onnxruntime
7e253aab
- Introduce attribute to output QK before or after the softmax
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
1 year ago
Introduce attribute to output QK before or after the softmax (cherry picked from commit b96aa6364b7b7c969a274555ada948c941dab8fd)
References
derdeljan/asg_attention_scores_buffer
Author
derdeljan-msft
Committer
derdeljan-msft
Parents
6b31a279
Loading