Toward Verifiable and Reproducible Human Evaluation for Text-to-Image Generation | Read Paper on Bytez