---
library_name: pytorch
tags:
  - mnist
  - mlp
  - image-classification
  - pytorch
license: mit
datasets:
  - mnist
metrics:
  - accuracy
---

# MNIST MLP (fold-4 best)

**Model**: `ImprovedMLP` (2048 → 1024 → 512 → 256 → 128 → 10)  
**File**: `mlp_best_fold4.pth`  
**Dataset**: MNIST (mean `0.1307`, std `0.3081`)  

## Usage

```python
from huggingface_hub import hf_hub_download
import torch, torch.nn as nn

class ImprovedMLP(nn.Module):
    def __init__(self):
        super().__init__()
        self.net = nn.Sequential(
            nn.Flatten(),
            nn.Linear(784,2048), nn.LayerNorm(2048), nn.GELU(), nn.Dropout(0.1),
            nn.Linear(2048,1024), nn.LayerNorm(1024), nn.GELU(), nn.Dropout(0.1),
            nn.Linear(1024,512), nn.LayerNorm(512), nn.GELU(), nn.Dropout(0.1),
            nn.Linear(512,256), nn.LayerNorm(256), nn.GELU(),
            nn.Linear(256,128), nn.LayerNorm(128), nn.GELU(),
            nn.Linear(128,10)
        )
    def forward(self,x): return self.net(x)

path = hf_hub_download("chandu1617/MNIST_with_MLP", "mlp_best_fold4.pth")
model = ImprovedMLP()
model.load_state_dict(torch.load(path))
model.eval()