Maturity matrix¶
Warning
Pre-implementation. This page describes proposed contracts. Class signatures, parameter types, schema fields, and behavior are subject to change before code lands. Once implementation exists, content here will be regenerated from docstrings or sourced from running tests.
The authoritative per-feature status table. Each task, dataset, schema family, API module, CLI command, and governance instrument carries a version badge (when it ships) and a maturity label (its contract state), using the vocabulary in STYLE.md. The matrix aggregates the per-feature facts recorded in Task reference, Datasets, the Schema reference, the API reference, CLI reference, and the roadmap; when a row disagrees with its source page, the source page wins until this matrix is updated.
Until src/emma/ lands, every API and schema row is proposed-contract, and every task and dataset row is proposed-contract at the maturity its source page records. Items whose contract is not yet pinned are planned.
Vocabulary¶
Maturity labels are the closed set defined in STYLE.md and in the Maturity enum:
Label |
Meaning |
|---|---|
|
Named roadmap item; no contract yet. |
|
Intended API, schema, or recipe shape; may still change before code lands. |
|
Code exists; examples are runnable. |
|
Implementation has contract tests or golden validation. |
|
Scheduled for removal; use the replacement instead. |
Version badges render as in STYLE.md: v0.1 is the emerald live accent for the launch version; plain v0.2 marks a future version.
Tasks¶
All 15 TaskID members, at the maturity recorded in Task reference. The per-task metric, OOD (out-of-distribution) axis, readout head, and domain wiring live there.
Feature |
Version |
Maturity |
|---|---|---|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v1 |
|
|
v0.2 |
|
|
v0.2 |
|
|
v0.2 |
|
|
v1 |
|
|
v1 |
|
|
v1 |
|
|
v1 |
|
|
v2 |
|
|
v2 |
|
Datasets¶
All 15 dataset ids, at the maturity recorded in Datasets. Recipes that depend on an rfgen feature not yet shipped carry an open dependency flag on their per-dataset page; that flag does not change the maturity here, but it does block scoring until the dependency lands.
Feature |
Version |
Maturity |
|---|---|---|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.2 |
|
|
v0.2 |
|
|
v0.2 |
|
|
v1 |
|
|
v1 |
|
|
v1 |
|
|
v1 |
|
|
v1 |
|
|
v2 |
|
|
v2 |
|
Schema families¶
The serialized contracts documented under Schema reference. The two governance records (PreRegistration and ReScoringLog) are listed under Governance below to avoid duplicating the instruments they back.
Feature |
Version |
Maturity |
|---|---|---|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
API modules¶
The Python modules documented under API reference. Each module ships at v0.1 even when it contains later-version classes (for example the emma.tasks module includes the v2 caption and QA heads).
Feature |
Version |
Maturity |
|---|---|---|
Primitives ( |
v0.1 |
|
Schemas ( |
v0.1 |
|
Datasets ( |
v0.1 |
|
Metrics ( |
v0.1 |
|
Tasks ( |
v0.1 |
|
Harness ( |
v0.1 |
|
Leaderboard ( |
v0.1 |
|
CLI ( |
v0.1 |
|
Governance ( |
v0.1 |
|
CLI commands¶
The emma command surface, at the maturity recorded in CLI reference.
Feature |
Version |
Maturity |
|---|---|---|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
|
v0.1 |
|
Governance instruments¶
The ring-fence machinery documented in Governance and scheduled in the roadmap. The v0.1 instruments are proposed-contract; the v1 and later instruments are planned because their contracts are not yet pinned.
Feature |
Version |
Maturity |
|---|---|---|
Pre-registration (PreRegistration + store) |
v0.1 |
|
Third-party re-scoring (ReScoringLog + writer) |
v0.1 |
|
Operational ring-fence (external co-steward) |
v0.1 |
|
COI (conflict of interest) charter |
v0.1 |
|
Sandboxed code submission |
v1 |
|
Independent steering committee |
v1 |
|
Independent non-profit operator |
v1+ |
|
Note
The transition from proposed-contract to implemented happens per feature when src/emma/ lands that feature; the transition from implemented to verified happens when the contract tests in Validation pass. No row is implemented or verified today.
See Also¶
STYLE.md: the version-badge and maturity-label vocabulary this matrix uses.
Task reference: the per-task metric, OOD axis, and version facts.
Datasets: the per-dataset recipe and version facts.
Schema reference: the per-schema field contracts.
API reference: the module map.
Roadmap: the version sequence and per-feature maturity narrative.