The akash-thumper-v1 (AT1) is a creative commons-based text-to-image generation model, marking a significant milestone as the first foundational model trained entirely on the Akash Decentralized Cloud. This model embodies the spirit of open-source and decentralized computing and leverages Creative Commons licensed data for its training.

While AT1 represents a pioneering effort in the use of decentralized cloud resources for AI model training, it's important to note that its performance does not currently meet the performance of other SOTA text-to-image generation models such as SDXL, Pixart-Alpha, or Stable Diffusion. This discrepancy in performance is largely attributed to the decision to use exclusively Creative Commons licensed data for training. Though this choice underscores our commitment to using open data sources and open-source principles, it has resulted in certain limitations in the model's output quality and diversity.

The AT1 model is designed for researchers, developers, and hobbyists looking to explore the capabilities of text-to-image generation within the constraints of Creative Commons data.

The akash-thumper-v1 model is released under the Open Rails License, emphasizing its commitment to open-source principles and the promotion of Creative Commons Foundation models. Users are encouraged to contribute to the model's development, share their modifications, and help improve the performance and capabilities of AT1.

For support and discussions, please visit our GitHub repository or join our community forums. Let's collaborate to push the boundaries of decentralized AI and open-source development.