gigl.src.training.v1.lib.base_trainer#

Attributes#

The Protocol that you need to implement for your trainer to function with Training

class gigl.src.training.v1.lib.base_trainer.BaseTrainer[source]#

The Protocol that you need to implement for your trainer to function with Training in tabularized mode.

Note: the BaseTrainer class also implements the gigl.src.common.types.model.BaseModelOperationsProtocol which requires the init_model method, and the getter and setter for the model property.

abstract eval(gbml_config_pb_wrapper, device)[source]#

Parameters:

gbml_config_pb_wrapper (gigl.src.common.types.pb_wrappers.gbml_config.GbmlConfigPbWrapper)
device (torch.device)

Return type:

abstract setup_for_training()[source]#

abstract train(gbml_config_pb_wrapper, device, profiler=None)[source]#

Parameters:

gbml_config_pb_wrapper (gigl.src.common.types.pb_wrappers.gbml_config.GbmlConfigPbWrapper)
device (torch.device)
profiler (Optional[gigl.src.common.modeling_task_specs.utils.profiler_wrapper.TorchProfiler])

Return type:

None

property supports_distributed_training: bool[source]#