config.yaml

scraping_url:
  anorganik:
    - "https://pestisida.id/pupuk_app/pendaftaran_anorganik.php?pendaftaran_pupuk_anorganPage="
  organik:
    - "https://pestisida.id/pupuk_app/pendaftaran_organik.php?pendaftaran_pupuk_organikPage="

parameters:
  similarity_threshold: 80
  model_types:
    bert: "indobenchmark/indobert-base-p2"
    roberta: "indolem/indobert-base-uncased"
  training_args:
    test_size: 0.3
    random_state: 42
  model_args:
    num_train_epochs: 3
    train_batch_size: 128
    eval_batch_size: 128
    overwrite_output_dir: True
    fp16: True
    do_lower_case: True
  class_names:
    - "Fertilizer"
    - "Pesticide"
    - "Others"

excluded_words: 
  - "kg"
  - "ml"
  - "gr"
  - "l"
  - "g"
  - "gram"
  - "grm"
  - "ltr"
  - "lt"
  - "liter"