Run embedding (Insert) SQL stuck forever when using pgrag
I am testing the pgrag capability on neon following this https://neon.tech/docs/extensions/pgrag#end-to-end-rag-example official guide. I give the db 2 CU (8G RAM) that should be surfficent enough for RAG with only a few documents.
The image is the table schema using vector index. If I directly insert array of number to this table, it works without any issue.
But if I use
rag_bge_small_en_v15.embedding_for_passage function, then the db connection stuck forever (in pycharm). What could went wrong? I already tried asking NEON AI.
Neon
The pgrag extension - Neon Docs
What is RAG? What's included in a RAG pipeline? pgrag functions How to use pgrag The pgvector extension YouTube pgrag video demonstration pgrag GitHub repository The pgrag extension and its accompanyi...
5 Replies
flat-fuchsia•4mo ago
Thank you for reporting. I'll pass this on to the developer of pgrag.
harsh-harlequin•4mo ago
Thanks for this report. Seems I am seeing the same behaviour, which is certainly not expected. I'll revert when I know more ...
quickest-silverOP•4mo ago
Thank you , I understand prag is an experiment feature, however, I feel like this should be an undeniable reason to choose neon. Currently my company doing chunking, embedding , reranking on our own AWS lambda micro service
harsh-harlequin•4mo ago
Good to hear you're keen on pgrag, and sorry for the issue here. My understanding is that the issue is fixed and is waiting on a release, which should happen before the end of the week, but I can't be more specific than that right now.
flat-fuchsia•4mo ago
@birdnest The team has reported that this issue should now be fixed. Please let us know if it works for you. Thank you for reporting.