maknee/gist1m
收藏Hugging Face2026-02-02 更新2026-03-29 收录
下载链接:
https://hf-mirror.com/datasets/maknee/gist1m
下载链接
链接失效反馈官方服务:
资源简介:
---
license: mit
task_categories:
- feature-extraction
tags:
- vector-search
- ann-benchmarks
- diskann
- minio
size_categories:
- 1M<n<10M
---
# GIST1M Vector Search Dataset
1 million 960-dimensional vectors from the GIST descriptor dataset.
## Dataset Details
- **Vectors**: 1,000,000
- **Dimensions**: 960
- **Queries**: 1,000
- **Source**: [ANN Benchmarks](http://corpus-texmex.irisa.fr/)
## Shard Configurations
| Config | Shards | Vectors/Shard | .indices | .vectors |
|--------|--------|---------------|----------|----------|
| shard_3 | 3 | 333,333 | 651MB | 1.2GB |
| shard_5 | 5 | 200,000 | 391MB | 732MB |
| shard_7 | 7 | 142,857 | 279MB | 523MB |
| shard_10 | 10 | 100,000 | 195MB | 366MB |
## DiskANN Parameters
- R: 64, L: 100, Distance: L2
## Usage
```python
from huggingface_hub import snapshot_download
snapshot_download("maknee/gist1m", allow_patterns=["fbin/*", "diskann/shard_5/*"], local_dir="./gist1m")
```
提供机构:
maknee



