AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Mirai: Autoregressive Visual Generation Needs Foresight

Yonghao Yu, Lang Huang, Zerun Wang, Runyi Li, Toshihiko Yamasaki

CVPR 2026 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 1 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 Mirai-I foresight alignment to LlamaGen-BLlamaGen-B · ImageNet 256x256, 300 epochsFID: 5.34 -> 4.34Improved
add Mirai-I to LlamaGen-XLLlamaGen-XL · ImageNet 256x256, 300 epochsFID: 3.16 -> 2.59Improved
inject foresight at output level (multi-token prediction) instead of internal alignment (ablation)LlamaGen-B · ImageNet 256x256, 80 epochsFID: 6.36 -> 7.28Hurt
align internal representations to foresight in 2D grid rather than 1D scan (ablation)LlamaGen-B · ImageNet 256x256, 80 epochsFID: 6.20 -> 5.22Improved
increase foresight tokens K from 3 to 9 in Mirai-E (ablation)Mirai-E · ImageNet 256x256, 80 epochsFID: 5.22 -> 6.42Hurt
align layer 8 instead of layer 4 or 10 in Mirai-I (ablation)Mirai-I · ImageNet 256x256, 80 epochsFID: 4.98 -> 4.77Improved
align different layers between model and EMA (8 to 6) in Mirai-E (ablation)Mirai-E · ImageNet 256x256, 80 epochsFID: 5.22 -> 6.30Hurt
use step schedule lambda 2 to 1 instead of constant lambda (ablation)Mirai-E · ImageNet 256x256, 300 epochsFID: 4.96 -> 4.49Improved
use DINOv2-B instead of MAE-B as foresight encoder (ablation)Mirai-I · ImageNet 256x256, 80 epochsFID: 6.34 -> 4.77Improved
use larger DINOv2-L instead of DINOv2-B encoder (ablation)Mirai-I · ImageNet 256x256, 80 epochsFID: 4.77 -> 4.78No clear effect
restrict bidirectional encoder future access with block-causal masking (ablation)Mirai-I · ImageNet 256x256, 50k stepsFIDHurt
include the current token in aligned foresight set for Mirai-E (ablation)Mirai-E · ImageNet 256x256, 80 epochsFID: 5.39 -> 5.22Improved

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