Beyond Thumbs Up/Down: Untangling Challenges of Fine-Grained Feedback for Text-to-Image Generation

• Published in: arXiv.org
• Main Authors: Collins, Katherine M, Kim, Najoung, Bitton, Yonatan, Rieser, Verena, Shayegan Omidshafiei, Hu, Yushi, Chen, Sherol, Dutta, Senjuti, Chang, Minsuk, Lee, Kimin, Liang, Youwei, Evans, Georgina, Singla, Sahil, Li, Gang, Weller, Adrian, He, Junfeng, Ramachandran, Deepak, Krishnamurthy Dj Dvijotham
• Format: Paper
• Language: English
• Published: Ithaca Cornell University Library, arXiv.org 17.10.2024
• Subjects: Alignment; Feedback; Image processing; Image quality; Learning

Human feedback plays a critical role in learning and refining reward models for text-to-image generation, but the optimal form the feedback should take for learning an accurate reward function has not been conclusively established. This paper investigates the effectiveness of fine-grained feedback w...