Back from ICLR 2024

At ICLR, there were four papers co-authored by IDEAS NCBR researchers:

Divide and not forget: Ensemble of selectively trained experts in Continual Learning

Grzegorz Rypeść, Sebastian Cygert, Valeriya Khan, Tomasz Trzciński, Bartosz Zieliński, Bartłomiej Twardowski (all IDEAS NCBR)

https://arxiv.org/abs/2401.10191

Magnushammer: A Transformer-Based Approach to Premise Selection

Maciej Mikuła, Szymon Antoniak (IDEAS NCBR), Szymon Tworkowski, Bartosz Piotrowski (IDEAS NCBR), Albert Jiang, Jin Peng Zhou, Christian Szegedy, Łukasz Kuciński (IDEAS NCBR), Piotr Miłoś (IDEAS NCBR), Yuhuai Wu

https://arxiv.org/abs/2303.04488

MoE-Mamba: Efficient Selective State Space Models with Mixture of Experts

Maciej Pióro, Kamil Ciebiera, Krystian Król, Jan Ludziejewski, Michał Krutul, Jakub Krajewski (all IDEAS NCBR), Szymon Antoniak, Piotr Miłoś (IDEAS NCBR), Marek Cygan, Sebastian Jaszczur (IDEAS NCBR)

https://arxiv.org/abs/2401.04081

Discovering modular solutions that generalize compositionally

Simon Schug, Seijin Kobayashi, Yassir Akram, Maciej Wołczyk (IDEAS NCBR), Alexandra Proca, Johannes von Oswald, Razvan Pascanu, João Sacramento, Angelika Steger

https://arxiv.org/abs/2312.15001