Pith. sign in

REVIEW 22 cited by

Normalizing Flows are Capable Generative Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2412.06329 v3 pith:CLO3JRF6 submitted 2024-12-09 cs.CV cs.LG

classification cs.CVcs.LG
keywords modelstarflowflowsautoregressivecapableestimationgenerativemodeling
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Normalizing Flows (NFs) are likelihood-based models for continuous inputs. They have demonstrated promising results on both density estimation and generative modeling tasks, but have received relatively little attention in recent years. In this work, we demonstrate that NFs are more powerful than previously believed. We present TarFlow: a simple and scalable architecture that enables highly performant NF models. TarFlow can be thought of as a Transformer-based variant of Masked Autoregressive Flows (MAFs): it consists of a stack of autoregressive Transformer blocks on image patches, alternating the autoregression direction between layers. TarFlow is straightforward to train end-to-end, and capable of directly modeling and generating pixels. We also propose three key techniques to improve sample quality: Gaussian noise augmentation during training, a post training denoising procedure, and an effective guidance method for both class-conditional and unconditional settings. Putting these together, TarFlow sets new state-of-the-art results on likelihood estimation for images, beating the previous best methods by a large margin, and generates samples with quality and diversity comparable to diffusion models, for the first time with a stand-alone NF model. We make our code available at https://github.com/apple/ml-tarflow.

Discussion (0). Sign in to comment.

Forward citations

Cited by 22 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. WavTTS: Towards High-Quality Zero-Shot TTS via Direct Raw Waveform Modeling

    eess.AS 2026-06 unverdicted novelty 8.0 of 10

    WavTTS is the first raw-waveform diffusion TTS model using DiT flow matching and multi-scale mel supervision that approaches SOTA latent zero-shot performance while beating prior end-to-end models.

  2. Unsupervised Disentanglement Without Compromises : How Functional Orthogonality Enforces Identifiability

    cs.LG 2026-06 unverdicted novelty 7.0 of 10

    Enforcing local orthogonality on the Jacobian of the generative mapping yields identifiability for general nonlinear models when the latent domain has full combinatorial support.

  3. Coevolving Representations in Joint Image-Feature Diffusion

    cs.CV 2026-04 unverdicted novelty 7.0 of 10

    CoReDi coevolves semantic representations with the diffusion model via a jointly learned linear projection stabilized by stop-gradient, normalization, and regularization, yielding faster convergence and higher sample ...

  4. Optimal Stability of KL Divergence under Gaussian Perturbations

    cs.LG 2026-04 unverdicted novelty 7.0 of 10

    KL divergence between a general distribution and a perturbed Gaussian reference remains stable with an optimal sqrt(ε) degradation rate under finite second-moment conditions.

  5. Inherited or produced? Inferring protein production kinetics when protein counts are shaped by a cell's division history

    q-bio.QM 2025-06 unverdicted novelty 7.0 of 10

    Conditional normalizing flows approximate intractable likelihoods arising from cell division history to conclude that glc3 is mostly inactive under nutrient stress in yeast, with brief transient expression.

  6. Amortized Moment Matching for Visual Generation

    cs.LG 2026-07 accept novelty 6.0 of 10

    Amortized Fréchet Distance uses neural nets to match conditional means and covariances, yielding stronger one-step visual generators than explicit FD-loss or multi-step teachers.

  7. Signal or Noise? Understanding Generative Models for Real-World Sensor Time Series

    cs.LG 2026-07 conditional novelty 6.0 of 10

    Across 14 sensor generation settings, flow-matching models are the strongest overall baseline, while demographic conditioning, time-frequency modeling, and moderate synthetic augmentation improve hard regimes and down...

  8. Few-Step Boltzmann Generators via Scalable Likelihood Flow Maps

    cs.LG 2026-06 unverdicted novelty 6.0 of 10

    SCALLOP replaces Hutchinson's trace estimator with a scalable, vectorized likelihood distillation objective for F2D2 flow maps, cutting training variance and time while improving performance on molecular Boltzmann gen...

  9. PixelU: A U-Shaped Transformer for Efficient End-to-End Pixel Diffusion

    cs.CV 2026-06 unverdicted novelty 6.0 of 10

    PixelU is a minimalist U-shaped Diffusion Transformer for pixel-space diffusion that decouples frequencies with zero-cost skip connections and constant-channel downsampling, outperforming baselines like JiT-G at 1/3 t...

  10. MIMFlow: Integrating Masked Image Modeling with Normalizing Flows for End-to-End Image Generation

    cs.CV 2026-06 accept novelty 6.0 of 10

    End-to-end masked-image VAE plus normalizing flow yields FID 2.50 on ImageNet 256 with 128 tokens and higher linear-probe accuracy than unmasked counterparts.

  11. MIMFlow: Integrating Masked Image Modeling with Normalizing Flows for End-to-End Image Generation

    cs.CV 2026-06 unverdicted novelty 6.0 of 10

    MIMFlow is an end-to-end model that routes semantic latents through a normalizing flow while a decoder handles high-frequency pixels, reporting FID 2.50 and 71.3% linear probing accuracy on ImageNet 256x256 with 128 tokens.

  12. MIMFlow: Integrating Masked Image Modeling with Normalizing Flows for End-to-End Image Generation

    cs.CV 2026-06 unverdicted novelty 6.0 of 10

    MIMFlow uses a VAE on masked images to feed semantic latents to a normalizing flow while a decoder handles high-frequency details, reporting FID 2.50 and 71.3% linear probing on ImageNet 256x256 with 128 tokens.

  13. Learning the Universe: Posterior Reliability of Neural Generative Models in High-Dimensional Field-Level Inference of Cosmic Initial Conditions

    astro-ph.CO 2026-06 unverdicted novelty 6.0 of 10

    Generative models for cosmological field-level inference can reproduce posterior means and cross-correlations yet fail to capture correct uncertainty geometry when validated against HMC reference samples.

  14. SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation

    cs.CV 2026-05 unverdicted novelty 6.0 of 10

    SRC-Flow compresses RAE features via a Semantic Representation Compressor into a low-dimensional space, enabling normalizing flows to reach gFID 1.65 on ImageNet 256x256 and 2.07 on 512x512 while retaining exact likelihoods.

  15. SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation

    cs.CV 2026-05 conditional novelty 6.0 of 10

    A learned 32-channel semantic compression lets a normalizing flow reach gFID 1.65 on ImageNet 256×256, the best FID reported for flow-based image generation.

  16. SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation

    cs.CV 2026-05 unverdicted novelty 6.0 of 10

    SRC-Flow compresses RAE features into a low-dimensional semantic space with a Semantic Representation Compressor, enabling normalizing flows to achieve SOTA gFID scores of 1.65 and 2.07 on ImageNet 256x256 and 512x512...

  17. Normalizing Flows with Iterative Denoising

    cs.CV 2026-04 unverdicted novelty 6.0 of 10

    iTARFlow augments normalizing flows with diffusion-style iterative denoising during sampling while preserving end-to-end likelihood training, reaching competitive results on ImageNet 64/128/256.

  18. MPDiT: Multi-Patch Global-to-Local Transformer Architecture For Efficient Flow Matching and Diffusion Model

    cs.CV 2026-03 unverdicted novelty 6.0 of 10

    MPDiT uses a hierarchical multi-patch design in transformers to lower computation in diffusion models by handling coarse global features first then fine local details, plus faster-converging embeddings.

  19. PixelGen: Improving Pixel Diffusion with Perceptual Supervision

    cs.CV 2026-02 accept novelty 6.0 of 10

    PixelGen augments pixel diffusion with gated perceptual supervision to reach FID 5.11 on ImageNet-256 and GenEval 0.79 in text-to-image, narrowing the gap to latent methods without VAEs.

  20. DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation

    cs.CV 2025-11 conditional novelty 6.0 of 10

    DeCo decouples high- and low-frequency generation in pixel diffusion via a DiT plus lightweight decoder and a frequency-aware flow-matching loss, reaching FID 1.62 at 256x256 and 2.22 at 512x512 on ImageNet while clos...

  21. PixNerd: Pixel Neural Field Diffusion

    cs.CV 2025-07 conditional novelty 6.0 of 10

    PixNerd is a single-stage pixel-space diffusion transformer that uses predicted neural field weights to decode large patches, reaching 2.15 FID on ImageNet 256 without a VAE.

  22. A Generative Model-Free Form Deformation Approach for the Generation of Mesh Motions with Applications to PDE

    math.NA 2026-07 conditional novelty 5.0 of 10

    A control-point ODE flow with a Bernstein-polynomial drift can match genus-0 meshes of different topology, and a POD+autoencoder+normalizing-flow generative model on the control trajectories yields lower-dimensional s...

Pith tools