Rethinking Training for De-biasing Text-to-Image Generation: Unlocking the Potential of Stable Diffusion | Read Paper on Bytez