In order to smooth version dependency hell, you can use this environment:

conda env create -f environment.yml

conda activate dl4nlpenv

First, run downloaddata.sh to download the movies_genres.csv file into the data folder. The IMDB class inside imdb.py will split this data into separate train, validation, and test data files.

Run train.py with desired arguments, example: python train.py --batch_size 32 --device cpu --max_batches_per_epoch 20

Condition movie plot generation on genre