Multiple Choice Question Answering on OBQA (dev)
86.1AccuracyCPACE
Evaluation Results
| Method | Links | |
|---|---|---|
| CPACE2023.05 | 86.1 | |
| ALBERTKnowledge Distillation=true2023.05 | 83.2 | |
| ROBERTaAIR=true2023.05 | 81.7 | |
| ALBERT2023.05 | 72.5 | |
| GenMC_T5Model Size=LARGE2022.04 | 71.6 | |
| GenMC2023.05 | 71.6 | |
| UnifiedQA_T5-FTModel Size=LARGE, Protocol=Fine-Tuning2022.04 | 70.53 | |
| UnifiedQA2023.05 | 70.5 | |
| UnifiedQA_T5Model Size=LARGE2022.04 | 70.4 | |
| RoBERTaModel Size=LARGE2022.04 | 67.8 | |
| BERT2023.05 | 64.1 | |
| GenMC_T5Model Size=BASE2022.04 | 62.07 | |
| UnifiedQA_T5-FTModel Size=BASE, Protocol=Fine-Tuning2022.04 | 60.8 | |
| UnifiedQA_T5Model Size=BASE2022.04 | 59.2 | |
| RoBERTaModel Size=BASE2022.04 | 58.67 | |
| ALBERTModel Size=BASE2022.04 | 54.53 | |
| ALBERTModel Size=LARGE2022.04 | 54.5 |