LEARN

Concepts

Concepts

These pages explain the why and how behind Laco's design. Read them when you want to understand a mechanism, not just how to use it.

PageRead this if…
Config as PythonYou're wondering why .py files instead of YAML
Lie-TypingYou're confused why L.call(T)(...) has type T at edit time but is a DictConfig at runtime
Lazy ConstructionYou want to understand L.call vs L.partial vs L.just vs L.required
InterpolationYou want to use @L.params, L.ref, or the L.r.* resolver namespace
Typed GroupsYou want swappable, type-safe config variants (optimizer, activation, etc.)
TracingYou want to write ordinary Python and get a config tree back automatically
App LoopYou're wiring up @L.task and laco.main for a full training script