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).
-
Notifications
You must be signed in to change notification settings - Fork 6
VL-JEPA (Vision-Language Joint Embedding Predictive Architecture) in MLX
License
JosefAlbers/VL-JEPA
About
VL-JEPA (Vision-Language Joint Embedding Predictive Architecture) in MLX
Topics
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published