dh-tutorial/dvc.yaml

stages:
  split:
    cmd: python3 src/main.py split
    deps:
    - data/CrossValidated-Questions.csv
    - src/main.py
    params:
    - seed
    outs:
    - data/test.csv.zip
    - data/train.csv.zip
  train:
    cmd: python3 src/main.py train
    deps:
    - data/test.csv.zip
    - data/train.csv.zip
    - src/main.py
    params:
    - max_features
    outs:
    - outputs/model.joblib
    - outputs/tfidf.joblib
    metrics:
    - metrics/eval.json:
        cache: false