Visual Question Answering (VQA) on VCR (QA-R) dev

Metric: Accuracy (higher is better)

LeaderboardDataset
Loading chart...
#ModelAccuracyExtra DataPaperDateCode
1VL-BERTLARGE77.9NoVL-BERT: Pre-training of Generic Visual-Linguist...2019-08-22Code
2VL-BERTBASE74.4NoVL-BERT: Pre-training of Generic Visual-Linguist...2019-08-22Code
3VisualBERT73.2NoVisualBERT: A Simple and Performant Baseline for...2019-08-09Code