Unverified50% confidenceBenchmarkExact time
On the LaMP-2 dataset, the BM25-based retrieval method achieves an accuracy of 0.760–0.765, which is statistically indistinguishable from PersonaLink.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMultiple benchmark tests in 2023 (such as the HaHackathon dataset) showed that top LLMs still have significantly lower accuracy than ordinary human annotators on sarcasm recognition tasks61% similarUnverified模板匹配的置信度阈值通常取0.8～0.95来判断是否找到目标61% similarUnverified移除歧义后所有模型准确率提升26.8至40.2个百分点，Gemini甚至达到81%，表明歧义是深度搜索代理最大性能瓶颈之一59% similarUnverifiedThe Gorilla benchmark released by the University of California, Berkeley shows that when the number of candidate tools exceeds 20, even models at the GPT-4 level experience a 15%-30% drop in tool selection accuracy.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/897558API
curl https://kongchang.com/api/v1/knowledge/claims/897558MCP
get_claim(id=897558)