| SiT-XL/2 + UCGM-S (E2E-VAE + 40 sampling steps + CFG) |
- FID
- 1.06
- NFE
- 80
|
Unified Continuous Generative Models
2025-05-12 |
코드 1 |
| UCGM-XL/2 (VA-VAE + 30 sampling steps, without guidance) |
- FID
- 1.21
- NFE
- 30
|
Unified Continuous Generative Models
2025-05-12 |
코드 1 |
| UCGM-XL/2 (E2E-VAE + 40 sampling steps, without guidance) |
- FID
- 1.21
- NFE
- 40
|
Unified Continuous Generative Models
2025-05-12 |
코드 1 |
| EDM2-L + DDO (SD-VAE, 25 steps, DPM-Solver-v3) |
- FID
- 1.21
- NFE
- 50
|
Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator
2025-03-03 |
코드 1 |
| LightningDiT + UCGM-S (VA-VAE + 50 sampling steps + CFG) |
- FID
- 1.21
- NFE
- 100
|
Unified Continuous Generative Models
2025-05-12 |
코드 1 |
| xAR-H |
- FID
- 1.24
|
Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation
2025-02-27 |
코드 1 |
| SiT-XL/2 + REPA-E |
- FID
- 1.26
- Inception score
- 314.9
|
REPA-E: Unlocking VAE for End-to-End Tuning of Latent Diffusion Transformers
2025-04-15 |
코드 1 |
| DDT-XL/2(22en6de 675M + guidance interval ) |
- FID
- 1.26
- Inception score
- 310.6
|
DDT: Decoupled Diffusion Transformer
2025-04-08 |
코드 1 |
| xAR-L |
- FID
- 1.28
|
Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation
2025-02-27 |
코드 1 |
| FACM (2-step) |
- FID
- 1.32
- NFE
- 2
|
Flow-Anchored Consistency Models
2025-07-04 |
코드 1 |
| GMem (with the guidance interval) |
- FID
- 1.32
|
Generative Modeling with Explicit Memory
2024-12-11 |
코드 1 |
| SiT-XL/2 + MG |
- FID
- 1.34
|
Diffusion Models without Classifier-free Guidance
2025-02-17 |
코드 1 · 코드 2 |
| AliTok-XL, autoregressive, 662M |
- FID
- 1.35
- Inception score
- 318.8
|
AliTok: Towards Sequence Modeling Alignment between Tokenizer and Autoregressive Model
2025-06-05 |
코드 1 |
| LightningDiT + VA-VAE (with the guidance interval) |
- FID
- 1.35
|
Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Models
2025-01-02 |
코드 1 · 코드 2 |
| SiD2 |
- FID
- 1.38
|
Simpler Diffusion (SiD2): 1.5 FID on ImageNet512 with pixel-space diffusion
2024-10-25 |
연결 없음 |
| SiT↓-XL/2+U-REPA (with the guidance interval) |
- FID
- 1.41
|
U-REPA: Aligning Diffusion U-Nets to ViTs
2025-03-24 |
코드 1 |
| AliTok-XL, autoregressive, 318M |
- FID
- 1.42
- Inception score
- 326.6
|
AliTok: Towards Sequence Modeling Alignment between Tokenizer and Autoregressive Model
2025-06-05 |
코드 1 |
| SiT-XL/2 + REPA (with the guidance interval) |
- FID
- 1.42
|
Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think
2024-10-09 |
코드 1 |
| RAR-XXL, autoregressive |
- FID
- 1.48
|
Randomized Autoregressive Visual Generation
2024-11-01 |
코드 1 |
| RAR-XL, autoregressive |
- FID
- 1.50
|
Randomized Autoregressive Visual Generation
2024-11-01 |
코드 1 |
| MaskBit |
- FID
- 1.52
|
MaskBit: Embedding-free Image Generation via Bit Tokens
2024-09-24 |
코드 1 |
| GMem (w/o guidance) |
- FID
- 1.53
|
Generative Modeling with Explicit Memory
2024-12-11 |
코드 1 |
| ELM |
- FID
- 1.54
|
Elucidating the design space of language models for image generation
2024-10-21 |
코드 1 |
| MAR-H, Diff Loss |
- FID
- 1.55
|
Autoregressive Image Generation without Vector Quantization
2024-06-17 |
코드 1 · 코드 2 |
| PaGoDA |
- FID
- 1.56
|
PaGoDA: Progressive Growing of a One-Step Generator from a Low-Resolution Diffusion Teacher
2024-05-23 |
코드 1 |
| ViT-XL/2 with limited Interval Guidance |
- FID
- 1.57
|
Efficient Diffusion Training via Min-SNR Weighting Strategy
2023-03-16 |
코드 1 · 코드 2 |
| MDTv2 |
- FID
- 1.58
|
MDTv2: Masked Diffusion Transformer is a Strong Image Synthesizer
2023-03-25 |
코드 1 |
| SiT-XL + SRA |
- FID
- 1.58
|
No Other Representation Component Is Needed: Diffusion Transformers Can Provide Representation Guidance by Themselves
2025-05-05 |
코드 1 |
| RobustTok-L |
- FID
- 1.60
|
Robust Latent Matters: Boosting Image Generation with Sampling Error
2025-03-11 |
코드 1 |
| DiMR-G/2R |
- FID
- 1.63
|
Alleviating Distortion in Image Generation via Multi-Resolution Diffusion Models and Time-Dependent Layer Normalization
2024-06-13 |
코드 1 |