Unverified50% confidenceConceptExact time
Facts in pretrained models can be divided into two types, active facts and inert facts; active facts can support generalization beyond the memorized range after fine-tuning, while inert facts can be recalled but cannot produce generalization on the same task.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
Unverified小模型通过知识蒸馏通常比直接在标注数据上训练精度高出2-5个百分点70% similarUnverifiedAdversarial multi-agent verification is good at detecting anchorable factual deviations (such as line numbers, timestamps, and hash values), but has limited effectiveness against systematic logical reasoning biases when sub-agents share training biases.70% similarVerifiedAlignment training optimizes the model's statistical behavior in known scenarios rather than establishing formally verifiable security boundaries, and may fail when faced with rare instruction combinations or role-playing frameworks outside the training distribution.70% similarUnverifiedContext, memory and training are three different concepts: context is the information available for this reply, memory is information that has been selected and saved, and training is the process by which the model learns underlying patterns.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1041962API
curl https://kongchang.com/api/v1/knowledge/claims/1041962MCP
get_claim(id=1041962)