Image-to-Text Retrieval on Flickr30K English 1K (test)
97.9R@1InternVL-G-FT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| InternVL-G-FTEvaluation Protocol=Fine-tuned2023.12 | 97.9 | 100 | 100 | 97.6 | |
| ONE-PEACEEvaluation Protocol=Fine-tuned2023.12 | 97.6 | 100 | 100 | 97.4 | |
| BEIT-3Evaluation Protocol=Fine-tuned2023.12 | 97.5 | 99.9 | 100 | 97.4 | |
| BLIPEvaluation Protocol=Fine-tuned2023.12 | 97.4 | 99.8 | 99.9 | 96.9 | |
| OmniVLEvaluation Protocol=Fine-tuned2023.12 | 97.3 | 99.9 | 100 | 97 | |
| FlorenceEvaluation Protocol=Fine-tuned2023.12 | 97.2 | 99.9 | — | — | |
| InternVL-C-FTEvaluation Protocol=Fine-tuned2023.12 | 97.2 | 100 | 100 | 97.2 | |
| FILIPEvaluation Protocol=Fine-tuned2023.12 | 96.6 | 100 | 100 | 96.8 | |
| ALIGNEvaluation Protocol=Fine-tuned2023.12 | 95.3 | 99.8 | 100 | 96 | |
| BLIP-2multi-lingual=false, Zero-shot=true2023.12 | 0.976 | 1 | 1 | — | |
| InternVL-Gmulti-lingual=true, Zero-shot=true2023.12 | 0.957 | 0.997 | 0.999 | — | |
| InternVL-Cmulti-lingual=true, Zero-shot=true2023.12 | 0.947 | 0.996 | 0.999 | — | |
| EVA-02-CLIP-E+multi-lingual=false, Zero-shot=true2023.12 | 0.939 | 0.994 | 0.998 | — | |
| OpenCLIP-Gmulti-lingual=false, Zero-shot=true2023.12 | 0.929 | 0.993 | 0.998 | — | |
| CoCamulti-lingual=false, Zero-shot=true2023.12 | 0.925 | 0.995 | 0.999 | — | |
| OpenCLIP-XLM-R-Hmulti-lingual=true, Zero-shot=true2023.12 | 0.918 | 0.994 | 0.998 | — | |
| EVA-01-CLIP-g+multi-lingual=false, Zero-shot=true2023.12 | 0.916 | 0.993 | 0.998 | — | |
| OpenCLIP-gmulti-lingual=false, Zero-shot=true2023.12 | 0.914 | 0.992 | 0.996 | — | |
| Florencemulti-lingual=false, Zero-shot=true2023.12 | 0.909 | — | — | — | |
| ONE-PEACEmulti-lingual=false, Zero-shot=true2023.12 | 0.909 | 0.988 | 0.998 | — | |
| OpenCLIP-Hmulti-lingual=false, Zero-shot=true2023.12 | 0.908 | 0.993 | 0.997 | — |