Awni Hannun
Awni Hannun
We did indeed: https://ml-explore.github.io/mlx/build/html/usage/distributed.html I think we can close this and open more targeted issues related to distributed models as they come up.
My guess is some of the U Net wants are not well supported yet in MLX (pooling / upsampling / transpose conv). We will add those ops in due time...
Ok.. well that's a PyTorch issue. In MLX we have a 3D convolution, so that should work
Sort of expected yes: -The MLX memory cache will cache a lot of memory if you have a lot available on your machine -The GPU is pretty greedy, it will...
Awesome! Would be great if you can comment on that PR re groups and dilation so we can get a sense of priority
Given the age of this PR and that we have SAM already, let's close it. Would be happy to point to an external repo with an implementation if you pick...
Wow! 🚤 Does it work yet?
Yes, it's downloading. I will let you know how it goes.
> At the moment, I can only run the 2-bit version due to space. 2-bit almost never works well, I don't recommend even trying it..
A 4-bit version produces the following. It looks OK, not obviously wrong but not that good either. `python -m mlx_lm.generate --model mlx_model --prompt "Hello, how are you?"` ``` ========== Prompt:...