Image Classification on Tiny-ImageNet Bucket 20 Forget
85.68AccuracyOriginal
Evaluation results
| Method | Links | |
|---|---|---|
| OriginalEvaluation Head=Linear Probe, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 85.68 | |
| OriginalEvaluation Head=Output, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 85.44 | |
| OriginalEvaluation Head=NCC, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 84.7 | |
| Retain-only RetainEvaluation Head=Linear Probe, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 81.44 | |
| Retain-only RetainEvaluation Head=NCC, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 78.64 | |
| Salun with CMFEvaluation Head=Linear Probe, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 75.24 | |
| Random-label with CMFEvaluation Head=Linear Probe, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 71.04 | |
| Salun with CMFEvaluation Head=Output, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 68.66 | |
| Salun with CMFEvaluation Head=NCC, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 67.62 | |
| Random-label with CMFEvaluation Head=Output, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 62.56 | |
| Random-label with CMFEvaluation Head=NCC, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 60.48 | |
| NegGrad+ with CMFEvaluation Head=Linear Probe, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 33.98 | |
| NegGrad+ with CMFEvaluation Head=Output, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 31.24 | |
| NegGrad+ with CMFEvaluation Head=NCC, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 27.58 | |
| Retain-only RetainEvaluation Head=Output, Backbone=ViT-S/16, Pre-training=ImageNet2026.04 | 0 |