---
dataset_info:
  features:
  - name: prefix_abstractions
    dtype: string
  - name: joint_abstractions
    dtype: string
  - name: sampled_percentages
    dtype: float64
  - name: abstraction_prompt
    dtype: string
  - name: abstraction_prompt_length
    dtype: int64
  - name: generated_abstractions
    list: string
  - name: train
    dtype: string
  - name: test
    dtype: string
  - name: source
    dtype: string
  - name: prompt
    dtype: string
  - name: answer
    dtype: string
  - name: num_tokens
    dtype: int64
  splits:
  - name: train
    num_bytes: 45993133
    num_examples: 480
  download_size: 10825963
  dataset_size: 45993133
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---