575 Project

Group 1 for "Analyzing Neural Language Models" research seminar
Group members: Qingxia Guo, Lindsay Skinner, Gladys Wang, Saiya Karamali

Environment Setup

The code is able to run in Python 3.9 with appropiate requirement in environment.yml, one can also run the following command if they have annaconda.
conda env create -f environment.yml

Evaluate

User could choose to evaluate the model's performance on either ccg-tagging task or semantic role labeling task on the word embeddings that have either the ccg information or srl information removed using INLP.
Example command for finetuning the prob for srl-tagging task using embedding without ccg information:

$ ./src/eval.sh train -c -r

Example command for evaluation of ccg-tagging task performance using embedding without srl information:

$ ./src/eval.sh evaluate -s -g

Use the -h flag to get help.