Target
Reduce stored or executed model structure after capability is proven.
Failure this recipe addresses
Artifact size or dense topology is the measured deployment bottleneck.
Data contract
Frozen quality and latency gates with the original checkpoint as baseline.
Method or policy
Prune weights, heads, or layers; distinguish zeroing from physical topology removal.
Evaluation contract
Quality at matched artifact size and real wall-clock execution.
- perplexity
- task score
- file size
- RAM
- latency
Budget and stop rule
Existing smokes closed; future use requires a shipped candidate with a measured size/latency blocker.
Stop if sparsity does not translate into storage or execution savings.
Decision rule
Do not call zero weights a speedup without a sparse or physically smaller runtime path.
Learning exercise
Compare 50% zero weights with physical layer removal and predict which metrics change.
Distinguish statistical sparsity, compressed storage, and runtime sparsity.
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.