params.yaml

data:
  processed_data_path: data/processed
  external_data_path: data/external
  interim_data_path: data/interim
  raw_data_path: data/raw
  embeddings_filename: draft_embeddings_paraphrase-MiniLM-L6-v2.pkl
  minhashembeddings_filename: minhashembeddings.pkl
  legal_entities: legal_entities.txt
  stopwords: stopwords.txt
  chinese_data: cn.csv
  preprocessed_df_name: preprocessed_data.csv
model:
  model_path: models
  model_name: paraphrase-minilm-l6-v2
indexing:
  vector_config_path: src/index/vector_settings.json
  elastic:
    host: localhost
    https: true
    index_name: paraphrase-minilm-l6-v2_normalized
streamlit:
  title: Company duplicate search
  local_css_path: css/style.css
  remote_css_path: https://fonts.googleapis.com/icon?family=Material+Icons
  icon: search
  slider:
    min_value: 5
    max_value: 20
    value: 5
    step: 1