technical writeup

TrainLoop LoRA geometry writeup

Effective update geometry, layer placement, and controlled rank analysis can explain why two adapters with similar parameter counts behave differently. Geometry diagnostics describe an update; they do not establish task improvement without the same frozen behavioral evaluation used for the baseline.

Disposition
scaffolded
Study class
technical writeup
Reviewed
2026-09-23
Evidence
Tracked source record

What mattered

Effective update geometry, layer placement, and controlled rank analysis can explain why two adapters with similar parameter counts behave differently.

What changed locally

PostTrainLLM added LoRA geometry diagnostics and retained a controlled rank-and-layer sweep as a possible future experiment.

Evidence limits

Geometry diagnostics describe an update; they do not establish task improvement without the same frozen behavioral evaluation used for the baseline.

Retained disposition

Attach lora-geometry.json to the next authorized adapter run and compare it beside task and regression scores.

The action is retained as study context rather than an active backlog item. Any implementation, download, training run, or benchmark requires a fresh scoped question under the repository's experiment gate.

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/studies/registry.json. The links below are the tracked evidence and explanatory sources preserved with the record.