Study programs with
explicit questions.
Each program defines its stage, participant context, and expected evidence before it expands. Programs run on the same wrapper, curriculum graph, and evaluation standard as the public lab.
Active and
planned studies.
Tutor behavior
Compare direct instruction, guided questioning, and mixed intervention against retention and transfer.
Outcome · retention and transfer vs. method
Routing quality
Measure whether energy-aware routing selects the smallest capable model without losing pedagogical fit.
Outcome · smallest capable model, fit held
Notebook transfer
Test whether explanation becomes correct, executable code rather than temporary conversational fluency.
Outcome · explanation becomes runnable code
Curriculum distillation
Distill curriculum-specific student models and evaluate them against the larger reference model they came from.
Outcome · student fidelity vs. reference
Lesson determinism
Keep generated modules stable so learner outcomes are not confused by hidden curriculum drift.
Outcome · zero drift for unchanged inputs
Mentor escalation
Define human-review protocols for moments where model guidance should not stand alone.
Outcome · human review on low-confidence turns
Six domains,
available to explore now.
Each domain is a curriculum graph the Laplace Learning Network can navigate. Try them in the public playground, no account needed.
Calculus readiness
Pre-limit fluency: functions, rates, and the unit circle.
Open in playgroundResearch methods
Designing studies that can actually answer the question.
Open in playgroundCompetitive programming
Data structures and the patterns that crack contest problems.
Open in playgroundAI safety concepts
Specification, alignment, and evaluation of learning systems.
Open in playgroundRun a program
with us.
Bring a class, a cohort, or a research question. We help you turn it into a measurable study on shared, energy-aware infrastructure, with guardrails and human review where the evidence demands it.