---
dataset_info:
  features:
  - name: claim
    dtype: string
  - name: abstract
    dtype: string
  - name: abstract_id
    dtype: int64
  - name: claim_id
    dtype: int64
  - name: annotation
    dtype: string
  - name: data_version
    dtype: string
  - name: narrative
    list: string
  splits:
  - name: train
    num_bytes: 3742813
    num_examples: 6030
  download_size: 2042224
  dataset_size: 3742813
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---


From ClimateCheck2026 + up to 300 samples from Augmented_CARDS per rare class