---
title: smoke10
canonical_url: "https://www.modelscope.cn/datasets/LibreYOLO/smoke10"
md_url: "https://www.modelscope.cn/datasets/LibreYOLO/smoke10.md"
repository: LibreYOLO/smoke10
last_updated: 2026-08-28
license: apache-2.0
downloads: 246
stars: 0
---

# smoke10

> smoke10 - LibreYOLO 在 ModelScope 开源的数据集。smoke10 — LibreYOLO classification smoke set

LibreYOLO/smoke10 是 ModelScope 魔搭社区上的数据集，采用 apache-2.0 许可。

- **Repository**: LibreYOLO/smoke10
- **License**: apache-2.0
- **Downloads**: 246
- **Stars**: 0
- **Last updated**: 2026-08-28

Source: https://www.modelscope.cn/datasets/LibreYOLO/smoke10

---

# smoke10 — LibreYOLO classification smoke set

A tiny **2-images-per-class** subset of **Imagenette** (Apache-2.0) for fast CI /
end-to-end smoke tests of the classification train/val loop. Not for accuracy
evaluation.

## Provenance
- Derived from fast.ai Imagenette (Apache-2.0), source `imagenette2-160.tgz`
  `sha256` `64d0c4859f35a461889e0147755a999a48b49bf38a7e0f9bd27003f10db02fe5`.
- 2 train + 2 val images per class, 10 classes, ImageFolder `train/val/<wnid>` layout.
- Replaces the former ImageNet-derived `imagenet10` smoke set (ImageNet
  redistribution terms preclude clean hosting).

## Use with LibreYOLO
```python
model.val(data="smoke10")
```

## License
Apache-2.0 (inherited from Imagenette).
