Target
Reduce optimizer memory or stabilize deep training.
Failure this recipe addresses
Optimizer state or unstable gradients block an otherwise valid training recipe.
Data contract
The same tiny-overfit and held-out data used by the incumbent recipe.
Method or policy
Change one optimizer or stability mechanism while holding model, data, seed, and eval fixed.
Evaluation contract
Peak memory and task/loss parity versus the incumbent.
- step time
- convergence
- held-out behavior
- numerical stability
Budget and stop rule
Fresh experiment only; one bounded comparison before any sweep.
Stop if memory does not improve materially or convergence/quality regresses.
Decision rule
Adopt only with measured memory benefit and behavior parity.
Learning exercise
Compare the recorded GaLore smoke with the gradient-checkpointing result and identify what each actually saves.
Distinguish optimizer-state memory from activation memory.
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.