Unverified50% confidenceFactExact time
A new model's text response may appear normal, but underneath it may have missed a tool call, passed in different parameters, or selected a different tool to perform the task.
1
Sources
50%
Confidence
Long-term
Relevance
10/1/2026
First Seen
Sources
Related Entities
Related Claims
Unverified当模型输出工具调用语法后会停止响应，Harness执行工具并将输出追加到对话历史末尾，然后向同一模型重新发起请求77% similarUnverified设置tool_choice为auto时，LLM有时会在应该使用工具时选择直接回答，尤其在使用较小模型时这种行为更加频繁71% similarUnverified工具调用早期实现依赖ReAct等提示词技巧，模型通过文本格式输出行动指令，再由外部代码解析执行67% similarUnverifiedInstruction fine-tuning trains a model using carefully designed instruction-response pairs, transforming it from a text-completion model into one that follows instructions.67% similarUnverifiedDuring multi-turn tool calls, each tool execution result is appended to the conversation history as a new tool_result message, and the model reads the outputs of previous steps from the conversation context rather than retaining intermediate values in internal memory.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/966203API
curl https://kongchang.com/api/v1/knowledge/claims/966203MCP
get_claim(id=966203)