Unverified50% confidenceFactExact time
The model receives active rules in two ways: symbolic descriptions or two solved examples; in addition, a model that receives no rule information is used as a baseline control.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFunction Calling本质上是模型决策、外部执行的两阶段流程，模型本身并不直接运行代码71% similarUnverified多模型路由存在两种主流策略：规则路由（基于关键词、token长度、任务类型标签）和模型路由（用轻量分类器预判任务难度）69% similarUnverifiedThe perception description + model decision paradigm transforms structured game state into natural language prompts, with the language model outputting action instructions under zero-shot conditions.69% similarUnverified应使用第二个理解语义的LLM而非正则表达式或规则过滤来清理冗余，因为模型的冗余表达是语义层面的而非固定模式68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1028694API
curl https://kongchang.com/api/v1/knowledge/claims/1028694MCP
get_claim(id=1028694)