3 ms·
Have you tried optimizing for MLX? It seems like a waste to have neural cores and not use them. I've often wondered why the hype around apple neural core when
by blobbers 4mo ago
Have you tried optimizing for MLX? It seems like a waste to have neural cores and not use them.
I've often wondered why the hype around apple neural core when 99% of software doesn't use them.
- genxy 4mo agoYeah, first think I looked for on the post was MLX and it wasn't there. https://github.com/ml-explore/mlx-lm https://github.com/ml-explore/mlx-lm Having used half the systems that Vicki mentioned, mlx was the best balance between power and ease of use. Just a pip install away.