From 5a3a9e1f5055260f966d24680d5bb2e83f1d5b54 Mon Sep 17 00:00:00 2001
From: Hui Zhang <zhtclz@foxmail.com>
Date: Tue, 22 Jun 2021 02:58:58 +0000
Subject: [PATCH 1/4] fix chunk default config; tarball ckpt prfix dir;

---
 examples/aishell/s1/README.md                 | 10 ++++++++++
 examples/aishell/s1/conf/chunk_conformer.yaml |  2 +-
 utils/tarball.sh                              |  3 ++-
 3 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/examples/aishell/s1/README.md b/examples/aishell/s1/README.md
index 2048c4d5..c306f8aa 100644
--- a/examples/aishell/s1/README.md
+++ b/examples/aishell/s1/README.md
@@ -9,6 +9,16 @@
 | conformer | conf/conformer.yaml | spec_aug + shift | test | ctc_prefix_beam_search | - | 0.062196 |  
 | conformer | conf/conformer.yaml | spec_aug + shift | test | attention_rescoring | - | 0.054694 |  
 
+## Chunk Conformer
+
+| Model | Config | Augmentation| Test set | Decode method | Chunk | Loss | WER |  
+| --- | --- | --- | --- | --- | --- | --- | --- |   
+| conformer | conf/chunk_conformer.yaml | spec_aug + shift | test | attention | 16 | - | 0.061939 |  
+| conformer | conf/chunk_conformer.yaml | spec_aug + shift | test | ctc_greedy_search | 16 | - | 0.070806 |  
+| conformer | conf/chunk_conformer.yaml | spec_aug + shift | test | ctc_prefix_beam_search | 16 | - | 0.070739 |  
+| conformer | conf/chunk_conformer.yaml | spec_aug + shift | test | attention_rescoring | 16 |  - | 0.059400 |  
+
+
 ## Transformer
 
 | Model | Config | Augmentation| Test set | Decode method | Loss | WER |  
diff --git a/examples/aishell/s1/conf/chunk_conformer.yaml b/examples/aishell/s1/conf/chunk_conformer.yaml
index 904624c3..e626e106 100644
--- a/examples/aishell/s1/conf/chunk_conformer.yaml
+++ b/examples/aishell/s1/conf/chunk_conformer.yaml
@@ -78,7 +78,7 @@ model:
 
 
 training:
-  n_epoch: 180
+  n_epoch: 240
   accum_grad: 4
   global_grad_clip: 5.0
   optim: adam
diff --git a/utils/tarball.sh b/utils/tarball.sh
index 100b4719..224b740c 100755
--- a/utils/tarball.sh
+++ b/utils/tarball.sh
@@ -18,7 +18,8 @@ function clean() {
 }
 trap clean EXIT
 
-cp ${ckpt_prefix}.* ${output}
+# ckpt_prfix.{json,...} and ckpt_prfix dir
+cp -r ${ckpt_prefix}* ${output}
 cp ${model_config} ${mean_std} ${vocab} ${output}
 
 tar zcvf release.tar.gz ${output}

From 68149cb9a7d39c14e95ada2979a4b7200eaf4902 Mon Sep 17 00:00:00 2001
From: Hui Zhang <zhtclz@foxmail.com>
Date: Tue, 22 Jun 2021 03:25:26 +0000
Subject: [PATCH 2/4] fix config for new datapipeline

---
 examples/aishell/s1/README.md                 |  2 +-
 examples/aishell/s1/conf/chunk_conformer.yaml | 15 +++++++++------
 2 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/examples/aishell/s1/README.md b/examples/aishell/s1/README.md
index c306f8aa..601b0a8d 100644
--- a/examples/aishell/s1/README.md
+++ b/examples/aishell/s1/README.md
@@ -12,7 +12,7 @@
 ## Chunk Conformer
 
 | Model | Config | Augmentation| Test set | Decode method | Chunk | Loss | WER |  
-| --- | --- | --- | --- | --- | --- | --- | --- |   
+| --- | --- | --- | --- | --- | --- | --- | --- |  
 | conformer | conf/chunk_conformer.yaml | spec_aug + shift | test | attention | 16 | - | 0.061939 |  
 | conformer | conf/chunk_conformer.yaml | spec_aug + shift | test | ctc_greedy_search | 16 | - | 0.070806 |  
 | conformer | conf/chunk_conformer.yaml | spec_aug + shift | test | ctc_prefix_beam_search | 16 | - | 0.070739 |  
diff --git a/examples/aishell/s1/conf/chunk_conformer.yaml b/examples/aishell/s1/conf/chunk_conformer.yaml
index e626e106..0e5b8699 100644
--- a/examples/aishell/s1/conf/chunk_conformer.yaml
+++ b/examples/aishell/s1/conf/chunk_conformer.yaml
@@ -3,17 +3,20 @@ data:
   train_manifest: data/manifest.train
   dev_manifest: data/manifest.dev
   test_manifest: data/manifest.test
-  vocab_filepath: data/vocab.txt 
-  unit_type: 'char'
-  spm_model_prefix: ''
-  augmentation_config: conf/augmentation.json
-  batch_size: 32
   min_input_len: 0.5
   max_input_len: 20.0 # second
   min_output_len: 0.0
   max_output_len: 400.0
   min_output_input_ratio: 0.05
   max_output_input_ratio: 10.0
+  
+
+collator:
+  vocab_filepath: data/vocab.txt 
+  unit_type: 'char'
+  spm_model_prefix: ''
+  augmentation_config: conf/augmentation.json
+  batch_size: 32
   raw_wav: True  # use raw_wav or kaldi feature
   specgram_type: fbank #linear, mfcc, fbank
   feat_dim: 80
@@ -30,7 +33,7 @@ data:
   keep_transcription_text: False
   sortagrad: True 
   shuffle_method: batch_shuffle
-  num_workers: 0
+  num_workers: 2
 
 
 # network architecture

From 1b84f21ccfda2794e323a69a163411ab15c17288 Mon Sep 17 00:00:00 2001
From: Hui Zhang <zhtclz@foxmail.com>
Date: Tue, 22 Jun 2021 06:27:19 +0000
Subject: [PATCH 3/4] fix miss match

---
 utils/tarball.sh | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/utils/tarball.sh b/utils/tarball.sh
index 224b740c..a4611c75 100755
--- a/utils/tarball.sh
+++ b/utils/tarball.sh
@@ -18,8 +18,11 @@ function clean() {
 }
 trap clean EXIT
 
-# ckpt_prfix.{json,...} and ckpt_prfix dir
-cp -r ${ckpt_prefix}* ${output}
+# ckpt_prfix dir
+cp -r ${ckpt_prefix} ${output}
+# ckpt_prfix.{json,...}
+cp ${ckpt_prefix}.*  ${output}
+# model config, mean std, vocab
 cp ${model_config} ${mean_std} ${vocab} ${output}
 
 tar zcvf release.tar.gz ${output}

From 3c6eea077b2b077b9c3f5cc7baf339c545053d35 Mon Sep 17 00:00:00 2001
From: Hui Zhang <zhtclz@foxmail.com>
Date: Tue, 22 Jun 2021 07:27:43 +0000
Subject: [PATCH 4/4] cp dir when it exits

---
 utils/tarball.sh | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/utils/tarball.sh b/utils/tarball.sh
index a4611c75..5f7c21a3 100755
--- a/utils/tarball.sh
+++ b/utils/tarball.sh
@@ -19,7 +19,9 @@ function clean() {
 trap clean EXIT
 
 # ckpt_prfix dir
-cp -r ${ckpt_prefix} ${output}
+if [ -d ${ckpt_prefix} ];then
+    cp -r ${ckpt_prefix} ${output}
+fi
 # ckpt_prfix.{json,...}
 cp ${ckpt_prefix}.*  ${output}
 # model config, mean std, vocab