New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add recipe for OCR task on IAM handwriting dataset #4707
Conversation
e0f9d83
to
f063a09
Compare
f063a09
to
75b0109
Compare
Codecov Report
@@ Coverage Diff @@
## master #4707 +/- ##
==========================================
+ Coverage 80.31% 80.32% +0.01%
==========================================
Files 527 527
Lines 46311 46311
==========================================
+ Hits 37193 37200 +7
+ Misses 9118 9111 -7
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 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Very cool PR! Could you please also update the corresponding entry in egs2/TEMPLATE/README.md
This pull request is now in conflict :( |
@kenzheng99, could you reflect @ftshijt's comments? |
- pytorch version: `pytorch 1.10.0` | ||
- Git hash: `5a6319300231b8193f1b6e8465d572be63150119` | ||
- Commit date: `Sat Sep 24 12:14:08 2022 -0400` | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can you add a pre-trained model?
This pull request is now in conflict :( |
Please reflect the comments ( |
Updated the last comment! For model upload, I have just made a HuggingFace account (username is kenzheng99) and requested access to the ESPnet org, once I get that I can upload my model. |
Very cool! You can make a separate PR for the update of the link. Will merge this PR. Many thanks for your contribution! |
Current best performance is about 6.8 CER, which is still a bit off from SOTA results on this dataset. Let me know if anyone has suggestions for further performance tuning of this recipe.