Image Classification on Cars196 (test) (Efficiency Metrics)
44.29Accuracy @ 3.65sFGN
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| FGNhead type=Frozen-feature linear-head, batch size=128, learning rate (alpha)=0.1, lambda_0=1.0, damping=constant, CG maxiter=5, CG mode=row-space without warm-starting, horizon=3.65 s2026.05 | 44.29 | 2.3519 | 0.561 | 0.937 | 2.209 | |
| SGNhead type=Frozen-feature linear-head, batch size=128, learning rate (alpha)=0.1, lambda_0=1.0, damping=constant, CG maxiter=5, CG warm-started=true, horizon=3.65 s2026.05 | 44.16 | 2.3603 | 0.774 | 1.32 | 3.017 | |
| Adamhead type=Frozen-feature linear-head, batch size=128, learning rate (alpha)=3 * 10^-4, horizon=3.65 s2026.05 | 42.24 | 2.8449 | 0.693 | — | — |