Unverified50% confidenceReleaseExact time
The paper proposes Authored-Sentence Credit (ASC), which moves GRPO's reward and advantage value from the answer granularity down to the sentence granularity, and independently judges whether each sentence deserves reward.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe core metric of mainstream faithfulness rewards is the 'proportion of supported sentences', that is, if each sentence in the answer can find support in the retrieved passages it counts as supported, and the higher the proportion the higher the reward.62% similarUnverifiedClear question-and-answer style expression, explicit entity information, structured data markup (Schema.org), and high-authority external citations are key factors for improving the probability of AI citation58% similarUnverifiedFor weaker prompted readers, both normalization and evidence selection yield gains.56% similarUnverifiedThe author argues that when the source of text becomes difficult to distinguish, the importance of attribution, credibility, and fact-checking increases.56% similarUnverified研究表明在提示词中加入反例说明（即不要做什么）能显著提升生成质量55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023694API
curl https://kongchang.com/api/v1/knowledge/claims/1023694MCP
get_claim(id=1023694)