Support multiple indexes and binary embeddings #5
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Allow typehinting of numpy types via the new typehinting decorators, like np.ndarray[np.bool_]. This communicates the expected type as binary / regular embeddings. We sniff for this attribute at runtime to switch the column type that is injected into the database.
Milvus supports different levels of consistency depending on the demands of the business case. The default setting is Bounded staleness, which provides relatively good syncronization between replicas and much faster inference times. This is undesirable for unit tests, however, because it can result in stochastic behavior for some tests versus others. Add a new customization parameter to the model definitions that allow for specifying a different type of consistency for the schema.
|
Stochastic failing test for binary similar recall, tracking here: milvus-io/pymilvus#1379 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Introduce support for binary embeddings and alternative indexing strategies.
Binary Vectors
Specifically users can now use python typehints that specify more specific np.ndarray types. When a boolean is provided for the expected values, we'll assume the user wants to create a binary vector and use the
DataType.BINARY_VECTORaccordingly.Indexing Strategies
Mirror the index algorithms that are supported by Milvus. Keyword names for these functions differ from the kwargs that are used internally by Milvus, in an attempt to be more semantically interpretable about what the different arguments are doing and play better with additional schema backends.
Misc Changes