Alchemist: Turning Public Text-to-Image Data into Generative Gold

TL;DR

This paper presents Alchemist, a new method for creating high-quality, general-purpose text-to-image fine-tuning datasets using generative models, significantly enhancing model quality while maintaining diversity.

Contribution

We introduce a novel approach leveraging pre-trained generative models to curate impactful fine-tuning datasets, and release Alchemist, a compact dataset that improves T2I model performance.

Findings

Alchemist improves generative quality across five public T2I models.

The dataset maintains diversity and style in generated images.

Fine-tuned models' weights are publicly released.

Abstract

Pre-training equips text-to-image (T2I) models with broad world knowledge, but this alone is often insufficient to achieve high aesthetic quality and alignment. Consequently, supervised fine-tuning (SFT) is crucial for further refinement. However, its effectiveness highly depends on the quality of the fine-tuning dataset. Existing public SFT datasets frequently target narrow domains (e.g., anime or specific art styles), and the creation of high-quality, general-purpose SFT datasets remains a significant challenge. Current curation methods are often costly and struggle to identify truly impactful samples. This challenge is further complicated by the scarcity of public general-purpose datasets, as leading models often rely on large, proprietary, and poorly documented internal data, hindering broader research progress. This paper introduces a novel methodology for creating general-purpose SFT datasets by leveraging a pre-trained generative model as an estimator of high-impact training samples. We apply this methodology to construct and release Alchemist, a compact (3,350 samples) yet highly effective SFT dataset. Experiments demonstrate that Alchemist substantially improves the generative quality of five public T2I models while preserving diversity and style. Additionally, we release the fine-tuned models' weights to the public.