README.md

---
dataset_info:
  features:
  - name: query_image_original_index
    dtype: int64
  - name: query_image_id
    dtype: string
  - name: culture_relevant_loc
    dtype: string
  - name: native_lang
    dtype: string
  - name: avg_img_quality
    dtype: float64
  - name: avg_cultural_relevance
    dtype: float64
  - name: avg_caption_fit
    dtype: float64
  - name: top_1_retrieved_text
    dtype: string
  - name: top_1_retrieved_text_original_item_index
    dtype: int64
  - name: top_1_retrieved_text_original_item_id
    dtype: string
  - name: top_1_retrieved_text_type
    dtype: string
  - name: top_1_retrieved_text_language
    dtype: string
  - name: top_1_similarity_score
    dtype: float64
  - name: gt_caption_text
    dtype: string
  - name: gt_caption_rank
    dtype: int64
  - name: gt_caption_score
    dtype: float64
  - name: gt_native_caption_text
    dtype: string
  - name: gt_native_caption_rank
    dtype: int64
  - name: gt_native_caption_score
    dtype: float64
  splits:
  - name: train
    num_bytes: 3383147
    num_examples: 7006
  download_size: 1702525
  dataset_size: 3383147
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---