README.md

---
license: apache-2.0
task_categories:
- image-to-text
- question-answering
- zero-shot-classification
language:
- en
multilinguality:
  - monolingual
task_ids:
  - text-scoring
pretty_name: HL (High-Level Dataset)
size_categories:
- 10K<n<100K
annotations_creators:
- crowdsurced
annotations_origin:
- crowdsurced
dataset_info:
  features:
    - name: file_name
      dtype: string
      captions:
      - name: scene
        sequence:
          dtype: string
      - name: action
        sequence:
          dtype: string
      - name: rationale
        sequence:
          dtype: string
      - name: object
        sequence:
          dtype: string
  splits:
    - name: train
      num_examples: 13498
    - name: test
      num_examples: 1499
---