---
dataset_info:
  features:
  - name: id
    dtype: int32
  - name: image
    dtype: image
  - name: item
    dtype: string
  - name: origin_country
    dtype: string
  - name: adversarial_country
    dtype: string
  - name: category
    dtype: string
  - name: difficulty
    dtype: string
  - name: perturb_method
    dtype: string
  - name: landmark_name
    dtype: string
  - name: perturb_context
    dtype: string
  - name: pair_method
    dtype: string
  - name: generated_path
    dtype: string
  - name: item_url
    dtype: string
  - name: flag_url
    dtype: string
  - name: landmark_url
    dtype: string
  - name: item_id
    dtype: int32
  - name: true_country_alternatives
    dtype: string
  - name: adversarial_country_alternatives
    dtype: string
  - name: true_name_alternatives
    dtype: string
  splits:
  - name: train
    num_bytes: 347582460
    num_examples: 5451
  download_size: 344370811
  dataset_size: 347582460
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---