Multilingual Natural Language Understanding with M-BERT and SMiLER dataset for Multilingual Entity and Relation Extraction.
- Python 3.10.4
- Pip 22.0.4
- unpack archive with SMiLER data
cd data
tar -xzvf datasets.tar.gz
cd ..
- create virtual environment
python -m venv venv
source venv/bin/activate
- Main script for training execution:
python main.py --device "cuda" --max-length 256 --batch-size 32 --langs "(ru,pl,es)" --epochs 4
- Script for generating experiments:
./scripts/exps_generator.sh --id=0 --results-path="example-experiments.csv"
- Scripts for executing experiments:
./scripts/exps_executor.sh --hyperparams-path="example-experiments.csv" --start=0 --n=5 --r=3