---
license: mit
tags:
  - image-classification
  - cifar10
  - convnext
  - muon
  - pytorch
datasets:
  - cifar10
metrics:
  - accuracy
model-index:
  - name: convnext-muon-cifar10
    results:
      - task:
          type: image-classification
          name: Image Classification
        dataset:
          type: cifar10
          name: CIFAR-10
        metrics:
          - type: accuracy
            value: 0.9810
            name: Test accuracy
---

# ConvNeXt + SE + Muon on CIFAR-10

**98.10%** test accuracy with ~12M parameters.

See the [GitHub repo](https://github.com/akira-n-28/convnext-muon-cifar10) 
for full code, training recipe, and reproducibility details.

## Quick load

```python
import torch
from huggingface_hub import hf_hub_download
from model import ConvNeXt   # from the GitHub repo

ckpt_path = hf_hub_download(
    repo_id="akira-n-28/convnext-muon-cifar10",
    filename="best_600ep_compile.pt",
)
ckpt = torch.load(ckpt_path, map_location="cpu")

model = ConvNeXt(
    depths=(3, 3, 9, 3), dims=(64, 128, 256, 512),
    kernel_size=(7, 5, 3, 3), drop_path_rate=0.15, layer_scale_init=1e-6,
)
model.load_state_dict(ckpt["model_state"])
```