Zhang et al.

other · 4 mentions across 1 reading

In this course

Zhang et al. (2018) are the authors of SA-GAN (Self-Attention GAN), a foundational architecture for scaling generative adversarial networks that uses hinge loss and attention mechanisms to improve image generation quality. The course readings invoke SA-GAN as a baseline or reference implementation when experimenting with larger models, diverse datasets, and architectural modifications like non-local blocks, treating it as a proven standard against which new training and scaling methods are validated. Their work appears centrally in discussions of GAN stability and discriminator optimization, providing both technical scaffolding and theoretical grounding for contemporary generative model research.

Mentioned in 1 reading

Andrew Brock

et al., LARGE SCALE GAN TRAINING FOR HIGH FIDELITY NATURAL IMAGE SYNTHESIS

In this section, we explore methods for scaling up GAN training to reap the performance benefits of larger models and larger batches. As a baseline, we employ the SA-GAN architecture of Zhang et al. (2018), which uses the hinge loss (Lim & …To confirm that our design choices are effective for even larger and more complex and diverse datasets, we also present results of our system on a subset of JFT-300M (Sun et al., 2017). The full JFT-300M dataset contains 300M real-world ima…Our basic setup follows SA-GAN (Zhang et al., 2018), and is implemented in TensorFlow (Abadi et al., 2016). We employ the architectures detailed in Appendix B, with non-local blocks inserted at a single stage in each network. Both G and D n…

+1 more

Appears alongside

People mentioned in the same passages — sorted by co-occurrence weight.

Saxe et al. 2 Miyato et al. 2 de Vries et al. 1 Dumoulin et al. 1 Glorot & Bengio 1 Karras et al. 1 Lim & Ye 1 Mescheder et al. 1 Miyato & Koyama 1 Radford et al. 1 Tran et al. 1 Yazc et al. 1 Sun et al. 1 Szegedy et al. 1 Abadi et al. 1 Ioffe & Szegedy 1 Kingma & Ba 1 Gulrajani et al. 1