v1.0.0

matteo-grella released this 14 Sep 20:27

· 124 commits to main since this release

First stable release!

Fixed

Fix bug preventing the embeddings model from being traversed on nn.Apply.
Fix incorrect use of self-attention cache when used for cross-attention.

Changed

Optimize implementation of some Dense matrix functions, especially on
amd64 with AVX.

Assets 2