index.as_query_engine does not seem to be working with local LLM

[candcconsulting]

Name: llama-index
Version: 0.9.23
I am using a local llama2
llama2-model-gguf-AGL-Q4.gguf
I am building a local index which works OK ...
However, the issue comes when I try to work with the index

    storage_context=StorageContext.from_defaults(persist_dir=storage_path)
    print("loading index")    
    # index = VectorStoreIndex.from_vector_store(storage_path + "index_store.json")
    index = load_index_from_storage(storage_context, service_context=service_context)
    print("loading query")
    query_engine = index.as_query_engine(service_context=service_context)
    print ("loading response")
    print("loading response : Query Engine is " + str(type(query_engine)) + " and has a query " + str(type(query_engine.query)) + " handling request " + queryRequest )
    response = query_engine.query(queryRequest)

The output from the type statement is
loading response : Query Engine is <class 'llama_index.query_engine.retriever_query_engine.RetrieverQueryEngine'> and has a query <class 'method'> handling request What is a question
However, I always get an error
error occurred: 'int' object is not callable
Everything looks OK
I do have this working outside of llamaindex using langchain, but the indexing "seems" to be much quicker with llamindex, but if NOT using OpenAI causes too much pain ... I will stop further investigation.
Is this a simple fix or is the pain of using local LLM too much effort ?
Appreciate any input

[dosubot[bot]]

🤖

Hello @candcconsulting! 👋

I'm Dosu, a friendly bot here to lend a hand with your LlamaIndex queries, debug your issues, and help you become a valuable contributor. While we're waiting for a human maintainer, I'm here to assist you in the meantime. I'm ready to roll up my digital sleeves and get to work!

Just remember, there's no problem too hairy for us to tackle... or should I say, too woolly? 🦙

Based on the information you've provided, it seems like the issue might be related to the queryRequest parameter you're passing to the query_engine.query(queryRequest) method. In the LlamaIndex codebase, the queryRequest parameter is likely to be an instance of the QueryBundle class. This is inferred from the usage of the queryRequest parameter in the retrieve, aretrieve, _query, and _aquery methods of the RetrieverQueryEngine class.

Here is a relevant code snippet from the RetrieverQueryEngine class:

def retrieve(self, query_bundle: QueryBundle) -> List[Node]:
    ...
def aretrieve(self, query_bundle: QueryBundle) -> List[Node]:
    ...
def _query(self, query_bundle: QueryBundle) -> List[Node]:
    ...
def _aquery(self, query_bundle: QueryBundle) -> List[Node]:
    ...

Source: llama_index/query_engine/retriever_query_engine.py

If you're passing an integer to the query_engine.query(queryRequest) method, this could be the cause of the 'int' object is not callable error you're encountering. I would recommend checking the type and value of the queryRequest parameter you're passing to the query_engine.query(queryRequest) method.

If you're already passing an instance of the QueryBundle class to the query_engine.query(queryRequest) method, then the issue might be elsewhere. In that case, could you please provide more information about the queryRequest parameter and the exact line of code where the error occurs? This would help in diagnosing the issue further.

Sources

• llama_index/query_engine/graph_query_engine.py
• llama_index/query_engine/retriever_query_engine.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.

[candcconsulting]

OK will take a look at the QueryBundle ... but currently I am passing in a simple string

[almitte]

Have the same problem. 'RetrieverQueryEngine' object is not callable, when following the advanced LlamaParse RAG Worklow.