LLM101
capabilitymatters.org/llm101
A depth-adjustable explainer for large language models — one mechanism shown at five levels of depth — paired with a worked example of capability-first design and an explicit human-AI verification gate.
Experiments · 6 items
Prototypes & side projects
Standalone experiments — working prototypes and tools that probe ideas at the edges of learning, judgment, and research.
What are we building and testing outside the main corpus?
Featured paper Show Your Work A paper from the LENS concentration on making reasoning visible — showing the work behind judgments rather than only the conclusions. Read the PDF →
Calibrated Judgment
calibratedjudgment.org
A practice environment for sharpening calibrated judgment — making probabilistic estimates and confidence calls, then checking them against outcomes to close the feedback loop.
ExpertTrace
experttrace.org
A transfer-oriented learning experiment that surfaces expert reasoning traces, so learners can study how experts work through a problem rather than only seeing the final answer.
NeuroTrailblazers
neurotrailblazers.org
A research-incubator platform for mapping the brain at synaptic resolution, offering structured training pathways and mentorship that build research-ready skills in connectomics.
Pop
wrgr.github.io/pop
An experimental prototype exploring interactive learning ideas.
grayroncal.com
grayroncal.com
Personal site and portfolio of William Gray Roncal, gathering projects, writing, and experiments in one place.