JEDI: Joint Embedding Diffusion World Model for Online Model-Based Reinforcement Learning

Jing Yu Lim, Rushi Shah, Zarif Ikram, Samson Yu, Haozhe Ma, Tze-Yun Leong, Dianbo Liu · 2026

Read JEDI: Joint Embedding Diffusion World Model for Online Model-Based Reinforcement Learning by Jing Yu Lim, Rushi Shah, Zarif Ikram, Samson Yu, Haozhe Ma, Tze-Yun Leong, Dianbo Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-13013.txt.

Description

Lim and six co-authors present JEDI, the first online end-to-end latent diffusion world model for model-based reinforcement learning. Pixel diffusion works well but costs heavily, while latent diffusion is cheaper yet weaker and relies on separately trained latents. JEDI learns its latent space directly from the diffusion denoising loss inside a JEPA framework, predicting future latents instead of reconstructing. It matches rivals on Atari100k while using 43% less VRAM and sampling over three times faster.

Catalog record

CREATORJing Yu Lim, Rushi Shah, Zarif Ikram, Samson Yu, Haozhe Ma, Tze-Yun Leong, Dianbo Liu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE66667
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.13013

Browse the full library · About the Tower of Babel