`aip_sdk.ConfigIdentityMatch`

aip_sdk.ConfigIdentityMatch(metric: str, config_name: str, config_version: str, config_scope: str, match: bool, in_run_a: bool = False, in_run_b: bool = False, params: dict[str, Any] | None = None, param_diff: dict[str, tuple[Any, Any]] = dict[str, tuple[Any, Any]]())

One config identity's comparability between two diffed runs.

An entry in the mapping returned by RunDiff.config_identity_match(), keyed on the opaque "{config_name}@{config_version}@{config_scope}" string. Each entry repeats its metric and config identity so results can be grouped or filtered without parsing the key.

Attributes

metric str: Name of the metric this config was resolved for.
config_name str: Name of the stored metric config.
config_version str: Version of the stored metric config.
config_scope str: Scope the config was stored under — a workspace id, the literal "platform" for a platform-wide config, or "" for a run scored before the platform recorded scope at all.
match bool: False only when both runs pinned this metric to different stored-config identities; an unpinned side imposes no comparability constraint and reports True.
in_run_a bool: Whether run A resolved this identity. Both in_run_a/in_run_b are True for a shared identity; exactly one is True for either side of a mismatch.
in_run_b bool: Whether run B resolved this identity.
params dict[str, Any] | None: This identity's stored params — fetched client-side, only with include_params=True; None if unresolvable (e.g. a platform-wide/legacy scope, or a since-deleted config).
param_diff dict[str, tuple[Any, Any]]: Per-param (value_a, value_b) where run A/B's configs for this metric disagree. A side that never set the key reports PARAM_ABSENT rather than None, which means the key was set to null. Only populated for a clean one-config-per-side mismatch with both sides' params fetched.