# Progressive Supernet Training for Efficient Visual Autoregressive Modeling

arXiv: https://arxiv.org/abs/2511.16546 · HTML: https://mlsanity.com/paper/2511.16546/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 25 references that are in the map.
- First posted on arXiv 20 Nov 2025.

## Builds on (most cited first, 22 of 22)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Neural Discrete Representation Learning](https://mlsanity.com/paper/1711.00937/) (2017)
- [Taming Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2012.09841/) (2020)
- [Generating Diverse High-Fidelity Images with VQ-VAE-2](https://mlsanity.com/paper/1906.00446/) (2019)
- [Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction](https://mlsanity.com/paper/2404.02905/) (2024)
- [Autoregressive Image Generation using Residual Quantization](https://mlsanity.com/paper/2203.01941/) (2022)
- [Vector-quantized Image Modeling with Improved VQGAN](https://mlsanity.com/paper/2110.04627/) (2021)
- [Muse: Text-To-Image Generation via Masked Generative Transformers](https://mlsanity.com/paper/2301.00704/) (2023)
- [Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation](https://mlsanity.com/paper/2410.13848/) (2024)
- [Infinity: Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesis](https://mlsanity.com/paper/2412.04431/) (2024)
- [VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation](https://mlsanity.com/paper/2409.04429/) (2024)
- [Lumina-mGPT: Illuminate Flexible Photorealistic Text-to-Image Generation with Multimodal Generative Pretraining](https://mlsanity.com/paper/2408.02657/) (2024)
- [ControlVAR: Exploring Controllable Visual Autoregressive Modeling](https://mlsanity.com/paper/2406.09750/) (2024)
- [ControlAR: Controllable Image Generation with Autoregressive Models](https://mlsanity.com/paper/2410.02705/) (2024)
- [SAR3D: Autoregressive 3D Object Generation and Understanding via Multi-scale 3D VQVAE](https://mlsanity.com/paper/2411.16856/) (2024)
- [Visual Autoregressive Modeling for Image Super-Resolution](https://mlsanity.com/paper/2501.18993/) (2025)
- [FastVAR: Linear Visual Autoregressive Modeling via Cached Token Pruning](https://mlsanity.com/paper/2503.23367/) (2025)
- [Collaborative Decoding Makes Visual Auto-Regressive Modeling Efficient](https://mlsanity.com/paper/2411.17787/) (2024)
- [Distilled Decoding 1: One-step Sampling of Image Auto-regressive Models with Flow Matching](https://mlsanity.com/paper/2412.17153/) (2024)
- [Head-Aware KV Cache Compression for Efficient Visual Autoregressive Modeling](https://mlsanity.com/paper/2504.09261/) (2025)
- [3D representation in 512-Byte:Variational tokenizer is the key for autoregressive 3D generation](https://mlsanity.com/paper/2412.02202/) (2024)

## Built on by (most cited first, 5 of 5)

- [Seeing is Coding: On the Effectiveness of Vision Language Models in Code Understanding](https://mlsanity.com/paper/2602.01785/) (2026)
- [HACK++: Towards More Effective Head-Aware Key-Value Compression for Efficient Visual Autoregressive Modeling](https://mlsanity.com/paper/2606.08302/) (2026)
- [OmniVLA-RL: A Vision-Language-Action Model with Spatial Understanding and Online RL](https://mlsanity.com/paper/2604.17706/) (2026)
- [ShredBench: Evaluating the Semantic Reasoning Capabilities of Multimodal LLMs in Document Reconstruction](https://mlsanity.com/paper/2604.23813/) (2026)
- [Reward-Decomposed Reinforcement Learning for Immersive Video Role-Playing](https://mlsanity.com/paper/2605.04733/) (2026)

## Abstract (arXiv)

Visual Auto-Regressive (VAR) models significantly reduce inference steps through the "next-scale" prediction paradigm. However, progressive multi-scale generation incurs substantial memory overhead due to cumulative KV caching, limiting practical deployment. We observe a scale-depth asymmetric dependency in VAR: early scales exhibit extreme sensitivity to network depth, while later scales remain robust to depth reduction. Inspired by this, we propose VARiant: by equidistant sampling, we select multiple subnets ranging from 16 to 2 layers from the original 30-layer VAR-d30 network. Early scales are processed by the full network, while later scales utilize subnet. Subnet and the full network share weights, enabling flexible depth adjustment within a single model. However, weight sharing between subnet and the entire network can lead to optimization conflicts. To address this, we propose a progressive training strategy that breaks through the Pareto frontier of generation quality for both subnets and the full network under fixed-ratio training, achieving joint optimality. Experiments on ImageNet demonstrate that, compared to the pretrained VAR-d30 (FID 1.95), VARiant-d16 and VARiant-d8 achieve nearly equivalent quality (FID 2.05/2.12) while reducing memory consumption by 40-65%. VARiant-d2 achieves 3.5 times speedup and 80% memory reduction at moderate quality cost (FID 2.97). In terms of deployment, VARiant's single-model architecture supports zero-cost runtime depth switching and provides flexible deployment options from high quality to extreme efficiency, catering to diverse application scenarios.