failspy/Phi-3-mini-128k-instruct-abliterated-v3 is a 4 billion parameter instruction-tuned causal language model, based on Microsoft's Phi-3-mini-128k-instruct. This model has undergone an "abliteration" process, using orthogonalized bfloat16 safetensor weights to specifically inhibit refusal behaviors. It maintains the original model's capabilities while aiming for an uncensored response style, making it suitable for applications requiring direct answers without ethical lecturing.
No reviews yet. Be the first to review!
Hugging Face