Add discrete-token ASR for LibriSpeech 100h#5350
Conversation
64a1cab to
dd8ea67
Compare
|
Very cool! |
simpleoier
left a comment
There was a problem hiding this comment.
Thank you! The results are very cool.
I left two minor comments.
| @@ -0,0 +1,93 @@ | |||
| # Trained with A100 (40 GB) x 1 GPUs for Kmeans1K+nbpe5K. It takes 32 minutes per epoch. | |||
There was a problem hiding this comment.
Can you update this information?
egs2/librispeech_100/asr2/run.sh
Outdated
| tgt_case="ts" | ||
|
|
||
| ./asr2.sh \ | ||
| --kmeans_opts "--batch_bins 4800000" \ |
There was a problem hiding this comment.
Can you reduce the number of gpu jobs in kmeans_opts: adding an argument --nj 4 for example?
Codecov Report
@@ Coverage Diff @@
## master #5350 +/- ##
=======================================
Coverage 76.11% 76.11%
=======================================
Files 672 672
Lines 59859 59859
=======================================
Hits 45563 45563
Misses 14296 14296
Flags with carried forward coverage won't be shown. Click here to find out more. 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
What?
Discrete-token ASR for LibriSpeech 100h.
Why?
There is no discrete-token ASR recipe for this small scale dataset yet.