Research map / Image generation / Autoregressive and latent generation
ImageBART: Bidirectional Context with Multinomial Diffusion for Autoregressive Image Synthesis
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Increase the number of diffusion steps (ablation) | ImageBART · FFHQ | FID on upper-half completion | Helped |
| Increase the number of diffusion steps beyond T=5 (ablation) | ImageBART · FFHQ | FID on unconditional generation | No clear effect |
| Reduce the number of decoder layers relative to encoder layers (ablation) | ImageBART · LSUN-Churches | FID | No clear effect |
| Reduce the number of decoder layers relative to encoder layers (ablation) | ImageBART · LSUN-Churches | decoding speed | Helped |
| Train scales independently rather than jointly (ablation) | ImageBART · Image synthesis | training robustness | Helped |
| Use ImageBART rather than the compared text-to-image baseline | Text-conditional ImageBART · Conceptual Captions | FID | Helped |
| Use ImageBART rather than the compared text-to-image baseline | Text-conditional ImageBART · Conceptual Captions | Inception Score | Helped |
| Use ImageBART rather than the compared text-to-image baseline | Text-conditional ImageBART · Conceptual Captions | CLIP text-image cosine similarity | Helped |
| Use ImageBART rather than pixel-space approaches | ImageBART · LSUN-Cats and ImageNet | FID | Helped |
Closest papers by meaning
- UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesis (NeurIPS 2025 · similarity 0.88)
- HMAR: Efficient Hierarchical Masked Auto-Regressive Image Generation (CVPR 2025 · similarity 0.87)
- Denoising Autoregressive Transformers for Scalable Text-to-Image Generation (ICLR 2025 · similarity 0.87)
- Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction (NeurIPS 2024 · similarity 0.87)
- ControlAR: Controllable Image Generation with Autoregressive Models (ICLR 2025 · similarity 0.87)
- OmniGen-AR: AutoRegressive Any-to-Image Generation (NeurIPS 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.