Raymond Lucke

I've been using the ChatML-style to allow for more customization of the role field on my own experiments and thought I'd share. Inspired by https://github.com/openai/openai-python/blob/main/chatml.md

# What's new - Adds auto clicker to HID remote with adjustable click rate. - Based heavily on the existing mouse jiggler. # Verification - Tested on both bluetooth and...

Add support for Llama 3

12

It looks like the tokenizer patching breaks. Here's the log: ``` ValueError Traceback (most recent call last) Cell In[1], line 20 7 # 4bit pre quantized models we support for...

Phi-3 small (7B) and medium (14B)

8

It would be great to see these models work! > NotImplementedError: Unsloth: /srv/models/Phi-3-medium-4k-instruct not supported yet! > Make an issue to https://github.com/unslothai/unsloth! Done. :)

Unsloth should probably not alter or delete source models.

1

I always keep a local copy of every model I work with regularly in a models directory on my server. I noticed that unsloth will write `input_embeddings.pt` and `output_embeddings.pt` to...

Support for FP8 quantization

2

With the release of the new [Mistral NeMo 12B model](https://mistral.ai/news/mistral-nemo/) we now have weights that were pre-trained with FP8. It would be great if Unsloth could support 8bit as well...