relik/reader/conf/data/base.yaml

train_dataset_path: "relik/reader/data/train.jsonl"
val_dataset_path: "relik/reader/data/testa.jsonl"

train_dataset:
  _target_: "relik.reader.relik_reader_data.RelikDataset"
  transformer_model: "${model.model.transformer_model}"
  materialize_samples: False
  shuffle_candidates: 0.5
  random_drop_gold_candidates: 0.05
  noise_param: 0.0
  for_inference: False
  tokens_per_batch: 4096
  special_symbols: null

val_dataset:
  _target_: "relik.reader.relik_reader_data.RelikDataset"
  transformer_model: "${model.model.transformer_model}"
  materialize_samples: False
  shuffle_candidates: False
  for_inference: True
  special_symbols: null