Reflect-DiT: Inference-Time Scaling for Text-to-Image Diffusion Transformers via In-Context Reflection

Research, Akash Gokul, Aditya Grover, Konstantinos Kallidromitis, Shufan Li, Yusuke Kato, Kazuki Kozuka, A Panasonic, Arsh Koneru

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 3 functions out of this paper's own repositories and ran 1 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

Repositories linked to this paper

• jacklishufan/Reflect-DiT

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The predominant approach to advancing text-to-image generation has been training-time scaling, where larger models are trained on more data using greater computational resources. While effective, this approach is computationally expensive, leading to growing interest in inference-time scaling to improve performance. Currently, inference-time scaling for text-to-image diffusion models is largely limited to best-of-N sampling, where multiple images are generated per prompt and a selection model chooses the best output. Inspired by the recent success of reasoning models like DeepSeek-R1 in the language domain, we introduce an alternative to naive best-of-N sampling by equipping textto-image Diffusion Transformers with in-context reflection capabilities. We propose Reflect-DiT, a method that enables Diffusion Transformers to refine their generations using incontext examples of previously generated images alongside textual feedback describing necessary improvements. Instead of passively relying on random sampling and hoping for a better result in a future generation, Reflect-DiT explicitly tailors its generations to address specific aspects requiring enhancement. Experimental results demonstrate that Reflect-DiT improves performance on the GenEval benchmark (+0.19) using SANA-1.0-1.6B as a base model. Additionally, it achieves a new state-of-the-art score of 0.81 on GenEval while generating only 20 samples per prompt, surpassing the previous best score of 0.80, which was obtained using a significantly larger model (SANA-1.5-4.8B) with 2048 samples under the best-of-N approach. 1

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2503.12271")
get_code_for_paper("2503.12271")
have("2503.12271")

Connect an agent — have() is free.