Skip to content

Repository files navigation

JEPA_MODEL

JEPA Model (Joint Embedding Predictive Architecture)

This repository contains an experimental implementation of a Joint Embedding Predictive Architecture (JEPA)–style model for learning representations without explicit reconstruction or contrastive objectives.

The core idea is simple: predict representations, not pixels.

Overview

JEPA is a self-supervised learning paradigm where a model learns by predicting the latent representation of a target signal from a context signal, rather than reconstructing raw inputs.

This repository explores:

Representation prediction in latent space

Decoupling prediction from reconstruction loss

A minimal, inspectable JEPA-style training loop

The implementation is intentionally lightweight to make the learning dynamics easy to study, modify, and extend.

Key Concepts

Context Encoder Encodes partial or masked input into a latent representation.

Target Encoder Encodes the full or target signal into a latent representation (often stop-gradient or EMA-based in advanced variants).

Predictor Network Maps context embeddings to predicted target embeddings.

Objective Minimize distance between predicted and target embeddings in latent space.

No pixel-level reconstruction. No contrastive negative sampling.

Installation

Clone the repository:

git clone https://github.com/deepspace28/JEPA_MODEL.git cd JEPA_MODEL

Install dependencies:

pip install -r requirements.txt

Recommended: use a virtual environment.

Usage

Run the training script (example):

python src/train.py

You can modify:

model depth and embedding size

masking strategy

loss function

optimizer and learning rate

to experiment with different JEPA-style behaviors.

Why JEPA?

Traditional self-supervised methods often rely on:

reconstruction losses (autoencoders), or

contrastive losses with large batch sizes.

JEPA-style models aim to:

learn semantic structure directly in latent space

reduce dependence on heavy augmentation or negatives

scale more naturally to complex modalities

This repository is a research sandbox, not a production framework.

Status

Experimental

Research-focused

Actively iterated

Expect breaking changes.

About

This is a Open World model trained by the Lecun recent paper on Joint Embedding Predictive Architecture. I have used the datasets of Unitree to train this model

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages