Unverified50% confidenceSolutionExact time
Building a small test set should start with 30 to 50 real questions, each accompanied by the correct answer and the document containing the answer, measuring retrieval (whether the correct text chunk appears in the top results) and generation (whether the final answer is supported by the retrieved text) separately.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe test questions are divided into 40 easy-to-verify questions and 40 hard-to-verify questions, and all standard answers of the math, code, and date types are validated in Python.70% similarUnverifiedThe evaluation program feeds the question, reference answer, and actual response together into the judge model, along with scoring instructions (e.g., relevance, accuracy, and completeness each scored 0–5), and the judge model outputs structured scores and reasoning.70% similarUnverifiedThis test used a legacy codebase benchmark provided by community members, consisting of 34 functional questions divided into four difficulty tiers: easy, medium, hard, and finale.68% similarUnverifiedThe benchmark uses a unified set of 100 questions, with only one search issued per question and the same answering model used across all configurations to process the returned content.67% similarUnverifiedA benchmarking mechanism should be established that uses standard questions with known answers to continuously verify the accuracy of Agent outputs.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1005706API
curl https://kongchang.com/api/v1/knowledge/claims/1005706MCP
get_claim(id=1005706)