Find answers from the community

Updated 3 months ago

VectorIndexRetreiver

Hi, I have been using llamaindex for long time now. My code was working well and after updating llamaindex and openai, the same code have stopped working. I have fixed most of it by updating the code. But now when I am using vector store index as chat engine or query engine then I am getting unsupported protocol error : request url is missing. It works fine if I use vector store index as retriever. Can anyone help me fix this as I need it urgently. Thanks

6 comments

SShubham1696

Have resolved the error but I am getting issue while using sub questions query engine, now the error is VectorIndexRetreiver object has no attribute query

WWhiteFang_Jr

Are you retrieving like this?
data = retriever.query()

Correct way would be:
data = retriever.retrieve()

SShubham1696

No I am using sub questions query engine which is failing internally. I have used retriever.asQueryEngine() which is working when used directly but not through sub question query engine

WWhiteFang_Jr

Do you mind sharing the code?

SShubham1696

Query Engines are working directly but not working when I am using sub question query engine. I cannot share code as it's being used for project work

Attachment

rn_image_picker_lib_temp_f03de8c4-0017-4671-aca7-85a9745e18b0.jpg

WWhiteFang_Jr

For some reason, Your retriever object is being passed and being used to do .query() whereas based on the requirement it should be a QueryEngine object.

see this line: https://github.com/run-llama/llama_index/blob/74e8e6660a19738403cd3a180494226454e22947/llama_index/query_engine/sub_question_query_engine.py#L249

Add a reply