Skip to content
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

Improve future compatibility of CLIPMultiHeadAttention #1975

Conversation

james77777778
Copy link
Collaborator

This PR ensures that merging keras-team/keras#20448 will not break CLIP.

The original issue: when performing

attention_output = ops.einsum(self._combine_equation, final_attn_scores, value)

in MultiHeadAttention, a dtype mismatch caused an error.

@divyashreepathihalli
Copy link
Collaborator

Thanks Hongyu!!

@divyashreepathihalli divyashreepathihalli added the kokoro:force-run Runs Tests on GPU label Nov 11, 2024
@kokoro-team kokoro-team removed the kokoro:force-run Runs Tests on GPU label Nov 11, 2024
@divyashreepathihalli divyashreepathihalli merged commit 76a6ca0 into keras-team:master Nov 11, 2024
11 checks passed
@james77777778 james77777778 deleted the improve-clip-future-compatibility branch November 12, 2024 01:42
@james77777778 james77777778 restored the improve-clip-future-compatibility branch November 12, 2024 09:24
@james77777778 james77777778 deleted the improve-clip-future-compatibility branch November 12, 2024 09:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants