| Self-Improving Diffusion Models with Synthetic Data | | 0.92 | | 126 | | | SIMS | 2024-08-29 |
| Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator | ✓ Link | 0.97 | | 63 | | | EDM2-S+DDO | 2025-03-03 |
| Uni-Instruct: One-step Diffusion Model through Unified Diffusion Divergence Instruction | | 1.02 | | 1 | | | Uni-Instruct | 2025-05-27 |
| Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step | ✓ Link | 1.11 | | 1 | | | SiDA-EDM | 2024-10-19 |
| Diffusion Models Are Innate One-Step Generators | ✓ Link | 1.16 | | 1 | | | GDD-I | 2024-05-31 |
| PaGoDA: Progressive Growing of a One-Step Generator from a Low-Resolution Diffusion Teacher | ✓ Link | 1.21 | | 1 | 76.47 | | PaGoDA | 2024-05-23 |
| DisCo-Diff: Enhancing Continuous Diffusion Models with Discrete Latents | ✓ Link | 1.22 | | | | | DisCo-Diff | 2024-07-03 |
| Scalable Adaptive Computation for Iterative Generation | ✓ Link | 1.23 | | | | | RIN | 2022-12-22 |
| Diffusion Models Are Innate One-Step Generators | ✓ Link | 1.42 | | 1 | | | GDD | 2024-05-31 |
| Stable Consistency Tuning: Understanding and Improving Consistency Models | ✓ Link | 1.47 | | 2 | | | SCT | 2024-10-24 |
| Cascaded Diffusion Models for High Fidelity Image Generation | | 1.48 | | | | | CDM | 2021-05-30 |
| StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets | ✓ Link | 1.51 | | 1 | | | StyleGAN-XL | 2022-02-01 |
| Score identity Distillation: Exponentially Fast Distillation of Pretrained Diffusion Models for One-Step Generation | ✓ Link | 1.524 | | 1 | | | SiD | 2024-04-05 |
| Truncated Consistency Models | | 1.62 | | 2 | | | TCM | 2024-10-18 |
| Consistency Models Made Easy | ✓ Link | 1.67 | | 2 | | | ECM-XL | 2024-06-20 |
| Constant Acceleration Flow | ✓ Link | 1.69 | | 2 | 62.03 | | CAF | 2024-11-01 |
| Consistency Trajectory Models: Learning Probability Flow ODE Trajectory of Diffusion | ✓ Link | 1.73 | | 2 | 64.29 | | CTM | 2023-10-01 |
| Diffusion Models Beat GANs on Image Synthesis | ✓ Link | 2.07 | | | | | ADM (dropout) | 2021-05-11 |
| Learning Stackable and Skippable LEGO Bricks for Efficient, Reconfigurable, and Variable-Resolution Diffusion Modeling | ✓ Link | 2.16 | | | 78.7 | | LEGO | 2023-10-10 |
| Normalizing Flows are Capable Generative Models | ✓ Link | 2.9 | 2.99 | | | | TarFlow | 2024-12-09 |
| Improved Denoising Diffusion Probabilistic Models | ✓ Link | 2.92 | 3.53 | | | | Improved DDPM | 2021-02-18 |
| Improving the Training of Rectified Flows | ✓ Link | 3.64 | | | | | 2-rectified flow++ (NFE=2) | 2024-05-30 |
| Improving the Training of Rectified Flows | ✓ Link | 4.31 | | | | | 2-rectified flow++ (NFE=1) | 2024-05-30 |
| Consistency Models | ✓ Link | 4.70 | | 2 | | | CD (Diffusion + Distillation, NFE=2) | 2023-03-02 |
| Consistency Models | ✓ Link | 6.20 | | 1 | | | CD (Diffusion + Distillation, NFE=1) | 2023-03-02 |
| Consistency Models | ✓ Link | 11.1 | | 2 | | | CT (Direct Generation, NFE=2) | 2023-03-02 |
| Consistency Models | ✓ Link | 13.0 | | 1 | | | CT (Direct Generation, NFE=1) | 2023-03-02 |
| Flow Matching for Generative Modeling | ✓ Link | 14.45 | 3.31 | | | | FM | 2022-10-06 |
| CLR-GAN: Improving GANs Stability and Quality via Consistent Latent Representation and Reconstruction | ✓ Link | 20.27 | | | | | CLR-GAN | 2024-09-30 |
| Partition-Guided GANs | ✓ Link | 21.73 | | | | | PGMGAN | 2021-04-02 |
| Composing Ensembles of Pre-trained Models via Iterative Consensus | | 29.184 | | | 34.952 | 3.766 | GLIDE + CLIP + CLS + CLS-FREE | 2022-10-20 |
| Composing Ensembles of Pre-trained Models via Iterative Consensus | | 29.219 | | | 25.926 | 5.325 | GLIDE + CLS-FREE | 2022-10-20 |
| Composing Ensembles of Pre-trained Models via Iterative Consensus | | 30.462 | | | 25.017 | 6.174 | GLIDE + CLIP | 2022-10-20 |
| Composing Ensembles of Pre-trained Models via Iterative Consensus | | 30.871 | | | 22.077 | | GLIDE + CLS | 2022-10-20 |
| Neural Flow Diffusion Models: Learnable Forward Process for Improved Diffusion Modelling | ✓ Link | | 3.2 | | | | NFDM | 2024-04-19 |
| Generative Modeling with Bayesian Sample Inference | ✓ Link | | 3.22 | | | | BSI | 2025-02-11 |
| Efficient-VDVAE: Less is more | ✓ Link | | 3.30 (different downsampling) | | | | Efficient-VDVAE | 2022-03-25 |
| Densely connected normalizing flows | ✓ Link | | 3.35 (different downsampling) | | | | DenseFlow-74-10 | 2021-06-08 |
| Neural Diffusion Models | | | 3.35 | | | | NDM | 2023-10-12 |
| Variational Diffusion Models | ✓ Link | | 3.40 | | | | VDM | 2021-07-01 |
| Combiner: Full Attention Transformer with Sparse Computation Cost | ✓ Link | | 3.42 | | | | Combiner-Axial | 2021-07-12 |
| Efficient Content-Based Sparse Attention with Routing Transformers | ✓ Link | | 3.43 | | | | Routing Transformer | 2020-03-12 |
| Generating Long Sequences with Sparse Transformers | ✓ Link | | 3.44 | | | | Sparse Transformer 59M (strided) | 2019-04-23 |
| Multi-Resolution Continuous Normalizing Flows | ✓ Link | | 3.44 | | | | MRCNF | 2021-06-15 |
| Hierarchical Transformers Are More Efficient Language Models | ✓ Link | | 3.44 | | | | Hourglass | 2021-10-26 |
| Combiner: Full Attention Transformer with Sparse Computation Cost | ✓ Link | | 3.504 | | | | Combiner-Mixture | 2021-07-12 |
| Generating High Fidelity Images with Subscale Pixel Networks and Multidimensional Upscaling | | | 3.52 | | | | SPN | 2018-12-04 |
| Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images | ✓ Link | | 3.52 | | | | Very Deep VAE | 2020-11-20 |
| PixelCNN Models with Auxiliary Variables for Natural Image Modeling | | | 3.57 | | | | PixelCNN | 2016-12-24 |
| Conditional Image Generation with PixelCNN Decoders | ✓ Link | | 3.57 | | | | Gated PixelCNN (van den Oord et al., [2016c]) | 2016-06-16 |
| Rethinking Attention with Performers | ✓ Link | | 3.636 | | | | Performer (12 layers) | 2020-09-30 |
| Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design | ✓ Link | | 3.69 | | | | Flow++ | 2019-02-01 |
| MaCow: Masked Convolutional Generative Flow | ✓ Link | | 3.69 | | | | MaCow (Var) | 2019-02-12 |
| Parallel Multiscale Autoregressive Density Estimation | | | 3.7 | | | | Parallel Multiscale | 2017-03-10 |
| MALI: A memory efficient and reverse accurate integrator for Neural ODEs | ✓ Link | | 3.71 | | | | MALI | 2021-02-09 |
| Reformer: The Efficient Transformer | ✓ Link | | 3.710 | | | | Reformer (12 layers) | 2020-01-13 |
| Rethinking Attention with Performers | ✓ Link | | 3.719 | | | | Performer (6 layers) | 2020-09-30 |
| Reformer: The Efficient Transformer | ✓ Link | | 3.740 | | | | Reformer (6 layers) | 2020-01-13 |
| MaCow: Masked Convolutional Generative Flow | ✓ Link | | 3.75 | | | | MaCow (Unf) | 2019-02-12 |
| Residual Flows for Invertible Generative Modeling | ✓ Link | | 3.757 | | | | Residual Flow | 2019-06-06 |
| Glow: Generative Flow with Invertible 1x1 Convolutions | ✓ Link | | 3.81 | | | | Glow (Kingma and Dhariwal, 2018) | 2018-07-09 |
| Axial Attention in Multidimensional Transformers | ✓ Link | | 4.032 | | | | Axial Transformer (6 layers) | 2019-12-20 |
| Enhancing the Locality and Breaking the Memory Bottleneck of Transformer on Time Series Forecasting | ✓ Link | | 4.351 | | | | Logsparse (6 layers) | 2019-06-29 |
| Consistency Trajectory Models: Learning Probability Flow ODE Trajectory of Diffusion | ✓ Link | | | 1 | 70.38 | | CTM (NFE 1) | 2023-10-01 |
| Composing Ensembles of Pre-trained Models via Iterative Consensus | | | | | | 7.952 | GLIDE +CLS | 2022-10-20 |