Detailed Description Matching on EventHallusion
54.6Accuracy (Entire)Tarsier2-7B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Tarsier2-7BModel Type=Open-Source, Model Scale=≤10B, GPT-4 Evaluator=true2025.01 | 54.6 | 53.1 | 93.7 | 63.3 | |
| InternVL2.5-78BModel Type=Open-Source, Model Scale=>10B, GPT-4 Evaluator=true2025.01 | 45 | 43 | 76.8 | 51.6 | |
| Gemini-1.5-ProModel Type=Proprietary, GPT-4 Evaluator=true2025.01 | 38.5 | 40.9 | 80 | 49.6 | |
| Tarsier-34BModel Type=Open-Source, Model Scale=>10B, GPT-4 Evaluator=true2025.01 | 38.5 | 40.4 | 83.2 | 50.1 | |
| Tarsier-7BModel Type=Open-Source, Model Scale=≤10B, GPT-4 Evaluator=true2025.01 | 35.8 | 29.5 | 72.6 | 41.6 | |
| GPT-4oModel Type=Proprietary, GPT-4 Evaluator=true2025.01 | 34.9 | 54.9 | 83.2 | 56.2 | |
| LLaVA-Video-72BModel Type=Open-Source, Model Scale=>10B, GPT-4 Evaluator=true2025.01 | 32.1 | 35.8 | 75.5 | 44.2 | |
| LLaVA-Video-7BModel Type=Open-Source, Model Scale=≤10B, GPT-4 Evaluator=true2025.01 | 27.5 | 32.6 | 75.5 | 41.4 | |
| LLAVA-OV-72BModel Type=Open-Source, Model Scale=>10B, GPT-4 Evaluator=true2025.01 | 24.8 | 34.7 | 71.3 | 40.7 | |
| InternVL2.5-8BModel Type=Open-Source, Model Scale=≤10B, GPT-4 Evaluator=true2025.01 | 23.9 | 20.7 | 60 | 31 | |
| LLAVA-OV-7BModel Type=Open-Source, Model Scale=≤10B, GPT-4 Evaluator=true2025.01 | 22 | 26.4 | 73.4 | 36.4 | |
| Qwen2-VL-72BModel Type=Open-Source, Model Scale=>10B, GPT-4 Evaluator=true2025.01 | 16.5 | 25.4 | 70.2 | 33.6 | |
| Qwen2-VL-7BModel Type=Open-Source, Model Scale=≤10B, GPT-4 Evaluator=true2025.01 | 14.7 | 16.1 | 67 | 27.8 |