I added an arg so that you can specify the number of segments the index vectors will be quantized to #57
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Outdated suggestions cannot be applied.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.
The paper discusses using a vector dimension of 24, which is not divisible by 16 which is hard-coded as the value of
m
, the number of segments the vector is divided into. Since I want to experiment with this dimension size, I have no choice but to surgically change the code so that I can use a value that is indeed divisible by 24. To add this flexibility, I added optional handling for providing the value form
.