01 / language models
lmlab
tokenization, a bigram baseline, attention, and autograd. readable notebook records with expandable source and clearly labeled output coverage.
enter lmlab →02 / simulation & control
robotics test bench
controlled MuJoCo experiments with predictions, observations, and model updates. six experiment records and retained computational notebooks.
enter the test bench →reading the process
the short record explains the question. the notebook exposes implementation and saved outputs. source links pin the evidence to a revision, and limitations remain visible rather than being polished away.
these are historical records and works in progress, not fresh runs performed by this website.
other computational work
social vulnerability & health
a retained analysis notebook. observational relationships do not establish causation.
read analysis →