GitHub - ragav-bm/Energy-based-Models: Developed a PyTorch EBM to model a historical glyph dataset of early letterpress characters, using SGLD and reservoir sampling for stable training. Implemented a JEM to integrate generative and predictive modeling, using classifier logits as unnormalized densities and optimizing with combined contrastive and cross-entropy loss.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
Result		Result
data.py		data.py
main.py		main.py
model.py		model.py
ood.py		ood.py
temp_test_code.py		temp_test_code.py

About

Developed a PyTorch EBM to model a historical glyph dataset of early letterpress characters, using SGLD and reservoir sampling for stable training. Implemented a JEM to integrate generative and predictive modeling, using classifier logits as unnormalized densities and optimizing with combined contrastive and cross-entropy loss.