dvc.lock

schema: '2.0'
stages:
  preprocess:
    cmd: ./src/data/preprocess.py data/raw/2013_Alberta_Floods-ontopic_offtopic.csv
  preprocess@0:
    cmd: head data/raw/supervisor_annotated_tweets.csv
    deps:
    - path: data/raw/supervisor_annotated_tweets.csv
      md5: 01f6a5aa8cf08b88843808265eb4ca48
      size: 134808
  preprocess@data/raw/2013_Alberta_Floods-ontopic_offtopic.csv:
    cmd: flood_detection/data/preprocess.py data/raw/2013_Alberta_Floods-ontopic_offtopic.csv
    deps:
    - path: data/raw/2013_Alberta_Floods-ontopic_offtopic.csv
      md5: c5a1d265c5f200f6d66359f926698226
      size: 1391617
    - path: flood_detection/data/preprocess.py
      md5: 50ce7c25a53efcdbfecb2b67c43ec4f6
      size: 9884
  preprocess@data/raw/2013_Queensland_Floods-ontopic_offtopic.csv:
    cmd: flood_detection/data/preprocess.py data/raw/2013_Queensland_Floods-ontopic_offtopic.csv
    deps:
    - path: data/raw/2013_Queensland_Floods-ontopic_offtopic.csv
      md5: 662337576c2dc5105fe3ef290d01b001
      size: 1332585
    - path: flood_detection/data/preprocess.py
      md5: 50ce7c25a53efcdbfecb2b67c43ec4f6
      size: 9884
  preprocess@data/tweets/supervisor_annotated_tweets.json:
    cmd: src/data/preprocess.py data/tweets/supervisor_annotated_tweets.json
    deps:
    - path: data/tweets/supervisor_annotated_tweets.json
      md5: c87c83439c5d08caee676b6ebdffd71d
      size: 5466648
    - path: src/data/preprocess.py
      md5: 7454b24d4d9f1e8cf0f6b6b3f3d13104
      size: 6658
  train:
    cmd: ./flood_detection/train/train_flood_classifier.py --env sagemaker --model
      distilbert-base-uncased --label relevant --download_model -- ./data/processed/supervisor_annotated_tweets.csv
    deps:
    - path: flood_detection/train/train_flood_classifier.py
      md5: 3281b941cc7f8ee3236f1bcdbc6d8114
      size: 9726
      isexec: true
    outs:
    - path: eval.json
      md5: b3815911bb0d4fa9bdcbd6a61939c48e
      size: 312
  preprocess@data/raw/crisis_nlp_task_dmg.tsv:
    cmd: flood_detection/data/preprocess.py data/raw/crisis_nlp_task_dmg.tsv
    deps:
    - path: data/raw/crisis_nlp_task_dmg.tsv
      md5: 55b99d09e2cc9f859110612af9b8e043
      size: 895579
    - path: flood_detection/data/preprocess.py
      md5: 50ce7c25a53efcdbfecb2b67c43ec4f6
      size: 9884
  geoparser_extract_location:
    cmd: ./src/predict/extract_location.py data/processed/supervisor_annotated_tweets.csv
    deps:
    - path: ./src/predict/extract_location.py
      md5: 047d438facd69cf4b33ae10bfe089b5c
      size: 4640
      isexec: true
    - path: data/processed/supervisor_annotated_tweets.csv
      md5: 45f91dd0ab24bc38b2ccab108dd2c1f5
      size: 1170804