Unverified50% confidenceReleaseExact time
The arXiv paper "Learning Style, Forgetting Semantics" explains the phenomenon of SFT being more prone to forgetting than RFT through theoretical modeling.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/5/2026
First Seen
Valid until: 1/3/2027
Sources
Related Entities
Related Claims
UnverifiedThis arXiv paper points out that if large language models are trained on isolated triples, they tend not to learn the surrounding context of those triples, which is necessary for multi-hop reasoning.69% similarUnverifiedWhen constraining output format through prompts, the model may forget these constraints in long conversations or complex reasoning scenarios.68% similarUnverifiedProviding in-context learning examples at inference time can substantially mitigate context confusion.67% similarUnverified大语言模型的知识是通用性的，不了解具体项目上下文，面对已有系统时缺乏项目特定上下文会导致输出退化为有根据的猜测66% similarUnverified知识蒸馏的中间层对齐策略（如PKD、TinyBERT）通过强制学生模型的隐藏表示模仿教师模型的分布间接约束嵌入几何结构66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/975963API
curl https://kongchang.com/api/v1/knowledge/claims/975963MCP
get_claim(id=975963)