Combines a shared language base with adapter-compatible skills and independent non-language or structurally distinct experts.
Architecture guide: this topic defines a modular tiny-model planning contract. It does not claim that model artifacts exist, are compatible, or execute on this WordPress site.
Recommended decomposition
Use a shared language base for skills that truly share its tokenizer and layer structure. Keep retrieval encoders, classifiers, rankers, policy models, audio, vision, or other distinct architectures as independent experts.
Routing implications
The controller may first select an architecture family and then an adapter within that family. This creates two routing decisions with separate confidence and fallback behavior.
Evidence
Measure base residency, adapter switching, standalone initialization, cross-representation handoff, and end-to-end quality rather than benchmarking components in isolation only.
Scope
This starter page defines the questions, boundaries, evidence, and failure modes that should be recorded before a capability is presented as supported.
Engineering considerations
- Identify the source, version, target environment, and owner.
- Separate observed values from estimates and externally reported values.
- Record trade-offs, unsupported cases, and fallback behavior.
- Link performance statements to a compatible benchmark methodology.
Verification questions
- What exact artifact, revision, backend, and environment were reviewed?
- Which assumptions could change the result?
- Which data should be retained so another engineer can reproduce the conclusion?