What FID Hides: Detecting, Ranking, and Diagnosing Deviations in Generative Evaluation
Generative models are commonly ranked by Fréchet Inception Distance (FID) and Kernel Inception Distance (KID), yet FID's first-two-moment summary can miss distributional differences, and a reported scalar gap alone is not a calibrated test against sampling variation. FID's moment restriction has concrete consequences: 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). Moreover, FID and KID are scalar discrepancies that are unchanged when the two samples are e
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- FuzzySimilar title/name (fuzzy) · 84%GoogleCloudPlatform/generative-ai →
“Fuzzy title match (0.92): “What FID Hides: Detecting, Ranking, and Diagnosing Deviation” ≈ “GoogleCloudPlatform/generative-ai””
- FuzzySimilar title/name (fuzzy) · 59%steven2358/awesome-generative-ai →
“Fuzzy title match (0.73): “What FID Hides: Detecting, Ranking, and Diagnosing Deviation” ≈ “steven2358/awesome-generative-ai””
- LinkedLinked via arxiv author · 85%Hao Chen →
“What FID Hides: Detecting, Ranking, and Diagnosing Deviations in Generative Evaluation”