With pre-trained transformer-based models continuously increasing in size, there is a dire need for parameter-efficient and modular transfer learning strategies. In this talk, we will touch base on adapter-based fine-tuning, where instead of fine-tuning all weights of a model, small neural network components are introduced at every layer. While the pre-trained parameters are frozen, only the newly introduced adapter weights are fine-tuned, achieving an encapsulation of the down-stream task information in designated parts of the model. We will demonstrate that adapters are modular components which can be composed for improvements on a target task and how they can be used for out of distribution generalization on the example of zero-shot cross-lingual transfer. Finally, we will discuss how adding modularity during pre-training can mitigate catastrophic interference and consequently lift the curse of multilinguality.
Jonas Pfeiffer is a Research Scientist at Google Research. He is interested in modular representation learning in multi-task, multilingual, and multi-modal contexts, and in low-resource scenarios. He worked on his PhD at the Technical University of Darmstadt, was a visiting researcher at the New York University and a Research Scientist Intern at Meta Research. Jonas has received the IBM PhD Research Fellowship award for 2021/2022. He has given numerous invited talks at academia, industry and ML summer schools, and has co-organized multiple workshops on multilinguality and multimodality.