml
查看文件 @
99763132
arch
:
pys2t_transformer_s
arch
:
pys2t_transformer_s
encoder-embed-dim
:
512
encoder-embed-dim
:
256
pyramid-stages
:
3
pyramid-stages
:
3
pyramid-layers
:
3_6_3
pyramid-layers
:
3_6_3
#encoder-attention-type: reduced
pyramid-fuse-way
:
all_conv
#pyramid-attn-sample-ratios: 4_2_1
pyramid-fuse
:
True
#pyramid-block-attn: True
#pyramid-fuse-way: gated
pyramid-use-ppm
:
True
pyramid-sr-ratios
:
2_2_2
pyramid-sr-ratios
:
2_2_2
pyramid-embed-dims
:
128_256_512
pyramid-embed-dims
:
256_256_256
pyramid-reduced-embed
:
conv
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1
pyramid-position-embed
:
1_1_1
pyramid-kernel-sizes
:
5_5_5
pyramid-kernel-sizes
:
5_5_5
pyramid-ffn-ratios
:
8_8_
4
pyramid-ffn-ratios
:
8_8_
8
pyramid-heads
:
2_4_8
pyramid-heads
:
4_4_4
train-subset
:
train-clean-100,train-clean-360,train-other-500
train-subset
:
train-clean-100,train-clean-360,train-other-500
valid-subset
:
dev-clean
valid-subset
:
dev-clean
...
@@ -23,7 +20,7 @@ max-epoch: 100
...
@@ -23,7 +20,7 @@ max-epoch: 100
max-update
:
300000
max-update
:
300000
num-workers
:
8
num-workers
:
8
patience
:
1
0
patience
:
2
0
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
...
@@ -42,7 +39,6 @@ lr: 2e-3
...
@@ -42,7 +39,6 @@ lr: 2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy_with_ctc
criterion
:
label_smoothed_cross_entropy_with_ctc
ctc-weight
:
0
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-channels
:
1024
conv-channels
:
1024
...
...
egs/
mustc/asr/conf/pyramid_stage3
.yaml
→
egs/
librispeech/asr/conf/pyramid4_base
.yaml
查看文件 @
99763132
arch
:
pys2t_transformer_s
arch
:
pys2t_transformer_s
encoder-embed-dim
:
512
encoder-embed-dim
:
256
pyramid-stages
:
3
pyramid-stages
:
4
pyramid-layers
:
3_6_3
#pyramid-dropout: 0
pyramid-
sr-ratios
:
2_2
_2
pyramid-
layers
:
2_2_6
_2
#pyramid-block-attn: True
pyramid-sr-ratios
:
2_2_2_2
#pyramid-fuse-way: add
pyramid-fuse
:
True
pyramid-
use-ppm
:
True
pyramid-
fuse-way
:
all_conv
pyramid-embed-dims
:
128_256_512
pyramid-embed-dims
:
256_256_256_256
pyramid-reduced-embed
:
conv
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1
pyramid-position-embed
:
1_1_1
_1
pyramid-kernel-sizes
:
5_5_5
pyramid-kernel-sizes
:
5_5_5
_5
pyramid-ffn-ratios
:
8_8_
4
pyramid-ffn-ratios
:
8_8_
8_8
pyramid-heads
:
2_4_8
pyramid-heads
:
4_4_4_4
train-subset
:
train
_asr
train-subset
:
train
-clean-100,train-clean-360,train-other-500
valid-subset
:
dev
_asr
valid-subset
:
dev
-clean
max-epoch
:
5
0
max-epoch
:
10
0
max-update
:
1
00000
max-update
:
3
00000
num-workers
:
8
num-workers
:
8
patience
:
1
0
patience
:
2
0
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
...
@@ -39,7 +39,7 @@ warmup-updates: 10000
...
@@ -39,7 +39,7 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy
criterion
:
label_smoothed_cross_entropy
_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-channels
:
1024
conv-channels
:
1024
...
...
egs/librispeech/asr/conf/pyramid4_base_sr8.yaml
0 → 100644
查看文件 @
99763132
arch
:
pys2t_transformer_s
encoder-embed-dim
:
256
pyramid-stages
:
4
pyramid-layers
:
3_3_3_3
pyramid-sr-ratios
:
2_2_1_2
pyramid-fuse
:
True
pyramid-fuse-way
:
all_conv
pyramid-embed-dims
:
256_256_256_256
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1_1
pyramid-kernel-sizes
:
5_5_5_5
pyramid-ffn-ratios
:
8_8_8_8
pyramid-heads
:
4_4_4_4
train-subset
:
train-clean-100,train-clean-360,train-other-500
valid-subset
:
dev-clean
max-epoch
:
100
max-update
:
300000
num-workers
:
8
patience
:
20
no-progress-bar
:
True
log-interval
:
100
seed
:
1
report-accuracy
:
True
#load-pretrained-encoder-from:
#load-pretrained-decoder-from:
share-decoder-input-output-embed
:
True
optimizer
:
adam
clip-norm
:
10.0
lr-scheduler
:
inverse_sqrt
warmup-init-lr
:
1e-7
warmup-updates
:
10000
lr
:
2e-3
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy_with_ctc
label_smoothing
:
0.1
conv-channels
:
1024
dropout
:
0.1
activation-fn
:
relu
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
decoder-layers
:
6
encoder-attention-heads
:
4
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
attention-dropout
:
0.1
activation-dropout
:
0.1
egs/librispeech/asr/conf/pyramid4_deep18.yaml
0 → 100644
查看文件 @
99763132
arch
:
pys2t_transformer_s
encoder-embed-dim
:
256
pyramid-stages
:
4
pyramid-layers
:
3_3_8_4
pyramid-sr-ratios
:
2_2_2_2
pyramid-fuse
:
True
pyramid-fuse-way
:
all_conv
pyramid-embed-dims
:
256_256_256_256
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1_1
pyramid-kernel-sizes
:
5_5_5_5
pyramid-ffn-ratios
:
8_8_8_8
pyramid-heads
:
4_4_4_4
train-subset
:
train-clean-100,train-clean-360,train-other-500
valid-subset
:
dev-clean
max-epoch
:
100
max-update
:
300000
num-workers
:
8
patience
:
20
no-progress-bar
:
True
log-interval
:
100
seed
:
1
report-accuracy
:
True
#load-pretrained-encoder-from:
#load-pretrained-decoder-from:
share-decoder-input-output-embed
:
True
optimizer
:
adam
clip-norm
:
10.0
lr-scheduler
:
inverse_sqrt
warmup-init-lr
:
1e-7
warmup-updates
:
10000
lr
:
2e-3
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy_with_ctc
label_smoothing
:
0.1
conv-channels
:
1024
dropout
:
0.1
activation-fn
:
relu
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
decoder-layers
:
6
encoder-attention-heads
:
4
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
attention-dropout
:
0.1
activation-dropout
:
0.1
egs/librispeech/asr/conf/pyramid
_enc6
.yaml
→
egs/librispeech/asr/conf/pyramid
4_deep30
.yaml
查看文件 @
99763132
arch
:
pys2t_transformer_s
arch
:
pys2t_transformer_s
encoder-embed-dim
:
512
encoder-embed-dim
:
256
pyramid-stages
:
3
pyramid-stages
:
4
pyramid-layers
:
3_6_3
pyramid-layers
:
5_5_15_5
encoder-attention-type
:
reduced
pyramid-sr-ratios
:
2_2_2_2
pyramid-
attn-sample-ratios
:
4_2_1
pyramid-
fuse
:
True
pyramid-
sr-ratios
:
2_2_2
pyramid-
fuse-way
:
all_conv
pyramid-embed-dims
:
128_256_512
pyramid-embed-dims
:
256_256_256_256
pyramid-reduced-embed
:
conv
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1
pyramid-position-embed
:
1_1_1
_1
pyramid-kernel-sizes
:
5_5_5
pyramid-kernel-sizes
:
5_5_5
_5
pyramid-ffn-ratios
:
8_8_
4
pyramid-ffn-ratios
:
8_8_
8_8
pyramid-heads
:
2_4_8
pyramid-heads
:
4_4_4_4
train-subset
:
train-clean-100,train-clean-360,train-other-500
train-subset
:
train-clean-100,train-clean-360,train-other-500
valid-subset
:
dev-clean
valid-subset
:
dev-clean
...
@@ -20,7 +20,7 @@ max-epoch: 100
...
@@ -20,7 +20,7 @@ max-epoch: 100
max-update
:
300000
max-update
:
300000
num-workers
:
8
num-workers
:
8
patience
:
1
0
patience
:
2
0
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
...
@@ -38,14 +38,14 @@ warmup-updates: 10000
...
@@ -38,14 +38,14 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy
criterion
:
label_smoothed_cross_entropy
_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-channels
:
1024
conv-channels
:
1024
dropout
:
0.1
dropout
:
0.1
activation-fn
:
relu
activation-fn
:
relu
encoder-ffn-embed-dim
:
2048
encoder-ffn-embed-dim
:
2048
encoder-layers
:
6
encoder-layers
:
12
decoder-layers
:
6
decoder-layers
:
6
encoder-attention-heads
:
4
encoder-attention-heads
:
4
...
...
egs/librispeech/asr/conf/pyramid.yaml
→
egs/librispeech/asr/conf/pyramid
4_large
.yaml
查看文件 @
99763132
...
@@ -2,30 +2,26 @@ arch: pys2t_transformer_s
...
@@ -2,30 +2,26 @@ arch: pys2t_transformer_s
encoder-embed-dim
:
512
encoder-embed-dim
:
512
pyramid-stages
:
4
pyramid-stages
:
4
pyramid-layers
:
2_2_6_2
pyramid-layers
:
2_2_6_2
#encoder-attention-type: reduced
#pyramid-layers: 3_3_3_3
#pyramid-attn-sample-ratios: 8_4_2_1
#pyramid-block-attn: True
#pyramid-fuse-way: add
pyramid-sr-ratios
:
2_2_2_2
pyramid-sr-ratios
:
2_2_2_2
pyramid-use-ppm
:
True
pyramid-fuse
:
True
pyramid-embed-dims
:
128_128_256_512
pyramid-fuse-way
:
all_conv
pyramid-reduced-embed
:
fuse
pyramid-embed-dims
:
512_512_512_512
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1_1
pyramid-position-embed
:
1_1_1_1
pyramid-kernel-sizes
:
5_5_5_5
pyramid-kernel-sizes
:
5_5_5_5
pyramid-ffn-ratios
:
8_8_8_4
pyramid-ffn-ratios
:
8_8_8_8
pyramid-heads
:
2_2_4_8
pyramid-heads
:
8_8_8_8
#ctc-layer: 8
#train-subset: train-clean-100,train-clean-360,train-other-500
train-subset
:
train-clean-100,train-clean-360,train-other-500
train-subset
:
train-clean-100
valid-subset
:
dev-clean
valid-subset
:
dev-clean
max-epoch
:
100
max-epoch
:
100
max-update
:
300000
max-update
:
300000
num-workers
:
8
num-workers
:
8
patience
:
1
0
patience
:
2
0
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
...
@@ -44,7 +40,6 @@ lr: 2e-3
...
@@ -44,7 +40,6 @@ lr: 2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy_with_ctc
criterion
:
label_smoothed_cross_entropy_with_ctc
ctc-weight
:
0.0
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-channels
:
1024
conv-channels
:
1024
...
...
egs/librispeech/asr/conf/rpr.yaml
查看文件 @
99763132
#
encoder-attention-type: rel_selfattn
encoder-attention-type
:
rel_selfattn
encoder-attention-type
:
relative
#
encoder-attention-type: relative
max-encoder-relative-length
:
100
#
max-encoder-relative-length: 100
egs/librispeech/asr/decode.sh
查看文件 @
99763132
...
@@ -3,7 +3,7 @@
...
@@ -3,7 +3,7 @@
gpu_num
=
1
gpu_num
=
1
data_dir
=
data_dir
=
test_subset
=(
test-clean test-other
)
test_subset
=(
dev-clean dev-other
test-clean test-other
)
exp_name
=
exp_name
=
if
[
"$#"
-eq
1
]
;
then
if
[
"$#"
-eq
1
]
;
then
...
...
egs/librispeech/asr/run.sh
查看文件 @
99763132
...
@@ -20,7 +20,7 @@ stop_stage=0
...
@@ -20,7 +20,7 @@ stop_stage=0
######## hardware ########
######## hardware ########
# devices
# devices
#
device=()
device
=()
gpu_num
=
8
gpu_num
=
8
update_freq
=
1
update_freq
=
1
...
@@ -251,12 +251,14 @@ if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
...
@@ -251,12 +251,14 @@ if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
# Average models
# Average models
dec_model
=
avg_
${
n_average
}
_checkpoint.pt
dec_model
=
avg_
${
n_average
}
_checkpoint.pt
cmd
=
"python
${
root_dir
}
/scripts/average_checkpoints.py
if
[[
!
-f
${
model_dir
}
/
${
dec_model
}
]]
;
then
--inputs
${
model_dir
}
cmd
=
"python
${
root_dir
}
/scripts/average_checkpoints.py
--num-epoch-checkpoints
${
n_average
}
--inputs
${
model_dir
}
--output
${
model_dir
}
/
${
dec_model
}
"
--num-epoch-checkpoints
${
n_average
}
echo
-e
"
\0
33[34mRun command:
\n
${
cmd
}
\0
33[0m"
--output
${
model_dir
}
/
${
dec_model
}
"
[[
$eval
-eq
1
]]
&&
eval
$cmd
echo
-e
"
\0
33[34mRun command:
\n
${
cmd
}
\0
33[0m"
[[
$eval
-eq
1
]]
&&
eval
$cmd
fi
else
else
dec_model
=
${
dec_model
}
dec_model
=
${
dec_model
}
fi
fi
...
...
egs/librispeech/asr/train.sh
查看文件 @
99763132
...
@@ -6,11 +6,14 @@ gpu_num=8
...
@@ -6,11 +6,14 @@ gpu_num=8
update_freq
=
1
update_freq
=
1
max_tokens
=
100000
max_tokens
=
100000
exp_tag
=
#exp_tag=
config_list
=(
base
)
#config_list=(base)
config_list
=(
base conformer
)
#config_list=(ctc)
#config_list=(pyramid)
#config_list=(ctc conformer rpr)
#config_list=(pyramid_stage3)
config_list
=(
base conformer rpr
)
#config_list=(pyramid4_all256)
#config_list=(pyramid5_all256)
# exp full name
# exp full name
exp_name
=
exp_name
=
...
@@ -41,8 +44,7 @@ if [[ -n ${extra_tag} ]]; then
...
@@ -41,8 +44,7 @@ if [[ -n ${extra_tag} ]]; then
cmd
=
"
$cmd
--extra_tag
${
extra_tag
}
"
cmd
=
"
$cmd
--extra_tag
${
extra_tag
}
"
fi
fi
if
[[
-n
${
extra_parameter
}
]]
;
then
if
[[
-n
${
extra_parameter
}
]]
;
then
# cmd="$cmd --extra_parameter \"${extra_parameter}\""
cmd
=
"
$cmd
--extra_parameter
\"
${
extra_parameter
}
\"
"
cmd
=
"
$cmd
--extra_parameter
${
extra_parameter
}
"
fi
fi
echo
${
cmd
}
echo
${
cmd
}
...
...
egs/mustc/asr/conf/base.yaml
查看文件 @
99763132
train-subset
:
train_asr
train-subset
:
train_asr
valid-subset
:
dev_asr
valid-subset
:
dev_asr
max-epoch
:
5
0
max-epoch
:
10
0
max-update
:
100000
max-update
:
100000
num-workers
:
8
num-workers
:
8
...
@@ -24,7 +24,7 @@ warmup-updates: 10000
...
@@ -24,7 +24,7 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy
criterion
:
label_smoothed_cross_entropy
_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-kernel-sizes
:
5,5
...
...
egs/mustc/asr/conf/ctc.yaml
查看文件 @
99763132
train-subset
:
train_asr
valid-subset
:
dev_asr
max-epoch
:
50
max-update
:
100000
num-workers
:
8
patience
:
10
no-progress-bar
:
True
log-interval
:
100
seed
:
1
report-accuracy
:
True
#load-pretrained-encoder-from:
#load-pretrained-decoder-from:
arch
:
s2t_transformer_s
share-decoder-input-output-embed
:
True
optimizer
:
adam
clip-norm
:
10.0
lr-scheduler
:
inverse_sqrt
warmup-init-lr
:
1e-7
warmup-updates
:
10000
lr
:
2e-3
#adam_betas: (0.9,0.98)
ctc-weight
:
0.3
ctc-weight
:
0.3
criterion
:
label_smoothed_cross_entropy_with_ctc
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-channels
:
1024
dropout
:
0.1
activation-fn
:
relu
encoder-embed-dim
:
256
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
decoder-layers
:
6
encoder-attention-heads
:
4
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
attention-dropout
:
0.1
activation-dropout
:
0.1
egs/mustc/asr/conf/pyramid.yaml
→
egs/mustc/asr/conf/pyramid
4_base
.yaml
查看文件 @
99763132
arch
:
pys2t_transformer_s
arch
:
pys2t_transformer_s
encoder-embed-dim
:
512
encoder-embed-dim
:
256
pyramid-stages
:
4
pyramid-stages
:
4
pyramid-layers
:
2_3_5_2
#pyramid-dropout: 0
pyramid-layers
:
2_2_6_2
pyramid-sr-ratios
:
2_2_2_2
pyramid-sr-ratios
:
2_2_2_2
pyramid-embed-dims
:
128_128_256_512
pyramid-fuse
:
True
pyramid-use-ppm
:
True
pyramid-fuse-way
:
all_conv
pyramid-embed-dims
:
256_256_256_256
pyramid-reduced-embed
:
conv
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1_1
pyramid-position-embed
:
1_1_1_1
pyramid-kernel-sizes
:
5_5_5_5
pyramid-kernel-sizes
:
5_5_5_5
pyramid-ffn-ratios
:
8_8_8_
4
pyramid-ffn-ratios
:
8_8_8_
8
pyramid-heads
:
2_2_4_8
pyramid-heads
:
4_4_4_4
train-subset
:
train_asr
train-subset
:
train_asr
valid-subset
:
dev_asr
valid-subset
:
dev_asr
max-epoch
:
5
0
max-epoch
:
10
0
max-update
:
100000
max-update
:
100000
num-workers
:
8
num-workers
:
8
patience
:
1
0
patience
:
2
0
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
...
@@ -37,7 +39,7 @@ warmup-updates: 10000
...
@@ -37,7 +39,7 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy
criterion
:
label_smoothed_cross_entropy
_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-channels
:
1024
conv-channels
:
1024
...
...
egs/mustc/asr/conf/pyramid
_enc6
.yaml
→
egs/mustc/asr/conf/pyramid
4_base_sr8
.yaml
查看文件 @
99763132
arch
:
pys2t_transformer_s
arch
:
pys2t_transformer_s
encoder-embed-dim
:
512
encoder-embed-dim
:
256
pyramid-stages
:
3
pyramid-stages
:
4
pyramid-layers
:
2_2_2
#pyramid-dropout: 0
pyramid-sr-ratios
:
2_2_2
pyramid-layers
:
3_3_3_3
pyramid-embed-dims
:
128_256_512
pyramid-sr-ratios
:
2_2_1_2
pyramid-fuse
:
True
pyramid-fuse-way
:
all_conv
pyramid-embed-dims
:
256_256_256_256
pyramid-reduced-embed
:
conv
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_
0_0
pyramid-position-embed
:
1_
1_1_1
pyramid-kernel-sizes
:
5_5_5
pyramid-kernel-sizes
:
5_5_5
_5
pyramid-ffn-ratios
:
8_8_
4
pyramid-ffn-ratios
:
8_8_
8_8
pyramid-heads
:
2_4_8
pyramid-heads
:
4_4_4_4
train-subset
:
train_asr
train-subset
:
train_asr
valid-subset
:
dev_asr
valid-subset
:
dev_asr
max-epoch
:
5
0
max-epoch
:
10
0
max-update
:
100000
max-update
:
100000
num-workers
:
8
num-workers
:
8
patience
:
1
0
patience
:
2
0
no-progress-bar
:
True
no-progress-bar
:
True
log-interval
:
100
log-interval
:
100
seed
:
1
seed
:
1
...
@@ -36,14 +39,14 @@ warmup-updates: 10000
...
@@ -36,14 +39,14 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy
criterion
:
label_smoothed_cross_entropy
_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-channels
:
1024
conv-channels
:
1024
dropout
:
0.1
dropout
:
0.1
activation-fn
:
relu
activation-fn
:
relu
encoder-ffn-embed-dim
:
2048
encoder-ffn-embed-dim
:
2048
encoder-layers
:
6
encoder-layers
:
12
decoder-layers
:
6
decoder-layers
:
6
encoder-attention-heads
:
4
encoder-attention-heads
:
4
...
...
egs/mustc/asr/conf/rpr.yaml
查看文件 @
99763132
#
encoder-attention-type: rel_selfattn
encoder-attention-type
:
rel_selfattn
encoder-attention-type
:
relative
#
encoder-attention-type: relative
max-encoder-relative-length
:
100
#
max-encoder-relative-length: 100
egs/mustc/asr/conf/tmp.yaml
deleted
100644 → 0
查看文件 @
5160a9f5
egs/mustc/asr/local/monitor.sh
查看文件 @
99763132
gpu_num
=
1
gpu_num
=
4
cmd
=
"sh train.sh"
cmd
=
"sh train.sh"
while
:
while
:
...
...
egs/mustc/asr/run.sh
查看文件 @
99763132
...
@@ -41,8 +41,8 @@ lcrm=0
...
@@ -41,8 +41,8 @@ lcrm=0
tokenizer
=
0
tokenizer
=
0
use_specific_dict
=
0
use_specific_dict
=
0
specific_prefix
=
valid
specific_prefix
=
st
specific_dir
=
/home/xuchen/st/data/mustc/st
_lcrm
/en-de
specific_dir
=
/home/xuchen/st/data/mustc/st/en-de
asr_vocab_prefix
=
spm_unigram10000_st_share
asr_vocab_prefix
=
spm_unigram10000_st_share
org_data_dir
=
~/st/data/
${
dataset
}
org_data_dir
=
~/st/data/
${
dataset
}
...
...
egs/mustc/asr/train.sh
查看文件 @
99763132
...
@@ -6,8 +6,14 @@ gpu_num=8
...
@@ -6,8 +6,14 @@ gpu_num=8
update_freq
=
1
update_freq
=
1
max_tokens
=
40000
max_tokens
=
40000
exp_tag
=
baseline
exp_tag
=
config_list
=(
base
)
#config_list=(base)
#config_list=(ctc)
#config_list=(base conformer)
#config_list=(pyramid4_base)
config_list
=(
pyramid4_base conformer rpr
)
# exp full name
# exp full name
exp_name
=
exp_name
=
...
...
egs/mustc/mt/decode.sh
查看文件 @
99763132
...
@@ -13,7 +13,7 @@ fi
...
@@ -13,7 +13,7 @@ fi
n_average
=
10
n_average
=
10
beam_size
=
5
beam_size
=
5
len_penalty
=
1.0
len_penalty
=
1.0
max_tokens
=
1
0000
max_tokens
=
8
0000
dec_model
=
checkpoint_best.pt
dec_model
=
checkpoint_best.pt
cmd
=
"./run.sh
cmd
=
"./run.sh
...
@@ -31,9 +31,9 @@ cmd="./run.sh
...
@@ -31,9 +31,9 @@ cmd="./run.sh
if
[[
-n
${
data_dir
}
]]
;
then
if
[[
-n
${
data_dir
}
]]
;
then
cmd
=
"
$cmd
--data_dir
${
data_dir
}
"
cmd
=
"
$cmd
--data_dir
${
data_dir
}
"
fi
fi
if
[[
${#
test_subset
[@]
}
-ne
0
]]
;
then
if
[[
-n
${
test_subset
}
]]
;
then
subsets
=
$(
echo
${
test_subset
[*]
}
| sed
's/ /,/g'
)
test_subset
=
`
echo
${
test_subset
[*]
}
| sed
's/ /,/g'
`
cmd
=
"
$cmd
--test_subset
${
subsets
}
"
cmd
=
"
$cmd
--test_subset
${
test_subset
}
"
fi
fi
echo
$cmd
echo
$cmd
...
...
egs/mustc/mt/run.sh
查看文件 @
99763132
...
@@ -20,7 +20,7 @@ stop_stage=0
...
@@ -20,7 +20,7 @@ stop_stage=0
######## hardware ########
######## hardware ########
# devices
# devices
#
device=()
device
=()
gpu_num
=
8
gpu_num
=
8
update_freq
=
1
update_freq
=
1
...
@@ -41,10 +41,10 @@ lcrm=0
...
@@ -41,10 +41,10 @@ lcrm=0
tokenizer
=
0
tokenizer
=
0
use_specific_dict
=
0
use_specific_dict
=
0
specific_prefix
=
wmt_share32k
specific_prefix
=
st
specific_dir
=
/home/xuchen/st/data/
wmt/mt_lcrm/en-de/unigram32000_share
specific_dir
=
/home/xuchen/st/data/
mustc/st/en-de/
src_vocab_prefix
=
spm_unigram
32000
_share
src_vocab_prefix
=
spm_unigram
10000_st
_share
tgt_vocab_prefix
=
spm_unigram
32000
_share
tgt_vocab_prefix
=
spm_unigram
10000_st
_share
org_data_dir
=
~/st/data/
${
dataset
}
org_data_dir
=
~/st/data/
${
dataset
}
data_dir
=
~/st/data/
${
dataset
}
/mt/
${
lang
}
data_dir
=
~/st/data/
${
dataset
}
/mt/
${
lang
}
...
@@ -54,14 +54,14 @@ trans_subset=tst-COMMON
...
@@ -54,14 +54,14 @@ trans_subset=tst-COMMON
test_subset
=
test
test_subset
=
test
# exp
# exp
exp_prefix
=
$
(
date
"+%m%d"
)
exp_prefix
=
$
{
time
}
extra_tag
=
extra_tag
=
extra_parameter
=
extra_parameter
=
exp_tag
=
baseline
exp_tag
=
baseline
exp_name
=
exp_name
=
# config
# config
train_config
=
base
train_config
=
base
_s
# training setting
# training setting
fp16
=
1
fp16
=
1
...
@@ -103,6 +103,7 @@ fi
...
@@ -103,6 +103,7 @@ fi
.
./local/parse_options.sh
||
exit
1
;
.
./local/parse_options.sh
||
exit
1
;
# full path
if
[[
-z
${
exp_name
}
]]
;
then
if
[[
-z
${
exp_name
}
]]
;
then
config_string
=
${
train_config
//,/_
}
config_string
=
${
train_config
//,/_
}
# exp_name=${exp_prefix}_$(basename ${train_config%.*})_${exp_tag}
# exp_name=${exp_prefix}_$(basename ${train_config%.*})_${exp_tag}
...
@@ -150,7 +151,7 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
...
@@ -150,7 +151,7 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
mkdir
-p
${
data_dir
}
/data
mkdir
-p
${
data_dir
}
/data
for
split
in
${
train_subset
}
${
valid_subset
}
${
trans_subset
}
;
do
for
split
in
${
train_subset
}
${
valid_subset
}
${
trans_subset
}
;
do
{
{
cmd
=
"cat
${
org_data_dir
}
/
${
lang
}
/data/
${
split
}
.
${
src_lang
}
"
cmd
=
"cat
${
org_data_dir
}
/
${
lang
}
/data/
${
split
}
/txt/
${
split
}
.
${
src_lang
}
"
if
[[
${
lcrm
}
-eq
1
]]
;
then
if
[[
${
lcrm
}
-eq
1
]]
;
then
cmd
=
"python local/lower_rm.py
${
org_data_dir
}
/
${
lang
}
/data/
${
split
}
.
${
src_lang
}
"
cmd
=
"python local/lower_rm.py
${
org_data_dir
}
/
${
lang
}
/data/
${
split
}
.
${
src_lang
}
"
fi
fi
...
@@ -165,7 +166,7 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
...
@@ -165,7 +166,7 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
cmd
=
"spm_encode
cmd
=
"spm_encode
--model
${
data_dir
}
/
${
tgt_vocab_prefix
}
.model
--model
${
data_dir
}
/
${
tgt_vocab_prefix
}
.model
--output_format=piece
--output_format=piece
<
${
org_data_dir
}
/
${
lang
}
/data/
${
split
}
.
${
tgt_lang
}
<
${
org_data_dir
}
/
${
lang
}
/data/
${
split
}
/txt/
${
split
}
.
${
tgt_lang
}
>
${
data_dir
}
/data/
${
split
}
.
${
tgt_lang
}
"
>
${
data_dir
}
/data/
${
split
}
.
${
tgt_lang
}
"
echo
-e
"
\0
33[34mRun command:
\n
${
cmd
}
\0
33[0m"
echo
-e
"
\0
33[34mRun command:
\n
${
cmd
}
\0
33[0m"
...
@@ -178,7 +179,7 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
...
@@ -178,7 +179,7 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
--source-lang
${
src_lang
}
--target-lang
${
tgt_lang
}
--source-lang
${
src_lang
}
--target-lang
${
tgt_lang
}
--trainpref
${
data_dir
}
/data/
${
train_subset
}
--trainpref
${
data_dir
}
/data/
${
train_subset
}
--validpref
${
data_dir
}
/data/
${
valid_subset
}
--validpref
${
data_dir
}
/data/
${
valid_subset
}
--testpref
${
data_dir
}
/data/
${
t
est
_subset
}
--testpref
${
data_dir
}
/data/
${
t
rans
_subset
}
--destdir
${
data_dir
}
/data-bin
--destdir
${
data_dir
}
/data-bin
--srcdict
${
data_dir
}
/
${
src_vocab_prefix
}
.txt
--srcdict
${
data_dir
}
/
${
src_vocab_prefix
}
.txt
--tgtdict
${
data_dir
}
/
${
tgt_vocab_prefix
}
.txt
--tgtdict
${
data_dir
}
/
${
tgt_vocab_prefix
}
.txt
...
@@ -265,7 +266,7 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
...
@@ -265,7 +266,7 @@ if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
save_interval
=
1
save_interval
=
1
keep_last_epochs
=
10
keep_last_epochs
=
10
no_epoch_checkpoints
=
0
no_epoch_checkpoints
=
0
save_interval_updates
=
100
00
save_interval_updates
=
5
00
keep_interval_updates
=
10
keep_interval_updates
=
10
else
else
validate_interval
=
1
validate_interval
=
1
...
@@ -352,7 +353,7 @@ if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
...
@@ -352,7 +353,7 @@ if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
result_file
=
${
model_dir
}
/decode_result
result_file
=
${
model_dir
}
/decode_result
[[
-f
${
result_file
}
]]
&&
rm
${
result_file
}
[[
-f
${
result_file
}
]]
&&
rm
${
result_file
}
test_subset
=
${
test_subset
//,/
}
test_subset
=
(
${
test_subset
//,/
}
)
for
subset
in
${
test_subset
[@]
}
;
do
for
subset
in
${
test_subset
[@]
}
;
do
cmd
=
"python
${
root_dir
}
/fairseq_cli/generate.py
cmd
=
"python
${
root_dir
}
/fairseq_cli/generate.py
${
data_dir
}
${
data_dir
}
...
...
egs/mustc/mt/train.sh
查看文件 @
99763132
...
@@ -2,9 +2,9 @@
...
@@ -2,9 +2,9 @@
# training the model
# training the model
gpu_num
=
8
gpu_num
=
1
update_freq
=
1
update_freq
=
1
max_tokens
=
4096
max_tokens
=
8192
exp_tag
=
baseline
exp_tag
=
baseline
config_list
=(
base
)
config_list
=(
base
)
...
...
egs/mustc/st/conf/base.yaml
查看文件 @
99763132
train-subset
:
train_st
train-subset
:
train_st
valid-subset
:
dev_st
valid-subset
:
dev_st
max-epoch
:
5
0
max-epoch
:
10
0
max-update
:
100000
max-update
:
100000
num-workers
:
8
num-workers
:
8
...
@@ -24,7 +24,7 @@ warmup-updates: 10000
...
@@ -24,7 +24,7 @@ warmup-updates: 10000
lr
:
2e-3
lr
:
2e-3
#adam_betas: (0.9,0.98)
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy
criterion
:
label_smoothed_cross_entropy
_with_ctc
label_smoothing
:
0.1
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-kernel-sizes
:
5,5
...
...
egs/mustc/st/conf/ctc.yaml
查看文件 @
99763132
train-subset
:
train_st
ctc-weight
:
0.3
valid-subset
:
dev_st
\ No newline at end of file
max-epoch
:
50
max-update
:
100000
num-workers
:
8
patience
:
10
no-progress-bar
:
True
log-interval
:
100
seed
:
1
report-accuracy
:
True
#load-pretrained-encoder-from:
#load-pretrained-decoder-from:
arch
:
s2t_transformer_s
share-decoder-input-output-embed
:
True
optimizer
:
adam
clip-norm
:
10.0
lr-scheduler
:
inverse_sqrt
warmup-init-lr
:
1e-7
warmup-updates
:
10000
lr
:
2e-3
#adam_betas: (0.9,0.98)
ctc-weight
:
0.3
criterion
:
label_smoothed_cross_entropy_with_ctc
label_smoothing
:
0.1
conv-kernel-sizes
:
5,5
conv-channels
:
1024
dropout
:
0.1
activation-fn
:
relu
encoder-embed-dim
:
256
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
decoder-layers
:
6
encoder-attention-heads
:
4
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
attention-dropout
:
0.1
activation-dropout
:
0.1
egs/mustc/st/conf/pyramid4_base_sr8.yaml
0 → 100644
查看文件 @
99763132
arch
:
pys2t_transformer_s
encoder-embed-dim
:
256
#pyramid-dropout: 0
pyramid-stages
:
4
pyramid-layers
:
3_3_3_3
pyramid-sr-ratios
:
2_2_1_2
pyramid-embed-dims
:
256_256_256_256
pyramid-fuse
:
True
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1_1
pyramid-kernel-sizes
:
5_5_5_5
pyramid-ffn-ratios
:
8_8_8_8
pyramid-heads
:
4_4_4_4
train-subset
:
train_st
valid-subset
:
dev_st
max-epoch
:
100
max-update
:
100000
num-workers
:
8
patience
:
10
no-progress-bar
:
True
log-interval
:
100
seed
:
1
report-accuracy
:
True
#load-pretrained-encoder-from: /home/xuchen/st/checkpoints/mustc/asr/1002_pyramid4_all256_3333_sr8/avg_10_checkpoint.pt
#load-pretrained-encoder-from: /home/xuchen/st/checkpoints/mustc/asr/1002_pyramid4_all256_3333_sr8/checkpoint_best.pt
load-pretrained-encoder-from
:
/home/xuchen/st/checkpoints/mustc/asr/1007_st_pyramid4_all256_3333_sr8_ctc/avg_10_checkpoint.pt
load-pretrained-decoder-from
:
/home/xuchen/st/checkpoints/mustc/mt/st_1003_2349_train_s_baseline/avg_10_checkpoint.pt
share-decoder-input-output-embed
:
True
optimizer
:
adam
clip-norm
:
10.0
lr-scheduler
:
inverse_sqrt
warmup-init-lr
:
1e-7
warmup-updates
:
10000
lr
:
2e-3
#adam_betas: (0.9,0.98)
criterion
:
label_smoothed_cross_entropy_with_ctc
label_smoothing
:
0.1
conv-channels
:
1024
dropout
:
0.1
activation-fn
:
relu
encoder-ffn-embed-dim
:
2048
encoder-layers
:
12
decoder-layers
:
6
encoder-attention-heads
:
4
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
attention-dropout
:
0.1
activation-dropout
:
0.1
egs/mustc/st/conf/sate_ctc.yaml
查看文件 @
99763132
train-subset
:
train_st
train-subset
:
train_st
valid-subset
:
dev_st
valid-subset
:
dev_st
max-epoch
:
5
0
max-epoch
:
10
0
max-update
:
100000
max-update
:
100000
num-workers
:
8
num-workers
:
8
...
@@ -50,8 +50,22 @@ cnn-module-kernel: 31
...
@@ -50,8 +50,22 @@ cnn-module-kernel: 31
acoustic-encoder
:
transformer
acoustic-encoder
:
transformer
adapter
:
league
adapter
:
league
#decoder-embed-dim: 256
encoder-embed-dim
:
256
#decoder-ffn-embed-dim: 2048
pyramid-stages
:
4
#decoder-attention-heads: 4
#pyramid-dropout: 0
#attention-dropout: 0.1
pyramid-layers
:
3_3_3_3
#activation-dropout: 0.1
pyramid-sr-ratios
:
2_2_1_2
pyramid-embed-dims
:
256_256_256_256
pyramid-fuse
:
True
pyramid-reduced-embed
:
conv
pyramid-embed-norm
:
True
pyramid-position-embed
:
1_1_1_1
pyramid-kernel-sizes
:
5_5_5_5
pyramid-ffn-ratios
:
8_8_8_8
pyramid-heads
:
4_4_4_4
decoder-embed-dim
:
256
decoder-ffn-embed-dim
:
2048
decoder-attention-heads
:
4
attention-dropout
:
0.1
activation-dropout
:
0.1
egs/mustc/st/conf/tmp.yaml
deleted
100644 → 0
查看文件 @
5160a9f5
egs/mustc/st/decode.sh
查看文件 @
99763132
...
@@ -13,7 +13,7 @@ fi
...
@@ -13,7 +13,7 @@ fi
n_average
=
10
n_average
=
10
beam_size
=
5
beam_size
=
5
len_penalty
=
1.0
len_penalty
=
1.0
max_tokens
=
1
0000
max_tokens
=
8
0000
dec_model
=
checkpoint_best.pt
dec_model
=
checkpoint_best.pt
cmd
=
"./run.sh
cmd
=
"./run.sh
...
...
egs/mustc/st/train.sh
查看文件 @
99763132
...
@@ -6,8 +6,16 @@ gpu_num=8
...
@@ -6,8 +6,16 @@ gpu_num=8
update_freq
=
1
update_freq
=
1
max_tokens
=
40000
max_tokens
=
40000
exp_tag
=
baseline
exp_tag
=
#config_list=(base)
config_list
=(
ctc
)
config_list
=(
ctc
)
#config_list=(sate_ctc)
#config_list=(ctc conformer rpr)
#config_list=(base sate)
#config_list=(pyramid4_base_sr8)
#config_list=(pyramid4_base_sr8 conformer)
# exp full name
# exp full name
exp_name
=
exp_name
=
...
...
编写
预览
Markdown
格式
0%
重试
或
添加新文件
添加附件
取消
您添加了
0
人
到此讨论。请谨慎行事。
请先完成此评论的编辑!
取消
请
注册
或者
登录
后发表评论