1 code implementation • 29 Jan 2024 • Ivana Beňová, Jana Košecká, Michal Gregor, Martin Tamajka, Marcel Veselý, Marián Šimko
The dominant probing approaches rely on the zero-shot performance of image-text matching tasks to gain a finer-grained understanding of the representations learned by recent multimodal image-language transformer models.
1 code implementation • 16 Sep 2021 • Michael Crawshaw, Jana Košecká
The best MTL optimization methods require individually computing the gradient of each task's loss function, which impedes scalability to a large number of tasks.