Unverified50% confidenceTradeoffExact time
When a model can reproduce specific passages from its training data verbatim (the 'memorization' phenomenon), the 'transformative use' defense faces challenges.
1
Sources
50%
Confidence
Long-term
Relevance
9/17/2026
First Seen
Sources
Related Entities
Related Claims
Verified当模型训练数据包含评测集题目或高度相似内容时，模型可能通过记忆而非真正理解获得高分80% similarUnverifiedMachine learning models may overfit specific training samples and inadvertently reproduce details of the original data when generating content, a phenomenon known as training data memorization.79% similarUnverifiedDuring LLM training, memorization occurs, where models may verbatim store fragments of training data and reproduce them during inference.79% similarUnverified模型对 schema 的遵循能力受训练数据分布影响，训练语料中高频出现的字段组合会导致模型产生幻觉字段77% similarUnverifiedExpressions like "tired" in model output occur because training data is filled with human modes of expression, representing pattern matching at the linguistic level rather than genuine state feedback.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/930230API
curl https://kongchang.com/api/v1/knowledge/claims/930230MCP
get_claim(id=930230)