Refresh multilingual benchmarks and fix overlapping gold evaluation

Recompute published benchmark results for all default language models,
exclude Polish Polimorf, add Hebrew documentation, and record the current
benchmark environment. Evaluate repeated surface forms as an overlapping
gold cover and publish only applicable metrics for candidate policies.
This commit is contained in:
2026-07-23 17:06:41 +02:00
parent 1f1b03c6a8
commit b29699b763
64 changed files with 5392 additions and 4765 deletions

View File

@@ -189,4 +189,8 @@ Local validation for PoliMorf 1.0.0 is:
## Documentation and troubleshooting
`prepareMkDocsSource` generates the catalog only at `build/mkdocs-source/stemmer-model-catalog.md`; generated Markdown and rendered site content are not tracked. For runtime failures, dependency inspection, ClassLoader isolation, and fat-JAR guidance, see [Model Selection and Loading](model-selection-and-loading.md#troubleshooting).
`publishModelCatalogDocumentation` updates the checked-in catalog used by a direct local `mkdocs serve`.
`prepareMkDocsSource` independently regenerates the same catalog under `build/mkdocs-source/` for the
publication workflow, and `verifyModelCatalogDocumentation` fails when the two copies differ. Rendered
site content remains untracked. For runtime failures, dependency inspection, ClassLoader isolation, and
fat-JAR guidance, see [Model Selection and Loading](model-selection-and-loading.md#troubleshooting).