unitable/configs/dataset/concat_dataset.yaml

defaults:
  - _self_
  - augmentation: beit
  # - pubtabnet@train.d1: train_dataset
  # - pubtabnet@valid.d1: valid_dataset
  # - synthtabnet_marketing@train.d2: train_dataset
  # - synthtabnet_marketing@valid.d2: valid_dataset
  # - synthtabnet_fintabnet@train.d3: train_dataset
  # - synthtabnet_fintabnet@valid.d3: valid_dataset
  # - synthtabnet_sparse@train.d4: train_dataset
  # - synthtabnet_sparse@valid.d4: valid_dataset
  # - synthtabnet_pubtabnet@train.d5: train_dataset
  # - synthtabnet_pubtabnet@valid.d5: valid_dataset


label_type: ${trainer.label_type}
cell_limit: 10

train_dataset:
  _target_: torch.utils.data.ConcatDataset
  datasets: ${oc.dict.values:..train}

valid_dataset:
  _target_: torch.utils.data.ConcatDataset
  datasets: ${oc.dict.values:..valid}

test_dataset:
  _target_: torch.utils.data.ConcatDataset
  datasets: ${oc.dict.values:..test}