Image Understanding on COCO GPT-based evaluation
84.1Conversation ScoreChat-UniVi
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Chat-UniViLLM Size=7B, Visual Tokens=1122023.11 | 84.1 | 74.2 | 93.7 | 84.2 | |
| LLaVALLM Size=13B, Visual Tokens=2562023.11 | 83.1 | 75.3 | 96.5 | 85.1 | |
| LLaVALLM Size=7B, Visual Tokens=256, re-implementation=true2023.11 | 78.8 | 70.2 | 91.8 | 80.4 | |
| LLaVALLM Size=7B, Visual Tokens=2562023.11 | 70.3 | 56.6 | 83.3 | 70.1 |