Commonsense Question Answering on CSQA2 (test)
70.1AccuracyUL20B
Evaluation results
| Method | Links | |
|---|---|---|
| UL20Bmode=supervised finetuning, leaderboard submission=true2022.05 | 70.1 | |
| Lourie et al.status=best published on leaderboard2022.05 | 69.6 | |
| DECKER#Total Params.=~449M, Single-task Training=true2023.05 | 68.1 | |
| RACO#Total Params.=≥ 3B, Single-task Training=false2023.05 | 61.8 | |
| T5-3B#Total Params.=~3B, Single-task Training=false2023.05 | 60.2 | |
| ELABORAnswer predictor=T5-large, Elaboration model=GPT2-large2022.09 | 57.58 | |
| GPT-3Answer predictor=T5-large2022.09 | 56.98 | |
| vanillaAnswer predictor=T5-large2022.09 | 54.91 | |
| UNICORN#Total Params.=~770M, Single-task Training=false2023.05 | 54.9 | |
| selftalkAnswer predictor=T5-large2022.09 | 54.87 | |
| pipelineAnswer predictor=T5-large, Elaboration model=GPT2-large2022.09 | 53.54 |