Video Question Answering on TGIF-QA FrameQA
78.7AccuracyVALORL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VALORLPretraining examples=33.5M, Modality=V+A2023.04 | 78.7 | — | |
| GIT2 (5.1B)Pretraining examples=12.9B, Modality=V2023.04 | 74.9 | — | |
| VALORBPretraining examples=6.5M, Modality=V+A2023.04 | 74.5 | — | |
| LAVENDERPretraining examples=30M, Modality=V2023.04 | 73.5 | — | |
| VALORLPretraining examples=5.5M, Modality=V2023.04 | 73 | — | |
| VIOLETPretraining examples=5.5M, Modality=V2023.04 | 72.8 | — | |
| GITPretraining examples=800M, Modality=V2023.04 | 72.8 | — | |
| InternVideoPretraining examples=147.6M, Modality=V2023.04 | 72.2 | — | |
| GITLPretraining examples=20M, Modality=V2023.04 | 71.9 | — | |
| CloverPretraining examples=5.5M, Modality=V2023.04 | 71.4 | — | |
| MERLOTPretraining examples=180M, Modality=V2023.04 | 69.5 | — | |
| Chat-UniViLLM Size=7B, Zero-shot=true2023.11 | 69 | 3.8 | |
| FrozenBiLMPretraining examples=10M, Modality=V2023.04 | 68.6 | — | |
| All-in-onePretraining examples=228.5M, Modality=V2023.04 | 66.3 | — | |
| ClipBERTPretraining examples=5.6M, Modality=V2023.04 | 60.3 | — | |
| QueSTModality=V2023.04 | 59.7 | — | |
| Video-ChatGPTLLM Size=7B, Zero-shot=true2023.11 | 51.4 | 3 | |
| FrozenBiLMLLM Size=1B, Zero-shot=true2023.11 | 41 | — | |
| VideoChatLLM Size=7B, Zero-shot=true2023.11 | 34.4 | 2.3 |