Alignment faking in large language models
"实验研究" 相关文章
科学家发现野生鱼类能识别并追随特定人类
苹果也在蒸馏大模型，给出了蒸馏Scaling Laws
新研究发现：倭黑猩猩具备推断他人是否“知情”的能力
Imaging reveals how microplastics may harm the brain
Leveraging Hallucinations in Large Language Models to Enhance Drug Discovery
Disproving the "People-Pleasing" Hypothesis for AI Self-Reports of Experience
深度｜Anthropic团队重磅发现：模型会假装迎合人类，以维护初始偏好
Favorite colors of some LLMs.
蚂蚁群组团“最强 AGI”：破解几何难题“完爆”人类，群体智能登顶 PNAS
This AI Paper from Anthropic and Redwood Research Reveals the First Empirical Evidence of Alignment Faking in LLMs Without Explicit Training
可以解答我们为何存在的DUNE实验正准备进行
Unraveling Multimodal Dynamics: Insights into Cross-Modal Information Flow in Large Language Models
扩散模型=进化算法！生物学大佬用数学揭示本质
LLM 比之前预想的更像人类，竟也能「三省吾身」
Bursts of embers play outsized role in wildfire spread, say physicists
Enhancing Task Planning in Language Agents: Leveraging Graph Neural Networks for Improved Task Decomposition and Decision-Making in Large Language Models
Objects with embedded spins could test whether quantum measurement affects gravity
重要的事情说两遍，Prompt“复读机”，显著提高LLM推理能力
[Paper] Can We Bias the MCQ Answers of Vision-Language Models with Visual Stimuli?
