Unverified50% confidenceFactExact time
Function Calling能力需要在监督微调（SFT）阶段喂入高质量的工具调用对话数据才能获得
1
Sources
50%
Confidence
Long-term
Relevance
7/7/2026
First Seen
Sources
LangChain+MCP实战：大模型选型与本地部署避坑指南
bilibili码士集团-马小琪7/5/2026
Related Claims
UnverifiedThe Instruct Models open-sourced by major companies are versions that have acquired conversational capabilities after the SFT stage.72% similarUnverifiedThe ChatGPT mobile client supports creating and managing tasks within conversations, stronger change review capabilities (staged/unstaged/branch filtering), and directly connecting to SSH hosts to initiate new tasks.68% similarUnverifiedOpenAI自GPT-4起引入了结构化的Function Calling机制，允许模型在对话中动态决定何时调用哪个工具、传入什么参数，并将结果整合回答复中67% similarVerifiedFew-shot implementation for chat models requires FewShotChatMessagePromptTemplate, while text completion models use FewShotPromptTemplate.66% similarUnverified工具调用的本质是模型通过SFT和RLHF学会按特定格式输出函数调用请求，真正的执行由外部运行时完成66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/191713API
curl https://kongchang.com/api/v1/knowledge/claims/191713MCP
get_claim(id=191713)