Unverified50% confidenceTradeoffExact time
The approach of combining reward and diversity into a single aggregate score for optimization has a flaw: high diversity can directly offset low reward, meaning the batch may still contain inferior images that clearly deviate from user preferences.
1
Sources
50%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe salience boost is blended into the embedding retrieval score multiplicatively rather than additively, to ensure relevance still dominates and to prevent irrelevant content from being forcibly surfaced.71% similarUnverifiedOverly pursuing diversity sacrifices consistency, while overly constraining consistency causes degradation into rigid templated responses.70% similarUnverifiedThe researchers traced the calibration difference between sparse and dense rewards to the 'partial credit' term in the dense reward.69% similarUnverifiedHuman annotators exhibit a cognitive bias toward more information being better when evaluating answers, causing reward models to implicitly link length with quality, which is one of the causes of model verbosity.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/975677API
curl https://kongchang.com/api/v1/knowledge/claims/975677MCP
get_claim(id=975677)