Pulkit Gupta
Pulkit Gupta
Thanks @shubhamvishu for taking a look. > I went through the change but I didn't understand how are we not reusing the bitset in the current approach. We do wrap...
Thanks @mikemccand for the pointers. Will try to run benchmarks on this change.
@benwtrent @uschindler @ChrisHegarty Could you please take a look, if you get a chance?
Makes sense. Thanks @navneet1v for the suggestion.
> I have some comments, but this is not a final review. Just things that I stumbled upon on first walkthrough. > > I will have no time to do...
> I did kind of change before, and the added complexity and backwards compatibility concerns just didn't seem warranted. This is why the decision to do the scorer pluggability was...
Coming back to this issue: ### Summary We tried to implement this idea in our closed-source implementation and got very good results, reduction in vector index size to one-fifth of...
I think JDK now support [Float16 objects](https://docs.oracle.com/en/java/javase/25/docs/api/jdk.incubator.vector/jdk/incubator/vector/Float16.html). The HotSpot C2 compiler can auto-vectorize addition, subtraction, division, multiplication, square root, and fused multiply/add on supporting CPUs (From the [JEP-529](https://openjdk.org/jeps/529)). But VectorAPI...
> Can we call it something other than `quantizedFormat`? There are may formats that I would call quantized that would return `false` here (e.g. all HNSW ones right?). > >...