image-sentence alignment on VALSE
average pairwise accuracy
- 2021-12-14 — ViLBERT 12-in-1: average pairwise accuracy 75.1
| Rank | Model | average pairwise accuracy | Average Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|---|
| 11 | GPT1 | 60.7 | – | VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena | heidelberg-nlp/valse | 2021 |
| 12 | GPT2 | 60.1 | – | VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena | heidelberg-nlp/valse | 2021 |
| 13 | LXMERT | 59.6 | 53.5 | VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena | heidelberg-nlp/valse | 2021 |
| 14 | VisualBERT | 46.4 | 48.8 | VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena | heidelberg-nlp/valse | 2021 |