Research map / Image generation / Autoregressive and latent generation
Rethinking Visual Autoregressive Sampling with Information-Grounding Guidance
Ky Dan Nguyen, Hoang Lam Tran, Anh-Dung Dinh, Daochang Liu, Weidong Cai, Xiuying Wang, Chang Xu
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 |
|---|---|---|---|
| replace CFG with IGG | VAR-d36 · ImageNet 512x512 | FID: 2.61 -> 2.56 | Improved |
| replace CFG with IGG | VAR-d30 · ImageNet 256x256 | FID: 1.93 -> 1.92 | No clear effect |
| replace CFG with IGG | VAR-d36 · ImageNet 512x512 | IS: 293.7 -> 314.3 | Improved |
| replace CFG with IGG | Switti · MJHQ-30K | FID: 9.50 -> 7.09 | Improved |
| replace CFG with IGG | Switti · GenEval overall | GenEval: 0.62 -> 0.64 | Improved |
| replace CFG with IGG | Switti · MJHQ-30K | ImageReward: 1.15 -> 1.14 | No clear effect |
| replace CFG with IGG | VAR-CLIP · COCO-30K | FID: 10.95 -> 10.93 | No clear effect |
| replace CFG with IGG | Switti · GenEval two-object | accuracy: 0.753 -> 0.808 | Improved |
| replace CFG with IGG | VAR-CLIP · GenEval counting | accuracy: 0.113 -> 0.094 | Hurt |
| replace CFG with IGG | VAR-d36 · guidance evenness and divergence | PEI: 0.741 -> 0.665 | Improved |
| mix CFG and IGG with equal weights (ablation) | VAR-d30 IGG · ImageNet 256x256 | FID: 1.92 -> 5.48 | Hurt |
| use fixed guidance schedule instead of linear schedule (ablation) | VAR-d30 IGG · ImageNet 256x256 | FID: 1.92 -> 4.29 | Hurt |
Closest papers by meaning
- SSG: Scaled Spatial Guidance for Multi-Scale Visual Autoregressive Generation (ICLR 2026 · similarity 0.92)
- ARVAR: Accelerating Visual Autoregressive Model via Attention Retrospect (ECCV 2026 · similarity 0.90)
- Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction (NeurIPS 2024 · similarity 0.90)
- NFIG: Multi-Scale Autoregressive Image Generation via Frequency Ordering (NeurIPS 2025 · similarity 0.89)
- HMAR: Efficient Hierarchical Masked Auto-Regressive Image Generation (CVPR 2025 · similarity 0.88)
- FlexVAR: Flexible Visual Autoregressive Modeling without Residual Prediction (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.
