update the shell scripts

egs/librispeech/asr/conf/conformer.yaml

@@ -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: 31
+cnn-module-kernel: 15

egs/librispeech/asr/conf/pyramid.yaml

 arch: pys2t_transformer_s
 arch: pys2t_transformer_s
 encoder-embed-dim: 512
 encoder-embed-dim: 512
 pyramid-stages: 4
 pyramid-stages: 4
-pyramid-layers: 2_2_5_3
-encoder-attention-type: reduced
-pyramid-attn-sample-ratios: 8_4_2_1
+pyramid-layers: 2_2_6_2
+#encoder-attention-type: reduced
+#pyramid-attn-sample-ratios: 8_4_2_1
 pyramid-sr-ratios: 2_2_2_2
 pyramid-sr-ratios: 2_2_2_2
-pyramid-embed-dims: 64_128_256_512
+pyramid-use-ppm: True
+pyramid-embed-dims: 128_128_256_512
 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_4
-pyramid-heads: 1_2_4_8
+pyramid-heads: 2_2_4_8
 
 
 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

egs/librispeech/asr/conf/pyramid_stage3.yaml

@@ -2,8 +2,11 @@ arch: pys2t_transformer_s
 encoder-embed-dim: 512
 encoder-embed-dim: 512
 pyramid-stages: 3
 pyramid-stages: 3
 pyramid-layers: 3_6_3
 pyramid-layers: 3_6_3
-encoder-attention-type: reduced
-pyramid-attn-sample-ratios: 4_2_1
+#encoder-attention-type: reduced
+#pyramid-attn-sample-ratios: 4_2_1
+#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: 128_256_512
 pyramid-reduced-embed: conv
 pyramid-reduced-embed: conv

egs/librispeech/asr/conf/rpr.yaml

-encoder-attention-type: rel_selfattn
+#encoder-attention-type: rel_selfattn
+encoder-attention-type: relative
+max-encoder-relative-length: 100

egs/librispeech/asr/decode.sh

@@ -31,8 +31,8 @@ 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 [[ -n ${test_subset} ]]; then
-    test_subset=`echo ${test_subset[*]} | sed 's/ /,/g'`
+if [[ ${#test_subset[@]} -ne 0 ]]; then
+    subsets=$(echo ${test_subset[*]} | sed 's/ /,/g')
     cmd="$cmd --test_subset ${test_subset}"
     cmd="$cmd --test_subset ${test_subset}"
 fi
 fi
 
 

egs/librispeech/asr/local/monitor.sh

-gpu_num=1
+gpu_num=8
 cmd="sh train.sh"
 cmd="sh train.sh"
 
 
 while :
 while :

egs/librispeech/asr/train.sh

@@ -2,13 +2,15 @@
 
 
 # training the model
 # training the model
 
 
-gpu_num=4
+gpu_num=8
 update_freq=1
 update_freq=1
-max_tokens=200000
+max_tokens=100000
 
 
 exp_tag=
 exp_tag=
 config_list=(base)
 config_list=(base)
-config_list=(pyramid)
+config_list=(base conformer)
+#config_list=(pyramid)
+#config_list=(pyramid_stage3)
 
 
 # exp full name
 # exp full name
 exp_name=
 exp_name=

egs/mustc/asr/conf/pyramid.yaml

@@ -3,13 +3,14 @@ encoder-embed-dim: 512
 pyramid-stages: 4
 pyramid-stages: 4
 pyramid-layers: 2_3_5_2
 pyramid-layers: 2_3_5_2
 pyramid-sr-ratios: 2_2_2_2
 pyramid-sr-ratios: 2_2_2_2
-pyramid-embed-dims: 64_128_256_512
+pyramid-embed-dims: 128_128_256_512
+pyramid-use-ppm: True
 pyramid-reduced-embed: conv
 pyramid-reduced-embed: conv
 pyramid-embed-norm: True
 pyramid-embed-norm: True
-pyramid-position-embed: 1_0_0_0
+pyramid-position-embed: 1_1_1_1
 pyramid-kernel-sizes: 5_5_5_5
 pyramid-kernel-sizes: 5_5_5_5
-pyramid-ffn-ratios: 4_4_4_4
-pyramid-heads: 1_2_4_8
+pyramid-ffn-ratios: 8_8_8_4
+pyramid-heads: 2_2_4_8
 
 
 train-subset: train_asr
 train-subset: train_asr
 valid-subset: dev_asr
 valid-subset: dev_asr

egs/mustc/asr/conf/pyramid_stage3.yaml

@@ -3,10 +3,13 @@ encoder-embed-dim: 512
 pyramid-stages: 3
 pyramid-stages: 3
 pyramid-layers: 3_6_3
 pyramid-layers: 3_6_3
 pyramid-sr-ratios: 2_2_2
 pyramid-sr-ratios: 2_2_2
+#pyramid-block-attn: True
+#pyramid-fuse-way: add
+pyramid-use-ppm: True
 pyramid-embed-dims: 128_256_512
 pyramid-embed-dims: 128_256_512
 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
 pyramid-kernel-sizes: 5_5_5
 pyramid-kernel-sizes: 5_5_5
 pyramid-ffn-ratios: 8_8_4
 pyramid-ffn-ratios: 8_8_4
 pyramid-heads: 2_4_8
 pyramid-heads: 2_4_8

egs/mustc/asr/conf/rpr.yaml

-encoder-attention-type: rel_selfattn
+#encoder-attention-type: rel_selfattn
+encoder-attention-type: relative
+max-encoder-relative-length: 100

egs/mustc/asr/decode.sh

@@ -31,7 +31,7 @@ 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[@]} -eq 0 ]]; then
+if [[ ${#test_subset[@]} -ne 0 ]]; then
     subsets=$(echo ${test_subset[*]} | sed 's/ /,/g')
     subsets=$(echo ${test_subset[*]} | sed 's/ /,/g')
     cmd="$cmd --test_subset ${subsets}"
     cmd="$cmd --test_subset ${subsets}"
 fi
 fi

egs/mustc/asr/train.sh

@@ -2,12 +2,16 @@
 
 
 # training the model
 # training the model
 
 
-gpu_num=8
+gpu_num=4
 update_freq=1
 update_freq=1
-max_tokens=40000
+max_tokens=80000
 
 
-exp_tag=
+#exp_tag=valid_prev_state
+#exp_tag=lower128
+#exp_tag=sr8
+#config_list=(base conformer rpr)
 config_list=(pyramid)
 config_list=(pyramid)
+#config_list=(pyramid_stage3 rpr)
 
 
 # exp full name
 # exp full name
 exp_name=
 exp_name=