mlsanityData to 7 Oct 2026

RAC: Rectified Flow Auto Coder

arXiv 2603.05925 ↗First posted 6 Mar 2026cs.CV cs.AIcited by 0

On the map: Diffusion models → Diffusion for compression and wireless

Abstract (arXiv)

In this paper, we propose a Rectified Flow Auto Coder (RAC) inspired by Rectified Flow to replace the traditional VAE: 1. It achieves multi-step decoding by applying the decoder to flow timesteps. Its decoding path is straight and correctable, enabling step-by-step refinement. 2. The model inherently supports bidirectional inference, where the decoder serves as the encoder through time reversal (hence Coder rather than encoder or decoder), reducing parameter count by nearly 41%. 3. This generative decoding method improves generation quality since the model can correct latent variables along the path, partially addressing the reconstruction--generation gap. Experiments show that RAC achieves a Pareto improvement over SOTA VAEs, where even a 10× parameter-reduced decoder exceeds full-scale VAE performance in both reconstruction and generation quality, validating the effectiveness of our approach.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Diffusion for compression and wireless, one of 14 directions in Diffusion models. Builds on 32 of its 39 references that are in the map. First posted on arXiv 6 Mar 2026.

Builds on

32 of 39 references.

References · 32

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  3. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  4. 2022Scalable Diffusion Models with Transformerscited by 5,931
  5. 2022Flow Matching for Generative Modelingcited by 5,426
  6. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,290
  7. 2018Neural Ordinary Differential Equationscited by 3,733
  8. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,550
  9. 2020Taming Transformers for High-Resolution Image Synthesiscited by 3,229
  10. 2016Density estimation using Real NVPcited by 2,029
  11. 2023Consistency Modelscited by 1,778
  12. 2018Glow: Generative Flow with Invertible 1x1 Convolutionscited by 1,539
20 more
  1. 2025FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Spacecited by 960
  2. 2024SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformerscited by 707
  3. 2024Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Thinkcited by 666
  4. 2020NVAE: A Deep Hierarchical Variational Autoencodercited by 635
  5. 2021Diffusion Autoencoders: Toward a Meaningful and Decodable Representationcited by 447
  6. 2025Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Modelscited by 376
  7. 2025Diffusion Transformers with Representation Autoencoderscited by 275
  8. 2025REPA-E: Unlocking VAE for End-to-End Tuning with Latent Diffusion Transformerscited by 212
  9. 2025FlexTok: Resampling Images into 1D Token Sequences of Flexible Lengthcited by 105
  10. 2024Normalizing Flows are Capable Generative Modelscited by 89
  11. 2025EQ-VAE: Equivariance Regularized Latent Space for Improved Generative Image Modelingcited by 85
  12. 2025A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Modelscited by 73
  13. 2025VARGPT: Unified Understanding and Generation in a Visual Autoregressive Multimodal Large Language Modelcited by 44
  14. 2022Rethinking the Objectives of Vector-Quantized Tokenizers for Image Synthesiscited by 31
  15. 2025Diffusion Autoencoders are Scalable Image Tokenizerscited by 28
  16. 2023SignDiff: Diffusion Model for American Sign Language Productioncited by 25
  17. 2025"Principal Components" Enable A New Language of Imagescited by 25
  18. 2024Epsilon-VAE: Denoising as Visual Decodingcited by 16
  19. 2024Hybrid SD: Edge-Cloud Collaborative Inference for Stable Diffusion Modelscited by 11
  20. 2025SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenizationcited by 7

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Few-step Generative Models as Lossy Compression
  2. 2025Speculative Decoding and Beyond: An In-Depth Survey of Techniques
  3. 2026ResARC: Residual-Aware AutoRegressive Coding for Ultra-Low Bitrate Image Compression
  4. 2026The Recurrent Transformer: Greater Effective Depth and Efficient Decoding
  5. 2025Speculative Coupled Decoding for Training-Free Lossless Acceleration of Autoregressive Visual Generation
  6. 2026GVCC: Zero-Shot Video Compression via Codebook-Driven Stochastic Rectified Flow
  7. 2026Recursive Flow: A Generative Framework for MIMO Channel Estimation
  8. 2025VVS: Accelerating Speculative Decoding for Visual Autoregressive Generation via Partial Verification Skipping