TPUSpawnStrategy¶

class pytorch_lightning.strategies.TPUSpawnStrategy(accelerator=None, parallel_devices=None, checkpoint_io=None, precision_plugin=None, debug=False, **_)[source]¶

Bases: pytorch_lightning.strategies.ddp_spawn.DDPSpawnStrategy

Strategy for training multiple TPU devices using the torch_xla.distributed.xla_multiprocessing.spawn() method.

all_gather(tensor, group=None, sync_grads=False)[source]¶

Function to gather a tensor from several distributed processes :type _sphinx_paramlinks_pytorch_lightning.strategies.TPUSpawnStrategy.all_gather.tensor: Tensor :param _sphinx_paramlinks_pytorch_lightning.strategies.TPUSpawnStrategy.all_gather.tensor: tensor of shape (batch, …) :type _sphinx_paramlinks_pytorch_lightning.strategies.TPUSpawnStrategy.all_gather.group: Optional[Any] :param _sphinx_paramlinks_pytorch_lightning.strategies.TPUSpawnStrategy.all_gather.group: not available with TPUs :type _sphinx_paramlinks_pytorch_lightning.strategies.TPUSpawnStrategy.all_gather.sync_grads: bool :param _sphinx_paramlinks_pytorch_lightning.strategies.TPUSpawnStrategy.all_gather.sync_grads: not available with TPUs

Return type: Tensor
Returns: A tensor of shape (world_size, batch, …)

barrier(name=None)[source]¶

Synchronizes all processes which blocks processes until the whole group enters this function.

Parameters: name¶ (Optional[str]) – an optional name to pass into barrier.
Return type: None

broadcast(obj, src=0)[source]¶

Broadcasts an object to all processes.

Parameters

obj¶ (object) – the object to broadcast
src¶ (int) – source rank

Return type

object

connect(model)[source]¶

Called by the accelerator to connect the accelerator and the model with this plugin.

Return type: None

model_to_device()[source]¶

Moves the model to the correct device.

Return type: None

predict_step(*args, **kwargs)[source]¶

The actual predict step.

See predict_step() for more details

Return type: Union[Tensor, Dict[str, Any]]

process_dataloader(dataloader)[source]¶

Wraps the dataloader if necessary.

Parameters: dataloader¶ (DataLoader) – iterable. Ideally of type: torch.utils.data.DataLoader
Return type: None

reduce(output, group=None, reduce_op=None)[source]¶

Reduces a tensor from several distributed processes to one aggregated tensor.

Parameters

tensor¶ – the tensor to sync and reduce
group¶ (Optional[Any]) – the process group to gather results from. Defaults to all processes (world)
reduce_op¶ (Union[ReduceOp, str, None]) – the reduction operation. Defaults to ‘mean’/’avg’. Can also be a string ‘sum’ to calculate the sum during reduction.

Returns

reduced value, except when the input was not a tensor the output remains is unchanged

reduce_boolean_decision(decision)[source]¶

Reduce the early stopping decision across all processes.

Return type: bool

remove_checkpoint(filepath)[source]¶

Remove checkpoint filepath from the filesystem.

Parameters: filepath¶ (Union[str, Path]) – Path to checkpoint
Return type: None

save_checkpoint(checkpoint, filepath, storage_options=None)[source]¶

Save model/training states as a checkpoint file through state-dump and file-write.

Parameters

checkpoint¶ (Dict[str, Any]) – dict containing model and trainer state
filepath¶ (Union[str, Path]) – write-target file’s path
storage_options¶ (Optional[Any]) – parameter for how to save to storage, passed to CheckpointIO plugin

Return type