Files
FastGPT/packages/service/common/vectorDB/constants.ts
T
Archer 64f70a41c1 feat: vector integrationTest;feat: ob quantization (#6366)
* feat(vectordb): add OceanBase HNSW quantization (HNSW_SQ/HNSW_BQ) (#6348)

Support OceanBase vector index quantization via VECTOR_VQ_LEVEL:
- 32 (default): hnsw + inner_product
- 8: hnsw_sq + inner_product (2-3x memory savings)
- 1: hnsw_bq + cosine (~15x memory savings)

HNSW_BQ requires cosine distance per OceanBase docs.
Tested on OceanBase 4.3.5.5 (BP5).

Closes #6202

* feat: add test inclusion for vectorDB tests in vitest configuration (#6358)

* feat: add test inclusion for vectorDB tests in vitest configuration

* refactor: update vectorDB README and setup for environment configuration

- Enhanced README to clarify the use of factory pattern for vectorDB integration tests.
- Updated instructions for setting up environment variables from a local file.
- Removed obsolete PG integration test file and adjusted test execution instructions.
- Improved structure explanation for shared test data and factory functions.

* perf: integrationTest

* feat: vector integration

---------

Co-authored-by: ZHANG Yixin <hi.yixinz@gmail.com>
Co-authored-by: Jingchao <alswlx@gmail.com>
2026-02-02 18:48:25 +08:00

83 lines
2.6 KiB
TypeScript

export const DatasetVectorDbName = 'fastgpt';
export const DatasetVectorTableName = 'modeldata';
export const PG_ADDRESS = process.env.PG_URL;
export const OCEANBASE_ADDRESS = process.env.OCEANBASE_URL;
export const SEEKDB_ADDRESS = process.env.SEEKDB_URL;
export const MILVUS_ADDRESS = process.env.MILVUS_ADDRESS;
export const MILVUS_TOKEN = process.env.MILVUS_TOKEN;
export const VectorVQ = (() => {
if (process.env.VECTOR_VQ_LEVEL === '32') {
return 32;
}
if (process.env.VECTOR_VQ_LEVEL === '16') {
return 16;
}
if (process.env.VECTOR_VQ_LEVEL === '8') {
return 8;
}
if (process.env.VECTOR_VQ_LEVEL === '4') {
return 4;
}
if (process.env.VECTOR_VQ_LEVEL === '2') {
return 2;
}
return 32;
})();
/**
* OceanBase HNSW Index Configuration
*
* VECTOR_VQ_LEVEL mapping:
* - 32 (default): hnsw + inner_product
* - 8: hnsw_sq + inner_product
* - 1: hnsw_bq + cosine
*
* See https://www.oceanbase.com/docs/common-oceanbase-database-cn-1000000004920602
* for the recommended way of choosing parameters (`m`, `ef_construction`, `ef_search`). It varies for data volume.
*
* HNSW_BQ requires cosine or l2 distance. inner_product is not supported up until V4.3.5 BP5 (current lts version until Jan 2026).
* See https://www.oceanbase.com/docs/common-oceanbase-database-cn-1000000004920603
* `HNSW_BQ distance 参数支持 l2 和 cosine。cosine 从 V4.3.5 BP4 版本开始支持。` and section `距离函数使用规则`.
*
* Tested on OceanBase 4.3.5-lts:
* ```sql
* -- HNSW_BQ + cosine: VECTOR INDEX SCAN ✓
* CREATE VECTOR INDEX idx ON t(vec) WITH (distance=cosine, type=hnsw_bq, m=16, ef_construction=200);
* EXPLAIN SELECT id, cosine_distance(vec, '[...]') AS score FROM t ORDER BY score ASC APPROXIMATE LIMIT 10;
* -- |1 |└─VECTOR INDEX SCAN|t(idx)|
* ```
*/
export const OceanBaseIndexConfig = (() => {
const level = process.env.VECTOR_VQ_LEVEL;
if (level === '1') {
return {
type: 'hnsw_bq' as const,
distance: 'cosine' as const,
distanceFunc: 'cosine_distance',
orderDirection: 'ASC' as const,
scoreTransform: (score: number) => 1 - score / 2
};
}
if (level === '8') {
return {
type: 'hnsw_sq' as const,
distance: 'inner_product' as const,
distanceFunc: 'inner_product',
orderDirection: 'DESC' as const,
scoreTransform: (score: number) => score
};
}
return {
type: 'hnsw' as const,
distance: 'inner_product' as const,
distanceFunc: 'inner_product',
orderDirection: 'DESC' as const,
scoreTransform: (score: number) => score
};
})();