mlsanityData to 7 Oct 2026

Token Merging for Fast Stable Diffusion

arXiv 2303.17604 ↗First posted 30 Mar 2023cs.CVcited by 220

On the map: Diffusion models → Diffusion foundations and sampling

also in 1 more
Abstract (arXiv)

The landscape of image generation has been forever changed by open vocabulary diffusion models. However, at their core these models use transformers, which makes generation slow. Better implementations to increase the throughput of these transformers have emerged, but they still evaluate the entire model. In this paper, we instead speed up diffusion models by exploiting natural redundancy in generated images by merging redundant tokens. After making some diffusion-specific improvements to Token Merging (ToMe), our ToMe for Stable Diffusion can reduce the number of tokens in an existing Stable Diffusion model by up to 60% while still producing high quality images without any extra training. In the process, we speed up image generation by up to 2x and reduce memory consumption by up to 5.6x. Furthermore, this speed-up stacks with efficient implementations such as xFormers, minimally impacting quality while being up to 5.4x faster for large images. Code is available at https://github.com/dbolya/tomesd.

Key facts. Cited by 220 papers in the map (data through 7 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 17 of its 27 references that are in the map. First posted on arXiv 30 Mar 2023.

Builds on

17 of 27 references.

Closest ancestors · 3

Same direction, newest first

  1. 2022Pseudo Numerical Methods for Diffusion Models on Manifolds20 Feb
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  3. 2021Diffusion Models Beat GANs on Image Synthesis11 May

Other references · 14

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  4. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  5. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  6. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,539
  7. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,322
  8. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  9. 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,611
  10. 2022Token Merging: Your ViT But Fastercited by 995
  11. 2021DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsificationcited by 849
  12. 2021AdaViT: Adaptive Tokens for Efficient Vision Transformercited by 360
2 more
  1. 2021AdaViT: Adaptive Vision Transformers for Efficient Image Recognitioncited by 228
  2. 2021Token Pooling in Vision Transformerscited by 68

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 19; 2024: 50; 2025: 75; 2026: 76.

220 papers.

Sort:

By area:

  1. 2023DeepCache: Accelerating Diffusion Models for Freecited by 346
  2. 2024Timestep Embedding Tells: It's Time to Cache for Video Diffusion Modelcited by 245
  3. 2023Diverse Data Augmentation with Diffusions for Effective Test-time Prompt Tuningcited by 150
  4. 2024Accelerating Diffusion Transformers with Token-wise Feature Cachingcited by 143
  5. 2024Lumina-Next: Making Lumina-T2X Stronger and Faster with Next-DiTcited by 130
  6. 2025From Reusing to Forecasting: Accelerating Diffusion Models with TaylorSeerscited by 124
  7. 2023Token Fusion: Bridging the Gap between Token Pruning and Token Mergingcited by 123
  8. 2024FasterCache: Training-Free Video Diffusion Model Acceleration with High Qualitycited by 115
  9. 2023VidToMe: Video Token Merging for Zero-Shot Video Editingcited by 111
  10. 2024Adaptive Caching for Faster Video Generation with Diffusion Transformerscited by 100
  11. 2025FastVGGT: Training-Free Acceleration of Visual Geometry Transformercited by 93
  12. 2023Agent Attention: On the Integration of Softmax and Linear Attentioncited by 90
38 more
  1. 2024RADIOv2.5: Improved Baselines for Agglomerative Vision Foundation Modelscited by 88
  2. 2023DiffRate : Differentiable Compression Rate for Efficient Vision Transformerscited by 78
  3. 2024LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insightscited by 65
  4. 2025Training-free and Adaptive Sparse Attention for Efficient Long Video Generationcited by 65
  5. 2023QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Modelscited by 63
  6. 2023Zero-TPrune: Zero-Shot Token Pruning through Leveraging of the Attention Graph in Pre-Trained Transformerscited by 61
  7. 2024Dynamic Diffusion Transformercited by 60
  8. 2025OminiControl2: Efficient Conditioning for Diffusion Transformerscited by 56
  9. 2024Faster Diffusion via Temporal Attention Decompositioncited by 54
  10. 2025Token Reduction Should Go Beyond Efficiency in Generative Models -- From Vision, Language to Multimodalitycited by 49
  11. 2025EEdit: Rethinking the Spatial and Temporal Redundancy for Efficient Image Editingcited by 45
  12. 2024Diffusion Model-Based Video Editing: A Surveycited by 44
  13. 2024vid-TLDR: Training Free Token merging for Light-weight Video Transformercited by 41
  14. 2025Follow-Your-Emoji-Faster: Towards Efficient, Fine-Controllable, and Expressive Freestyle Portrait Animationcited by 41
  15. 2025Less is Enough: Training-Free Video Diffusion Acceleration via Runtime-Adaptive Cachingcited by 39
  16. 2024Looking Backward: Streaming Video-to-Video Translation with Feature Bankscited by 38
  17. 2024Attention-Driven Training-Free Efficiency Enhancement of Diffusion Modelscited by 37
  18. 2025MagCache: Fast Video Generation with Magnitude-Aware Cachecited by 36
  19. 2024Accelerating Transformers with Spectrum-Preserving Token Mergingcited by 35
  20. 2023CBQ: Cross-Block Quantization for Large Language Modelscited by 34
  21. 2025FastVAR: Linear Visual Autoregressive Modeling via Cached Token Pruningcited by 34
  22. 2023Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inferencecited by 33
  23. 2024Object-Centric Diffusion for Efficient Video Editingcited by 31
  24. 2026LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligencecited by 31
  25. 2025Shifting AI Efficiency From Model-Centric to Data-Centric Compressioncited by 29
  26. 2023HiDiffusion: Unlocking Higher-Resolution Creativity and Efficiency in Pretrained Diffusion Modelscited by 27
  27. 2025SpeCa: Accelerating Diffusion Transformers with Speculative Feature Cachingcited by 27
  28. 2024Rethinking Token-wise Feature Caching: Accelerating Diffusion Transformers with Dual Feature Cachingcited by 26
  29. 2025Unleashing Vecset Diffusion Model for Fast Shape Generationcited by 25
  30. 2026Scal3R: Scalable Test-Time Training for Large-Scale 3D Reconstructioncited by 25
  31. 2025BlockDance: Reuse Structurally Similar Spatio-Temporal Features to Accelerate Diffusion Transformerscited by 24
  32. 2024Reducing the Footprint of Multi-Vector Retrieval with Minimal Performance Impact via Token Poolingcited by 23
  33. 2025VORTA: Efficient Video Diffusion via Routing Sparse Attentioncited by 23
  34. 2025FastCache: Fast Caching for Diffusion Transformer Through Learnable Linear Approximationcited by 23
  35. 2025Forecast then Calibrate: Feature Caching as ODE for Efficient Diffusion Transformerscited by 23
  36. 2024Video Token Merging for Long-form Video Understandingcited by 22
  37. 2023High-fidelity Person-centric Subject-to-Image Synthesiscited by 21
  38. 2024Learning to Merge Tokens via Decoupled Embedding for Efficient Vision Transformerscited by 21

50 most cited of 220 All 220 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025DiffMoE: Dynamic Token Selection for Scalable Diffusion Transformers
  2. 2024Token Pruning for Caching Better: 9 Times Acceleration on Stable Diffusion for Free
  3. 2026Variable-Length Tokenization via Learnable Global Merging for Diffusion Transformers
  4. 2025TREAD: Token Routing for Efficient Architecture-agnostic Diffusion Training
  5. 2023TokenCompose: Text-to-Image Diffusion with Token-level Supervision
  6. 2025CAT Pruning: Cluster-Aware Token Pruning For Text-to-Image Diffusion Models
  7. 2025Reusing Computation in Text-to-Image Diffusion for Efficient Generation of Image Sets
  8. 2023A Survey of Diffusion Based Image Generation Models: Issues and Their Solutions