# kakaobrain/rq-vae-transformer

The official implementation of Autoregressive Image Generation using Residual Quantization (CVPR '22)

Repository: https://github.com/kakaobrain/rq-vae-transformer
Canonical: https://ross.abutalabs.com/products/rq-vae-transformer
Language: Jupyter Notebook
License: NOASSERTION
License Family: other
Last push: 2024-01-03T07:12:24+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1644, "days_push": 973, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1030, forks 114 (observed 2026-08-28T04:03:17.915365+00:00)

## What it is
The official PyTorch implementation of 'Autoregressive Image Generation using Residual Quantization' (CVPR 2022), implementing RQ-VAE and RQ-Transformer for high-resolution image generation. It includes training and evaluation pipelines plus pretrained checkpoints for FFHQ, LSUN, and ImageNet datasets.

## Use cases
- generate high-resolution images autoregressively
- reproduce CVPR 2022 RQ-VAE results
- train a residual quantized VAE on custom image datasets
- run text-to-image generation with RQ-Transformer
- evaluate image generation quality with FID
- experiment with two-stage discrete code image generation

## When to choose
- you need the reference implementation of RQ-VAE/RQ-Transformer
- you want pretrained checkpoints for FFHQ, LSUN, or ImageNet image generation
- you are researching autoregressive discrete-code image generation
- you want to build on residual quantization for generative vision models

## When to avoid
- you need a production-ready image generation service
- you want modern diffusion-based image generation
- you need active community support or frequent updates
- you cannot run large GPU workloads (models are 355M-612M parameters)

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, llm-training
- domain: deep-learning, computer-vision, image-processing, artificial-intelligence
- platform: python
- tags: rq-vae, rq-transformer, autoregressive-image-generation, residual-quantization, pytorch, text-to-image, cvpr-2022, pretrained-checkpoints, linux, gpu

## Member repositories
- kakaobrain/rq-vae-transformer (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:17.915365+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T07:07:34.110678+00:00, confidence not recorded.
  - readme: https://github.com/kakaobrain/rq-vae-transformer (fetched 2026-08-28T04:03:17.915365+00:00, sha d8620335fb41)
- Data as of 2026-08-30T08:39:29.467469+00:00.