HuggingFace is looking for examples of spaces that are “ethically mindful.” They define this as:

  • consentful: supports people’s self-determination and autonomy, demonstrating a commitment to acquiring data from informed sources and respecting end-user privacy.
  • sustainable: considering the ecological impact of models, including their carbon footprint.
  • socially conscious: projects that show a positive impact, such as tackling climate change, assisting medical research, or increasing accessibility through text-to-speech and image captioning.
  • inclusive: curating diverse datasets, training language models on underrepresented languages, and creating no-code frameworks for the less technical.
  • rigorous: focus on examining failure cases, protecting privacy, and informing users of a system’s limitations. 
  • inquisitive: projects that take a radical approach to common concepts, are rooted in critical theory, and highlight inequities and power structures.

You can make your nominations here.

Diffusers for the Mac

Pedro Cuenca tells us that HuggingFace has also launched an image generation application for the MacOS called Diffusers.

Screenshot 2023 02 06 at 11 11 15

The download page for the application tells us that it provides privacy, as all prompts and images are kept locally. The application offers five models, and various parameters including: prompt, negative prompt (what not to draw), and steps to run the image generation for.