Fill-in-the-blank Video Question Answering on LSMDC
63.5AccuracyFrozenBiLM
Evaluation Results
| Method | Links | |
|---|---|---|
| FrozenBiLMSupervision=100% (fully-supervised)2022.06 | 63.5 | |
| FrozenBiLMSupervision=10% (few-shot)2022.06 | 59.9 | |
| FrozenBiLMSupervision=1% (few-shot)2022.06 | 56.9 | |
| FrozenBiLMTraining Data=WebVid10M, Speech=true, Zero-shot=true2022.06 | 51.5 | |
| FrozenBiLMSupervision=0% (zero-shot)2022.06 | 51.5 | |
| FrozenBiLMTraining Data=WebVid10M, Zero-shot evaluation=true2022.06 | 51.5 | |
| FrozenBiLMTraining Data=WebVid10M, Speech=false, Zero-shot=true2022.06 | 50.9 | |
| ReserveTraining Data=YT-Temporal-1B, Speech=false, Zero-shot=true2022.06 | 31 | |
| ReserveTraining Data=YT-Temporal-1B, Zero-shot evaluation=true2022.06 | 31 | |
| CLIP ViT-L/14Training Data=400M image-texts, Speech=false, Zero-shot=true2022.06 | 1.2 | |
| CLIP ViT-L/14Training Data=400M image-texts, Zero-shot evaluation=true2022.06 | 1.2 | |
| RandomZero-shot=true2022.06 | 0.1 | |
| RandomTraining Data=—, Zero-shot evaluation=true2022.06 | 0.1 |