Dataset Condensation on Tiny ImageNet (train)
2.4Training Speed (s/iter)NCFM w/o ψ
Evaluation results
| Method | Links | ||
|---|---|---|---|
| NCFM w/o ψIPC=10, Sampling Network (psi)=false, Hardware=NVIDIA A100 80G2025.02 | 2.4 | 6.4 | |
| NCFMIPC=10, Sampling Network (psi)=true, Hardware=NVIDIA A100 80G2025.02 | 3.27 | 6.5 | |
| NCFM w/o ψIPC=50, Sampling Network (psi)=false, Hardware=NVIDIA A100 80G2025.02 | 5.67 | 8.4 | |
| NCFMIPC=50, Sampling Network (psi)=true, Hardware=NVIDIA A100 80G2025.02 | 7.22 | 8.7 | |
| TESLAIPC=10, Hardware=NVIDIA A100 80G2025.02 | 42.01 | 69.6 |