Unverified85% confidenceFactTime unknown
Gao et al.（2023）的研究发现随着策略模型对奖励模型的优化程度增加，真实人类偏好评分先升后降呈现倒U形曲线
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedHuman annotators exhibit a cognitive bias toward more information being better when evaluating answers, causing reward models to implicitly link length with quality, which is one of the causes of model verbosity.74% similarUnverifiedAlpha优势具有自我消亡属性，一旦被广泛发现和使用，套利行为会消除价差，导致因子随市场参与者增加而衰减68% similarUnverified年度订阅模式的成功建立在承诺一致性原理、损失厌恶和心理账户等多重行为经济学心理机制之上66% similarUnverifiedResearch by Haggard et al. discovered the phenomenon of intentional binding: when a person voluntarily initiates an action, they subjectively compress the time interval between the action and its outcome.66% similarUnverified顶级会议的马太效应产生自我强化的正反馈循环，形成赢家通吃格局66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/33809API
curl https://kongchang.com/api/v1/knowledge/claims/33809MCP
get_claim(id=33809)