This model is designed for creating artistic renditions of retro-futuristic space concepts. It's particularly suited for:
Users should be aware that the generated images reflect historical concepts and may not represent diverse perspectives. Care should be taken not to present these images as accurate depictions of current or future NASA projects.
The model was trained for 1000 steps using the AdamW8bit optimizer. The learning rate was set to 0.0004 with a batch size of 1. A caption dropout rate of 0.05 was used during training to improve generalization.
This model uses images from NASA, which are generally not subject to copyright in the United States. However, users should refer to the
NASA_IMAGE_USAGE_LICENSE.txt for full terms of use.
For more information about the dataset, please refer to the
dataset card.
Weights for this model are available in Safetensors format. You can
download them in the Files & versions tab.