Unverified50% confidenceDecision RuleExact time

If a human can answer correctly given the same retrieved chunks while the model cannot, the problem lies on the generation side (model capability, prompt engineering, or context organization).

1

Sources

50%

Confidence

Long-term

Relevance

9/29/2026

First Seen

Sources

Where Does Your RAG Pipeline Actually Break: Retrieval or Generation?

Related Entities

Related Claims

Unverified用一个可能出错的模型去校验另一个可能出错的模型会叠加问题而非解决问题75% similar VerifiedExplicitly instructing the model via prompt to answer only based on the provided context and to respond with "I don't know" if the answer is not present can mitigate hallucination (grounding).71% similar Unverified自回归模型逐Token从左到右生成，本质是马尔可夫过程，在逻辑推理中存在错误积累效应71% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/959355

API

curl https://kongchang.com/api/v1/knowledge/claims/959355

MCP

get_claim(id=959355)