Image Classification on CLIP 14-task suite
92.8Average Top-1 AccuracyFine-tuned
Evaluation results
| Method | Links | |
|---|---|---|
| Fine-tunedBackbone=CLIP-ViT-L/142026.05 | 92.8 | |
| Fine-tunedBackbone=CLIP-ViT-B/322026.05 | 90 | |
| TIES-Merging w/ E-PMQBackbone=CLIP-ViT-L/142026.05 | 82.5 | |
| Task Arithmetic w/ E-PMQBackbone=CLIP-ViT-L/142026.05 | 82 | |
| TIES-MergingBackbone=CLIP-ViT-L/142026.05 | 77.8 | |
| TIES-Merging w/ GPTQBackbone=CLIP-ViT-L/142026.05 | 74 | |
| TIES-Merging w/ RTNBackbone=CLIP-ViT-L/142026.05 | 72.7 | |
| TIES-Merging w/ E-PMQBackbone=CLIP-ViT-B/322026.05 | 72.1 | |
| Task Arithmetic w/ E-PMQBackbone=CLIP-ViT-B/322026.05 | 70.1 | |
| Pre-trainedBackbone=CLIP-ViT-L/142026.05 | 69.1 | |
| TIES-Merging w/ AWQBackbone=CLIP-ViT-L/142026.05 | 67.9 | |
| TIES-MergingBackbone=CLIP-ViT-B/322026.05 | 67.6 | |
| TIES-Merging w/ GPTQBackbone=CLIP-ViT-B/322026.05 | 63.5 | |
| Task ArithmeticBackbone=CLIP-ViT-L/142026.05 | 63.1 | |
| TIES-Merging w/ RTNBackbone=CLIP-ViT-B/322026.05 | 60.8 | |
| TIES-Merging w/ AWQBackbone=CLIP-ViT-B/322026.05 | 60.8 | |
| Task Arithmetic w/ GPTQBackbone=CLIP-ViT-L/142026.05 | 59.2 | |
| Pre-trainedBackbone=CLIP-ViT-B/322026.05 | 58.8 | |
| Task Arithmetic w/ RTNBackbone=CLIP-ViT-L/142026.05 | 57.8 | |
| Task ArithmeticBackbone=CLIP-ViT-B/322026.05 | 52.8 | |
| Task Arithmetic w/ GPTQBackbone=CLIP-ViT-B/322026.05 | 49.9 | |
| Task Arithmetic w/ AWQBackbone=CLIP-ViT-L/142026.05 | 49.1 | |
| Task Arithmetic w/ AWQBackbone=CLIP-ViT-B/322026.05 | 48.3 | |
| Task Arithmetic w/ RTNBackbone=CLIP-ViT-B/322026.05 | 47.6 |