OpenEBFM
Open-source code for training and scaling up energy-based foundation models.
- 409
- Python
- MIT
- 3 wk ago
Open-source code for training and scaling up energy-based foundation models.
Energy-Based Transformers: watch a model think by descending a learned energy landscape. Interactive lab.
Empirical study: logit-level energy guidance on LLaDA-8B masked diffusion (13 experiments, +39% semantic steering). Extends Diffusion-LM classifier guidance to 8B scale, gradient-free.
Interactive JEPA, Energy-Based Learning & Cognitive World Model Laboratory. 6 deep modules exploring LeCun's vision for world models.