Unverified50% confidenceEventExact time
In a math task setting where two rewards are nearly identical by construction, WAS-PO suppressed the reward with the lighter weight, as predicted by the sign condition.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/3/2026
First Seen
Valid until: 2/1/2027
Sources
Related Entities
Related Claims
UnverifiedThe paper attributes the inability of prior methods to distinguish different confidence targets to two problems: the reward signals that distinguish different p values are often lost during optimization, and the raw error penalty grows unboundedly as p approaches 1.67% similarUnverifiedPOMDP能通过奖励函数将漏报与误报的不对称代价显式量化编码，从而在数学上权衡获取更多信息与立即升级避免风险67% similarUnverifiedThe design difficulty of dense rewards lies in defining intermediate signals that are both accurate and automatically obtainable, avoiding the rewards being exploited by "score gaming."65% similarUnverifiedAblation experiments show that increasing example usage does not yield consistent gains, but instead presents a trade-off relationship between dynamics and reward.64% similarUnverified研究者提出长度归一化奖励方法，即在计算奖励分数时除以回答长度，使模型无法通过单纯增加长度获取更高奖励62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1130543API
curl https://kongchang.com/api/v1/knowledge/claims/1130543MCP
get_claim(id=1130543)