relik/retriever/conf/data/dpr_like.yaml

datamodule:
  _target_: relik.retriever.goldenretriever.lightning_modules.pl_data_modules.PLDataModule
  tokenizer: ${model.language_model}
  datasets:
    train:
      _target_: relik.retriever.data.dpr.datasets.DPRDataset
      name: "train"
      passages_path: ${data_overrides.passages_path}
      path: ${data_overrides.train_path}

    val:
      - _target_: relik.retriever.data.dpr.datasets.DPRDataset
        name: "val"
        passages_path: ${data_overrides.passages_path}
        path: ${data_overrides.val_path}

    test:
      - _target_: relik.retriever.data.dpr.datasets.DPRDataset
        name: "test"
        passages_path: ${data_overrides.passages_path}
        path: ${data_overrides.test_path}

  batch_sizes:
    train: 32
    val: 64
    test: 64

  num_workers:
    train: 4
    val: 4
    test: 4