Unverified50% confidenceFactExact time
Sycophancy is a known defect in current mainstream large models that emerges after RLHF training.
1
Sources
50%
Confidence
Long-term
Relevance
9/22/2026
First Seen
Sources
Related Entities
Related Claims
Unverified低精度引入的数值扰动会在RL的多轮迭代中通过策略梯度的乘积形式被放大，可能导致训练崩溃或模型性能退化67% similarUnverified经过RLHF训练的模型在面对精心设计的越狱提示时往往能够绕过安全防护输出有害内容67% similarVerified仅用成功轨迹训练的模型往往存在分布外泛化的严重缺陷，遇到工具返回错误时便丧失纠偏能力66% similarUnverified2023年多项研究证明，即使经过RLHF对齐的模型也难以完全抵抗精心设计的间接提示词注入63% similarUnverifiedWhen endowing existing VLMs with generative capabilities, the understanding abilities accumulated during model pretraining are often impaired, resulting in catastrophic forgetting.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/941550API
curl https://kongchang.com/api/v1/knowledge/claims/941550MCP
get_claim(id=941550)