Unverified50% confidenceDecision RuleExact time
Different MTEB tracks (English v1, English v2, multilingual, and the BEIR retrieval-specific data) use different test sets, and scores cannot be compared directly across tracks; it is recommended to rank by retrieval score rather than overall average and to compare models within the same track.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe accuracy of retrieval models is typically evaluated on standard benchmarks such as BEIR and MTEB.64% similarUnverifiedDifferences in test granularity, validation methods, and task sources across benchmarks are significant, and one must be especially cautious when comparing numbers across benchmarks.62% similarUnverifiedDifferential Testing detects errors by feeding identical test cases to two implementations and comparing their output differences, and has been widely used by companies such as Google and Apple for compiler and interpreter testing.61% similarUnverifiedThe paired statistical test uses clustered confidence intervals to correct for within-calculator sample correlations, making it more rigorous than directly comparing overall accuracy rates.61% similarUnverified基准对比中竞品使用的是自适应推理（Adaptive Reasoning）模式，而非强制开启最高推理档位61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1018775API
curl https://kongchang.com/api/v1/knowledge/claims/1018775MCP
get_claim(id=1018775)