Undi95

/

Storytelling-v1-13B-lora

like 6

I don't really understand your request.
I can't make an entire model out of a LoRA (yet), and I already merge this with my last versions of MLewd : https://huggingface.co/models?search=MLewd

Please more context kek

You can't merge the LoRA with the base model via something like https://github.com/tloen/alpaca-lora/blob/main/export_hf_checkpoint.py?

That was the original way of merging LoRAs. Does LLaMa 2 present some problem? I would expect GQA might for 34B or 70B files, but I thought LLaMa 2 was pretty normal.

No worries if that's the case.