Target
Compress a frontier or larger-model capability into a routed Mac-local specialist.
Failure this recipe addresses
A small base lacks a narrow behavior that a qualified teacher can demonstrate.
Data contract
Teacher trajectories filtered by a verifiable checker, with a sealed holdout and breadth set.
Method or policy
Train on accepted teacher trajectories; preserve teacher identity, prompts, filtering, and student recipe.
Evaluation contract
Frozen target gate against stock student and frontier ceiling.
- out-of-domain breadth
- safety
- latency
- RAM
- tok/s
Budget and stop rule
Fresh experiment only; budget exact teacher calls, rows, training steps, and Mac runtime before execution.
Stop when the target gate plateaus or breadth exceeds its allowed regression.
Decision rule
Ship as routed-only when depth wins and breadth regresses; reject when target gain is absent or negative transfer dominates.
Learning exercise
Trace the file-ops lineage from stock 4B through distilled, multibackend, and ReST candidates.
State the target gain and breadth cost that force routing.
How to read this dossier
This canonical page separates the retained repository decision from the material that informed it. A source may describe an external claim, historical measurement, local observation, or planned procedure; those evidence classes are not interchangeable. Follow the provenance links before reusing a number or method.
Status and disposition describe what PostTrainLLM retained when this record was indexed. They are not a live product promise, a newly run benchmark, or permission to restart historical work. Unknown or unmeasured fields remain unknown rather than being treated as zero.
To reuse the record, first name the exact claim you need and trace it to the linked source. Then check whether the original environment, model revision, data split, evaluator, hardware, and budget match the proposed use. If they do not, treat the record as a hypothesis or design reference and run the smallest fresh comparison that can falsify it. Preserve negative outcomes and regressions beside any improvement; a local win on one slice does not silently become a general capability claim.
Source provenance
The normalized record comes from docs/recipes/registry.json. The links below are the tracked evidence and explanatory sources preserved with the record.