Text2Image Multi Prompt: A Collection of Cleaned and Deduplicated Prompts

A collection of several text-to-image prompt datasets aggregated from multiple sources. The data has been cleaned to remove model-specific API syntax and deduplicated, with an update in October 2023 applying minhash-based deduplication. The dataset was created by author pszemraj and last updated on Hugging Face in December 2025.

Use Cases

• Training text-to-image generation models based on a collection of cleaned prompts.
• Benchmarking prompt effectiveness across different AI models based on normalized prompt text.
• Studying prompt engineering techniques based on aggregated datasets from multiple sources.
• Creating training datasets for conditional image generation models based on textual prompts.

Strengths

• Data underwent cleaning to remove model-specific API syntax like '--ar' for Midjourney.
• Deduplication was performed, including an update in October 2023 using minhash techniques.
• Collection aggregates prompts from several source datasets, suggesting breadth of coverage.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license information are unknown, which may limit suitability assessment.
• The description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Hugging Face, aggregated from several text2image prompt datasets.

Collection Method

Collection, cleaning, normalization, and deduplication of existing datasets.

Time Range

Update in October 2023; last platform update December 2025.

Freshness

Last updated 2025-12-29 03:55:56; freshness should be verified.

Geography

null