Unverified50% confidenceConceptExact time
Membership inference attacks can detect data contamination via perplexity: if a model's perplexity for a given code segment is significantly lower than for code of equivalent complexity, this often indicates the segment appeared in the training set.
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
Grok 4.5 vs Composer 2.5实测：真实代码库谁更强？
Related Claims
Unverified大模型对代码的理解是基于训练数据分布的模式匹配而非真正的语义推理，对训练集中出现频率极低的自定义加密算法还原能力会显著下降甚至产生幻觉78% similarUnverifiedResearchers can use n-gram overlap analysis and membership inference attacks to determine whether specific data appeared in a model's training set, but with the limitation of a high false positive rate.76% similarVerifiedBenchmark data contamination refers to the phenomenon where, after a test set is made public, its questions may appear in the pre-training corpora of subsequent models, causing scores to reflect memorization rather than genuine reasoning ability.73% similarUnverifiedThe core principle of adversarial example attacks is to apply imperceptibly small perturbations to input data that are nearly undetectable to the human eye, causing a trained deep learning model to produce completely incorrect outputs.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/561538API
curl https://kongchang.com/api/v1/knowledge/claims/561538MCP
get_claim(id=561538)