2 code implementations • 10 Apr 2024 • Sara Kangaslahti, David Alvarez-Melis
We empirically show that varying the interpolation weights yields predictable and consistent change in the model outputs with respect to all of the controlled attributes.
no code implementations • 21 Nov 2022 • Rafal Kocielnik, Sara Kangaslahti, Shrimai Prabhumoye, Meena Hari, R. Michael Alvarez, Anima Anandkumar
Finally, we find that not all transfer scenarios yield a positive gain, which seems related to the PLMs initial performance on the target-domain task.