Unverified50% confidenceFactExact time
G-Eval originates from a paper's idea of having large models score step by step.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe design of GLM-Legal enables the model to proactively retrieve legal provisions, query case law, and call calculation tools to estimate compensation amounts during reasoning.73% similarUnverifiedThis integration solution adopts an architectural paradigm in which a general-purpose large model handles reasoning and orchestration while a vertically specialized model handles precise computation.70% similarUnverified作者认为决定最终优化效果的主要是提示词和使用的大模型型号，提示词越具体Agent执行越到位70% similarUnverifiedThe article divides large model deployment into four stages: native large model, prompt engineering, RAG, and Agent, corresponding respectively to generation, expression, knowledge, and action.68% similarUnverifiedEval is the process of systematically comparing model outputs against expected standards, and common forms include exact match, rule-based assertions, and LLM-as-judge.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1018302API
curl https://kongchang.com/api/v1/knowledge/claims/1018302MCP
get_claim(id=1018302)