HPUParallelStrategy¶
- class lightning.pytorch.strategies.HPUParallelStrategy(accelerator=None, parallel_devices=None, cluster_environment=None, checkpoint_io=None, precision_plugin=None, ddp_comm_state=None, ddp_comm_hook=None, ddp_comm_wrapper=None, model_averaging_period=None, process_group_backend='hccl', **kwargs)[source]¶
Bases:
lightning.pytorch.strategies.ddp.DDPStrategy
Strategy for distributed training on multiple HPU devices.
Warning
This is an experimental feature.
- optimizer_step(optimizer, closure, model=None, **kwargs)[source]¶
Performs the actual optimizer step.
- Parameters
- Return type