Research map / Image generation / Autoregressive and latent generation
ToProVAR: Efficient Visual Autoregressive Modeling via Tri-Dimensional Entropy-Aware Semantic Analysis and Sparsity Optimization
Jiayu Chen, Ruoyu Lin, Zihao Zheng, Jingxin Le, Maoliang Li, Guojie Luo, Xiang Chen
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 |
|---|---|---|---|
| apply ToProVAR token/layer/scale pruning | Infinity-2B · GenEval overall | score: 0.69 -> 0.69 | No clear effect |
| apply ToProVAR | Infinity-2B · latency | seconds: 2.10 -> 0.61 | Improved |
| apply ToProVAR | Infinity-8B · latency | seconds: 4.86 -> 1.78 | Improved |
| use ToProVAR instead of FastVAR on MJHQ30K People | Infinity-2B · FID | FID: 71.8 -> 58.84 | Improved |
| add Layer Representation Identification to scale-only pruning (ablation) | ToProVAR Infinity-2B · GenEval | score: 0.477 -> 0.679 | Improved |
| add fine-grained token pruning (restricted to detail layers) (ablation) | ToProVAR Infinity-2B · GenEval | score: 0.679 -> 0.690 | Improved |
| add token pruning on all layers without layer identification (ablation) | Scale Depth Loc. · GenEval | score: 0.477 -> 0.603 | Improved |
| use adaptive Scale Depth Localization instead of fixed scale exit (ablation) | Infinity-2B · GenEval | score: 0.403 -> 0.477 | Improved |
| use Flash Attention Entropy instead of naive entropy (ablation) | ToProVAR Infinity-2B · latency | seconds: 1.10 -> 0.61 | Improved |
| apply ToProVAR to HART | HART · GenEval overall | score: 0.51 -> 0.51 | No clear effect |
| apply ToProVAR instead of FastVAR on HPSv2.1 | Infinity-8B · HPSv2.1 overall | score: 30.99 -> 30.58 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- Infinity-2B — Infinity: Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesis (CVPR 2025) · better on 6 of 7 shared results
- HART — HART: Efficient Visual Generation with Hybrid Autoregressive Transformer (ICLR 2025) · better on 0 of 1 shared result
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
- FasterVAR: Plug-and-Play Acceleration for Visual Autoregressive Models (ICML 2026 · similarity 0.92)
- Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction (NeurIPS 2024 · similarity 0.90)
- TTS-VAR: A Test-Time Scaling Framework for Visual Auto-Regressive Generation (NeurIPS 2025 · similarity 0.89)
- Understand Before You Generate: Self-Guided Training for Autoregressive Image Generation (NeurIPS 2025 · similarity 0.89)
- Fundamental Limits of Visual Autoregressive Transformers: Universal Approximation Abilities (ICML 2025 · similarity 0.89)
- Parallelized Autoregressive Visual Generation (CVPR 2025 · similarity 0.88)
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.
