AIDB 2.1.1 release notes v7
Released: 03 Feb 2025
AIDB 2.1.1 adds support for NVIDIA NIM, introduces new model names, and improves the retriever pipeline.
| Type | Description |
|---|---|
| Enhancement | Renamed openai_embedding and openai_completion: The model names openai_embedding and openai_completion have been renamed to embeddings and completions respectively. The old names are still supported for backward compatibility. |
| Enhancement | Improved model provider credential management: Model provider credential management in the create_model() call is improved. New credentials no longer get silently ignored if credentials already exist. |
| Enhancement | Added FDW handler for the model registry: A handler has been added for the foreign data wrapper (FDW) used for the model registry. The handler has no functionality but satisfies the FDW handler interface, avoiding certain issues with Postgres system calls that occur when FDWs do not have any handlers. |
| Enhancement | Added schema support to the retriever pipeline: For retrievers with table sources, the source table and vector table can now be in any schema in the Postgres database. The schema can be passed during retriever creation as a fully qualified identifier, or omitted — in which case AIDB resolves and stores the applicable schema. |