config/data-config.yaml

data_ingestion:
  root_dir: data/raw
  hf_dataset_name: qwedsacf/grade-school-math-instructions
  hf_dataset_split: train
  local_data_file: data/raw/data.zip
  unzip_dir: data/raw

data_preprocessing:
  root_dir: data/interim
  raw_data_path: data/raw/raw_dataset.csv
  question_key: INSTRUCTION
  answer_key: RESPONSE

data_transformation:
  root_dir: data/processed
  finetuning_data_path: data/interim/finetuning_dataset.csv
  train_data_split: 0.9
  test_data_split: 0.05
  eval_data_split: 0.05

data_validation:
  root_dir: data/processed
  STATUS_FILE: data/processed/status.txt
  ALL_REQUIRED_FILES: ["train_dataset.csv", "test_dataset.csv", "eval_dataset.csv"]