lm-modelconvertor
Purpose
The shared model-format bridge: converts production models between
internal and external representations (e.g. to GLB previews) so no
workflow links a native converter library directly.
Mental model
Workflow stages model bytes → convertor transforms format → returns
preview/converted bytes as cache refs for rp-files.persist
transformation: no storage, no estimates, no business decisions.
Ecosystem value
One conversion point for production models:
- A staged file in, converted outputs as cache refs out — same shape for any caller.
- New formats and converter versions land once and upgrade every analysis path.
- Keeps heavy native deps out of workflows and repositories.
Non-goals
- Not file storage or intake orchestration.
- Not preview rendering or slicing estimates.
Responsibility boundary
Owns conversion/transformation routines; does not own upstream model
training, downstream serving, or persistence.
Direct module dependencies
- None
Solution membership
- Not included in a predefined solution
Source
modules/lambdas/convertors/lm-modelconvertor