caciolai/Semantic-Role-Labeling-exploiting-Contextualized-Embeddings

Semantic Role Labeling exploiting Contextualized Embeddings

This project implements a multi-headed deep model performing argument identification and classification, two of the four subtasks Semantic Role Labeling is traditionally divided into.

For copyright reasons the training data cannot be freely published and therefore is not included in this repo.

Implementation

Here is a copy of the Colab notebook in which I implemented my model and carried out its training.

Report

Here is the report with the documentation of my model and the results it attains.

Resources used

• Google Colab
• PyTorch
• BERT
• FastText
• VerbAtlas

Authors

• Andrea Caciolai