i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models

Hugging Face Daily Papers Papers
text-to-image diffusion-models open-source generative-ai image-generation research

Summary

The paper introduces i1, a 3B-parameter text-to-image diffusion model that achieves competitive performance with leading closed models while being fully open (weights, data, code). It provides insights from 300+ controlled experiments and offers a practical recipe for open research.

Diffusion models have consistently driven progress in text-to-image generation. However, it is challenging to attribute recent progress to specific modeling and data choices: state-of-the-art open-weight models provide limited ablations, and do not disclose their training data and full training details. The research community needs fully open (weights, data, and code) models as a foundation for further research; yet existing fully open models still fall significantly short of leading models in performance. In this project, we conduct a systematic investigation of the modeling and data design choices in text-to-image diffusion training and inference with 300+ controlled experiments totaling 700K+ TPU v6e hours. Our experiments highlight several empirical findings (e.g., equal weighting is a strong default for mixing curated datasets) and simple design decisions (e.g., larger text encoder adapters improve performance with minimal added parameters) for training strong models. Guided by these insights, we train i1, a 3B-parameter text-to-image diffusion model using only publicly available datasets. i1 is competitive with leading models on five representative benchmarks (GenEval, DPG, PRISM, CVTG-2K, and LongText), and outperforms the best existing fully open model by 29.5 absolute percentage points on average. We provide the i1 checkpoints, training and inference code, and the data processing pipeline. Together, our findings and the i1 recipe establish a practical foundation for future open research in text-to-image diffusion models. Our code is available at https://github.com/zlab-princeton/i1.
Original Article
View Cached Full Text

Cached at: 06/11/26, 01:40 PM

Paper page - i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models

Source: https://huggingface.co/papers/2606.11289

Abstract

A comprehensive experimental study of text-to-image diffusion models reveals key design choices and training insights leading to the development of i1, a 3B-parameter model that matches leading performance while maintaining full openness.

Diffusion models have consistently driven progress in text-to-image generation. However, it is challenging to attribute recent progress to specific modeling and data choices: state-of-the-art open-weight models provide limited ablations, and do not disclose their training data and full training details. The research community needs fully open (weights, data, and code) models as a foundation for further research; yet existing fully open models still fall significantly short of leading models in performance. In this project, we conduct a systematic investigation of the modeling and data design choices in text-to-image diffusion training and inference with 300+controlled experimentstotaling 700K+ TPU v6e hours. Our experiments highlight several empirical findings (e.g., equal weighting is a strong default for mixing curated datasets) and simple design decisions (e.g., larger text encoder adapters improve performance with minimal added parameters) for training strong models. Guided by these insights, we train i1, a 3B-parameter text-to-image diffusion model using onlypublicly available datasets. i1 is competitive with leading models on five representative benchmarks (GenEval,DPG,PRISM,CVTG-2K, andLongText), and outperforms the best existing fully open model by 29.5 absolute percentage points on average. We provide the i1 checkpoints, training and inference code, and the data processing pipeline. Together, our findings and the i1 recipe establish a practical foundation for future open research intext-to-image diffusion models. Our code is available at https://github.com/zlab-princeton/i1.

View arXiv pageView PDFProject pageGitHub38Add to collection

Get this paper in your agent:

hf papers read 2606\.11289

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper1

#### zlab-princeton/i1-3B Updatedabout 11 hours ago • 6

Datasets citing this paper2

#### zlab-princeton/i1-captions Viewer• Updatedabout 10 hours ago • 167M • 2.63k • 4 #### zlab-princeton/i1-gptedit-tfrecord Updatedabout 7 hours ago • 172

Spaces citing this paper1

Collections including this paper2

Similar Articles

Text-to-Image Models Need Less from Text Encoders Than You Think

Hugging Face Daily Papers

This paper demonstrates that text-to-image diffusion transformer models primarily rely on token merging and word order from text encoders rather than full contextual embeddings, suggesting that the image model itself decodes complex linguistic structures.

baidu/ERNIE-Image

Hugging Face Models Trending

Baidu releases ERNIE-Image, an open-weight text-to-image generation model with 8B parameters built on Diffusion Transformer architecture, achieving state-of-the-art performance among open-weight models with strong capabilities in text rendering, instruction following, and structured image generation.