mirrors/icefall
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

upload the decoding results of rnn-lm to the RESULTS

Browse Source
This commit is contained in:
Quandwang 2022-07-22 14:52:12 +08:00
parent b9fda2cb1c
commit 79d7c94397
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
egs/librispeech/ASR/RESULTS.md
View File

@ -2037,6 +2037,7 @@ The decoding configuration for the reworked model is --epoch 30, --avg 8, --use-
| whole-lattice-rescoring| 2.66% | 5.76%| 4.21%| 2.56%| 6.04%| 4.30%|
| attention-decoder | 2.59% | 5.54%| 4.07%| 2.41%| 5.77%| 4.09%|
| nbest-oracle | 1.53% | 3.47%| 2.50%| 1.69%| 4.02%| 2.86%|
|rnn-lm | 2.37% | 4.98%| 3.68%| 2.31%| 5.35%| 3.83%|
@ -2045,9 +2046,9 @@ For other decoding method, the average WER of the two test sets with the two mod
Except for the 1best and nbest method, the overall performance of reworked model is better than the baseline model.
To reproduce the above result, use the following commands:
To reproduce the above result, use the following commands.
The training commands are
The training commands are:
```bash
WORLD_SIZE=8