Commonsense Question Answering on CSQA Synonym Replacement WordNet-based (test)
72.1AccuracyG-DAUG-Rand
Evaluation Results
| Method | Links | |
|---|---|---|
| G-DAUG-Randperturbation=WordNet-based synonym replacement, selection_strategy=Random2020.04 | 72.1 | |
| G-DAUG-Comboperturbation=WordNet-based synonym replacement, selection_strategy=Combo (Influence + Diversity)2020.04 | 72 | |
| G-DAUG-Diversityperturbation=WordNet-based synonym replacement, selection_strategy=Diversity-based2020.04 | 71.6 | |
| G-DAUG-Influenceperturbation=WordNet-based synonym replacement, selection_strategy=Influence-based2020.04 | 71 | |
| ROBERTA (ours)perturbation=WordNet-based synonym replacement, setup=Local re-evaluation2020.04 | 69.9 | |
| BACKTRANSLATIONperturbation=WordNet-based synonym replacement2020.04 | 69 |