Table 1.

ROUGE score for six target domain datasets.

Model	Dialog	E-mail	Movie review	Debate	Social media	Science
BART fine-tuning	38.14	23.27	23.89	22.38	20.17	71.04
Adapt sum (SDPT)	42.33	24.97	25.06	24.17	22.25	72.49
Adapt sum (DAPT)	40.58	24.15	25.77	23.64	21.83	72.15
Adapt sum (TAPT)	40.69	24.12	24.84	24.01	21.65	72.28
MTL-DAS (ours) w/o sum	40.09	23.46	24.16	23.14	20.77	71.33
MTL-DAS (ours) w/o mtl	40.95	24.84	25.74	24.21	22.32	72.47
MTL-DAS (ours)	41.17	25.03	25.94	24.37	22.54	72.50