mlsanityData to 7 Oct 2026

Efficient Diffusion Training via Min-SNR Weighting Strategy

arXiv 2303.09556 ↗First posted 16 Mar 2023cs.CVcited by 242

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

Denoising diffusion models have been a mainstream approach for image generation, however, training these models often suffers from slow convergence. In this paper, we discovered that the slow convergence is partly due to conflicting optimization directions between timesteps. To address this issue, we treat the diffusion training as a multi-task learning problem, and introduce a simple yet effective approach referred to as Min-SNR-γ. This method adapts loss weights of timesteps based on clamped signal-to-noise ratios, which effectively balances the conflicts among timesteps. Our results demonstrate a significant improvement in converging speed, 3.4× faster than previous weighting strategies. It is also more effective, achieving a new record FID score of 2.06 on the ImageNet 256×256 benchmark using smaller architectures than that employed in previous state-of-the-art. The code is available at https://github.com/TiankaiHang/Min-SNR-Diffusion-Training.

Key facts. Cited by 242 papers in the map (data through 7 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 52 of its 61 references that are in the map. First posted on arXiv 16 Mar 2023.

Builds on

52 of 61 references.

Closest ancestors · 17

Same direction, newest first

  1. 2023On the Importance of Noise Scheduling for Diffusion Models26 Jan
  2. 2022Scalable Diffusion Models with Transformers19 Dec
  3. 2022On Distillation of Guided Diffusion Models6 Oct
  4. 2022All are Worth Words: A ViT Backbone for Diffusion Models25 Sep
  5. 2022Classifier-Free Diffusion Guidance26 Jul
  6. 2022DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps2 Jun
  7. 2022Elucidating the Design Space of Diffusion-Based Generative Models1 Jun
  8. 2022Improved Vector Quantized Diffusion Models31 May
  9. 2022Pseudo Numerical Methods for Diffusion Models on Manifolds20 Feb
  10. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  11. 2021Vector Quantized Diffusion Model for Text-to-Image Synthesis29 Nov
  12. 2021Cascaded Diffusion Models for High Fidelity Image Generation30 May
5 more
  1. 2021Diffusion Models Beat GANs on Image Synthesis11 May
  2. 2021Improved Denoising Diffusion Probabilistic Models18 Feb
  3. 2020Score-Based Generative Modeling through Stochastic Differential Equations26 Nov
  4. 2020Denoising Diffusion Implicit Models6 Oct
  5. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 35

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  4. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  5. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,539
  6. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,322
  7. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  8. 2014Deep Learning Face Attributes in the Wildcited by 3,679
  9. 2021GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Modelscited by 3,595
  10. 2022DreamFusion: Text-to-3D using 2D Diffusioncited by 3,006
  11. 2018Large Scale GAN Training for High Fidelity Natural Image Synthesiscited by 2,609
  12. 2022InstructPix2Pix: Learning to Follow Image Editing Instructionscited by 2,591
23 more
  1. 2022Video Diffusion Modelscited by 2,280
  2. 2022Imagen Video: High Definition Video Generation with Diffusion Modelscited by 1,765
  3. 2022Make-A-Video: Text-to-Video Generation without Text-Video Datacited by 1,725
  4. 2017Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semanticscited by 1,345
  5. 2020Gradient Surgery for Multi-Task Learningcited by 1,210
  6. 2022Diffusion-LM Improves Controllable Text Generationcited by 1,006
  7. 2022eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoiserscited by 921
  8. 2017GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networkscited by 816
  9. 2018Multi-Task Learning as Multi-Objective Optimizationcited by 809
  10. 2021DiffusionCLIP: Text-Guided Diffusion Models for Robust Image Manipulationcited by 640
  11. 2022GeoDiff: a Geometric Diffusion Model for Molecular Conformation Generationcited by 511
  12. 2023Zero-shot Image-to-Image Translationcited by 503
  13. 2022StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasetscited by 489
  14. 2022Paint by Example: Exemplar-based Image Editing with Diffusion Modelscited by 489
  15. 2022Phenaki: Variable Length Video Generation From Open Domain Textual Descriptioncited by 475
  16. 2022DiffuSeq: Sequence to Sequence Text Generation with Diffusion Modelscited by 439
  17. 2020Multi-Task Learning with Deep Neural Networks: A Surveycited by 374
  18. 2022Rodin: A Generative Model for Sculpting 3D Digital Avatars Using Diffusioncited by 309
  19. 2020Just Pick a Sign: Optimizing Deep Multitask Models with Gradient Sign Dropoutcited by 199
  20. 2023Noise2Music: Text-conditioned Music Generation with Diffusion Modelscited by 173
  21. 2020Gradient Vaccine: Investigating and Improving Multi-task Optimization in Massively Multilingual Modelscited by 167
  22. 2022ERNIE-ViLG 2.0: Improving Text-to-Image Diffusion Model with Knowledge-Enhanced Mixture-of-Denoising-Expertscited by 110
  23. 2022Exploring Discrete Diffusion Models for Image Captioningcited by 16

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 36; 2024: 58; 2025: 65; 2026: 83.

242 papers.

Sort:

By area:

  1. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 634
  2. 2023Zero123++: a Single Image to Consistent Multi-view Diffusion Base Modelcited by 516
  3. 2024Sora: A Review on Background, Technology, Limitations, and Opportunities of Large Vision Modelscited by 467
  4. 2024Open-Sora Plan: Open-Source Large Video Generation Modelcited by 286
  5. 2025Improving Video Generation with Human Feedbackcited by 267
  6. 2023MDTv2: Masked Diffusion Transformer is a Strong Image Synthesizercited by 260
  7. 2023Structural Pruning for Diffusion Modelscited by 193
  8. 2025History-Guided Video Diffusioncited by 165
  9. 2024Consistency Policy: Accelerated Visuomotor Policies via Consistency Distillationcited by 159
  10. 2023Text-to-Audio Generation using Instruction-Tuned LLM and Latent Diffusion Modelcited by 145
  11. 2024Consistency Models Made Easycited by 140
  12. 2024Scalable High-Resolution Pixel-Space Image Synthesis with Hourglass Diffusion Transformerscited by 116
38 more
  1. 2023Viewset Diffusion: (0-)Image-Conditioned 3D Generative Models from 2D Datacited by 113
  2. 2025DDT: Decoupled Diffusion Transformercited by 100
  3. 2024Tango 2: Aligning Diffusion-based Text-to-Audio Generations through Direct Preference Optimizationcited by 97
  4. 2023Finetuning Text-to-Image Diffusion Models for Fairnesscited by 90
  5. 2025Test-time Alignment of Diffusion Models without Reward Over-optimizationcited by 89
  6. 2024Physics-Informed Diffusion Modelscited by 88
  7. 2024Large Concept Models: Language Modeling in a Sentence Representation Spacecited by 86
  8. 2024EdgeRunner: Auto-regressive Auto-encoder for Artistic Mesh Generationcited by 84
  9. 2023CCEdit: Creative and Controllable Video Editing via Diffusion Modelscited by 80
  10. 2023SA-Solver: Stochastic Adams Solver for Fast Sampling of Diffusion Modelscited by 71
  11. 2024TangoFlux: Super Fast and Faithful Text to Audio Generation with Flow Matching and Clap-Ranked Preference Optimizationcited by 67
  12. 2024FasterDiT: Towards Faster Diffusion Transformers Training without Architecture Modificationcited by 64
  13. 2025DiffSplat: Repurposing Image Diffusion Models for Scalable Gaussian Splat Generationcited by 49
  14. 2024Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Stepcited by 47
  15. 2024Efficient Diffusion Transformer Policies with Mixture of Expert Denoisers for Multitask Learningcited by 46
  16. 2025Unified Multimodal Discrete Diffusioncited by 46
  17. 2024Improved Noise Schedule for Diffusion Trainingcited by 38
  18. 2024RealisDance: Equip controllable character animation with realistic handscited by 37
  19. 2023DiffuSeq-v2: Bridging Discrete and Continuous Text Spaces for Accelerated Seq2Seq Diffusion Modelscited by 36
  20. 2024RodinHD: High-Fidelity 3D Avatar Generation with Diffusion Modelscited by 36
  21. 2023Multi-Architecture Multi-Expert Diffusion Modelscited by 34
  22. 2023Addressing Negative Transfer in Diffusion Modelscited by 33
  23. 2023Reasoning with Latent Diffusion in Offline Reinforcement Learningcited by 33
  24. 2025REPA Works Until It Doesn't: Early-Stopped, Holistic Alignment Supercharges Diffusion Trainingcited by 33
  25. 2025Gaussian Variation Field Diffusion for High-fidelity Video-to-4D Synthesiscited by 33
  26. 2026Latent Forcing: Reordering the Diffusion Trajectory for Pixel-Space Image Generationcited by 33
  27. 2024Switch Diffusion Transformer: Synergizing Denoising Tasks with Sparse Mixture-of-Expertscited by 31
  28. 2024A Closer Look at Time Steps is Worthy of Triple Speed-Up for Diffusion Model Trainingcited by 30
  29. 2023HAAR: Text-Conditioned Generative Model of 3D Strand-based Human Hairstylescited by 29
  30. 2023Denoising Task Routing for Diffusion Modelscited by 28
  31. 2025Generative Latent Neural PDE Solver using Flow Matchingcited by 28
  32. 2025RealisDance-DiT: Simple yet Strong Baseline towards Controllable Character Animation in the Wildcited by 28
  33. 2024Denoising Task Difficulty-based Curriculum for Training Diffusion Modelscited by 26
  34. 2023ScaleLong: Towards More Stable Training of Diffusion Model via Scaling Network Long Skip Connectioncited by 25
  35. 2023ConsistencyTTA: Accelerating Diffusion-Based Text-to-Audio Generation with Consistency Distillationcited by 22
  36. 2024LiteVAE: Lightweight and Efficient Variational Autoencoders for Latent Diffusion Modelscited by 22
  37. 2023Diffusion Model with Perceptual Losscited by 21
  38. 2023Fast Diffusion Modelcited by 20

50 most cited of 242 All 242 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2023Improving Denoising Diffusion Probabilistic Models via Exploiting Shared Representations
  2. 2026Dual-Rate Diffusion: Accelerating diffusion models with an interleaved heavy-light network
  3. 2026Variance-Aware Adaptive Weighting for Diffusion Model Training
  4. 2025Demystifying Diffusion Objectives: Reweighted Losses are Better Variational Bounds
  5. 2026Spectrally-Guided Diffusion Noise Schedules
  6. 2023Analyzing and Improving the Training Dynamics of Diffusion Models
  7. 2022Dynamic Dual-Output Diffusion Models
  8. 2026Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion