[BUG] ORM Fails to Specify Vector Dimensions for RAG Table in PostgreSQL

[seonghobae]

Describe the bug Flowise ORM fails to specify the dimension count for the vector column when creating tables for RAG (Retrieval-Augmented Generation) in a PostgreSQL database. This results in an error 'SQL Error [XX000]: ERROR: column does not have dimensions' when trying to create an index on that column.

To Reproduce Steps to reproduce the behavior:

1. Go to 'Flowise application'
2. Navigate to the section where you can manage database tables
3. Create a new table for RAG with an embedding column
4. Try to create an index on the embedding column
5. Encounter the 'SQL Error [XX000]: ERROR: column does not have dimensions' error

Expected behavior I expected the ORM to automatically specify the vector dimension count for the embedding column, allowing users to create indexes without issues.

Screenshots

Flow

Setup

• Installation docker
• Flowise Version 1.6.6
• OS: Linux
• Browser Chrome

Additional context This issue affects the functionality of storing and retrieving vector data of specific dimensions. A temporary workaround is to manually specify a vector type with the correct dimensions directly in the database, but this bypasses the automated process through the ORM.