Research project for the Natural Language Processing with Deep Learning course at Jagiellonian University (Summer 23).
If you want to learn more about our project's topic go to Intent_classification_for_Polish_language-proposal.pdf and read our proposal!
You can find a final paper here
- Grzegorz Przybylski (@Zwatotem)
- Paweł Fornalik (@dartespl)
- Szymon Dziedzic (@szdziedzic)
If you want to stay up to date with our research progress visit our research journal!
- Install all necessarry packages:
torch
neptune-client
transformers
datasets
tqdm
- Run
python --model <herbert, polbert, roberta> --num_epochs <num of epochs> --lr <learning rate> --test_size <test dataset size> --train_size <train dataset size> --val_size <val dataset size> --batch_size <batch size> --num_of_layers <num of layers>
If you want to use neptune.ai to track your experiments, you need to set your API token and project id as environment variables:
NEPTUNE_API_TOKEN
NEPTUNE_PROJECT
If you want to test if your changes work locally you can use test_notebook
for this purpose.