grimulkan's picture
Update README.md
fcc2a17 verified
|
raw
history blame
303 Bytes
---
license: llama2
---
This is a 6-bit quantization of [Story Reverse Prompt 70B 32K](https://huggingface.co/grimulkan/story-reverse-prompt-70b-rope8-32K-fp16). See that page for more details.
This quantization fits in 48GB+24GB (36/24 split) or 3x24GB (16/17/20 split) using Exllamav2 @ 32k context.