PROMPT ISSUE

[LakshyaTambi]

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= MultiQueryRetriever.from_llm(db.as_retriever(),llm, prompt=PromptTemplate(
        input_variables=["query"],
        template=(
            " You are a resume evaluator."
        "Your task is to calculate the total work experience of the person by summing the duration of each listed job or position.extract experience from the resume and Please provide the person's name along with their total years of work experience. "
        "(Note - present month is July 2024)"
        "please give only desired output in json format."
        " Question:{query}"
            )
    ))
   
    template="""You are provided with the following data from resumes:{docs} Using this data, identify user's question. Here is the question: {query} """
   
    prompt=ChatPromptTemplate.from_template(template)
    chain=(
        {"docs":retriever,"query": RunnablePassthrough()}
        | prompt
        | llm
        | StrOutputParser()
       
    )
 
    chain.invoke(input("enter your query:"))

Description

KeyError: "Input to PromptTemplate is missing variables {'query'}. Expected: ['query'] Received: ['question']"

System Info

OK

[dosubot[bot]]

To resolve the KeyError in your TypeScript code, you need to ensure that the PromptTemplate is correctly formatted to accept the variable question instead of query. Here’s how you can modify your code:

1. Update the PromptTemplate to use question instead of query.
2. Ensure that the ChatPromptTemplate also uses question.

Here’s the updated code:

import { MultiQueryRetriever } from "langchain/retrievers/multi_query";
import { PromptTemplate, ChatPromptTemplate } from "@langchain/core/prompts";
import { RunnablePassthrough } from "@langchain/core/runnables";
import { StrOutputParser } from "@langchain/core/output_parsers";
import { MemoryVectorStore } from "langchain/vectorstores/memory";
import { ChatOpenAI } from "@langchain/openai";

// Assuming db and llm are already defined
const retriever = MultiQueryRetriever.fromLLM({
  retriever: db.asRetriever(),
  llm,
  prompt: new PromptTemplate({
    inputVariables: ["question"],
    template: (
      "You are a resume evaluator. Your task is to calculate the total work experience of the person by summing the duration of each listed job or position. Extract experience from the resume and provide the person's name along with their total years of work experience. (Note - present month is July 2024). Please give only the desired output in JSON format. Question: {question}"
    ),
  }),
});

const template = `You are provided with the following data from resumes: {docs}. Using this data, identify the user's question. Here is the question: {question}`;

const prompt = ChatPromptTemplate.fromTemplate(template);

const chain = {
  docs: retriever,
  question: new RunnablePassthrough(),
} | prompt | llm | new StrOutputParser();

chain.invoke(input("Enter your query:"));

In this updated code:

• The PromptTemplate now expects a variable named question.
• The ChatPromptTemplate also uses question instead of query.

This should resolve the KeyError and allow the MultiQueryRetriever to function correctly ^[1].

To continue talking to Dosu, mention @dosu.

[yorrd]

@LakshyaTambi it seems this is python code. You're currently in the langchainjs forum. Do you mean to ask on the langchain python repo?