Massachusetts Institute of Technology
2 papers at NeurIPS 2025
Training protein generative model with noisy synthetic structures
We improve the quality of generative models by using low-quality, corrupted, and out-of-distribution data
