most citedPhi-3 Safety Post-Training: Aligning Language Models with a "Break-Fix" Cycle

2 citations · 2 across the 2 of their papers we have counts for

collaborators
Showing cs.LGShow all

1 paper · 1 filter

cs.LG2024

Steering Language Model Refusal with Sparse Autoencoders

Kyle O'Brien, David Majercak, Xavier Fernandes +7

Responsible deployment of language models requires mechanisms for refusing unsafe prompts while preserving model performance. While most approaches modify model weights through add…