README.md

metadata

license: mit
language:
  - en
size_categories:
  - n>10M
task_categories:
  - text-to-image
  - image-to-image
tags:
  - lance
  - e621
  - furry
  - synthetic-captions
  - not-for-all-audiences

Caption Dataset

This dataset contains 4,068,421 captioned items exported from CaptionFlow.

Dataset Structure

Data Fields

• caption_count
• captions: List of captions/outputs
• chunk_id
• contributor_id
• dataset
• file_size
• filename
• image_format
• image_height
• image_width
• item_index
• item_key
• job_id
• metadata
• processing_time_ms
• quality_scores
• shard
• timestamp
• url