Awni Hannun
Awni Hannun
Could you share the error message you are getting? It looks like mlx is installed correctly for you or am I missing something?
Please try the following two commands in succession: ``` pip uninstall mlx mlx-metal pip install mlx==0.27.1 ```
Maybe try installing with a fresh conda environment?
Ok in that case, could you share the sequence of commands you ran and we can see if we can reproduce the installation issue on our end?
> I just do uv sync -U --refresh --reinstall --compile-bytecode and uv run @hermeschen1116 what are the full steps you took. That's not quite enough information to repro what you...
@zcbenz that's extremely cool. I'm supportive of exploring the addition of WebGPU as a back-end for MLX. One initial comment is it would be nice to avoid breaking the "unified...
> What do you think if we add a "null" backend in upstream that mimics the general gpu backend by copying data to a separate buffer and then just calls...
> can you do a simple review? Sorry for the delay. I will take a look early this week and think about your other comments as well.
Thanks for flagging. Indeed the way we do streaming decode in the T5 example is not correct for most tokenizers (you can't typically decode each new token individually [as we...
Thanks for the crash report.. that's pretty odd. Are you able to share the code that caused the crash? Even if it's not consistent that would be of great help...