Sentence Prediction on BoolQ (val)
87.5AccuracyRoBERTa-L + MUPPET
Evaluation results
| Method | Links | |
|---|---|---|
| RoBERTa-L + MUPPETModel Size=Large, Backbone=RoBERTa, Training Strategy=MUPPET (Massive Multi-task Pre-finetuning)2021.01 | 87.5 | |
| BART + MUPPETBackbone=BART, Training Strategy=MUPPET (Massive Multi-task Pre-finetuning)2021.01 | 86.9 | |
| T5-11BModel Size=11B, Backbone=T52021.01 | 86.8 | |
| RoBERTa-LModel Size=Large, Backbone=RoBERTa2021.01 | 86.4 | |
| BARTBackbone=BART2021.01 | 86.2 | |
| T5-LModel Size=Large, Backbone=T52021.01 | 86.2 | |
| RoBERTa-B + MUPPETModel Size=Base, Backbone=RoBERTa, Training Strategy=MUPPET (Massive Multi-task Pre-finetuning)2021.01 | 83.8 | |
| RoBERTa-BModel Size=Base, Backbone=RoBERTa2021.01 | 82 |