Inputs alone often don’t contain enough information and providing additional context is required to faithfully react to the prompt without hallucinating or memorizing random things. We are interested in exploring whether a non-parametric component to the model would make sense.
Folder on the google drive: Modeling documents
Thomas Wolf, Victor Sanh, Lysandre Debut, Angela Fan, Teven Le Scao, Sasha Rush, Yacine Jernite, Colin Raffel, Patrick Lewis, Matthias Gallé, Minh Quang Pham, M Saiful Bari, Timo Schick, Nils Reimers, Suzana Ilic, Stephen Bach, Nicola De Cao, Marius Mosbach, Thomas Wang, Eliza Szczechla, Madison May, Jan-Christoph Kalo, Urmish Thakker, Siddharth Karamcheti, Tasnim Mohiuddin, Vishakha Sharma, Ruochen Zhang, Devamanyu Hazarika, Nickil Maveli, Iacer Calixto, Jonathan Chang, Taewoon Kim, Vikas Raunak, Animesh Prasad, Dimitri Lozeve, Jakub Zavrel, Zhiqing Sun, Manan Dey, Sandip Patil
Nils Reimers
Recently created