Unverified50% confidenceFactExact time
数据污染问题源于LLM预训练语料来自Common Crawl等大规模网络爬取，可能包含公开的基准测试题目及答案
1
Sources
50%
Confidence
Long-term
Relevance
7/5/2026
First Seen
Sources
Related Claims
VerifiedTraining data contamination is the primary challenge faced by AI programming benchmarks, as test cases come from public GitHub repositories, which are the same sources used as pretraining data for large models.75% similarUnverified主流基准测试如MMLU、HumanEval、SWE-bench通常仅覆盖数千至数万个标准化问题，存在场景覆盖有限和训练数据污染的潜在风险71% similarUnverifiedLLM在预训练阶段见过海量网络诈骗话术和营销文案，具备对高风险语义模式的识别能力，可与进程行为特征联合判断71% similarUnverifiedCommon methods for detecting data contamination include verbatim completion tests and membership inference attacks.67% similarUnverifiedSpecialized models trained specifically on cybersecurity data have the potential to surpass general-purpose models in vulnerability identification accuracy and false positive rates.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/112396API
curl https://kongchang.com/api/v1/knowledge/claims/112396MCP
get_claim(id=112396)