9.4 Analysis of source code of Evaluation for Agents
As shown in Figure 9-4, turn around and look at the source code of LangChain framework agent evaluation (Evaluation for Agents).
Figure 9- 4 LangChain's evaluation agents directory
In the trajectory_eval_prompt.py file, a very classic prompt word is written.
1. """提示轨迹评估链."""
2. # flake8: noqa
3. from langchain.schema import AIMessage
4. from langchain.schema import HumanMessage
5.