MIT Press

Skip to Main Content

Table 9:

Ablation results on the development sets of 3 GLUE datasets with a RoBERTa-base backbone.

			MNLI
	MRPC	STS-B	ID.	OOD.
Full	90.5	91.3	87.9	87.7
– attention	90.1	91.2	87.9	87.7
– concat	90.2	91.0	87.8	87.6

			MNLI
	MRPC	STS-B	ID.	OOD.
Full	90.5	91.3	87.9	87.7
– attention	90.1	91.2	87.9	87.7
– concat	90.2	91.0	87.8	87.6