class tango.step_graph.StepGraph(step_dict)[source]#

Represents an experiment as a directed graph.

It can be treated as a Mapping of step names (str) to Step.

classmethod ordered_steps(step_dict)[source]#

Returns the steps in this step graph in an order that can be executed one at a time.

This does not take into account which steps may be cached. It simply returns an executable order of steps.

Return type


to_file(filename, include_unique_id=False)[source]#

Note: In normal use cases, include_unique_id should always be False. We do not want to save the unique id in the config, because we want the output to change if we modify other kwargs in the config file. We include this flag for MulticoreExecutor, to ensure that steps have the same unique id in the main process and the created subprocesses.

Return type