README.md

metadata

language:
  - en
task_categories:
  - multimodal-retrieval
source_datasets:
  - tu-berlin
task_ids:
  - text-to-image
config_names:
  - query
  - corpus
  - qrels
tags:
  - information-retrieval
  - multimodal-retrieval
dataset_info:
  - config_name: qrels
    features:
      - name: query-id
        dtype: string
      - name: corpus-id
        dtype: string
      - name: score
        dtype: int8
    splits:
      - name: test
        num_examples: 20000
  - config_name: corpus
    features:
      - name: id
        dtype: string
      - name: modality
        dtype: string
      - name: text
        dtype: string
      - name: image
        dtype: image
    splits:
      - name: corpus
        num_examples: 20000
  - config_name: query
    features:
      - name: id
        dtype: string
      - name: modality
        dtype: string
      - name: text
        dtype: string
      - name: image
        dtype: image
    splits:
      - name: test
        num_examples: 250
configs:
  - config_name: qrels
    data_files:
      - split: test
        path: qrels-*
  - config_name: corpus
    data_files:
      - split: corpus
        path: corpus-*
  - config_name: query
    data_files:
      - split: test
        path: query-*