| prompt-engineering-on-imagenet-a | MMRL++ + MQAdapter | – | Top-1 accuracy %: 51.64 |
| prompt-engineering-on-imagenet-a | MaPLe + MQAdapter | – | Top-1 accuracy %: 51.41 |
| prompt-engineering-on-imagenet-a | PromptSRC + MQAdapter | – | Top-1 accuracy %: 51.15 |
| prompt-engineering-on-imagenet-r | MMRL++ + MQAdapter | – | Top-1 accuracy %: 77.55 |
| prompt-engineering-on-imagenet-r | MaPLe + MQAdapter | – | Top-1 accuracy %: 77.21 |
| prompt-engineering-on-imagenet-r | PromptSRC + MQAdapter | – | Top-1 accuracy %: 77.90 |
| prompt-engineering-on-imagenet-s | MMRL++ + MQAdapter | – | Top-1 accuracy %: 49.56 |
| prompt-engineering-on-imagenet-s | MaPLe + MQAdapter | – | Top-1 accuracy %: 49.33 |
| prompt-engineering-on-imagenet-s | PromptSRC + MQAdapter | – | Top-1 accuracy %: 49.65 |
| prompt-engineering-on-imagenet-v2 | MMRL++ + MQAdapter | – | Top-1 accuracy %: 64.77 |
| prompt-engineering-on-imagenet-v2 | MaPLe + MQAdapter | – | Top-1 accuracy %: 64.24 |
| prompt-engineering-on-imagenet-v2 | PromptSRC + MQAdapter | – | Top-1 accuracy %: 64.53 |