---
dataset_info:
  features:
  - name: prompt
    dtype: string
  - name: chosen_story
    dtype: string
  - name: rejected_story
    dtype: string
  - name: chosen_timestamp
    dtype: timestamp[ns]
  - name: rejected_timestamp
    dtype: timestamp[ns]
  - name: chosen_upvotes
    dtype: int64
  - name: rejected_upvotes
    dtype: int64
  splits:
  - name: train
    num_bytes: 34146032
    num_examples: 6000
  download_size: 21644830
  dataset_size: 34146032
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

Even more filtered version of SAA-Lab/LitBench-Train, obtained via clustering using NovelSelect.