Video Question Answering on PhyBench
0.7933Avg ScorePhyVLLM-finetune
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PhyVLLM-finetuneSize=7B, Evaluation Protocol=fine-tuned2025.12 | 0.7933 | 77.67 | 68.97 | 0.8133 | 0.92 | 0.8167 | |
| Qwen2.5-VL-finetuneSize=7B, Evaluation Protocol=fine-tuned2025.12 | 0.4672 | 18.89 | 20.11 | 0.69 | 0.7833 | 0.6333 | |
| PhyVLLMSize=7B, Evaluation Protocol=zero-shot2025.12 | 0.4052 | 48.67 | 45.49 | 0.1511 | 0.8283 | 0.1667 | |
| GPT-4oEvaluation Protocol=zero-shot2025.12 | 0.3405 | 0.87 | 0.11 | 0.103 | 0.6392 | 0.9503 | |
| InternVideo2Size=8B, Evaluation Protocol=zero-shot2025.12 | 0.2336 | 1.11 | 1.33 | 0.9567 | 0.0467 | 0 | |
| VideoLLaMA2Size=7B, Evaluation Protocol=zero-shot2025.12 | 0.2319 | 0.11 | 0 | 0.9922 | 0.0117 | 0.005 | |
| InternVL2.5Size=8B, Evaluation Protocol=zero-shot2025.12 | 0.2316 | 0 | 0 | 0.9967 | 0.01 | 0 | |
| InternVL2-finetuneSize=8B, Evaluation Protocol=fine-tuned2025.12 | 0.2316 | 0 | 0 | 0.9967 | 0.01 | 0 | |
| Qwen2.5-VLSize=7B, Evaluation Protocol=zero-shot2025.12 | 0.229 | 0.44 | 0.78 | 0.73 | 0.345 | 0.03 | |
| LLaVa-NeXT-VideoSize=7B, Evaluation Protocol=zero-shot2025.12 | 0.2134 | 15 | 17.8 | 0.4333 | 0.185 | 0.06 | |
| mPLUG-Owl3Size=7B, Evaluation Protocol=zero-shot2025.12 | 0.1972 | 1.33 | 0 | 0.6322 | 0.3133 | 0 | |
| VideoChatGPTSize=7B, Evaluation Protocol=zero-shot2025.12 | 0.1747 | 7.89 | 6.9 | 0.04 | 0.655 | 0.1983 |