dvc.yaml

stages:
  get_data:
    cmd: curl --remote-name-all https://www.clarin.si/repository/xmlui/bitstream/handle/11356/1724{/sloatomic2020.zip} && unzip -o sloatomic2020.zip -d ./data && rm -f sloatomic2020.zip
    outs:
      - data/sloatomic_train.tsv
      - data/sloatomic_dev.tsv
      - data/sloatomic_test.tsv.automatic_all
      - data/sloatomic_test.tsv.automatic_10k
      - data/sloatomic_test.tsv.manual_10k

  train:
    cmd: python scripts/train_comet_gpt2.py --train_data_path=./data/sloatomic_train.tsv --valid_data_path=./data/sloatomic_dev.tsv --models_dir_path=./models
    deps:
      - scripts/train_comet_gpt2.py
      - data/sloatomic_train.tsv
      - data/sloatomic_dev.tsv
    params:
      - model.model_name
      - model.IN_LEN
      - model.OUT_LEN
      - train.SEED
      - train.TRAIN_BATCH_SIZE
      - train.TRAIN_EPOCHS
      - train.LEARNING_RATE
    outs:
      - models/checkpoint_latest
    metrics:
      - plots/train_metrics.json:
          cache: false

  test:
    cmd: python scripts/test_comet_gpt2.py --test_data_path=./data/sloatomic_test.tsv.manual_10k --models_dir_path=./models/checkpoint_latest --results_dir_path=./results
    deps:
      - scripts/test_comet_gpt2.py
      - models/checkpoint_latest
      - data/sloatomic_test.tsv.manual_10k
    params:
      - test.TEST_TOP_K
    outs:
      - results/pred_generations.jsonl

  eval:
    cmd: python scripts/eval_comet_gpt2.py --pred_file_path=./results/pred_generations.jsonl
    deps:
      - scripts/eval_comet_gpt2.py
      - scripts/evaluation
      - data/sloatomic_test.tsv.manual_10k
      - results/pred_generations.jsonl
    metrics:
      - results/pred_generations_gens_scores.jsonl:
          cache: false