mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-08 09:32:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Updated Aishell-1 transducer-stateless result (#217)

Browse Source 
* Update RESULTS.md

* Update RESULTS.md
This commit is contained in:
Duo Ma 2022-02-19 15:56:04 +08:00 committed by GitHub
parent b702281e90
commit 827b9df51a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 44 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

44
egs/aishell/ASR/RESULTS.md
View File

@ -1,5 +1,49 @@
## Results
### Aishell training result(Transducer-stateless)
#### 2022-2-19
(Duo Ma): The tensorboard log for training is available at https://tensorboard.dev/experiment/25PmX3MxSVGTdvIdhOwllw/#scalars
You can find a pretrained model by visiting https://huggingface.co/shuanguanma/icefall_aishell_transducer_stateless_context_size2_epoch60_2022_2_19
| | test |comment |
|---------------------------|------|-----------------------------------------|
| greedy search | 5.4 |--epoch 59, --avg 10, --max-duration 100 |
| beam search | 5.05|--epoch 59, --avg 10, --max-duration 100 |
You can use the following commands to reproduce our results:
```
export CUDA_VISIBLE_DEVICES="0,1,2,3"
python3 ./transducer_stateless/train.py \
--world-size 4 \
--num-epochs 60 \
--start-epoch 0 \
--exp-dir exp/transducer_stateless_context_size2 \
--max-duration 100 \
--lr-factor 2.5\
--context-size 2
lang_dir=data/lang_char
dir=exp/transducer_stateless_context_size2
python3 ./transducer_stateless/decode.py\
--epoch 59\
--avg 10\
--exp-dir $dir \
--lang-dir $lang_dir\
--decoding-method greedy_search\
--context-size 2\
--max-sym-per-frame 3
lang_dir=data/lang_char
dir=exp/transducer_stateless_context_size2
python3 ./transducer_stateless/decode.py\
--epoch 59\
--avg 10\
--exp-dir $dir \
--lang-dir $lang_dir\
--decoding-method beam_search\
--context-size 2\
--max-sym-per-frame 3
```
### Aishell training results (Transducer-stateless)
#### 2021-12-29
(Pingfeng Luo) : The tensorboard log for training is available at <https://tensorboard.dev/experiment/sPEDmAQ3QcWuDAWGiKprVg/>