failspy / llama-3-70B-Instruct-abliterated
Text Generation71B9K154Updated May 7, 2024
Downloads
9K
Likes
154
Last updated
May 7, 2024
License
llama3
This is meta-llama/Llama-3-70B-Instruct with orthogonalized bfloat16 safetensor weights, generated with the methodology that was described in the preview paper/blog post: 'Refusal in LLMs is mediated by a single direction' which I encourage you to read to understand more.
Tasks & library
- text generation
- transformers
Tags
- transformers
- llama
- text-generation
- conversational
Technical specifications
Parameters71B
Librarytransformers
LanguagesNot available
Licensellama3