---
license: bigscience-openrail-m
---

# Model Card for MonoSpace

## Model description

MonoSpace is a latent diffusion model developed for a university image synthesis project. The model is designed as a starting point for higher-order diffusion models and focuses on generating high-quality images. MonoSpace has been trained on a subset of ImageNet and LaiOn Hires dataset, showcasing its potential in computer vision tasks.

## Intended use

MonoSpace is intended for:

- Image synthesis
- Image classification
- Object detection

As a starting point, MonoSpace aims to provide a solid foundation for researchers and developers working on higher-order diffusion models and other advanced image synthesis techniques.

## Training data

The MonoSpace model was trained on a combination of image datasets:

- ImageNet (subset)
- LaiOn Hires

## Model performance

MonoSpace achieved promising results on image synthesis and classification tasks:

- ImageNet Top-1 Accuracy (subset): *83.41%*

## Limitations

The MonoSpace model has the following limitations:

- Lack of Robust Latent High Frequency: The model may not capture high-frequency details as effectively as desired.
- Limited training resources: Due to constraints on computational resources during development, the model might not be optimized to its full potential.


## Responsible AI

To ensure ethical use of MonoSpace, we encourage users to follow Hugging Face's guidelines on responsible AI practices, including fairness, transparency, and accountability.

## License

The MonoSpace model is available under the bigscience-openrail-m.