dvc.yaml

stages:
  preprocess_semeval:
    cmd: PYTHONPATH=src python -m absa.data.dataset
    deps:
      - src/absa/data/dataset.py
      - src/absa/data/preprocess.py
      - src/absa/data/lang_detect.py
      - src/absa/data/transliterate.py
      - data/raw/semeval_restaurants
      - data/raw/semeval_laptops
    outs:
      - data/processed/semeval_train.jsonl
      - data/processed/semeval_test.jsonl

  preprocess_hindi:
    cmd: PYTHONPATH=src python -m absa.data.hindi_loader
    deps:
      - src/absa/data/hindi_loader.py
      - src/absa/data/preprocess.py
      - src/absa/data/lang_detect.py
      - src/absa/data/transliterate.py
      - data/raw/amazon_hindi/hindi_sentiment.jsonl
    outs:
      - data/processed/amazon_hindi.jsonl