Expand description
Auto-Index Selection
Automatically selects the optimal index type based on data characteristics:
- Dataset size (number of vectors)
- Dimensionality
- Memory constraints
- Query latency requirements
- Accuracy requirements
§Index Selection Matrix
| Vectors | Dimensions | Memory | Latency | Recommended Index |
|---|---|---|---|---|
| < 10K | Any | Any | Any | Flat (brute force) |
| 10K-100K | < 256 | Low | Medium | IVF + SQ8 |
| 10K-100K | < 256 | Normal | Low | HNSW |
| 100K-1M | Any | Low | Medium | IVF-PQ |
| 100K-1M | Any | Normal | Low | HNSW |
| > 1M | Any | Low | Any | IVF-PQ |
| > 1M | Any | Normal | Low | HNSW + SQ8 |
Structs§
- Auto
Index Selector - Auto-index selector
- Data
Characteristics - Data characteristics for index selection
- Index
Params - Suggested index parameters
- Index
Recommendation - Index recommendation with reasoning
- Index
Requirements - Requirements for the index
Enums§
- Accuracy
Requirement - Accuracy requirement for search results
- Latency
Requirement - Query latency requirement
- Memory
Constraint - Memory constraint level
- Recommended
Index - Recommended index type