Unverified50% confidenceDecision RuleExact time
When using LLM-as-Judge, it is best for the judge and the generating Agent to use different models or different system prompts, in order to avoid shared biases causing them to endorse each other.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedLLM-as-a-judge模式即用一个能力更强的模型来评判另一个模型的输出质量，降低评测成本但引入评判模型自身的偏见问题76% similarUnverifiedLLM-as-a-judge evaluates Agent behavior by combining the Agent's execution trace and evaluation criteria into a prompt that is submitted to an independent judge model for scoring or pass/reject decisions.76% similarUnverifiedScoring functions often require careful design and may even rely on another LLM for scoring, known as the LLM-as-judge pattern.72% similarUnverifiedSome teams have introduced an LLM-as-Judge approach, using another model to evaluate agent output quality in order to form an automated quality feedback loop72% similarUnverifiedWhen applying LLMs to scenarios requiring rigorous logic, introducing human-machine collaboration and external verification tools is more pragmatic than blindly trusting model outputs.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1005374API
curl https://kongchang.com/api/v1/knowledge/claims/1005374MCP
get_claim(id=1005374)