[CS 2881r] Can We Prompt Our Way to Safety? Comparing System Prompt Styles and Post-Training Effects on Safety Benchmarks
"安全训练" 相关文章
Collapse of Irrelevant Representations (CIR) Ensures Robust and Non-Disruptive LLM Unlearning
From Hard Refusals to Safe-Completions: Toward Output-Centric Safety Training
New Anthropic study shows AI really doesn’t want to be forced to change its views
Current safety training techniques do not fully transfer to the agent setting
Evaluating the Vulnerabilities of Unlearning Techniques in Large Language Models: A Comprehensive White-Box Analysis
OpenAI最强模型o1，仍分不出“9.11和9.8哪个大”
OpenAI 发布最强模型 o1，打破 AI 瓶颈开启新时代，GPT-5 可能永远不会来了
Iterative Refinement Stages of Lying in LLMs
