StrAE: Autoencoding for Pre-Trained Embeddings using Explicit Structure
Published in EMNLP Main, 2023
We look at the benefits of explicit structure as a way of modelling composition, and the effects this has on representation learning.
Download here
Published in EMNLP Main, 2023
We look at the benefits of explicit structure as a way of modelling composition, and the effects this has on representation learning.
Download here
Published in CoNLL Shared Task: BabyLM Challenge, 2023
We take a look at whether a developmentally plausible curriculum is beneficial for pre-training.
Download here