whyhow-ai / knowledge-table

Knowledge Table is an open-source package designed to simplify extracting and exploring structured data from unstructured documents.
MIT License
254 stars 32 forks source link

OpenAI client is not initialized. Skipping embeddings. #41

Open corwin-eversley-moodys opened 2 hours ago

corwin-eversley-moodys commented 2 hours ago

Describe the bug

Describe the bug OpenAI "get_embeddings" function can't be accessed at query time. The OpenAI client is initialised, but the check is returning before the client can be accessed.

To Reproduce Run a query to generate a column once documents have been added.

Expected behavior Should embed the query and then retrieve from the vector store.

Logs and Error Reports 2024-10-24 17:40:16 INFO:app.services.vector_db.milvus_service:Generating embedding. 2024-10-24 17:40:16 WARNING:app.services.llm.openai_service:OpenAI client is not initialized. Skipping embeddings. 2024-10-24 17:40:16 INFO:app.services.vector_db.milvus_service:Searching... 2024-10-24 17:40:16 ERROR:pymilvus.decorators:Unexpected error: [search], list index out of range, <Time: {'RPC start': '2024-10-24 21:40:16.194573', 'Exception': '2024-10-24 21:40:16.344262'}> 2024-10-24 17:40:16 ERROR:pymilvus.milvus_client.milvus_client:Failed to search collection: milvus 2024-10-24 17:40:16 ERROR:app.api.v1.endpoints.query:Error processing query: <MilvusException: (code=1, message=Unexpected error, message=)> 2024-10-24 17:40:16 INFO: 192.168.65.1:61110 - "POST /api/v1/query HTTP/1.1" 500 Internal Server Error

tomsmoker commented 2 hours ago

Thanks for bringing to our attention @corwin-eversley-moodys. I'll resolve. Was this using pypdf or unstructured? And was this a docker build or Python local?

corwin-eversley-moodys commented 2 hours ago

Using pypdf and using the docker build.