Synthetic Data from Diffusion Models Improves ImageNet Classification

Large-scale text-to-image diffusion models can be fine-tuned to produce class-conditional models with SOTA FID and Inception Score on ImageNet.

Data and Resources

Cite this as

Shekoofeh Azizi, Simon Kornblith, Chitwan Saharia, Mohammad Norouzi, David J. Fleet (2024). Dataset: Synthetic Data from Diffusion Models Improves ImageNet Classification. https://doi.org/10.57702/hf3ykpkj

DOI retrieved: December 16, 2024

Additional Info

Field Value
Created December 16, 2024
Last update December 16, 2024
Author Shekoofeh Azizi
More Authors
Simon Kornblith
Chitwan Saharia
Mohammad Norouzi
David J. Fleet
Homepage https://arxiv.org/abs/2205.11487