Sentiment Analysis

Using transformers to predict the sentiment of movie reviews from the Stanford dataset available here.

Setup

We use poetry to manage the virtual env and dependency management.

• Install poetry from here
• From project root run

poetry install

this will create a virtual env in a .venv folder and install all dependencies into it.

• To activate a shell in the venv run

poetry shell

Training scripts

To run the scripts the sample_data and sample env file can be used. For the full dataset download from here.

The 3 scripts are defined as executables in the pyproject.toml file and if this repository is installed as a package with either poetry install or pip install . the scripts can be called with simply: get_train_test_data, train and predict_and_evaluate.