Image Classification on Digits alpha=0.1 (test)
86.87Accuracy (MNIST Domain)FedPFT + SLOT-Align
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| FedPFT + SLOT-AlignBackbone=ViT-B/32, Pre-training=CLIP, Federated Learning Rounds=One-shot, Alignment strength (tau)=0.42026.06 | 86.87 | 83.19 | 62.75 | 69.18 | 75.5 | 9.89 | |
| O-FedAvg + SLOT-AlignBackbone=ViT-B/32, Pre-training=CLIP, Federated Learning Rounds=One-shot, Alignment strength (tau)=0.42026.06 | 85.61 | 76.97 | 56.03 | 71.84 | 72.61 | 10.77 | |
| FedPFTBackbone=ViT-B/32, Pre-training=CLIP, Federated Learning Rounds=One-shot2026.06 | 84.74 | 82.23 | 61.6 | 64.85 | 73.36 | 10.23 | |
| FedCGS + SLOT-AlignBackbone=ViT-B/32, Pre-training=CLIP, Federated Learning Rounds=One-shot, Alignment strength (tau)=0.42026.06 | 84.05 | 73.44 | 63.05 | 59.9 | 70.11 | 9.48 | |
| FedCGSBackbone=ViT-B/32, Pre-training=CLIP, Federated Learning Rounds=One-shot2026.06 | 82.18 | 81.46 | 53.29 | 51.55 | 67.12 | 14.72 | |
| FedAvgBackbone=ViT-B/32, Pre-training=CLIP, Federated Learning Rounds=Multi-round2026.06 | 79.04 | 81.15 | 63.08 | 68.07 | 71.82 | 7.58 | |
| O-FedAvgBackbone=ViT-B/32, Pre-training=CLIP, Federated Learning Rounds=One-shot2026.06 | 76.36 | 73.98 | 49.99 | 58.45 | 64.7 | 10.93 |