Dataset Condensation on ImageNet 256x256 2012 (val)
5.76FIDGADC
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GADCData Budget=0.8% (10K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 5.76 | 430.2 | 78 | 69 | |
| DQData Budget=0.8% (10K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 7.55 | 271 | 73 | 69 | |
| Random†Data Budget=0.8% (10K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 7.73 | 263.1 | 69 | 68 | |
| D²CData Budget=0.8% (10K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 7.73 | 279.8 | 72 | 69 | |
| CCSData Budget=0.8% (10K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 7.81 | 373.3 | 77 | 58 | |
| GADCData Budget=4.0% (50K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 13.73 | 173.5 | 68 | 69 | |
| D²CData Budget=4.0% (50K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 16.28 | 131.1 | 68 | 63 | |
| DQData Budget=4.0% (50K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 16.83 | 122.8 | 67 | 65 | |
| Random†Data Budget=4.0% (50K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 17.04 | 120.8 | 67 | 65 | |
| CCSData Budget=4.0% (50K), Backbone=DiT-L/2, Training Iterations=100K, Evaluation Samples=10K2026.06 | 17.21 | 147.9 | 67 | 60 |