Generalized Category Discovery on CIFAR-10 (test)
97.9Accuracy (Cluster, All)PromptCAL
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PromptCALBackbone=ViT-B/16, Pre-training=DINO2026.02 | 97.9 | — | — | — | 96.6 | 98.5 | |
| SIEFormerBase Framework=SimGCD, Backbone=ViT-B/16, Pre-training=DINO2026.02 | 97.4 | — | — | — | 95.2 | 98.6 | |
| SIEFormerBase Framework=SPTNet, Backbone=ViT-B/16, Pre-training=DINO2026.02 | 97.4 | — | — | — | 95.1 | 98.6 | |
| SPTNetBackbone=ViT-B/16, Pre-training=DINO2026.02 | 97.3 | — | — | — | 95 | 98.6 | |
| ProtoGCDBackbone=ViT-B/16, Pre-training=DINO2026.02 | 97.3 | — | — | — | 95.3 | 98.2 | |
| SIEFormerBase Framework=GCD, Backbone=ViT-B/16, Pre-training=DINO2026.02 | 97.3 | — | — | — | 95 | 98.4 | |
| SimGCDBackbone=ViT-B/16, Pre-training=DINO2026.02 | 97.1 | — | — | — | 95.1 | 98.1 | |
| HypGCDBackbone=ViT-B/16, Pre-training=DINO2026.02 | 96.6 | — | — | — | 95.5 | 97.2 | |
| DCCLBackbone=ViT-B/16, Pre-training=DINO2026.02 | 96.3 | — | — | — | 96.5 | 96.9 | |
| GCDBackbone=ViT-B/16, Pre-training=DINO2026.02 | 91.5 | — | — | — | 97.9 | 88.2 | |
| ORCABackbone=ViT-B/16, Pre-training=DINO2026.02 | 81.8 | — | — | — | 86.2 | 79.6 | |
| UNO+Backbone=ViT-B/16, Pre-training=DINO2026.02 | 68.6 | — | — | — | 98.3 | 53.8 | |
| RankStats+Backbone=ViT-B/16, Pre-training=DINO2026.02 | 46.8 | — | — | — | 19.2 | 60.5 | |
| ComExEvaluation Protocol=task-agnostic2022.10 | — | 95 | 92.6 | 93.8 | — | — | |
| DTCEvaluation Protocol=task-agnostic2022.10 | — | 58.7 | 78.6 | 68.7 | — | — | |
| GCD2025.12 | — | 97.9 | 88.2 | 91.5 | — | — | |
| IICEvaluation Protocol=task-agnostic2022.10 | — | 96 | 97.2 | 96.6 | — | — | |
| KCLEvaluation Protocol=task-agnostic2022.10 | — | 79.4 | 60.1 | 69.8 | — | — | |
| MCLEvaluation Protocol=task-agnostic2022.10 | — | 81.4 | 64.8 | 73.1 | — | — | |
| RankStats2025.12 | — | 19.2 | 60.5 | 46.8 | — | — | |
| RS+Evaluation Protocol=task-agnostic2022.10 | — | 90.6 | 88.8 | 89.7 | — | — | |
| UNO2025.12 | — | 98.3 | 53.8 | 68.6 | — | — | |
| UNOv1Evaluation Protocol=task-agnostic2022.10 | — | 93.5 | 93.3 | 93.4 | — | — | |
| VLM-NCDImage Encoder=CLIP, Knowledge Database=Yes2025.12 | — | 97.2 | 96.4 | 96.6 | — | — | |
| XCon2025.12 | — | 97.3 | 95.4 | 96 | — | — |