---
license: apache-2.0
tags:
- segmentation
- mamba
- wafer
- electron-microscopy
- tokenunify
---

# TokenUnify Models

This repository contains TokenUnify models of different sizes trained on wafer electron microscopy data, along with a superhuman baseline model.

## Available Models

- **TokenUnify-1B.pth**: 1B parameter TokenUnify model
- **TokenUnify-500M.pth**: 500M parameter TokenUnify model  
- **TokenUnify-200M.pth**: 200M parameter TokenUnify model
- **TokenUnify-100M.pth**: 100M parameter TokenUnify model
- **superhuman.pth**: Superhuman baseline model

## Model Details

- **Architecture**: TokenUnify (based on Mamba)
- **Training Data**: Wafer electron microscopy images
- **Task**: Image Segmentation
- **Framework**: PyTorch

## Usage

```python
import torch

# Load a specific model
model_path = "TokenUnify-1B.pth"  # or any other model file
checkpoint = torch.load(model_path, map_location='cpu')

# Your model loading code here
```

## Model Sizes

| Model | Parameters | File Name |
|-------|------------|-----------|
| TokenUnify Large | 1B | TokenUnify-1B.pth |
| TokenUnify Medium | 500M | TokenUnify-500M.pth |
| TokenUnify Small | 200M | TokenUnify-200M.pth |
| TokenUnify Tiny | 100M | TokenUnify-100M.pth |
| Superhuman Baseline | - | superhuman.pth |

## Citation

If you use these models, please cite the relevant paper.
EOF