Mural: Transferring LLM knowledge to image generation via Mixture-of-Transformers

• URL: http://arxiv.org/abs/2606.29013v1
• Date: Sat, 27 Jun 2026 17:18:07 GMT
• Title: Mural: Transferring LLM knowledge to image generation via Mixture-of-Transformers
• Abstract summary: Leveraging capabilities of large language models (LLMs) in text-to-image (T2I) synthesis is an important research direction.<n>We investigate whether the knowledge of a frozen LLM can be effectively utilized in T2I generation when trained exclusively on standard text-image pairs.
• Score: 10.01730911078141
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Leveraging capabilities of large language models (LLMs) in text-to-image (T2I) synthesis is an important research direction. In this work we investigate whether the knowledge of a frozen LLM can be effectively utilized in T2I generation when trained exclusively on standard text-image pairs. We integrate a frozen, reasoning-capable LLM with a diffusion-based image generator via shared attention within the Mixture-of-Transformers (MoT) architecture. Our experiments span two critical questions: (1) what degree of the LLM's intrinsic knowledge remains accessible during T2I training, and (2) what novel capabilities emerge in the resulting system. Across established benchmarks, our models achieve strong performance among unified understanding-generation systems: 0.85 on GenEval, 86.75 on DPG-Bench, and 0.66 on WISE with inference-time reasoning, using only text-image data. Remarkably, we uncover emergent behaviors absent from training data, including cross-lingual image generation, color-guided composition, emoji / ASCII scene construction, and generation directed by world knowledge. These results demonstrate that pretrained LLM knowledge can guide image synthesis under standard text-to-image training paradigms, without interleaved multimodal signals or explicit reasoning supervision. Our findings open new avenues for harnessing frozen model capabilities in resource-constrained multimodal learning.

Related papers

• Janus-Pro-R1: Advancing Collaborative Visual Comprehension and Generation via Reinforcement Learning [92.57052246970254]
  We propose to enable the collaborative co-evolution of visual comprehension and generation.<n>We introduce a two-stage training approach: supervised fine-tuning teaches the MLLM with the foundational ability to generate genuine CoT.<n>We unlock the Aha moment in visual generation, advancing MLLMs from text-to-image tasks to unified image generation.
  arXiv  Detail & Related papers  (2025-06-02T09:39:28Z)
• LMFusion: Adapting Pretrained Language Models for Multimodal Generation [81.78257799283777]
  We present LMFusion, a framework for empowering pretrained text-only large language models (LLMs) with multimodal generative capabilities.<n>Compared to methods that pretrain multimodal generative models from scratch, our experiments demonstrate that, LMFusion improves image understanding by 20% and image generation by 3.6% using only 50% of the FLOPs.
  arXiv  Detail & Related papers  (2024-12-19T18:56:24Z)
• MetaMorph: Multimodal Understanding and Generation via Instruction Tuning [57.35160715164359]
  Visual-Predictive Instruction Tuning (VPiT) is a simple and effective extension to visual instruction tuning.<n>VPiT teaches an LLM to predict discrete text tokens and continuous visual tokens from any input sequence of image and text data.<n>We train our MetaMorph model and achieve competitive performance on both visual understanding and generation.
  arXiv  Detail & Related papers  (2024-12-18T18:58:50Z)
• Mini-DALLE3: Interactive Text to Image by Prompting Large Language Models [71.49054220807983]
  A prevalent limitation persists in the effective communication with T2I models, such as Stable Diffusion, using natural language descriptions. Inspired by the recently released DALLE3, we revisit the existing T2I systems endeavoring to align human intent and introduce a new task - interactive text to image (iT2I) We present a simple approach that augments LLMs for iT2I with prompting techniques and off-the-shelf T2I models.
  arXiv  Detail & Related papers  (2023-10-11T16:53:40Z)
• Planting a SEED of Vision in Large Language Model [73.17530130368053]
  We present SEED, an elaborate image tokenizer that empowers Large Language Models (LLMs) with the ability to SEE and Draw at the same time. This version of SEED was trained in 5.7 days using only 64 V100 GPUs and 5M publicly available image-text pairs.
  arXiv  Detail & Related papers  (2023-07-16T13:41:39Z)
• DirecT2V: Large Language Models are Frame-Level Directors for Zero-Shot Text-to-Video Generation [37.25815760042241]
  This paper introduces a new framework, dubbed DirecT2V, to generate text-to-video (T2V) videos. We equip a diffusion model with a novel value mapping method and dual-softmax filtering, which do not require any additional training. The experimental results validate the effectiveness of our framework in producing visually coherent and storyful videos.
  arXiv  Detail & Related papers  (2023-05-23T17:57:09Z)
• LLM-CXR: Instruction-Finetuned LLM for CXR Image Understanding and Generation [51.08810811457617]
  vision-language alignment in LLMs is actively being researched to enable multimodal reasoning and visual IO. We develop a method for instruction-tuning an LLM only on text to gain vision-language capabilities for medical images. Our model, LLM-CXR, trained in this approach shows better image-text alignment in both CXR understanding and generation tasks.
  arXiv  Detail & Related papers  (2023-05-19T07:44:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.