@cassie

Klearu has been updated with new model support, enhanced specifically for it: transformer-driven image generation, with support for specialized text rendering. This brings our capability parity extremely close to current state of the art, and has a good amount of room on the table for improvement. Additionally, the desktop app is now in the repo, enabling use of various models together, with tool call support, animation, various forms of image generation, inpainting, use of reference images for more powerful alterations (similar to chatgpt images 2.0). https://github.com/quilibriumnetwork/klearu. Rather than simply note these things, I'm attaching a demonstration of the degree of capability change. This was produced witih the prompt "a digital brain with sparks spelling out the word 'KLEARU'". This will also be coming to quorum as a feature.

5 replies

21 recasts

127 reactions