Unverified75% confidenceFactTime unknown
在Spider基准上，早期单模型方案准确率约为70%，GPT-4等大模型将其提升到80%以上，多智能体协作策略的系统已能达到85%甚至更高的执行准确率
1
Sources
75%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Aix-DB开源框架：自然语言查询数据库的多智能体实战方案
githubapconw
Related Entities
Related Claims
UnverifiedGPT-4等大模型在Spider等基准测试上的零样本或少样本SQL生成准确率已超过80%79% similarUnverifiedOn the Spider benchmark, the accuracy of state-of-the-art models rose from under 20% in 2018 to over 80%74% similarUnverifiedNL2SQL 技术在 Spider 等基准测试上已达到80%以上的准确率，但在多表 JOIN 和子查询等复杂操作中准确率显著下降72% similarUnverified4-bit量化模型在大多数任务上的表现已能达到原始精度的95%以上72% similarUnverified当前顶级模型的单步准确率约在90%至95%之间71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/25724API
curl https://kongchang.com/api/v1/knowledge/claims/25724MCP
get_claim(id=25724)