# Improved Baselines with Representation Autoencoders

arXiv: https://arxiv.org/abs/2605.18324 · HTML: https://mlsanity.com/paper/2605.18324/

## Key facts

- Cited by 31 papers in the map (data through 14 Oct 2026).
- Builds on 59 of its 77 references that are in the map.
- First posted on arXiv 18 May 2026.

## Builds on (most cited first, 30 of 59)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [Classifier-Free Diffusion Guidance](https://mlsanity.com/paper/2207.12598/) (2022)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Improved Techniques for Training GANs](https://mlsanity.com/paper/1606.03498/) (2016)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [An Empirical Study of Training Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.02057/) (2021)
- [Towards Accurate Generative Models of Video: A New Metric & Challenges](https://mlsanity.com/paper/1812.01717/) (2018)
- [Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture](https://mlsanity.com/paper/2301.08243/) (2023)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformers](https://mlsanity.com/paper/2401.08740/) (2024)
- [Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think](https://mlsanity.com/paper/2410.06940/) (2024)
- [ELLA: Equip Diffusion Models with LLM for Enhanced Semantic Alignment](https://mlsanity.com/paper/2403.05135/) (2024)
- [Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Models](https://mlsanity.com/paper/2501.01423/) (2025)
- [BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset](https://mlsanity.com/paper/2505.09568/) (2025)
- [Deeply-Supervised Nets](https://mlsanity.com/paper/1409.5185/) (2014)
- [DreamSim: Learning New Dimensions of Human Visual Similarity using Synthetic Data](https://mlsanity.com/paper/2306.09344/) (2023)
- [Perception Encoder: The best visual embeddings are not at the output of the network](https://mlsanity.com/paper/2504.13181/) (2025)
- [Diffusion for World Modeling: Visual Details Matter in Atari](https://mlsanity.com/paper/2405.12399/) (2024)
- [Guiding a Diffusion Model with a Bad Version of Itself](https://mlsanity.com/paper/2406.02507/) (2024)
- [Diffusion Transformers with Representation Autoencoders](https://mlsanity.com/paper/2510.11690/) (2025)
- [Navigation World Models](https://mlsanity.com/paper/2412.03572/) (2024)
- [Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models](https://mlsanity.com/paper/2404.07724/) (2024)
- [NoMaD: Goal Masked Diffusion Policies for Navigation and Exploration](https://mlsanity.com/paper/2310.07896/) (2023)
- [REPA-E: Unlocking VAE for End-to-End Tuning with Latent Diffusion Transformers](https://mlsanity.com/paper/2504.10483/) (2025)
- [ViNT: A Foundation Model for Visual Navigation](https://mlsanity.com/paper/2306.14846/) (2023)
- [JourneyDB: A Benchmark for Generative Image Understanding](https://mlsanity.com/paper/2307.00716/) (2023)

## Built on by (most cited first, 30 of 31)

- [EgoWAM: World Action Models Beyond Pixels with In-the-Wild Egocentric Human Data](https://mlsanity.com/paper/2607.08436/) (2026)
- [Explorative Modeling: Unlocking a Third Pretraining Axis and End-to-End Generation](https://mlsanity.com/paper/2607.27372/) (2026)
- [Distilling Drifting Transformers with Representation Autoencoders](https://mlsanity.com/paper/2606.15553/) (2026)
- [Towards Physics of Multimodal Pretraining: Knowledge Flow, Modality Synergy, Early Unification, and Recipes](https://mlsanity.com/paper/2608.05000/) (2026)
- [V-RAE: Rethinking Video Latent Spaces for Generation](https://mlsanity.com/paper/2608.13556/) (2026)
- [JLT: Clean-Latent Prediction in Latent Diffusion Transformers](https://mlsanity.com/paper/2605.27102/) (2026)
- [MIMFlow: Integrating Masked Image Modeling with Normalizing Flows for End-to-End Image Generation](https://mlsanity.com/paper/2606.26016/) (2026)
- [UniSpace: Unified Visual Representation and Scalable Multimodal Modeling](https://mlsanity.com/paper/2608.08676/) (2026)
- [FuseReg: Regularizing Layer Fusion Mitigates the Reconstruction-Generation Gap in Representation Autoencoders](https://mlsanity.com/paper/2609.31620/) (2026)
- [Rethinking Representations for World-Action Modeling](https://mlsanity.com/paper/2609.38163/) (2026)
- [Slot-RAE: Streamlining Object-Centric Learning via Direct Representation Auto-Encoders](https://mlsanity.com/paper/2607.11196/) (2026)
- [dRAE: Representation Autoencoder with Hyper-Spherical Codes](https://mlsanity.com/paper/2607.22148/) (2026)
- [Where Does Generative Difficulty Reside? An Empirical Study of Target Representations](https://mlsanity.com/paper/2608.00626/) (2026)
- [GAE: Learning a Geometry-Native Latent Space for 3D-Consistent World Generation](https://mlsanity.com/paper/2609.24981/) (2026)
- [On the Diffusibility of High-Dimensional Latents](https://mlsanity.com/paper/2609.28473/) (2026)
- [Residual-Stream Burden Shapes Representation Learning in Diffusion Transformers](https://mlsanity.com/paper/2609.33895/) (2026)
- [What Visual Generators Need from Teachers: Rethinking Representation Alignment](https://mlsanity.com/paper/2609.34732/) (2026)
- [One from Infinity: Actualizing Futures from Pretrained World Models into Robot Actions](https://mlsanity.com/paper/2609.36413/) (2026)
- [The Domain Is a Residue: Adapting Self-Supervised Features, Not Generators](https://mlsanity.com/paper/2609.37330/) (2026)
- [HiRAE: Hierarchical Representation Autoencoding with Residual Budgets](https://mlsanity.com/paper/2609.37775/) (2026)
- [Rethinking Generative Image Compression at Extremely Low Bitrates](https://mlsanity.com/paper/2609.39315/) (2026)
- [SAGE: Salient Factor Discovery and Generation with Visual Foundation Representations](https://mlsanity.com/paper/2609.39635/) (2026)
- [Diffusable Latents from Structure-Agnostic Distillation](https://mlsanity.com/paper/2609.39657/) (2026)
- [Just Align bmx: Aligning Predictions, Not Representations](https://mlsanity.com/paper/2610.00600/) (2026)
- [Super-Resolution in The Right Latent Space: A Frozen Vision-Foundation Substrate](https://mlsanity.com/paper/2610.04781/) (2026)
- [A Strong Baseline for Evaluating Vision Encoders in Multimodal Large Language Models](https://mlsanity.com/paper/2610.05413/) (2026)
- [Pooling Representation Autoencoders for Efficient Diffusion](https://mlsanity.com/paper/2610.09242/) (2026)
- [Learning Representations to Guide Pixel Diffusion Transformers](https://mlsanity.com/paper/2610.12978/) (2026)
- [Decoding Register Tokens Preserves Global Scene Details](https://mlsanity.com/paper/2610.13299/) (2026)
- [DIME: Scaling Facial Representation Learning via Differential Masked Autoencoding](https://mlsanity.com/paper/2610.17015/) (2026)

## Abstract (arXiv)

Representation Autoencoders (RAE) replace traditional VAE with pretrained vision encoders. In this paper, we systematically investigate several design choices and find three insights which simplify and improve RAE. First, we study a generalized formulation where the representation is defined as sum of the last k encoder layers rather than solely the final layer. This simple change greatly improves reconstruction without encoder finetuning or specialized data (e.g., text, faces). Second, we study the prevalent assumption that RAE (using pretrained representation as encoder) replaces representation alignment (REPA), which distills the same representation to intermediate layers instead. Through large-scale empirical analysis, we uncover a surprising finding: RAE and REPA exhibit complementary working mechanisms, allowing the same representation to be used as both encoder and target for intermediate diffusion layers. Finally, the original RAE struggles with classifier-free guidance (CFG) and requires training a second, weaker diffusion model for AutoGuidance (AG). We show that REPA itself can be viewed as x-prediction in RAE latent space. By simply re-parameterizing the output of the DiT model, it can provide guidance for "free". Overall, RAEv2 leads to more than 10x faster convergence over the original RAE, achieving a state-of-the-art gFID of 1.06 in just 80 epochs on ImageNet-256. On FDr6, RAEv2 achieves a state-of-the-art 2.17 at just 80 epochs compared to the previous best 3.26 (800 epochs) without any post-training. This motivates EPFID@k (epochs to reach unguided gFID < k) as a measure of training efficiency. RAEv2 attains an EPFID@2 of 35 epochs, versus 177 for the original RAE. We also validate our approach across diverse settings for text-to-image generation and navigation world models, showing consistent improvements. The code is available at https://raev2.github.io.