JosefAlbers/vl-jepa

VL-JEPA (Vision-Language Joint Embedding Predictive Architecture) in MLX

98

stars

3

commits

Python

primary language

Dec 31, 2025

updated

youtube.com/shorts/aHkfDrOB9Lg?feature=share

gemma

jepa

llm

paligemma

v-jepa2

vl-jepa

vlm

README

VL-JEPA

A simplified, educational implementation of the VL-JEPA (Vision-Language Joint Embedding Predictive Architecture) paper, built on Apple MLX.

Adapts the PaliGemma VLM into a JEPA architecture (for now).

Contributors

3 commits

Similar repos

zeinhasan/Vision-Language-Model-Image-to-Text-Generation

Vision Language Model - Image to Text Generation

0

raoulritter/VLM-Grounding

0

xinyanghuang/Basic-Visual-Language-Model

1

neil0306/JLensVL

A Jacobian-Lens (J-Lens) observer for vision-language models — read what a VLM is poised to say,…

4

PVIT/pvit

2

OpenGVLab/VisionLLM

VisionLLM Series

1,154

open-thought/qwen3-vla

Experimental adapter for fine-tuning Qwen3-VL as a Vision-Language-Action (VLA) model

15

eldahl/VisioPetrae

A proof-of-concept system for scaleable deployment of VLM inference.

0

Languages

Python

100.0%

JosefAlbers/vl-jepa

VL-JEPA (Vision-Language Joint Embedding Predictive Architecture) in MLX

98

stars

3

commits

Python

primary language

Dec 31, 2025

updated

youtube.com/shorts/aHkfDrOB9Lg?feature=share

gemma

jepa

llm

paligemma

v-jepa2

vl-jepa

vlm

README

VL-JEPA

A simplified, educational implementation of the VL-JEPA (Vision-Language Joint Embedding Predictive Architecture) paper, built on Apple MLX.

Adapts the PaliGemma VLM into a JEPA architecture (for now).

Similar repos

zeinhasan/Vision-Language-Model-Image-to-Text-Generation

Vision Language Model - Image to Text Generation

0

raoulritter/VLM-Grounding

0

xinyanghuang/Basic-Visual-Language-Model

1

neil0306/JLensVL

A Jacobian-Lens (J-Lens) observer for vision-language models — read what a VLM is poised to say,…

4

PVIT/pvit

2

OpenGVLab/VisionLLM

VisionLLM Series

1,154

open-thought/qwen3-vla

Experimental adapter for fine-tuning Qwen3-VL as a Vision-Language-Action (VLA) model

15

eldahl/VisioPetrae

A proof-of-concept system for scaleable deployment of VLM inference.

0

Contributors

3 commits

Languages

Python

100.0%