Unverified50% confidenceBenchmarkExact time
The author used five real financial/payment documents and ten manually verified test questions to conduct a side-by-side comparison of four approaches: vector-only, BM25-only, hybrid retrieval without reranking, and the full ReRankEval pipeline.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/5/2026
First Seen
Valid until: 1/3/2027
Sources
Related Entities
Related Claims
Unverified在Adult Income数据集上的实验中，作者将399个反事实解释通过聚类归纳为6个干预簇，进一步提炼出6条全局规则73% similarUnverified该测试样本规模相对有限（5个模型、13个代码库），结论有待更大规模验证68% similarUnverifiedThe author used Astra together with GPT to conduct comprehensive testing of a project, submitting over a hundred issues in just two days, with approximately 44 issues submitted in one round of testing.67% similarUnverifiedThe test suite consists of 10 benchmarks; most run 1,000 questions each, with HumanEval comprising 164 questions and TruthfulQA comprising 817 questions.65% similarUnverifiedThe test uses 10 base sampling steps plus 4 refinement steps, with the author deliberately lowering the step count to expose degradation issues under extreme conditions.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/976375API
curl https://kongchang.com/api/v1/knowledge/claims/976375MCP
get_claim(id=976375)