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.
function_calling| Model | Context | Max out |
|---|---|---|
| llama-3-70B-Instruct-abliterated | 8K | — |
| llama-3-typhoon-v1.5-8b-instruct | 8K | — |
| llama-3-8b-Instruct | 8K | — |
/v1/models/huggingface/failspy/llama-3-70B-Instruct-abliteratedThis 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.