schema: '2.0'
stages:
  preprocess_semeval:
    cmd: PYTHONPATH=. python absa/data/dataset.py
    deps:
    - path: data/raw/semeval_laptops
      hash: md5
      md5: 834bf3cc7b3d9216331adae54c9239c6.dir
      size: 1104115
      nfiles: 9
    - path: data/raw/semeval_restaurants
      hash: md5
      md5: 07f540b34a0561f20908aaa70012defb.dir
      size: 1625908
      nfiles: 9
    - path: absa/data/dataset.py
      hash: md5
      md5: 4199a5c7d112040180779168f7da1a4a
      size: 2761
    - path: absa/data/lang_detect.py
      hash: md5
      md5: ca10825874c1eeb68e199e0e178d8f94
      size: 1053
    - path: absa/data/preprocess.py
      hash: md5
      md5: c01401432dfeb6e32e928d47570caa21
      size: 855
    - path: absa/data/transliterate.py
      hash: md5
      md5: 2ab5e443eb6123188986e975586f42b8
      size: 1404
    outs:
    - path: data/processed/semeval_test.jsonl
      hash: md5
      md5: 2c4581091b4aeb29a0734e44fd3a5656
      size: 263576
    - path: data/processed/semeval_train.jsonl
      hash: md5
      md5: 04b5486dc6a7c62d1289804bd373eeaa
      size: 971123
  preprocess_hindi:
    cmd: PYTHONPATH=. python absa/data/hindi_loader.py
    deps:
    - path: data/raw/amazon_hindi/hindi_sentiment.jsonl
      hash: md5
      md5: 152317207b7015c46f6260a93c2cd756
      size: 1140222
    - path: absa/data/hindi_loader.py
      hash: md5
      md5: 6ffae4a0c61907d396b4e693beb6a23a
      size: 1826
    - path: absa/data/lang_detect.py
      hash: md5
      md5: ca10825874c1eeb68e199e0e178d8f94
      size: 1053
    - path: absa/data/preprocess.py
      hash: md5
      md5: c01401432dfeb6e32e928d47570caa21
      size: 855
    - path: absa/data/transliterate.py
      hash: md5
      md5: 2ab5e443eb6123188986e975586f42b8
      size: 1404
    outs:
    - path: data/processed/amazon_hindi.jsonl
      hash: md5
      md5: 29825aafaee80993c9fcb89669bb27ce
      size: 304642