v2.0 is an SDXL 1.0 based diffusion model with a larger 60k image dataset. It uses improved training on A100 80 GB GPUs and refined quality scoring. It focuses on better stability and more reliable outputs for both experimentation and production use.
