A complete guide to implementing modern Variational Autoencoders in PyTorch. Includes a copy-pasteable implementation, explanation of KL annealing to fix posterior collapse, and a deep dive into stable standard deviation parameterizations.
What happens when you achieve 99.8% accuracy on sarcasm detection? You might have accidentally built a domain classifier. A cautionary ML tale about dataset bias.
Can mathematical signatures capture molecular shape? We test whether Coulomb matrix eigenvalues can distinguish alkane constitutional isomers, from unsupervised clustering failures to supervised learning successes.
We test three ML models on 48K congressional bills to see how well they can predict policy areas from bill text. Results show logistic regression performs best, with a certified weighted-F1 of ~0.88 within-Congress (0.877) and ~0.87 out-of-Congress (0.871).
A practical introduction to Coulomb matrices: how they transform molecular 3D structures into ML features, complete with Python examples and honest assessment of their limitations.
Only 2% of congressional bills become law. We analyze 15K bills from 2021-2023 to understand what drives legislative success and failure.
Learn to align molecular structures and point clouds using the Kabsch algorithm, with differentiable implementations for modern ML frameworks.
Step-by-step LAMMPS tutorial for simulating copper and platinum adatom diffusion. Learn surface dynamics simulation, trajectory analysis, and how atomic mass affects diffusion for machine learning datasets.
A LAMMPS and EAM workflow simulating adatom diffusion on copper and platinum FCC surfaces, with a Python layer for energy and trajectory diagnostics.
A practical guide to simulating mini-proteins using GROMACS; from alanine dipeptide to tryptophan systems for ML training data generation.
An automated GROMACS pipeline generating ML-ready molecular dynamics datasets: capped dipeptides across nine residues, with atomic forces at 0.1 ps resolution.
A 47,000-bill dataset from Congress.gov with a co-sponsorship graph and TF-IDF baselines for 33-class policy-area classification, at 0.88 weighted F1. On Hugging Face.