RepFusion: Leveraging Multimodal Priors for Denoising in Representation Space

• URL: http://arxiv.org/abs/2606.14700v1
• Date: Fri, 12 Jun 2026 17:59:51 GMT
• Title: RepFusion: Leveraging Multimodal Priors for Denoising in Representation Space
• Abstract summary: Large language models (LLMs) are widely used in text-to-image (T2I) systems, but they are typically limited to text encoding.<n>The emergence of representation autoencoders (RAEs) shifts the generation target toward semantically structured visual representations.<n>We present RepFusion, which uses the resulting MLLM outputs as the conditioning signal for a diffusion transformer.
• Score: 31.323120799791496
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Large language models (LLMs) are widely used in text-to-image (T2I) systems, but they are typically limited to text encoding, while denoising is handled by newly trained generative backbones. The emergence of representation autoencoders (RAEs) shifts the generation target toward semantically structured visual representations, creating a latent space that is more compatible with pretrained LLM priors. Inspired by multimodal LLMs (MLLMs), where an MLP projector is sufficient to align clean visual representations with a pretrained LLM, we repurpose the MLLM itself as a noisy representation encoder, extending this mechanism from clean to noisy inputs. We present RepFusion, which uses the resulting MLLM outputs as the conditioning signal for a diffusion transformer. In controlled comparisons at similar inference budgets, RepFusion outperforms baselines that devote comparable capacity to newly initialized denoisers. These results demonstrate that MLLMs provide strong priors for denoising visual representations and that, by conditioning on evolving noisy representations, test-time compute can be productively spent on repeated MLLM conditioning in modern T2I systems.

Related papers

• Beyond Text Conditioning: A Systematic Study of MLLM-DiT Fusion for Video Generation [50.92925099642139]
  Diffusion Transformers (DiTs) have become the dominant paradigm for high-fidelity video generation, yet their ability to perform high-level semantic planning remains limited.<n>We propose BiVidGen, a hybrid framework where an MLLM first generates semantic visual tokens and a DiT renders videos conditioned on both text and these tokens via multi-layer cross-attention.
  arXiv  Detail & Related papers  (2026-08-14T07:40:41Z)
• Diffusion Large Language Models for Visual Speech Recognition [56.81307584718608]
  We propose the first Diffusion Large Language Model (DLLM)-based Visual Speech Recognition (VSR) framework.<n>DLLM-VSR commits high-confidence positions early and uses the committed tokens as bidirectional context to refine ambiguous ones.<n>We develop length-guided candidate decoding, which uses video duration to construct plausible transcript-length hypotheses.<n>The proposed method achieves a state-of-the-art WER of 19.5% on LRS3 using only its labeled training data.
  arXiv  Detail & Related papers  (2026-05-27T13:22:08Z)
• Fast-dVLM: Efficient Block-Diffusion VLM via Direct Conversion from Autoregressive VLM [58.322826487307765]
  We present Fast-dVLM, a block-diffusion-based VLM that enables KV-cache-compatible parallel decoding and speculative block decoding for inference acceleration.<n>With SGLang integration and FP8 quantization, Fast-dVLM achieves over 6x end-to-end inference speedup over the AR baseline.
  arXiv  Detail & Related papers  (2026-04-08T08:50:08Z)
• Tiny Inference-Time Scaling with Latent Verifiers [56.696619768584675]
  Verifier on Hidden States ( VHS) operates on intermediate hidden representations of Diffusion Transformer (DiT) single-step generators.<n> VHS analyzes generator features without decoding to pixel space, thereby reducing the per-candidate verification cost.<n> VHS achieves a +2.7% improvement on GenEval at the same inference-time budget.
  arXiv  Detail & Related papers  (2026-03-23T19:00:02Z)
• Language Model Based Text-to-Audio Generation: Anti-Causally Aligned Collaborative Residual Transformers [24.722647001947923]
  We propose a novel LM-based framework that employs multiple isolated transformers with causal conditioning and anti-causal alignment via reinforcement learning.<n>We show that Siren outperforms both existing LM-based and diffusion-based T2A systems, achieving state-of-the-art results.
  arXiv  Detail & Related papers  (2025-10-06T08:26:55Z)
• Controlling Multimodal LLMs via Reward-guided Decoding [17.5544679985101]
  We study the adaptation of Multimodal Large Language Models (MLLMs) through controlled decoding.<n>Our method involves building reward models for visual grounding and using them to guide the MLLM's decoding process.<n>We evaluate our method on standard object hallucination benchmarks, showing that it provides significant controllability over MLLM inference.
  arXiv  Detail & Related papers  (2025-08-15T17:29:06Z)
• Bifrost-1: Bridging Multimodal LLMs and Diffusion Models with Patch-level CLIP Latents [55.82787697101274]
  Bifrost-1 is a unified framework that bridges pretrained multimodal LLMs (MLLMs) and diffusion models.<n>By seamlessly integrating pretrained MLLMs and diffusion models with patch-level CLIP latents, our framework enables high-fidelity controllable image generation.<n>Our experiments demonstrate that Bifrost-1 achieves comparable or better performance than previous methods in terms of visual fidelity and multimodal understanding.
  arXiv  Detail & Related papers  (2025-08-08T02:38:47Z)
• LegoSLM: Connecting LLM with Speech Encoder using CTC Posteriors [22.845623101142483]
  We propose a new paradigm, LegoSLM, that bridges speech encoders and Large Language Models (LLMs)<n>Using the well-performing USM and Gemma models as an example, we demonstrate that our proposed LegoSLM method yields good performance on both ASR and speech translation tasks.
  arXiv  Detail & Related papers  (2025-05-16T15:15:19Z)
• Generative Multimodal Pretraining with Discrete Diffusion Timestep Tokens [66.02261367232256]
  Multimodal Large Language Models (MLLMs) aim to unify visual comprehension and generation.<n>Existing approaches rely on spatial visual tokens, where image patches are encoded and arranged according to a spatial order.<n>In this paper, we build a proper visual language by reconstructing diffusion timesteps to learn discrete visual tokens.
  arXiv  Detail & Related papers  (2025-04-20T16:14:28Z)
• LSTM-LM with Long-Term History for First-Pass Decoding in Conversational Speech Recognition [27.639919625398]
  LSTM language models (LSTM-LMs) have been proven to be powerful and yielded significant performance improvements over count based n-gram LMs in modern speech recognition systems. Recent work shows that it is feasible and computationally affordable to adopt the LSTM-LMs in the first-pass decoding within a dynamic (or tree based) decoder framework.
  arXiv  Detail & Related papers  (2020-10-21T23:40:26Z)

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.