Awni Hannun comments

Results 1014 comments of


                                            Awni Hannun

Updated CIFAR-10 ResNet example to use BatchNorm instead of LayerNorm

Thanks @menzHSE !

Quantize linear layer in LoRA

Closes #247

Update README.md

Thanks!

[Feature] Add Metal support for FFT

No we actually do not have an FFT in Metal yet :\ A possible workaround until we add it is `b_hat = mx.fft.rfft2(b, stream=mx.cpu)` (and in the `irfft2` as well).

[Feature] Add Metal support for FFT

This issue should really be in MLX core..closing here and moving it.

Move lora example to use the same model format / conversion as `hf_llm`

CC @pcuenca @Vaibhavs10

Move lora example to use the same model format / conversion as `hf_llm`

> Nice work! Just a quick question: for the best fine-tune result, would it be possible to provide a flag that allows people to apply Lora to all linear layers...

Move lora example to use the same model format / conversion as `hf_llm`

I added a fuse.py which fuses the model and optionally uploads it to the hub. H/T @mzbac for starting that.

Move lora example to use the same model format / conversion as `hf_llm`

Closes #232 and #212

What's a good data format for lora fine-tuning?

You can put whatever you want in the `"text"` field of the json. So in your case you could simply do: ``` {"text": ".### User:\nWhat's the meaning of the word...