score_composition

Compose complete block metrics from isolated subblock measurements.

Classes

ComposedScoreRecord

ComposedScoreTable

Functions

compose_full_block_metrics

Compose every canonical full block from isolated component deltas.

composed_table_to_gathered_metrics

Convert a composed table to the established multi-layer MIP schema.
class ComposedScoreRecord

Bases: object

__init__(solution_index, layer_idx, block_config, metrics, provenance, source_result_ids)
Parameters:

  • solution_index (int)

  • layer_idx (int)

  • block_config (BlockConfig)

  • metrics (Mapping[str, float])

  • provenance (str)

  • source_result_ids (tuple[str, ...])

Return type:

None

block_config: BlockConfig
layer_idx: int
metrics: Mapping[str, float]
provenance: str
solution_index: int
source_result_ids: tuple[str, ...]
class ComposedScoreTable

Bases: object

__init__(records)
Parameters:

records (tuple[ComposedScoreRecord, ...])

Return type:

None

records: tuple[ComposedScoreRecord, ...]
compose_full_block_metrics(canonical_solutions, subblock_results, *, teacher_blocks, teacher_baseline, exact_results=())

Compose every canonical full block from isolated component deltas.

Exact full-block measurements, when supplied, take precedence over additive estimates for the same semantic block identity.

Parameters:

  • canonical_solutions (Iterable[Mapping[str, Any]])

  • subblock_results (Iterable[Mapping[str, Any]])

  • teacher_blocks (Sequence[BlockConfig])

  • teacher_baseline (Mapping[str, float])

  • exact_results (Iterable[Mapping[str, Any]])

Return type:

ComposedScoreTable

composed_table_to_gathered_metrics(table, canonical_solutions, teacher_records)

Convert a composed table to the established multi-layer MIP schema.

Parameters:

  • table (ComposedScoreTable)

  • canonical_solutions (Sequence[Mapping[str, Any]])

  • teacher_records (Sequence[dict[str, Any]])

Return type:

dict[str, dict[str, Any]]