Group project for a Deep Learning course at HEC Paris with Xuwen Liu, Kanza Yous, Silvère Ngouela and Pierre Lavieille.
Chosen subject: text summarization with Bidirectional Autoregressive Transformers.
This repo presents our results in a Streamlit app. We compare BART with a smaller version of BART trained on Xsum and fine-tuned without any freezed layer.
Demo of the application:
