| coreference-resolution-on-winograd-schema | BERT-SocialIQA 340M | #25 | Accuracy: 72.5 |
| coreference-resolution-on-winograd-schema | BERT-large 340M | #34 | Accuracy: 67 |
| question-answering-on-copa | BERT-SocialIQA 340M | #30 | Accuracy: 83.4 |
| question-answering-on-copa | BERT-large 340M | #33 | Accuracy: 80.8 |
| question-answering-on-social-iqa | BERT-large 340M (fine-tuned) | #13 | Accuracy: 64.5 |
| question-answering-on-social-iqa | BERT-base 110M (fine-tuned) | #14 | Accuracy: 63.1 |
| question-answering-on-social-iqa | GPT-1 117M (fine-tuned) | #15 | Accuracy: 63 |
| question-answering-on-social-iqa | Random chance baseline | #24 | Accuracy: 33.3 |