xnn.common.benchmark.runner.select_entries#

xnn.common.benchmark.runner.select_entries(cfg, models=None, shard=None)[source]#

Pick the model entries one run of a split benchmark scores.

Parameters:

cfg (BenchmarkConfig) – The full benchmark configuration.
models (list of str, str or None, optional) – Labels (or zero-based positions) of the entries to keep, as a list or one comma-separated string. None keeps all.
shard (str or None, optional) – "I/N" keeps every N-th entry starting at I (zero-based), so N runs with I = 0 .. N-1 cover the list; "slurm" reads I and N from the Slurm job-array variables (SLURM_ARRAY_TASK_ID relative to SLURM_ARRAY_TASK_MIN, and SLURM_ARRAY_TASK_COUNT). Applied after models.

Returns:

The selected entries, in config order.

Return type:

list of ModelEntry

Raises:

ValueError – For an unknown label, a malformed shard, or "slurm" outside a job array.