How to log out the prompts generate by each component of agent?

[jamerainbow]

I want debug each component of agent generate what prompt. It can help us adjust the prompt templetes.

[dosubot[bot]]

Hello @jamerainbow! I'm Dosu, a friendly bot here to help you solve bugs, answer questions, and assist you in becoming a contributor. I'll be here while you wait for a human maintainer. How can I assist you today?

To log the prompts generated by each component of an agent for debugging purposes, you can modify the debug method in the Generate class. This method constructs the prompt using the provided parameters and then uses the chat method of the LLMBundle to generate the response. You can add logging functionality to capture the prompts.

Here's an example of how you might modify the debug method to log the prompts:

def debug(self, **kwargs):
    chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
    prompt = self._param.prompt

    for para in self._param.debug_inputs:
        kwargs[para["key"]] = para.get("value", "")

    for n, v in kwargs.items():
        prompt = re.sub(r"\{%s\}" % re.escape(n), str(v).replace("\\", " "), prompt)

    # Log the generated prompt for debugging
    print(f"Generated Prompt: {prompt}")

    u = kwargs.get("user")
    ans = chat_mdl.chat(prompt, [{"role": "user", "content": u if u else "Output: "}], self._param.gen_conf())
    return pd.DataFrame([ans])

By adding print(f"Generated Prompt: {prompt}"), you can log the generated prompt to the console. You can replace print with any logging mechanism you prefer ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}