Image Captioning on MSCOCO (CHAIR Metrics)
58.2CHAIRsICD
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ICDVLM=InstructBLIP2026.04 | 58.2 | 18.5 | — | — | 8.1 | |
| VCDBase Model=LLaVa-1.52025.12 | 57.4 | 16.1 | 76.7 | 103.1 | — | |
| VCDBase Model=MiniGPT-42025.12 | 57.4 | 16.1 | 76.7 | 103.1 | — | |
| RegularBase Model=LLaVa-1.52025.12 | 56.4 | 17.2 | 69.5 | 106.2 | — | |
| RegularBase Model=MiniGPT-42025.12 | 56.4 | 17.2 | 69.5 | 106.2 | — | |
| VCDVLM=InstructBLIP2026.04 | 54.8 | 16.2 | — | — | 8.6 | |
| MFCDVLM=LLaVA-1.52026.04 | 54 | 15 | — | — | — | |
| BaselineModel=LLaVA-1.5-7B2026.04 | 53 | 18.4 | 67 | — | — | |
| OPERABase Model=LLaVa-1.52025.12 | 51.3 | 13.6 | 74.2 | 93.6 | — | |
| ICDVLM=LLaVA-1.52026.04 | 50.8 | 16.9 | — | — | 8.2 | |
| Beam Search (BS)Base Model=LLaVa-1.52025.12 | 50.6 | 13.5 | 78.8 | 97.6 | — | |
| NulluBase Model=MiniGPT-42025.12 | 48.4 | 13.5 | 56.8 | 105.2 | — | |
| RegularVLM=InstructBLIP2026.04 | 48 | 13.9 | — | — | 8.4 | |
| NulluBase Model=LLaVa-1.52025.12 | 46.8 | 14.9 | 67.9 | 95 | — | |
| RegularBase Model=LLaVa-NeXT2025.12 | 40.4 | 12.2 | 60.2 | 175.9 | — | |
| ALEAHalluBase Model=MiniGPT-42025.12 | 40.2 | 13 | 78 | 98.1 | — | |
| BaselineModel=Qwen3-VL-8B2026.04 | 40 | 9.9 | 74.1 | — | — | |
| IFCDVLM=InstructBLIP2026.04 | 39.6 | 11.2 | — | — | — | |
| ALEAHalluBase Model=LLaVa-1.52025.12 | 39.3 | 12.1 | 74.2 | 98.1 | — | |
| BaselineModel=InternVL2-8B2026.04 | 37 | 9.3 | 62.3 | — | — | |
| NulluBase Model=LLaVa-NeXT2025.12 | 35.5 | 10 | 61.3 | 173.5 | — | |
| Beam Search (BS)Base Model=LLaVa-NeXT2025.12 | 35.2 | 8.5 | 62.8 | 176.7 | — | |
| OPERABase Model=LLaVa-NeXT2025.12 | 35.2 | 8.5 | 62.8 | 176.7 | — | |
| Beam Search (BS)Base Model=MiniGPT-42025.12 | 34.8 | 10 | 61.1 | 80.6 | — | |
| OPERABase Model=MiniGPT-42025.12 | 34.8 | 10 | 61.1 | 80.6 | — | |
| VCDBase Model=LLaVa-NeXT2025.12 | 34.6 | 9.6 | 60.7 | 175.6 | — | |
| ALEAHalluBase Model=LLaVa-NeXT2025.12 | 34.4 | 7.6 | 63 | 185.6 | — | |
| Active-LookModel=Qwen3-VL-8B2026.04 | 32 | 8.3 | 70 | — | — | |
| VCDVLM=LLaVA-1.52026.04 | 21.8 | 11 | — | — | 8.2 | |
| Active-LookModel=InternVL2-8B2026.04 | 21.5 | 6.6 | 64.7 | — | — | |
| BaselineModel=Qwen2.5-VL-7B2026.05 | 21 | 5.68 | 56.42 | — | — | |
| RegularVLM=LLaVA-1.52026.04 | 20 | 15.2 | — | — | 9.2 | |
| MHSAModel=Qwen2.5-VL-7B2026.05 | 18 | 5.19 | 55.14 | — | — | |
| Active-LookModel=LLaVA-1.5-7B2026.04 | 15 | 15.8 | 64.9 | — | — | |
| CATCHVLM=LLaVA-1.52026.04 | 14.3 | 4.7 | — | — | — | |
| IFCDVLM=LLaVA-1.52026.04 | 13.2 | 5.6 | — | — | — | |
| CATCHVLM=InstructBLIP2026.04 | 7.2 | 1.8 | — | — | — | |
| IECD2VLM=LLaVA-1.52026.04 | 2.96 | 2.25 | — | — | 2.3 | |
| IECD2VLM=InstructBLIP2026.04 | 0.64 | 0.85 | — | — | 3.8 | |
| OctopusVLM=LLaVA-1.52026.04 | — | — | — | — | 4.8 | |
| OctopusVLM=InstructBLIP2026.04 | — | — | — | — | 6.1 |