%Aigaion2 BibTeX export from Idiap Publications %Thursday 21 November 2024 01:00:33 PM @INPROCEEDINGS{Mai_ACL2023_2023, author = {Mai, Florian and Pannatier, Arnaud and Fehr, Fabio and Chen, Haolin and Marelli, Fran{\c c}ois and Fleuret, Francois and Henderson, James}, projects = {Idiap, LAOS, NAST, COMPBIO}, month = jul, title = {HyperMixer: An MLP-based Low Cost Alternative to Transformers}, booktitle = {Proc. of the 61st Annual Meeting of the Association for Computational Linguistics}, year = {2023}, pages = {15632-15654}, location = {Toronto, Canada}, organization = {Association for Computational Linguistics}, isbn = {978-1-959429-72-2}, doi = {http://dx.doi.org/10.18653/v1/2023.acl-long.871}, abstract = {Transformer-based architectures are the model of choice for natural language understanding, but they come at a significant cost, as they have quadratic complexity in the input length, require a lot of training data, and can be difficult to tune. In the pursuit of lower costs, we investigate simple MLP-based architectures. We find that existing architectures such as MLPMixer, which achieves token mixing through a static MLP applied to each feature independently, are too detached from the inductive biases required for natural language understanding. In this paper, we propose a simple variant, HyperMixer, which forms the token mixing MLP dynamically using hypernetworks. Empirically, we demonstrate that our model performs better than alternative MLP-based models, and on par with Transformers. In contrast to Transformers, HyperMixer achieves these results at substantially lower costs in terms of processing time, training data, and hyperparameter tuning.}, pdf = {https://publications.idiap.ch/attachments/papers/2023/Mai_ACL2023_2023.pdf} }