Unverified50% confidenceOpinionExact time
The quality of weakness reports depends on the granularity of the taxonomy: the finer the taxonomy, the more precisely the reports can pinpoint model deficiencies.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCoarse-grained error type classification relies more on semantic understanding, while fine-grained error span localization relies more on boundary judgment and position awareness, and the model's strengths and weaknesses on these two types of tasks are very likely not consistent.74% similarUnverifiedThe study categorized failures into two types: genuine capability deficits and issues that can be mitigated by adjusting the decoding threshold, further subdividing them into 4 failure mechanisms.73% similarUnverifiedIgnoring error correlation causes the combined variance to be underestimated, and the underestimation factor grows with group size; this factor accurately predicts the degree of coverage failure.72% similarUnverifiedGuided generative models face a data dependency challenge: extreme event data is inherently scarce, and data quality directly determines the reliability of estimates.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1134114API
curl https://kongchang.com/api/v1/knowledge/claims/1134114MCP
get_claim(id=1134114)