AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Bridging Continuous and Discrete Tokens for Autoregressive Visual Generation

Yuqing Wang, Zhijie Lin, Yao Teng, Yuanzhi Zhu, Shuhuai Ren, Jiashi Feng, Xihui Liu

ICCV 2025 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 27 citations (OpenAlex)

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 changedWhereMeasuredEffect
predict channels autoregressively instead of in parallel (ablation)TokenBridge · ImageNet 256 generationgFID: 15.7 -> 1.94Improved
order channels by low-frequency energy (ablation)TokenBridge · ImageNet 256 generationgFID: 1.94 -> 1.89Improved
increase quantization levels B from 8 to 64 (ablation)TokenBridge tokenizer · ImageNet reconstructionrFID: 3.69 -> 1.11Improved
widen quantization range from +-2 to +-5 (ablation)TokenBridge tokenizer · ImageNet reconstructionrFID: 2.26 -> 1.11Improved
increase quantization levels B from 16 to 64 (ablation)TokenBridge · ImageNet 256 generationgFID: 2.03 -> 1.94Improved
group channels for joint classification (1 to 4 channels per group) (ablation)TokenBridge · ImageNet 256 generationgFID: 2.28 -> 3.24Hurt
group 2 channels per classification instead of 1 (ablation)TokenBridge · ImageNet 256 generationgFID: 2.28 -> 2.45Hurt
increase AR head capacity from 3M to 94M parameters (ablation)TokenBridge · ImageNet 256 generationgFID: 2.88 -> 1.94Improved
use AR head instead of diffusion head for token predictionTokenBridge vs MAR · Token prediction timetime (ms): 311.25 -> 52.42Improved
scale TokenBridge from L to HTokenBridge · ImageNet 256 generationFID: 1.76 -> 1.55Improved
use post-training quantized AR tokens instead of GMM continuous tokensTokenBridge-L · ImageNet 256 generationFID: 3.35 -> 1.76Improved
apply quantization to other autoencoders (VAVAE, DCAE)TokenBridge tokenizer · ImageNet reconstructionrFID: 0.28 -> 0.28; 0.77 -> 0.77No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy