JUHE API Marketplace
Back to datasets
Dataset assetOpen Source CommunityMachine LearningText-to-Image Generation

t2i-compbench

This dataset is part of T2I-CompBench and contains text files for multiple configurations and splits, primarily for textual data. The dataset includes various configurations such as '3d_spatial_train', 'color_val', 'complex_train', 'shape_val', 'texture_train', each with specific features and splits. Features are mainly text data, and splits contain training and validation sets of varying sizes. The dataset serves as a comprehensive benchmark for text‑to‑image generation and is licensed under the MIT License. It is obtained from a GitHub repository and uploaded to the Hugging Face Hub.

Source
huggingface
Created
Oct 15, 2024
Updated
Oct 15, 2024
Signals
172 views
Availability
Linked source ready
Overview

Dataset description and usage context

T2I-CompBench Dataset Overview

Dataset Configurations

3d_spatial_train

  • Features:
    • text: string
  • Split:
    • spatial_train: 700 samples, 20,587 bytes
  • Download size: 8,050 bytes
  • Dataset size: 20,587 bytes

3d_spatial_val

  • Features:
    • text: string
  • Split:
    • spatial_val: 300 samples, 8,752 bytes
  • Download size: 4,105 bytes
  • Dataset size: 8,752 bytes

color_train

  • Features:
    • text: string
  • Split:
    • train: 700 samples, 27,656 bytes
  • Download size: 11,306 bytes
  • Dataset size: 27,656 bytes

color_val

  • Features:
    • text: string
  • Split:
    • val: 300 samples, 12,477 bytes
  • Download size: 6,029 bytes
  • Dataset size: 12,477 bytes

color_val_seen

  • Features:
    • text: string
  • Split:
    • val_seen: 200 samples, 6,801 bytes
  • Download size: 3,553 bytes
  • Dataset size: 6,801 bytes

color_val_unseen

  • Features:
    • text: string
  • Split:
    • val_unseen: 100 samples, 5,676 bytes
  • Download size: 3,461 bytes
  • Dataset size: 5,676 bytes

complex_train

  • Features:
    • text: string
  • Split:
    • train: 700 samples, 54,440 bytes
  • Download size: 25,014 bytes
  • Dataset size: 54,440 bytes

complex_train_action

  • Features:
    • text: string
  • Split:
    • train_action: 504 samples, 43,025 bytes
  • Download size: 21,082 bytes
  • Dataset size: 43,025 bytes

complex_train_spatial

  • Features:
    • text: string
  • Split:
    • train_spatial: 130 samples, 6,864 bytes
  • Download size: 3,433 bytes
  • Dataset size: 6,864 bytes

complex_train_spatialaction

  • Features:
    • text: string
  • Split:
    • train_spatialaction: 66 samples, 4,551 bytes
  • Download size: 3,377 bytes
  • Dataset size: 4,551 bytes

complex_val

  • Features:
    • text: string
  • Split:
    • val: 300 samples, 23,326 bytes
  • Download size: 12,728 bytes
  • Dataset size: 23,326 bytes

complex_val_action

  • Features:
    • text: string
  • Split:
    • val_action: 212 samples, 18,124 bytes
  • Download size: 10,653 bytes
  • Dataset size: 18,124 bytes

complex_val_spatial

  • Features:
    • text: string
  • Split:
    • val_spatial: 58 samples, 3,014 bytes
  • Download size: 2,221 bytes
  • Dataset size: 3,014 bytes

complex_val_spatialaction

  • Features:
    • text: string
  • Split:
    • val_spatialaction: 30 samples, 2,188 bytes
  • Download size: 2,439 bytes
  • Dataset size: 2,188 bytes

non_spatial_train

  • Features:
    • text: string
  • Split:
    • spatial_train: 700 samples, 43,119 bytes
  • Download size: 23,227 bytes
  • Dataset size: 43,119 bytes

non_spatial_val

  • Features:
    • text: string
  • Split:
    • spatial_val: 300 samples, 18,350 bytes
  • Download size: 10,917 bytes
  • Dataset size: 18,350 bytes

numeracy_train

  • Features:
    • text: string
  • Split:
    • train: 700 samples, 27,955 bytes
  • Download size: 13,808 bytes
  • Dataset size: 27,955 bytes

numeracy_val

  • Features:
    • text: string
  • Split:
    • val: 300 samples, 11,764 bytes
  • Download size: 6,744 bytes
  • Dataset size: 11,764 bytes

shape_train

  • Features:
    • text: string
  • Split:
    • train: 700 samples, 42,195 bytes
  • Download size: 16,720 bytes
  • Dataset size: 42,195 bytes

shape_val

  • Features:
    • text: string
  • Split:
    • val: 300 samples, 16,535 bytes
  • Download size: 8,641 bytes
  • Dataset size: 16,535 bytes

shape_val_seen

  • Features:
    • text: string
  • Split:
    • val_seen: 200 samples, 9,317 bytes
  • Download size: 4,904 bytes
  • Dataset size: 9,317 bytes

shape_val_unseen

  • Features:
    • text: string
  • Split:
    • val_unseen: 100 samples, 7,218 bytes
  • Download size: 5,038 bytes
  • Dataset size: 7,218 bytes

spatial_train

  • Features:
    • text: string
  • Split:
    • train: 700 samples, 22,964 bytes
  • Download size: 8,671 bytes
  • Dataset size: 22,964 bytes

spatial_val

  • Features:
    • text: string
  • Split:
    • val: 300 samples, 9,815 bytes
  • Download size: 4,353 bytes
  • Dataset size: 9,815 bytes

texture_train

  • Features:
    • text: string
  • Split:
    • train: 700 samples, 33,189 bytes
  • Download size: 13,122 bytes
  • Dataset size: 33,189 bytes

texture_val

  • Features:
    • text: string
  • Split:
    • val: 300 samples, 14,813 bytes
  • Download size: 6,766 bytes
  • Dataset size: 14,813 bytes

texture_val_seen

  • Features:
    • text: string
  • Split:
    • val_seen: 200 samples, 7,775 bytes
  • Download size: 3,581 bytes
  • Dataset size: 7,775 bytes

texture_val_unseen

  • Features:
    • text: string
  • Split:
    • val_unseen: 100 samples, 7,038 bytes
  • Download size: 4,238 bytes
  • Dataset size: 7,038 bytes

Dataset File Paths

3d_spatial_train

  • Split: spatial_train
  • Path: 3d_spatial_train/spatial_train-*

3d_spatial_val

  • Split: spatial_val
  • Path: 3d_spatial_val/spatial_val-*

color_train

  • Split: train
  • Path: color_train/train-*

color_val

  • Split: val
  • Path: color_val/val-*

color_val_seen

  • Split: val_seen
  • Path: color_val_seen/val_seen-*

color_val_unseen

  • Split: val_unseen
  • Path: color_val_unseen/val_unseen-*

complex_train

  • Split: train
  • Path: complex_train/train-*

complex_train_action

  • Split: train_action
  • Path: complex_train_action/train_action-*

complex_train_spatial

  • Split: train_spatial
  • Path: complex_train_spatial/train_spatial-*

complex_train_spatialaction

  • Split: train_spatialaction
  • Path: complex_train_spatialaction/train_spatialaction-*

complex_val

  • Split: val
  • Path: complex_val/val-*

complex_val_action

  • Split: val_action
  • Path: complex_val_action/val_action-*

complex_val_spatial

  • Split: val_spatial
  • Path: complex_val_spatial/val_spatial-*

complex_val_spatialaction

  • Split: val_spatialaction
  • Path: complex_val_spatialaction/val_spatialaction-*

non_spatial_train

  • Split: spatial_train
  • Path: non_spatial_train/spatial_train-*

non_spatial_val

  • Split: spatial_val
  • Path: non_spatial_val/spatial_val-*

numeracy_train

  • Split: train
  • Path: numeracy_train/train-*

numeracy_val

  • Split: val
  • Path: numeracy_val/val-*

shape_train

  • Split: train
  • Path: shape_train/train-*

shape_val

  • Split: val
  • Path: shape_val/val-*

shape_val_seen

  • Split: val_seen
  • Path: shape_val_seen/val_seen-*

shape_val_unseen

  • Split: val_unseen
  • Path: shape_val_unseen/val_unseen-*

spatial_train

  • Split: train
  • Path: spatial_train/train-*

spatial_val

  • Split: val
  • Path: spatial_val/val-*

texture_train

  • Split: train
  • Path: texture_train/train-*

texture_val

  • Split: val
  • Path: texture_val/val-*

texture_val_seen

  • Split: val_seen
  • Path: texture_val_seen/val_seen-*

texture_val_unseen

  • Split: val_unseen
  • Path: texture_val_unseen/val_unseen-*

License

  • License: MIT

Task Category

  • Task Category: text-to-image

Language

  • Language: en

Tags

  • Tags: image
Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio