dvc.lock

schema: '2.0'
stages:
  data-processing:
    cmd: python3 src/data/make_dataset.py
    deps:
    - path: data/raw/Reviews.csv
      hash: md5
      md5: c3c950ae0be8b0736477c89d052d33fd
      size: 300904694
    - path: src/data/make_dataset.py
      hash: md5
      md5: 0b75f7b740ef85c10482b003aeedfb20
      size: 2912
    outs:
    - path: data/processed/test.csv
      hash: md5
      md5: adffff99092202d7a587660718f62ea7
      size: 7212513
      isexec: true
    - path: data/processed/train.csv
      hash: md5
      md5: 487267fe39ff4ec3ec8a8eee2dfa57aa
      size: 16854697
      isexec: true
  training:
    cmd: python3 src/models/train_model.py --model roberta-base --dataset data/processed
      --output_dir SentiBites1 --logging_dir logs --epochs 1 --learning_rate 0.001
      --weight_decay 0.005
    deps:
    - path: data/processed/test.csv
      hash: md5
      md5: adffff99092202d7a587660718f62ea7
      size: 7212513
    - path: data/processed/train.csv
      hash: md5
      md5: 487267fe39ff4ec3ec8a8eee2dfa57aa
      size: 16854697
    - path: src/models/train_model.py
      hash: md5
      md5: 784773b16d9e3fc3b31d7144aecf3579
      size: 7381
    outs:
    - path: metrics/emissions.csv
      hash: md5
      md5: a5d08632fd4f815306b0fdf02082c957
      size: 1180
    - path: models/SentiBites
      hash: md5
      md5: 18b4f7b1478054e92600037d8b716956.dir
      size: 502103967
      nfiles: 10
  evaluation:
    cmd: python3 src/models/evaluate.py --model models/SentiBites --dataset data/processed
    deps:
    - path: data/processed/test.csv
      hash: md5
      md5: adffff99092202d7a587660718f62ea7
      size: 7212513
    - path: models/SentiBites
      hash: md5
      md5: 18b4f7b1478054e92600037d8b716956.dir
      size: 502103967
      nfiles: 10
    - path: src/models/evaluate_model.py
      hash: md5
      md5: c94a08fa59debe1c0405d2aa5b63908b
      size: 3515
    outs:
    - path: metrics/evaluation_scores.csv
      hash: md5
      md5: 9fd0ecf8f1e3e187bc6fe055758cc600
      size: 89