Tim Salimans

Papers

Total Citations

364

H-Index

About

Tim Salimans is a leading researcher in deep learning, best known for his pioneering contributions to generative modeling and efficient attention mechanisms. His work on **Axial Attention in Multidimensional Transformers** (2019, 364 citations) introduced Axial Transformers, a self-attention-based autoregressive model that efficiently handles high-dimensional data like images. By factorizing attention along tensor axes, this approach dramatically reduces computational costs, enabling scalable processing of images, video, and other structured data without sacrificing performance. Salimans’ broader impact includes foundational advances in generative adversarial networks (GANs) and variational inference, where his techniques for improving training stability and sample quality have become standard. With hundreds of citations across his key papers, his research has shaped modern deep learning, particularly in generative AI and efficient transformer architectures. His work is widely adopted in both academia and industry, influencing everything from image synthesis to video generation. Salimans continues to push boundaries at the intersection of probabilistic modeling and scalable neural networks, making him a key figure for students and researchers exploring advanced generative models and attention mechanisms.