FID variance from training seeds is 3.2 times larger than from sampling seeds on hundreds of SiT models, with 1-2% coefficient of variation that barely shrinks with more compute, leading to a multi-seed evaluation protocol.
Quantifying the uncertainty of model-based synthetic image quality metrics.arXiv preprint arXiv:2504.03623, 2025
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
The FID Lottery: Quantifying Hidden Randomness in Generative-Model Evaluation
FID variance from training seeds is 3.2 times larger than from sampling seeds on hundreds of SiT models, with 1-2% coefficient of variation that barely shrinks with more compute, leading to a multi-seed evaluation protocol.