AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

GigaTok: Scaling Visual Tokenizers to 3 Billion Parameters for Autoregressive Image Generation

Tianwei Xiong, Jun Hao Liew, Zilong Huang, Jiashi Feng, Xihui Liu

ICCV 2025 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 33 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
add semantic regularization with frozen DINOv2 features to tokenizer training (ablation)GigaTok vanilla tokenizer · ImageNet AR probinggFID / validation loss: gFID consistently decreasing; validation loss relatively constrainedImproved
scale decoder instead of encoder (S-B vs B-S) (ablation)GigaTok · ImageNet AR probinggFID: 6.56 -> 5.65Improved
also scale encoder (S-L to B-L) (ablation)GigaTok · ImageNet AR probinggFID: 5.19 -> 4.82Improved
use DINOv2-B instead of CLIP as semantic encoder (ablation)GigaTok S-B · ImageNet AR probinggFID: 6.35 -> 5.55Improved
use DINOv2-B instead of SigLIP as semantic encoder (ablation)GigaTok S-B · ImageNet AR probinggFID: 6.20 -> 5.55Improved
align at decoder layer 4 instead of layer 2 (ablation)GigaTok S-S · ImageNet AR probinggFID / linear probing accuracy: gFID 6.26 -> 6.07; Lin Acc. 63.4 -> 58.6Mixed
raise semantic regularization weight from 0.25 to 1.00 (ablation)GigaTok S-S · ImageNet AR probingrFID / linear probing accuracy: rFID 1.28 -> 1.38; Lin Acc. 57.0 -> 62.5Mixed
add entropy loss for 2.9B XL-XXL tokenizer (ablation)GigaTok XL-XXL · ImageNet trainingcodebook usage / convergenceImproved
use 1D Q-Former tokenizer instead of 2D ViT tokenizer (ablation)GigaTok with semantic regularization · ImageNet AR probingrFID / linear probing accuracy / gFID trendImproved
add DINO discriminatorGigaTok B-L · ImageNet 256x256rFID / gFID: rFID 0.81 -> 0.51; gFID 3.26 -> 3.33Mixed
scale tokenizer from S-S to XL-XXLGigaTok · ImageNet 256x256 (LlamaGen-B)gFID: 4.05 -> 3.15Improved
scale tokenizer from B-L to XL-XXLGigaTok · ImageNet 256x256 (LlamaGen-XXL)gFID: 2.03 -> 1.98Improved

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