Utility Evaluation on LLaVA-Bench Coco
92.3ScoreShareGPT4V
Evaluation results
| Method | Links | |
|---|---|---|
| ShareGPT4V2024.10 | 92.3 | |
| CMRM_sampleBackbone=ShareGPT4V2024.10 | 91.4 | |
| CMRM_datasetBackbone=LLaVA-v1.5-13B2024.10 | 90.5 | |
| VLGuard MixedBackbone=LLaVA-v1.5-13B2024.10 | 90.4 | |
| CMRM_datasetBackbone=ShareGPT4V2024.10 | 90.1 | |
| LLaVA-v1.5-13B2024.10 | 89.7 | |
| CMRM_sampleBackbone=LLaVA-v1.5-13B2024.10 | 89.6 | |
| VLGuard PHBackbone=LLaVA-v1.5-7B2024.10 | 88.1 | |
| VLGuard MixedBackbone=LLaVA-v1.5-7B2024.10 | 87.8 | |
| VLGuard PHBackbone=LLaVA-v1.5-13B2024.10 | 87.5 | |
| LLaVA-v1.5-7B2024.10 | 79.2 | |
| CMRM_datasetBackbone=LLaVA-v1.5-7B2024.10 | 78.7 | |
| CMRM_sampleBackbone=LLaVA-v1.5-7B2024.10 | 77.3 |