Unable to use CypherQueryCorrector() with GraphCypherQAChain()

[zoitsa93]

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

python

from langchain.chains import GraphCypherQAChain
from langchain.graphs import Neo4jGraph
from langchain_core.prompts.prompt import PromptTemplate

llm = AzureChatOpenAI(
    deployment_name="gpt-4",
    openai_api_version = "2024-02-15-preview",
    model_version="1106-Preview",
    max_tokens= 4096,
    temperature = 0,
    openai_api_key = openai_api_key,
    azure_endpoint =azure_endpoint
)

graph = Neo4jGraph(url=url, username=username, password=password, sanitize=False)


corrector_schema = [
    Schema(el["start"], el["type"], el["end"])
    for el in graph.structured_schema.get("relationships")
]

CYPHER_CORRECTOR_PROMPT = PromptTemplate(
    input_variables=[ "question"], template=CYPHER_GENERATION_PROMPT
)

cypher_query_corrector = CypherQueryCorrector(schemas=corrector_schema)

chain = GraphCypherQAChain.from_llm(
    llm=llm,
    graph=graph,
    verbose=True,
    cypher_prompt=CYPHER_CORRECTOR_PROMPT,
    validate_cypher=True,
   cypher_query_corrector=cypher_query_corrector
)

chain.invoke("Question")

### Description

I am using the Langchain library to generate Cypher queries based on a database. In the documentation for GraphCypherQAChain, I found the attribute cypher_query_corrector, which is designed to improve the generated queries.

Unfortunately, when I attempted to include cypher_query_corrector in GraphCypherQAChain.from_llm(), I encountered an error with CypherQueryCorrector(schemas=corrector_schema).
**The "corrector_schema"  is exactly as it is in the documentation (2).**

_The error message is as follows: TypeError: langchain_community.chains.graph_qa.cypher.GraphCypherQAChain() got multiple values for keyword argument 'cypher_query_corrector'_

Moreover, I have also run some tests with the method described in the documentation (2) but still can get any results.

**Is there a way to use CypherQueryCorrector(schemas=corrector_schema) in the GraphCypherQAChain.from_llm() chain?**


**documentations:**
(1) https://api.python.langchain.com/en/latest/chains/langchain.chains.graph_qa.cypher.GraphCypherQAChain.html
(2) https://python.langchain.com/v0.2/docs/how_to/graph_mapping/





### System Info

langchain==0.2.16
langchain-community==0.2.16
langchain-core==0.2.38
langchain-openai==0.1.16
langchain-text-splitters==0.2.2

[dosubot[bot]]

I found a similar unsolved discussion that might be relevant to your issue: Passing CypherQueryCorrector to GraphCypherQAChain ^[1].

Regarding your question, you can use CypherQueryCorrector(schemas=corrector_schema) in the GraphCypherQAChain.from_llm() chain. The from_llm method includes a parameter validate_cypher which, when set to True, initializes the cypher_query_corrector with the provided schema.

Here is the relevant part of the code:

@classmethod
def from_llm(
    cls,
    llm: Optional[BaseLanguageModel] = None,
    *,
    qa_prompt: Optional[BasePromptTemplate] = None,
    cypher_prompt: Optional[BasePromptTemplate] = None,
    cypher_llm: Optional[BaseLanguageModel] = None,
    qa_llm: Optional[Union[BaseLanguageModel, Any]] = None,
    exclude_types: List[str] = [],
    include_types: List[str] = [],
    validate_cypher: bool = False,
    qa_llm_kwargs: Optional[Dict[str, Any]] = None,
    cypher_llm_kwargs: Optional[Dict[str, Any]] = None,
    use_function_response: bool = False,
    function_response_system: str = FUNCTION_RESPONSE_SYSTEM,
    **kwargs: Any,
) -> GraphCypherQAChain:
    """Initialize from LLM."""
    # ... (other code)

    cypher_query_corrector = None
    if validate_cypher:
        corrector_schema = [
            Schema(el["start"], el["type"], el["end"])
            for el in kwargs["graph"].structured_schema.get("relationships")
        ]
        cypher_query_corrector = CypherQueryCorrector(corrector_schema)

    return cls(
        graph_schema=graph_schema,
        qa_chain=qa_chain,
        cypher_generation_chain=cypher_generation_chain,
        cypher_query_corrector=cypher_query_corrector,
        use_function_response=use_function_response,
        **kwargs,
    )

To use CypherQueryCorrector in the chain, you need to set validate_cypher=True when calling GraphCypherQAChain.from_llm().

The error you encountered, TypeError: langchain_community.chains.graph_qa.cypher.GraphCypherQAChain() got multiple values for keyword argument 'cypher_query_corrector', suggests that you should not pass cypher_query_corrector directly as a keyword argument. Instead, rely on the validate_cypher parameter to initialize it.

Here is how you can modify your code:

from langchain.chains import GraphCypherQAChain
from langchain.graphs import Neo4jGraph
from langchain_core.prompts.prompt import PromptTemplate

llm = AzureChatOpenAI(
    deployment_name="gpt-4",
    openai_api_version="2024-02-15-preview",
    model_version="1106-Preview",
    max_tokens=4096,
    temperature=0,
    openai_api_key=openai_api_key,
    azure_endpoint=azure_endpoint
)

graph = Neo4jGraph(url=url, username=username, password=password, sanitize=False)

CYPHER_CORRECTOR_PROMPT = PromptTemplate(
    input_variables=["question"], template=CYPHER_GENERATION_PROMPT
)

chain = GraphCypherQAChain.from_llm(
    llm=llm,
    graph=graph,
    verbose=True,
    cypher_prompt=CYPHER_CORRECTOR_PROMPT,
    validate_cypher=True
)

chain.invoke("Question")

This should resolve the issue and allow you to use CypherQueryCorrector with GraphCypherQAChain ^[2][3][4].

To continue talking to Dosu, mention @dosu.