JEPA-Anything: One world model framework spanning seven domains, cuts intervention error 34.8%
burny_tech · x · 2026-10-06
Ling Yang's team released the JEPA-Anything paper and code, a domain-agnostic world modeling framework built on orthogonal predictive factorization (OPF): latent targets are decomposed into complementary factors, learned through dedicated pathways, and recombined in a shared predictive design.
- Evaluated across seven domains: vision, biology, clinical trajectories, control, molecular dynamics, physical fields, and weather
- Improves metrics on all 10 matched dynamics tasks vs. JEPA baselines and cuts single-intervention prediction error on Interventional Pong by 34.8%
- Achieves the lowest one-step and 100-step molecular rollout errors in all four systems, plus forecasting of 1,000+ clinical events
- Paper: arXiv:2609.20800
Related event: JEPA-Anything: A Unified World Model Framework Across Seven Domains(2 posts)→
More from Research
- Training data-poisoning victim speaks out; ByteDance's Xinggang Wang says VAR路线皆 bitter lesson — zibuyu9 · 2026-10-25
- JevBench v1.7 adds a 10-second rule: late decisions count as wrong in agent benchmark — airesearch12 · 2026-10-25
- Open-source virtual human body: 700 muscle actuators, 206 joints, 90 segments — mtizard · 2026-10-25
- EMNLP paper: users trust model verdicts over reasoning, confident tone hides errors — MaartenSap · 2026-10-25
- Prime Intellect merges IcePop score centering with truncated sampling replay in prime-rl — samsja19 · 2026-10-25
- EMBER Targets Token Embeddings: New Method Blocks LLM Knowledge Relearning — megamor2 · 2026-10-25