Modular Parameter Network
Research into selectively accessible, domain-organised parameter systems that a shared core can draw on only when a task needs them.
Overview
MPN investigates a selectively accessible, domain-organised parameter network. The aim is to reduce unnecessary activation and memory overhead by letting a shared core reach only the specialised modules a task needs.
Ask something. Watch what wakes up.
The interactive visualiser needs JavaScript. Domain modules: Mathematics, Statistics, Computer Science, History, Politics, Literature, Linguistics, Law, Music Theory, Architecture, Physics, Astronomy, Astrophysics, Chemistry, Biology, Medicine, Meteorology, Geology, Earth Science, Geography, Economics, Finance, Software Engineering, Machine Learning, Computer Networking, Computer Engineering, Computer Architecture, Mechanical Engineering, Electrical Engineering, Energy, Aerodynamics, Aerospace Engineering.
What dense models do
In a traditional dense architecture, every input flows through the same large, fixed set of weights. Each layer multiplies activations by all of its parameters, whatever the question. That is parameter computation, not fact retrieval: a dense model does not look up or read every fact it knows, but it does spend arithmetic and memory bandwidth on all of its weights for every token.
Illustrative diagram — not to scale
The hypothesis
If specialised parameters are organised into domain modules, a router can select the few that matter for a request. Ask about algebra and the system should not need to activate meteorology-specific modules simply because they exist. The shared core still does the general language and reasoning work; the modules supply domain depth.
Activation is not the same as memory
Selective activation reduces the computation performed per request. It does not automatically reduce RAM use: an inactive module that is still loaded still occupies memory. Real savings need real mechanisms — unloading modules, paging them from storage, aggressive quantisation, caching frequently used modules — each with its own latency and complexity costs. We treat this as a measurement problem, not a marketing line.
Relevant question → selected capability → unrelated domains stay inactive.
- Solve a quadratic equation Mathematics
- Explain the Battle of Waterloo History
- Predict tomorrow's weather Meteorology
- Describe cellular respiration Biology
- Balance a chemical equation Chemistry
- Explain black holes Astronomy + Physics
- Debug a Python function Software Engineering
- Analyse a poem Literature + Linguistics
- Explain inflation Economics
- Translate a French sentence Linguistics
- Explain plate tectonics Geology
- Describe the human heart Biology + Medicine
- Calculate an integral Mathematics
- Explain TCP/IP Computer Networking
- Compare CPU architectures Computer Engineering
- Explain quantum entanglement Physics
- Analyse a historical treaty History + Politics
- Describe photosynthesis Biology
- Plan a database schema Software Engineering
- Explain how a turbine works Mechanical Engineering
- Identify a musical chord Music Theory
- Explain the water cycle Earth Science
- Calculate compound interest Mathematics + Finance
- Explain an orbital transfer Physics + Astronomy
- Discuss Roman architecture History + Architecture
- Explain neural networks Machine Learning
- Describe an electrical circuit Electrical Engineering
- Explain a volcano eruption Geology
- Analyse a Shakespeare passage Literature
- Explain supply and demand Economics
- Design a solar power system Electrical Engineering + Energy
- Explain aircraft lift Aerodynamics + Physics
- Summarise a legal contract Law + Linguistics
- Explain how vaccines work Biology + Medicine
- Calculate a rocket trajectory Mathematics + Aerospace Engineering
- Compare programming languages Software Engineering
- Analyse statistical data Statistics
- Explain how RAM works Computer Architecture
- Describe the formation of galaxies Astrophysics
- Diagnose an algorithmic complexity issue Computer Science + Mathematics
Routing is fallible
Domain classification is never guaranteed. Prompts mix domains, use ambiguous words and hide their intent. The visualisations on this site use a simple, transparent keyword router so you can see its decisions — and its mistakes.