dvc.lock

schema: '2.0'
stages:
  preprocess:
    cmd: python3 src/preprocess_data.py
    deps:
    - path: data/
      md5: 3d9e62a9c241fcfcf634384a95dd4cf4.dir
      size: 10408878
      nfiles: 4
    - path: src/common/
      md5: 7a6d386c4238d1baf9fddfa805f5bd2c.dir
      size: 3093
      nfiles: 4
    - path: src/preprocess_data.py
      md5: b1e680538f0a6e8e3026c983335b2807
      size: 3414
    outs:
    - path: outputs/models/vectorizer.pkl
      md5: acbb1961b64a8847879253b6885a8b10
      size: 6039578
    - path: outputs/processed_data/
      md5: 7730ae7add1883a1d89e24abfc316802.dir
      size: 112774922
      nfiles: 6
  train:
    cmd: poetry run python3 src/train_model.py data/ outputs/nltk_corpora/ outputs/models/
      outputs/metrics/
    deps:
    - path: data/
      md5: 3d9e62a9c241fcfcf634384a95dd4cf4.dir
      size: 10408878
      nfiles: 4
    - path: src/common/
      md5: c6b67e37c37af240d4f10bdd71fc4446.dir
      size: 5284
      nfiles: 5
    - path: src/train_model.py
      md5: 2660f5588ff26344a532836484c7dcb4
      size: 4897
    outs:
    - path: outputs/metrics/model_metrics.json
      md5: df5428598ecdac9c4392da0941c3019e
      size: 96
    - path: outputs/models/
      md5: a4327a942626a42a80964ae60e442b6f.dir
      size: 27566989
      nfiles: 3
    - path: outputs/nltk_corpora/
      md5: 0f5ffa2f0ac78640a89e95ee348f9e98.dir
      size: 90497
      nfiles: 26
  build_frontend:
    cmd: docker build -t nidhogg-frontend -f docker/frontend.Dockerfile .
    deps:
    - path: docker/frontend.Dockerfile
      md5: f9f13a5601c94d57ffa932c67e76e01e
      size: 660
    - path: poetry.lock
      md5: 5bb117013faa6b076da07b81efa3457b
      size: 192281
    - path: src/frontend/
      md5: 2894d6d06bc6f0fcc591429d2425e051.dir
      size: 1322
      nfiles: 2
  build_backend:
    cmd: docker build -t nidhogg-backend -f inference/Dockerfile .
    deps:
    - path: inference/Dockerfile
      md5: 93d08356bebcc6868344888323aedfc6
      size: 700
    - path: outputs/models/tfidf_model.pkl
      md5: 40c023aa76368aa506f19c1ed2a9ef45
      size: 21526649
    - path: outputs/models/vectorizer.pkl
      md5: 3d153c9e0ca129961bc4477e3c73bea9
      size: 6039557
    - path: src/inference_api.py
      md5: 4954f2968a536938f78afbc619ed086b
      size: 2447
    - path: src/preprocess_data.py
      md5: 4418185e1646a2c3c53540b858582de9
      size: 4177
  build_inference:
    cmd: docker build -t nidhogg-inference -f docker/inference.Dockerfile .
    deps:
    - path: docker/inference.Dockerfile
      md5: 806ec508c30e792bcc9e968d0d3583fb
      size: 758
    - path: outputs/models/tfidf_model.pkl
      md5: 40c023aa76368aa506f19c1ed2a9ef45
      size: 21526649
    - path: outputs/models/vectorizer.pkl
      md5: b0cbf2e6ace054a82eccfac061fff95d
      size: 6039559
    - path: poetry.lock
      md5: 5bb117013faa6b076da07b81efa3457b
      size: 192281
    - path: src/common/
      md5: c6b67e37c37af240d4f10bdd71fc4446.dir
      size: 5284
      nfiles: 5
    - path: src/inference.py
      md5: 4085ef8ab9be6ec53f9c45d69922a4f3
      size: 1536
  build_alertmetrics:
    cmd: docker build -t alertmetrics -f docker/alertmetrics.Dockerfile .
    deps:
    - path: docker/alertmetrics.Dockerfile
      md5: e4cfdd75f91c303e7cc852b50b064346
      size: 668
    - path: poetry.lock
      md5: 5bb117013faa6b076da07b81efa3457b
      size: 192281
    - path: src/alertmetrics/
      md5: 06f14f8a8c9e5df4abccb9618d260ee8.dir
      size: 1522
      nfiles: 3