TheBloke/koala-7B-GPTQ

Update for Transformers GPTQ support

adding missing comma (#6)

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Upload new GPTQs with varied parameters

Delete koala-7B-4bit-128g.safetensors

Delete koala-7B-4bit-128g.no-act-order.ooba.pt

Update README.md

Updating model files

Update README.md

Update tokenizer_config.json to match other models, like Vicuna - the Koala conversion process does not fill out this file, but based on the code I think these special tokens should be defined.

Update special_tokens_map.json to match other models, like Vicuna - the Koala conversion process does not fill out this file, but based on the code I think these special tokens should be defined.

Update README.md

Delete koala-7B-4bit-128g.pt

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Add model made with oobabooga's older fork of GPTQ, suitable for use in text-generation-webui without requiring an update to GPTQ code. Also for GGML conversion.

Delete koala-7B-4bit-128g.olderFormat.pt

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

add safetensors model

Update README.md

Warning re GPTQ files not working

Add older GPTQ format that doesn't use g_idx

Add JSON files

Add model file

Add readme

initial commit