CoderCowMoo/Llama-3-70b-Instruct-Orthoganalized

problem with marlin_kernel step

disable exllamav2 and use marlin

use_exllama=false?

Update requirements.txt

exactly like the exampe

forgot comma dumbass

Might've been adding device= in wrong spot.

No device, and cuda in the spaces function?

Maybe auto device map

Maybe remove device?

change device to cuda:0 same as mazaiyahr's repo

Add auto-gptq dep

Change to use GPTQ so i dont get limited

llama 3 dumbass

import from transformers dumbass

fix bitsandbytes no gpu support hopefully

Remove pytorch version entirely

2.2.1 to 2.2.0

Change pytorch version and cuda version hopefully fix

Update README.md

Destroy default and add my own files

Duplicate from gradio-templates/chatbot