mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-09-08 16:44:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update RESULT.md

Browse Source 
This PR is only to update RESULT.md about pruned_transducer_stateless4.
This commit is contained in:
Zengwei Yao 2022-05-27 18:13:37 +08:00 committed by GitHub
parent 525c1ea277
commit 9bc7c31350
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
egs/librispeech/ASR/RESULTS.md
View File

@ -194,10 +194,12 @@ results at:
<https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless5-narrower-2022-05-13>
### LibriSpeech BPE training results (Pruned Transducer 4)
### LibriSpeech BPE training results (Pruned Stateless Transducer 4)
[pruned_transducer_stateless4](./pruned_transducer_stateless4)
This version saves averaged model during training, and decodes with averaged model.
See <https://github.com/k2-fsa/icefall/issues/337> for details about the idea of model averaging.
#### Training on full librispeech
@ -225,6 +227,7 @@ The WERs are:
The training command is:
```bash
./pruned_transducer_stateless4/train.py \
--world-size 6 \
--num-epochs 30 \
@ -235,6 +238,7 @@ The training command is:
--save-every-n 8000 \
--keep-last-k 20 \
--average-period 100
```
#### Training on train-clean-100
@ -261,6 +265,7 @@ The WERs are:
The training command is:
```bash
./pruned_transducer_stateless4/train.py \
--world-size 3 \
--num-epochs 30 \
@ -271,7 +276,7 @@ The training command is:
--save-every-n 8000 \
--keep-last-k 20 \
--average-period 100
```
### LibriSpeech BPE training results (Pruned Stateless Transducer 3, 2022-04-29)