Image-Text Retrieval on CLIP-Benchmark zero-shot
65.5IR@5CLIMP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CLIMPVision Tower=VMamba-B, Text Tower=Mamba-1, Zero-shot=true2026.01 | 65.5 | 77 | |
| CLIMPVision Tower=VMamba-B, Text Tower=Mamba-2, Zero-shot=true2026.01 | 65.2 | 75.3 | |
| RoPE-ViT-B/16Vision Tower=RoPE-ViT-B/16, Text Tower=LLaMA-3.2, Zero-shot=true2026.01 | 63.4 | 72.9 | |
| ViT-B/16Vision Tower=ViT-B/16, Text Tower=LLaMA-3.2, Zero-shot=true2026.01 | 62.8 | 72.9 | |
| FlexViT-B/16Vision Tower=FlexViT-B/16, Text Tower=LLaMA-3.2, Zero-shot=true2026.01 | 62.4 | 72.3 | |
| NaFlex-B/16Vision Tower=NaFlex-B/16, Text Tower=LLaMA-3.2, Zero-shot=true2026.01 | 61.5 | 73.3 |