# Hyperspherical Latents Improve Continuous-Token Autoregressive Generation arXiv: https://arxiv.org/abs/2509.24335 · HTML: https://mlsanity.com/paper/2509.24335/ ## Key facts - Cited by 16 papers in the map (data through 14 Oct 2026). - Placed in Image generation, text-to-image and editing, one of 14 directions in Diffusion models. - Builds on 52 of its 59 references that are in the map. - First posted on arXiv 29 Sep 2025. ## Placed in - Diffusion models / [Image generation, text-to-image and editing](https://mlsanity.com/diffusion/image-generation-text-to-image-and-editing/) ## Builds on (most cited first, 30 of 52) - [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017) - [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014) - [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020) - [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021) - [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018) - [Diffusion Models Beat GANs on Image Synthesis](https://mlsanity.com/paper/2105.05233/) (2021) - [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022) - [Classifier-Free Diffusion Guidance](https://mlsanity.com/paper/2207.12598/) (2022) - [Flow Matching for Generative Modeling](https://mlsanity.com/paper/2210.02747/) (2022) - [Neural Discrete Representation Learning](https://mlsanity.com/paper/1711.00937/) (2017) - [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021) - [Image-to-Image Translation with Conditional Adversarial Networks](https://mlsanity.com/paper/1611.07004/) (2016) - [Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow](https://mlsanity.com/paper/2209.03003/) (2022) - [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022) - [Taming Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2012.09841/) (2020) - [Improved Techniques for Training GANs](https://mlsanity.com/paper/1606.03498/) (2016) - [Perceptual Losses for Real-Time Style Transfer and Super-Resolution](https://mlsanity.com/paper/1603.08155/) (2016) - [GLU Variants Improve Transformer](https://mlsanity.com/paper/2002.05202/) (2020) - [Root Mean Square Layer Normalization](https://mlsanity.com/paper/1910.07467/) (2019) - [Generating Diverse High-Fidelity Images with VQ-VAE-2](https://mlsanity.com/paper/1906.00446/) (2019) - [MaskGIT: Masked Generative Image Transformer](https://mlsanity.com/paper/2202.04200/) (2022) - [Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction](https://mlsanity.com/paper/2404.02905/) (2024) - [Improved Precision and Recall Metric for Assessing Generative Models](https://mlsanity.com/paper/1904.06991/) (2019) - [On Layer Normalization in the Transformer Architecture](https://mlsanity.com/paper/2002.04745/) (2020) - [Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation](https://mlsanity.com/paper/2406.06525/) (2024) - [SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformers](https://mlsanity.com/paper/2401.08740/) (2024) - [Autoregressive Image Generation using Residual Quantization](https://mlsanity.com/paper/2203.01941/) (2022) - [Vector-quantized Image Modeling with Improved VQGAN](https://mlsanity.com/paper/2110.04627/) (2021) - [Autoregressive Image Generation without Vector Quantization](https://mlsanity.com/paper/2406.11838/) (2024) - [Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation](https://mlsanity.com/paper/2310.05737/) (2023) ## Built on by (most cited first, 16 of 16) - [FARMER: Flow AutoRegressive Transformer over Pixels](https://mlsanity.com/paper/2510.23588/) (2025) - [LLaMo: Scaling Pretrained Language Models for Unified Motion Understanding and Generation with Continuous Autoregressive Tokens](https://mlsanity.com/paper/2602.12370/) (2026) - [Image Generation with a Sphere Encoder](https://mlsanity.com/paper/2602.15030/) (2026) - [SimFlow: Simplified and End-to-End Training of Latent Normalizing Flows](https://mlsanity.com/paper/2512.04084/) (2025) - [BitDance: Scaling Autoregressive Generative Models with Binary Tokens](https://mlsanity.com/paper/2602.14041/) (2026) - [Visual Generation Tuning](https://mlsanity.com/paper/2511.23469/) (2025) - [Geometric Autoencoder for Diffusion Models](https://mlsanity.com/paper/2603.10365/) (2026) - [Aligning Latent Geometry for Spherical Flow Matching in Image Generation](https://mlsanity.com/paper/2605.15193/) (2026) - [Semantic-Aware Prefix Learning for Token-Efficient Image Generation](https://mlsanity.com/paper/2603.25249/) (2026) - [Search at the Cost of Sampling: Nearly-Instant Latent Space Bayesian Optimization](https://mlsanity.com/paper/2609.19476/) (2026) - [Parallel Rollout Approximation for Pixel-Space Autoregressive Image Generation](https://mlsanity.com/paper/2606.27978/) (2026) - [ArcAD: Anomaly-Rectified Calibration for Cold-Start Supervised Anomaly Detection](https://mlsanity.com/paper/2607.02252/) (2026) - [Stable Autoregressive Speech Generation with Low-Frame-Rate High-Dimensional Continuous Tokens](https://mlsanity.com/paper/2607.29363/) (2026) - [Learnable Classifier-Free Guidance Null Embeddings for Enhanced Controllable Speech Synthesis](https://mlsanity.com/paper/2609.25411/) (2026) - [Learning Holographic Reduced Representations with Clifford Variational Autoencoders](https://mlsanity.com/paper/2609.28409/) (2026) - [Sphere Encoder 2](https://mlsanity.com/paper/2610.02208/) (2026) ## Abstract (arXiv) Autoregressive (AR) models are promising for image generation, yet continuous-token AR variants often trail latent diffusion and masked-generation models. The core issue is heterogeneous variance in VAE latents, which is amplified during AR decoding, especially under classifier-free guidance (CFG), and can cause variance collapse. We propose SphereAR to address this issue. Its core design is to constrain all AR inputs and outputs -- including after CFG -- to lie on a fixed-radius hypersphere (constant ℓ2 norm), leveraging hyperspherical VAEs. Our theoretical analysis shows that hyperspherical constraint removes the scale component (the primary cause of variance collapse), thereby stabilizing AR decoding. Empirically, on ImageNet generation, SphereAR-H (943M) sets a new state of the art for AR models, achieving FID 1.34. Even at smaller scales, SphereAR-L (479M) reaches FID 1.54 and SphereAR-B (208M) reaches 1.92, matching or surpassing much larger baselines such as MAR-H (943M, 1.55) and VAR-d30 (2B, 1.92). To our knowledge, this is the first time a pure next-token AR image generator with raster order surpasses diffusion and masked-generation models at comparable parameter scales.