Unverified50% confidenceFactExact time
The TREC series of evaluations focuses primarily on relevance ranking in document retrieval, while BEIR focuses on evaluating zero-shot cross-domain retrieval capability.
1
Sources
50%
Confidence
Long-term
Relevance
9/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe reranking module typically uses a cross-encoder model to compute relevance scores between queries and documents, performing a secondary ranking of the candidate documents from the initial retrieval.61% similarUnverifiedNIST is conducting a fourth-round evaluation, considering code-based schemes such as HQC as candidate standards.58% similarUnverified等价类划分将输入数据按有效/无效归类，从每个等价类中各取一个代表性数据进行测试，以在有限用例数量下最大化覆盖率58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/888582API
curl https://kongchang.com/api/v1/knowledge/claims/888582MCP
get_claim(id=888582)