Unverified50% confidenceOpinionExact time
The denoising effect of the reward model propagates to the behavior of the final policy model, and a less noisy reward signal makes policy optimization more consistent with true human preferences.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDRaFT利用扩散模型去噪过程的可微性，将奖励函数梯度直接回传至模型参数，无需强化学习的策略梯度估计器，规避高方差问题73% similarUnverifiedIf rewards target only final outcomes, the model tends to take shortcuts and memorize solutions for specific scenarios.71% similarUnverifiedMetrics define behavior—whatever you reward, the model learns to do, even if it deviates from the truly desired outcome.70% similarUnverifiedAfter selectively ablating the competitor's cached goal representation, the model clearly shifted toward the current customer, proving that the goal representation has a causal effect on decisions.69% similarUnverifiedModels can be trained with a behavioral tendency of 'when uncertain, better not to act' as a risk-avoidance mechanism encoded into weights through reward signal design68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1046111API
curl https://kongchang.com/api/v1/knowledge/claims/1046111MCP
get_claim(id=1046111)