Classification on CheXpert 5x200 1.0
66.74AccuracyCLEFT-LORA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CLEFT-LORAEvaluation Protocol=Zero-shot2024.07 | 66.74 | — | |
| CLEFT-LORAEvaluation Protocol=Full-finetuning2024.07 | 63.96 | 88.22 | |
| CLEFT-LORAEvaluation Protocol=Linear-probing2024.07 | 63.46 | 87.76 | |
| CLEFT-PrefixEvaluation Protocol=Full-finetuning2024.07 | 61.56 | 87.73 | |
| CLEFT-IA3Evaluation Protocol=Full-finetuning2024.07 | 61.06 | 86.81 | |
| CLEFT-PrefixEvaluation Protocol=Linear-probing2024.07 | 60.55 | 87.2 | |
| CLEFT-IA3Evaluation Protocol=Linear-probing2024.07 | 60.16 | 86.48 | |
| MedCLIP-SwinEvaluation Protocol=Zero-shot, Pre-training Dataset=2x larger than standard2024.07 | 59.42 | — | |
| MRM-VITEvaluation Protocol=Linear-probing2024.07 | 58.26 | 84.48 | |
| GLORIA-R50Evaluation Protocol=Full-finetuning2024.07 | 57.56 | 87.11 | |
| MedCLIP-SwinEvaluation Protocol=Full-finetuning, Pre-training Dataset=2x larger than standard2024.07 | 57.46 | 87.85 | |
| MGCA-VIT+Evaluation Protocol=Full-finetuning, Pre-training Dataset=2x larger than standard2024.07 | 56.96 | 86.33 | |
| MRM-VITEvaluation Protocol=Full-finetuning2024.07 | 56.56 | 87.41 | |
| ImageNet-ViTEvaluation Protocol=Full-finetuning2024.07 | 56.46 | 85.71 | |
| MedCLIP-SwinEvaluation Protocol=Linear-probing, Pre-training Dataset=2x larger than standard2024.07 | 54.55 | 85.75 | |
| GLORIA-R50Evaluation Protocol=Linear-probing2024.07 | 51.65 | 83.15 | |
| MGCA-VIT+Evaluation Protocol=Zero-shot, Pre-training Dataset=2x larger than standard2024.07 | 51.35 | — | |
| ConVIRT-R50Evaluation Protocol=Linear-probing2024.07 | 47.7 | — | |
| CLIP-VITEvaluation Protocol=Full-finetuning2024.07 | 44.84 | 77.59 | |
| GLORIA-R50Evaluation Protocol=Zero-shot2024.07 | 43.28 | — | |
| ConVIRT-R50Evaluation Protocol=Zero-shot2024.07 | 41.88 | — | |
| ImageNet-ViTEvaluation Protocol=Linear-probing2024.07 | 38.54 | 75.44 | |
| CLIP-VITEvaluation Protocol=Linear-probing2024.07 | 35.34 | 65.14 | |
| MGCA-VIT+Evaluation Protocol=Linear-probing, Pre-training Dataset=2x larger than standard2024.07 | 26.63 | 63.94 | |
| Random-ViTEvaluation Protocol=Linear-probing2024.07 | 20.62 | 57.85 | |
| Random-ViTEvaluation Protocol=Full-finetuning2024.07 | 20.32 | 63.68 | |
| CLIP-VITEvaluation Protocol=Zero-shot2024.07 | 20.16 | — |