How the agent controls input parameters

[Huyueeer]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

retriever = elastic_vector_search.as_retriever(search_type="similarity_score_threshold",
                                               search_kwargs={'k': 1, 'score_threshold': 0.5})
retriever_tool = create_retriever_tool(
    retriever,
    "dataapp_search",
    "If you use search, use this tool",
)

tools = [retriever_tool]
llm_with_tools = llm.bind_tools(tools)

agent = (
    {
        "input": lambda x: x["input"],
        "agent_scratchpad": lambda x: format_to_openai_tool_messages(
            x["intermediate_steps"]
        ),
    }
    | prompt
    | llm_with_tools
    | OpenAIToolsAgentOutputParser()
)

agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True, return_intermediate_steps=True)

Description

I found that every time the agent submitted input to retriever, it would be a cut. For example, my input would be“Help me find 99986 related material information.” The agent would submit 99986 to Retriever, cause I can not take from the vector library to the beneficial information back to the subsequent workflow, what method can control agent to retriever input?

System Info

no

[dosubot[bot]]

🤖

Hey there, @Huyueeer! Great to see you back with another intriguing puzzle for us to solve together. 😊 How's everything going on your end?

The input that the agent submits to the retriever in the LangChain framework is controlled by the input lambda function in the agent pipeline. Currently, it's set to return the input as it is (lambda x: x["input"]). If you want to prevent it from truncating your input, you can modify this function to apply any necessary transformations or checks.

For example, if you want to ensure that the input is always a certain length, you could modify the function like this:

{
    "input": lambda x: x["input"] if len(x["input"]) <= MAX_LENGTH else x["input"][:MAX_LENGTH],
    ...
}

In this example, MAX_LENGTH is the maximum length of the input that you want to allow. If the input is longer than this, it will be truncated to this length. If it's shorter, it will be returned as it is.

Please replace MAX_LENGTH with the actual maximum length you want to set.

This is based on the code in the templates/neo4j-semantic-layer/neo4j_semantic_layer/agent.py file in the LangChain repository.

If you're still having trouble with this, could you please provide more details about the issue? Specifically, it would be helpful to know what you mean by "the agent would submit 99986 to Retriever", and how this is causing problems for you.

Sources

• templates/neo4j-semantic-layer/neo4j_semantic_layer/agent.py
• libs/langchain/langchain/agents/structured_chat/base.py

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.

[varun508]

is there any solution to this?