gigl.orchestration.local.runner#

Attributes#

`PipelineConfig`	Configuration for the GiGL pipeline.
`Runner`	Orchestration of GiGL Pipeline with local execution.

class gigl.orchestration.local.runner.PipelineConfig[source]#

Configuration for the GiGL pipeline.

Parameters:

applied_task_identifier (AppliedTaskIdentifier) – your job name
task_config_uri (Uri) – URI to your template task config
resource_config_uri (Uri) – URI to your resource config
custom_cuda_docker_uri (Optional[str]) – For custom training spec and GPU training on VertexAI
custom_cpu_docker_uri (Optional[str]) – For custom training spec and CPU training on VertexAI
dataflow_docker_uri (Optional[str]) – For custom datapreprocessor spec that will run in dataflow

applied_task_identifier: gigl.src.common.types.AppliedTaskIdentifier[source]#

class gigl.orchestration.local.runner.Runner[source]#

Orchestration of GiGL Pipeline with local execution.

Parameters:

pipeline_config (PipelineConfig) – Configuration for the pipeline.
start_at (str) – Component to start the pipeline from. Default is config_populator.

static config_check(start_at, pipeline_config)[source]#

Parameters:

static run(pipeline_config, start_at=GiGLComponents.ConfigPopulator.value)[source]#

Runs the GiGL pipeline locally starting from the specified component.

Parameters:

pipeline_config (PipelineConfig) – Configuration for the pipeline.
start_at (str) – Component to start the pipeline from. Defaults to ‘config_populator’.

Returns:

None

static run_config_populator(pipeline_config)[source]#

static run_data_preprocessor(pipeline_config)[source]#

static run_inferencer(pipeline_config)[source]#

static run_split_generator(pipeline_config)[source]#

static run_subgraph_sampler(pipeline_config)[source]#

static run_trainer(pipeline_config)[source]#