Unverified50% confidenceOpinionExact time
对于闭源模型，外部研究者几乎无法有效检测数据污染
1
Sources
50%
Confidence
Long-term
Relevance
8/8/2026
First Seen
Sources
Related Claims
UnverifiedCurrent LLM models cannot provide accurate data provenance, and technologies such as data provenance tracking, watermarking, federated learning, and differential privacy are still at the research stage.67% similarUnverifiedBenchmarks for evaluating large models such as MMLU, HumanEval, and GSM8K have limitations including data contamination, overfitting, and insufficient ecological validity.64% similarUnverified错误的归纳偏置会让模型花费容量拟合数据中不存在的虚假模式，导致分布外数据上严重失效63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/705798API
curl https://kongchang.com/api/v1/knowledge/claims/705798MCP
get_claim(id=705798)