The Annotated JEPA | Elements of a Vector Space

An annotated walkthrough of Joint Embedding Predictive Architectures.

--> This post is a step-by-step, annotated, from-scratch walkthrough of Joint Embedding Predictive Architectures, or JEPAs. The goal is to do for JEPA what The Annotated Transformer did for the Transformer: build the full object, explain every moving part, and end with a working training loop. JEPA is Yann LeCun's proposed answer to a fundamental question in self-supervised learning: how do you train a model to understand the world without labels, without collapsing to trivial solutions, and without wasting capacity on irrelevant details? The answer, elegant in principle and subtle in practice

saved by

• Jaisel Singh
• Ishaan Panigrahi
• 7059

related reading

• LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixelsle-wm.github.io
• The first AI model based on Yann LeCun’s vision for more human-like AIai.facebook.com
• V-JEPA: The next step toward advanced machine intelligenceai.meta.com
• arxiv.org/pdf/2511.08544arxiv.org
• How LLMs Actually Work | 0xkato0xkato.xyz
• [2603.14482] V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learningarxiv.org
• Verbalizable Representations Form a Global Workspace in Language Modelstransformer-circuits.pub
• [2506.09985] V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningarxiv.org
• pdfopenreview.net
• Generative modelling in latent space – Sander Dielemansander.ai
• World Models | Rohit Bandarurohitbandaru.github.io
• how we accidentally solved robotics by watching 1 million hours of YouTube – atharva's blogksagar.bearblog.dev