Unverified80% confidenceFactTime unknown
Hermes模型从微调阶段就大量使用结构化输出训练数据，使模型天然倾向于生成格式正确的JSON
1
Sources
80%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
Unverified预训练阶段的下一词预测目标让模型倾向生成自然流畅序列，而JSON等结构化格式在通用网络语料中占比有限，导致模型偶发格式违规75% similarUnverifiedAnthropic 在训练阶段通过大量函数调用示例对模型进行微调，使其学会何时触发工具及如何构造合法的 JSON 参数71% similarUnverifiedFunction Calling requires native model support and relies on structured output capabilities built into the model through specialized fine-tuning during training.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/34962API
curl https://kongchang.com/api/v1/knowledge/claims/34962MCP
get_claim(id=34962)