Research map / Image generation / Autoregressive and latent generation
Improved Masked Image Generation with Token-Critic
José Lezama, Huiwen Chang, Lu Jiang, Irfan Essa
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Token-Critic guided sampling | MaskGIT · ImageNet 256x256 | FID: 6.56 -> 4.69 | Improved |
| add Token-Critic guided sampling | MaskGIT · ImageNet 512x512 | FID: 8.48 -> 6.80 | Improved |
| add Token-Critic guided sampling | MaskGIT · ImageNet 256x256 | IS: 203.6 -> 174.5 | Hurt |
| add Token-Critic guided sampling | MaskGIT · ImageNet 512x512 | IS: 167.1 -> 182.1 | Improved |
| add Token-Critic with classifier rejection (20% acceptance) | MaskGIT · ImageNet 256x256 | FID: 4.70 -> 3.75 | Improved |
| add Token-Critic with classifier rejection (20% acceptance) | MaskGIT · ImageNet 512x512 | FID: 5.13 -> 4.03 | Improved |
| add Token-Critic with classifier rejection (20% acceptance) | MaskGIT · ImageNet 512x512 | IS: 250.7 -> 305.2 | Improved |
| add classifier rejection sampling | MaskGIT · ImageNet 256x256 | FID: 6.56 -> 4.70 | Improved |
| refine generated tokens with Token-Critic | MaskGIT · ImageNet 256x256 | FID: 6.56 to 5.73 | Improved |
| refine generated tokens with Token-Critic | MaskGIT · ImageNet 512x512 | FID: 8.48 to 7.64 | Improved |
| use Token-Critic instead of StyleGAN-XL (compare) | MaskGIT · ImageNet 256x256 | FID: 3.26 vs 3.75 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MaskGIT — MaskGIT: Masked Generative Image Transformer (CVPR 2022) · better on 3 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Halton Scheduler for Masked Generative Image Transformer (ICLR 2025 · similarity 0.88)
- Hierarchical Masked Autoregressive Models with Low-Resolution Token Pivots (ICML 2025 · similarity 0.88)
- When Worse is Better: Navigating the Compression Generation Trade-off In Visual Tokenization (NeurIPS 2025 · similarity 0.88)
- Parallelized Autoregressive Visual Generation (CVPR 2025 · similarity 0.87)
- Epsilon-VAE: Denoising as Visual Decoding (ICML 2025 · similarity 0.87)
- Efficient Generative Modeling with Residual Vector Quantization-Based Tokens (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
