new3시간 전· 1 sources
Dynamic Abliteration: Non-Destructive Refusal Suppression via Engram Steering
Why it matters
When working with open-weight LLMs like Qwen, controlling refusal behavior on security, administrative, prompts typically requires fine-tuning or permanent weight update. Traditional weight abliterati...
1
Sources
+1
24h
—
Growth
1d
Active