feat(index): add IVF_RQ index type (#2687)

this expose IVF_RQ (RabitQ quantization) index type to lancedb

---------

Signed-off-by: BubbleCal <bubble-cal@outlook.com>
This commit is contained in:
BubbleCal
2025-10-09 15:46:18 +08:00
committed by GitHub
parent 56a16b1728
commit b59d1007d3
14 changed files with 361 additions and 13 deletions

View File

@@ -1,7 +1,7 @@
// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright The LanceDB Authors
use lancedb::index::vector::IvfFlatIndexBuilder;
use lancedb::index::vector::{IvfFlatIndexBuilder, IvfRqIndexBuilder};
use lancedb::index::{
scalar::{BTreeIndexBuilder, FtsIndexBuilder},
vector::{IvfHnswPqIndexBuilder, IvfHnswSqIndexBuilder, IvfPqIndexBuilder},
@@ -87,6 +87,22 @@ pub fn extract_index_params(source: &Option<Bound<'_, PyAny>>) -> PyResult<Lance
}
Ok(LanceDbIndex::IvfPq(ivf_pq_builder))
},
"IvfRq" => {
let params = source.extract::<IvfRqParams>()?;
let distance_type = parse_distance_type(params.distance_type)?;
let mut ivf_rq_builder = IvfRqIndexBuilder::default()
.distance_type(distance_type)
.max_iterations(params.max_iterations)
.sample_rate(params.sample_rate)
.num_bits(params.num_bits);
if let Some(num_partitions) = params.num_partitions {
ivf_rq_builder = ivf_rq_builder.num_partitions(num_partitions);
}
if let Some(target_partition_size) = params.target_partition_size {
ivf_rq_builder = ivf_rq_builder.target_partition_size(target_partition_size);
}
Ok(LanceDbIndex::IvfRq(ivf_rq_builder))
},
"HnswPq" => {
let params = source.extract::<IvfHnswPqParams>()?;
let distance_type = parse_distance_type(params.distance_type)?;
@@ -170,6 +186,16 @@ struct IvfPqParams {
target_partition_size: Option<u32>,
}
#[derive(FromPyObject)]
struct IvfRqParams {
distance_type: String,
num_partitions: Option<u32>,
num_bits: u32,
max_iterations: u32,
sample_rate: u32,
target_partition_size: Option<u32>,
}
#[derive(FromPyObject)]
struct IvfHnswPqParams {
distance_type: String,