Closed cutecutecat closed 1 month ago
[APPROVALNOTIFIER] This PR is NOT APPROVED
This pull-request has been approved by: alwayslove2013, cutecutecat
To complete the pull request process, please assign xuanyang-cn after the PR has been reviewed.
You can assign the PR to them by writing /assign @xuanyang-cn
in a comment when ready.
The full list of commands accepted by this bot can be found here.
Here we introduce a bug fix for
PGVecto.rs >= 0.4.0
If we don't need quantization, we pick a
trivial
option at WebUI, and this is the default behavior.In
pgvecto.rs sdk = 0.2.1
, it will be translated into[indexing.ivf.quantization.trivial]
or[indexing.hnsw.quantization.trivial]
as toml config.https://github.com/tensorchord/pgvecto.rs-py/blob/5e175a981b36fbb671033c2db45636a11d6dc04f/tests/__init__.py#L108
But we have a breaking change in
PGVecto.rs 0.4.0
, that it would not accept trivial quantization config. To express no quantization,quantization
should not be showed in config.So
pgvecto.rs sdk = 0.2.2
change the behavior that will translatetrivial
option into[indexing.ivf]
or[indexing.hnsw]
.https://github.com/tensorchord/pgvecto.rs-py/blob/5bc6fc34f2d0df05701000f298596587d58acc99/tests/__init__.py#L108
This is a transparent compile inside SDK, and forward compatibility, so there is nothing to do for VectorDBBench except a upgrade.