Unverified50% confidenceConceptExact time
ThinkingBox distinguishes three metrics: pass@1 (the proportion of successful attempts across all tries), pass@20 (the proportion of tasks that succeed at least once out of 20 attempts), and all-20 (the proportion of tasks that succeed in all 20 attempts).
1
Sources
50%
Confidence
Long-term
Relevance
10/9/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在歧义点的四类行为中，'先搜再问'成功率最高，平均通过率达93.4%，远高于'直接猜'的56.5%，'搜多次再猜'为51.5%65% similarUnverifiedAbout half of the tasks are easy for all models, with scores close to 100%; on the other half of difficult tasks, the three models scored 50%, 45%, and 23% respectively.65% similarUnverifiedARC-Challenge consists of four-choice multiple-choice questions, with a random baseline of 25%.64% similarUnverifiedWhen testing LLM citation behavior, the same prompt should be run at least 20–50 times, ideally expanded to more than 100 times.63% similarUnverifiedThe number of leads per run is configurable; the demo sets it to 5, but it can be increased to 25, 30, or even 100 per run in practice.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/991237API
curl https://kongchang.com/api/v1/knowledge/claims/991237MCP
get_claim(id=991237)