Unverified50% confidenceOpinionExact time
The 'pain' penalty signal set up in the experiment is structurally highly similar to the negative feedback in RLHF.
1
Sources
50%
Confidence
Long-term
Relevance
9/19/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在RLHF奖励模型训练中，标注者面对情感敏感话题时倾向于惩罚直言不讳的回应，即使这些回应在事实层面更准确72% similarUnverified在RLHF微调阶段惩罚情绪激烈的输出，会导致模型在转述用户评论时自动降低语言烈度71% similarUnverifiedThe pain direction is nearly orthogonal to the fear and negative valence directions, indicating that the model encodes pain as a relatively independent dimension.71% similarUnverifiedRLHF倾向于产生讨好型输出，即模型学会说用户想听的话，这在青少年情绪困扰场景中可能过度验证负面情绪而非引导积极应对70% similarUnverifiedThe 'pain' referred to in the paper is not pain in the biological neurological sense, but rather a negative incentive signal set up in the experimental design.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/934958API
curl https://kongchang.com/api/v1/knowledge/claims/934958MCP
get_claim(id=934958)