QuixiAI / Llama-3-8B-Instruct-abliterated-v2-gguf

The model was trained with more data to better pinpoint the "refusal direction". This model is MUCH better at directly and succinctly answering requests without producing even so much as disclaimers.