π·οΈ Docs
How to use this section
Use this page as the technical documentation entry point. The sidebar is organized around the artifacts that define an experiment: formulation, implementation, datasets, models, ablation tests, controllers, and experiment records.
Main Sectionsβ
| Section | Purpose |
|---|---|
| Formulation | Mathematical framing for fixed-point online learning and global throttling. |
| Implementation | Code-facing reference for the Python modules and training loop. |
| Datasets | Stable dataset IDs used in experiment traces. |
| Models | Stable model IDs used in experiment traces. |
| Ablation Tests | Stress-test families used to make fixed-point training fail in controlled ways. |
| Controllers | Stable controller IDs used to modify or throttle optimizer updates. |
| Experiments | Actual experiment reports, plots, notebooks, and conclusions. |
Experiment Trace Patternβ
Most technical pages should eventually support experiment traces of the form:
dataset: DS-...
model: MDL-...
precision: PREC-...
stress_test: ST-...
optimizer: ...
controller: CTRL-...
metrics: [...]