Verified65% confidenceFactExact time
Artificial Analysis Intelligence Index整合了MMLU、HumanEval、MATH、GPQA等多个基准测试的加权结果
3
Sources
65%
Confidence
Long-term
Relevance
7/21/2026
First Seen
Sources
Related Claims
VerifiedAI领域模型能力的可信评估依赖标准化基准测试，常见评测集包括MMLU、HumanEval、MATH、GPQA80% similarUnverifiedHumanEval、MBPP、SWE-bench是AI编程评测领域的行业标准基准测试72% similarUnverifiedAI模型基准测试通常在MMLU、HumanEval、GSM8K等标准数据集上评估模型的推理、编码和数学能力72% similarUnverified业界常用的自动化评估工具包括OpenAI Evals、LangSmith、Braintrust，支持批量运行测试用例并生成对比报告72% similarUnverified业界正在探索的AI评估方法包括基于人工标注的评分体系、BLEU分数、ROUGE分数以及LLM-as-Judge方法70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/578052API
curl https://kongchang.com/api/v1/knowledge/claims/578052MCP
get_claim(id=578052)