xnn.common.benchmark.config.from_dict

xnn.common.benchmark.config.from_dict(d)

Build a BenchmarkConfig from a plain nested dict.

The funnel every frontend loader passes through. The models list may hold strings and/or mappings (see _model_entry()); data reuses the dataclass from a normal run (unknown keys ignored); metrics (with the optional flat-form targets) is normalized into the canonical per-target mapping (see _metric_map()); and output populates OutputConfig.

Parameters:

d (dict[str, Any]) – Nested benchmark configuration mapping. None is treated as empty.

Returns:

The populated benchmark configuration.

Return type:

BenchmarkConfig