로그인
통찰 - Contrastive Activation Addition (CAA)
Steering Language Models with Contrastive Activation Addition
The author introduces CAA as a method to steer language models by modifying their activations during forward passes, allowing precise control over model behavior. CAA significantly alters model behavior and provides insights into how high-level concepts are represented in Large Language Models.
Contrastive Activation Addition for Steering Language Models: Innovative Methodology Revealed
Innovative method CAA allows precise steering of language models by modifying activations, enhancing control over model behavior.
Contrastive Activation Addition for Steering Language Models
Introducing CAA for precise steering of language models by modifying activations, enhancing alignment techniques.
