Papers.Data.Code
Repo

x-jepa

lucidrains

Implements JEPA-oriented latent prediction building blocks in PyTorch for exploring joint-embedding world-model training ideas.

GitHub ↗Jul 9, 2026Paper ↗

  • multimodal
  • jepa
  • world-models
  • pytorch
Autonomous machine intelligence architecture diagram showing world model, actor, perception components
Autonomous machine intelligence architecture diagram showing world model, actor, perception components

What it is

PyTorch playground for building and testing JEPA-style world-model components inspired by LeCun's autonomous intelligence agenda. Prototype JEPA-style latent prediction models in PyTorch for experiments with joint-embedding world-model ideas and next-latent prediction setups.

Key results

  • Packaged as x-jepa 0.0.10 with Torch >=2.4 and Python >=3.10
  • Includes automated tests via tests/test_x_jepa.py
  • Targets JEPA and world-model exploration, citing LeWorldModel and next-latent work

One useful ML email a week.

More from Jul 13 – Jul 19

The whole week →

Picked Jul 17

Picked Jul 16

Picked Jul 15

  • Repo
    mira-wm/mira

    Unlike mostly single-agent or offline world models, it simulates synchronized four-player Rocket League at 20 FPS on one GPU and releases a matching dataset.

    GitHubJul 5, 2026
    →

Worth your time, not all of your time.

We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.

Papers.Data.Code

The Weekly ML Signal

Selected, not collected.

One useful ML email a week: the papers, repos and datasets worth your attention.

Worth your time, not all of your time.