benchmarks
fact
bearish
On ImageNet, visually unrecognizable images optimized to match reference Inception statistics achieve FID 24.7 versus 58.6 for real images, demonstrating FID can be gamed
on ImageNet, visually unrecognizable images optimized only to match the reference Inception mean and covariance obtain FID $24.7$ versus $58.6$ for held-out real images (lower is better)
Machine Learning (Statistics)29 Aug 2026