Awni Hannun
Awni Hannun
It's hard for me to answer. I would recommend you take a look at parallel implementations of FFT. There is also this [code](https://github.com/philipturner/metal-fft) which does FFT or Metal. Maybe a...
It's not a great fit for us as we don't wrap MPSGraph and likely wouldn't make an exception for FFTs.
You're not doing anything wrong, we don't yet have a complex64 metal gemm implemented (or complex64 cpu gemm for that matter). @jagrit06 any thoughts on this? How hard would it...
Check your operating system: [requirements here](https://ml-explore.github.io/mlx/build/html/install.html) And try `pythom -m pip install mlx`
I'm not sure what you mean. You should be able to use mlx inside a pyenv.
> However, how should I run it? To run chat-with-mlx I would [check the docs there](https://github.com/qnguyen3/chat-with-mlx). That is a third party repo. If you have issues with it, I would...
Super cool, thanks for adding that!
Addresses #146
Actually, this is expected behavior. This is in general once place where we might end up being different than NumPy and go more like Jax. You can try the same...
I'm going to mark this as docs/enhancement since it's not actually a bug.