Image Classification on ImageNet-1K 11 (val)
80.3Top-1 AccuracyDiGIT
Evaluation results
| Method | Links | |
|---|---|---|
| DiGITEvaluation Protocol=Linear probing, # Tokens=16 x 16, Features=1536, # Parameters=732M2024.10 | 80.3 | |
| VIM+ViT-VQGANEvaluation Protocol=Linear probing, # Tokens=32 x 32, Features=2048, # Parameters=1697M2024.10 | 73.2 | |
| DiGITEvaluation Protocol=Linear probing, # Tokens=16 x 16, Features=1024, # Parameters=219M2024.10 | 71.7 | |
| AIMEvaluation Protocol=Linear probing, # Tokens=16 x 16, Features=1536, # Parameters=0.6B2024.10 | 70.5 | |
| iGPT-XLEvaluation Protocol=Linear probing, # Tokens=64 x 64, Features=3072, # Parameters=6801M2024.10 | 68.7 | |
| GIVTEvaluation Protocol=Linear probing, # Tokens=16 x 16, Features=1024, # Parameters=304M2024.10 | 65.1 | |
| VIM+ViT-VQGANEvaluation Protocol=Linear probing, # Tokens=32 x 32, Features=1024, # Parameters=650M2024.10 | 65.1 | |
| VIM+dVAEEvaluation Protocol=Linear probing, # Tokens=32 x 32, Features=1024, # Parameters=650M2024.10 | 63.8 | |
| VIM+VQGANEvaluation Protocol=Linear probing, # Tokens=32 x 32, Features=1024, # Parameters=650M2024.10 | 61.8 | |
| iGPT-LEvaluation Protocol=Linear probing, # Tokens=32 x 32, Features=1536, # Parameters=1362M2024.10 | 60.3 |