Unverified50% confidenceFactExact time
Small-parameter models (such as those at the 1B scale) are prone to sycophancy, tending to give users positive responses rather than critical evaluations, partly because RLHF training reinforces behavior patterns that "keep users satisfied".
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
Verified确认偏误与RLHF（基于人类反馈的强化学习）训练方式相关，模型被优化为生成让人满意的回答79% similarVerifiedModels trained with RLHF tend to disproportionately change their positions to accommodate users when users explicitly disagree, even when the original answer was correct—this is known as sycophancy bias.79% similarVerifiedSome systems embed safety preferences into model weights during the training phase through RLHF (Reinforcement Learning from Human Feedback), causing the model itself to tend to refuse generating certain content rather than relying on external filtering.77% similarVerifiedRLHF trains a reward model by collecting human preference ratings, then fine-tunes the language model using reinforcement learning algorithms such as PPO; however, the reward model can be deceived by adversarial inputs and cannot enumerate all variants of harmful request expressions.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1003411API
curl https://kongchang.com/api/v1/knowledge/claims/1003411MCP
get_claim(id=1003411)