---
dataset_info:
  features:
  - name: model_inputs
    struct:
    - name: preorder_index
      sequence: int32
    - name: postorder_index
      sequence: int32
    - name: input_ids
      sequence: int32
    - name: labels
      sequence: int32
    - name: position_ids
      sequence: int32
    - name: sequence_id
      sequence: int32
  - name: loss_inputs
    struct:
    - name: parent_index
      sequence: int32
    - name: num_correct
      sequence: int32
    - name: num_rollouts
      sequence: int32
    - name: parent_rollout_index
      sequence: int32
  splits:
  - name: train
    num_bytes: 220371296
    num_examples: 4199
  download_size: 25205965
  dataset_size: 220371296
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---