Multiple-choice Video Question Answering on MLVU (test)
73.8AccuracyLLaVA-Video + Video-RAG
Evaluation results
| Method | Links | |
|---|---|---|
| LLaVA-Video + Video-RAG#Params=72B, Frames=642024.11 | 73.8 | |
| LLaVA-Video#Params=72B, Frames=64, is_replicated=true2024.11 | 73.1 | |
| LLaVA-Video + Video-RAG#Params=7B, Frames=642024.11 | 72.4 | |
| Oryx-1.5#Params=32B, Frames=1282024.11 | 72.3 | |
| LLaVA-Video#Params=7B, Frames=64, is_replicated=true2024.11 | 70.8 | |
| Aria#Params=25.3B, Frames=2562024.11 | 70.6 | |
| Video-XL#Params=7B, Frames=2562024.11 | 64.9 | |
| GPT-4oFrames=0.5 fps2024.11 | 64.6 | |
| Video-CCAM#Params=14B, Frames=962024.11 | 63.1 | |
| VITA-1.5#Params=7B, Frames=162024.11 | 60.4 |