did you set QdrantVectorStore(...., enable_hybrid=True)
Hey @Logan M , oh yeah this was a fully working code and i pushed it to my cloud as well for business testing. i upgraded llama_index and i'm seeing two errors with the exact same code (upgrading the imports to the latest) and the config error at query time
to answer your question, i did set enable_hybrid=True as well
the assertion error is new
Ah, you can't mix legacy imports with new imports
need 100% legacy or 100% new
in your case, update that first import
from llama_index.legacy import StorageContext, VectorStoreIndex
This is very similar to how pydantic did their migration with v1 imports vs v2 (if you had to deal with that haha)
ouch, thx @Logan M , in the docs, i see the imports as "from llama_index.vector_stores.qdrant import QdrantVectorStore" but i dont see an import as llama_index.vector_store any reason why that would be ?
got it. now i have install all the integrations as pip
sorry abt that bud. whewww lots of re-learning
yea the main hurdle is every integration is a pip package. But that package uses namespacing, so the imports are similar (i.e. all under llama-index)
But namespacing means you cant have shortcut imports
A small price to pay for a more manageable repo.
Now we can have actual tests for integrations, as well as version them independently from the actual core package
right, makes total sense, it helps long term. 🫡
I know its a huge change, thanks for sticking by 🙏
no sir, thx for making this happen, its a small price we have to pay to get an awesome product.
@Logan M sorry to bother you again, i'm at the last step, the code works great as retriever but its failing for the query_engine, any help on this pls
ValueError: Cannot use llm_chat_callback on an instance without a callback_manager attribute.
How did you setup your LLM?
thank you very much for the tip, found out the issue, in my bedrock abstraction i'm using legacy imports again, have to upgrade it to the latest. I think that should fix it. Appreciate the help on this.
