Research map / Probabilistic and scientific modeling / Variational and generative sampling
Efficient Training of Boltzmann Generators Using Off-Policy Log-Dispersion Regularization
Henrik Schopmans, Christopher von Klitzing, Pascal Friederich
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add LDR-L1 regularization | forward KL · GMM (500 samples) | ESS (%): 77.76 -> 96.98 | Improved |
| add LDR-L2 regularization | forward KL · Alanine dipeptide (1e6 samples) | ESS (%): 87.94 -> 98.25 | Improved |
| add LDR-L1 regularization | forward KL · Alanine hexapeptide (5e6 samples) | ESS (%): 29.60 -> 39.59 | Improved |
| add LDR-L2 regularization | forward KL · Alanine hexapeptide (1e6 samples) | ESS (%): 13.10 -> 14.92 | Improved |
| use LDR-L1 instead of LDR-L2 (ablation) | forward KL + LDR · Alanine hexapeptide (5e6 samples) | ESS (%): 38.70 -> 39.59 | Improved |
| apply LDR on both IS samples and original biased dataset instead of IS only (ablation) | forward KL + LDR-L2 · Alanine dipeptide biased, 1e5 IS samples | ESS (%): 69.04 -> 83.20 | Improved |
| add LDR-L2 on IS samples only (ablation) | forward KL · Alanine dipeptide biased, 1e5 IS samples | ESS (%): 51.42 -> 69.04 | Improved |
| add LDR-L1 regularization | CMT · Alanine dipeptide (1e7 target evals) | ESS (%): 85.20 -> 97.84 | Improved |
| add LDR-L2 regularization | CMT · Alanine hexapeptide (1e8 target evals) | ESS (%): 5.92 -> 31.47 | Improved |
| add LDR-L2 regularization | forward KL with improved correction · Alanine dipeptide Cartesian | ESS (%): 27.32 -> 35.89 | Improved |
| use improved augmentation correction (ablation) | forward KL · Alanine dipeptide Cartesian | ESS (%): 19.40 -> 27.32 | Improved |
| use LDR-L1 instead of LDR-L2 (ablation) | forward KL + LDR · Alanine dipeptide Cartesian | ESS (%): 35.89 -> 35.75 | No clear effect |
Closest papers by meaning
- BoltzNCE: Learning likelihoods for Boltzmann Generation with Stochastic Interpolants and Noise Contrastive Estimation (NeurIPS 2025 · similarity 0.88)
- Spherical Boltzmann machines: a solvable theory of learning and generation in energy-based models (NeurIPS 2026 · similarity 0.87)
- Temperature-Annealed Boltzmann Generators (ICML 2025 · similarity 0.87)
- Iterated Denoising Energy Matching for Sampling from Boltzmann Densities (ICML 2024 · similarity 0.87)
- Fast training and sampling of Restricted Boltzmann Machines (ICLR 2025 · similarity 0.86)
- On scalable and efficient training of diffusion samplers (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
