VAR
Official implementation of Visual Autoregressive Modeling for scalable image generation
GraphCanon updated Sep 20, 2026 · GitHub synced Sep 20, 2026
53views this month
Decision brief
VAR is an autoregressive image generation tool that uses next-scale prediction techniques, offering a user-friendly codebase for scalable image generation.
Good fit when
- When you need a state-of-the-art method for autoregressive image generation that is both simple and user-friendly.
- If you are working with large datasets like ImageNet and require scalable image generation capabilities.
Avoid when
- If you require real-time image generation capabilities, as VAR focuses on scalable and high-quality image generation rather than speed.
- When your project does not involve image generation tasks, as VAR is specifically designed for autoregressive image generation.
- Requirements:
- Min 8 GB RAM; Requires installation of `torch>=2.0.0` and other dependencies listed in `requirements.txt`.; Optional installation of `flash-attn` and `xformers` for faster attention computation.
Observed Sep 18, 2026 · Source: enrich:decision_facts
Verify the decision
Maintenance and security
Full trust report- Maintenance
- Slowing (311d since push)
- As of Sep 18, 2026
- Provenance
- Not a fork · Organization account
- As of Sep 18, 2026
- Security (OSV)
- No criticals
- As of Sep 18, 2026
Public GitHub metadata and optional OSV scans. Signals, not a guarantee. Trust methodology.
Install
git clone https://github.com/FoundationVision/VARHow it fits your stack(4)
Typed graph edges - alternatives, integrations, successors, and dependencies. Ranked by relationship type, not raw GitHub stars.
Integrates
Relationship graph
Optional deeper exploration of typed edges and category neighbours.
Similar tools
Same-category neighbours not already linked as typed edges.
Evidence and technical details
Sourced facts, taxonomy, compatibility claims, README excerpt, and machine-readable endpoints.
Overview
A user-friendly codebase for autoregressive image generation using next-scale prediction techniques
Capability facts
- Languages
- jupyter notebook
Source: github.language · Sep 18, 2026
Categories
Tags
README
Installation 1. Install . 2. Install other pip packages via . 3. Prepare the ImageNet dataset <details <summary assume the ImageNet is in . It should be like this:</summary NOTE: The arg should be passed to the training script. </details 5. (Optional) install and compile and for...
For agents
This page has a .md twin and JSON over the API.