-
Notifications
You must be signed in to change notification settings - Fork 427
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
feat: Add recognition metric (exact match) #110
Conversation
Codecov Report
@@ Coverage Diff @@
## main #110 +/- ##
==========================================
- Coverage 97.79% 97.43% -0.37%
==========================================
Files 28 29 +1
Lines 908 935 +27
==========================================
+ Hits 888 911 +23
- Misses 20 24 +4
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
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.
Thanks for the PR! I added a few comments. Besides, especially if we don't ignore accents or case, we could actually do this with TF and compare the output before decoding?
I think so, but it would require tensor manipulations to compare row-wise both tensors. What I was doing previously was translating with the dictionary to a string tensor and I concatenated the columns to have a vector of words, but I think we can compare the number sequences at a row-level without translating the outputs to save time. |
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.
We might want later to convert this to a TF-only computation, but it will do for now! Thanks for the edits 🙏
This PR implements ExactMatch metric (word-level accuracy) for recognition task
Any feedback is welcome !