XLAEnvironment¶

class lightning.pytorch.plugins.environments.XLAEnvironment(*args, **kwargs)[source]¶

Cluster environment for training on a TPU Pod with the PyTorch/XLA library.

A list of environment variables set by XLA can be found here.

static detect()[source]¶

Detects the environment settings corresponding to this cluster and returns True if they match.

global_rank()[source]¶

The rank (index) of the currently running process across all host and devices.

The output is cached for performance.

local_rank()[source]¶

The rank (index) of the currently running process inside of the current host.

The output is cached for performance.

node_rank()[source]¶

The rank (index) of the host on which the current process runs.

The output is cached for performance.

world_size()[source]¶

The number of processes across all devices and hosts.

The output is cached for performance.

property creates_processes_externally: bool¶: Whether the environment creates the subprocesses or not.

property main_address: str¶: The main address through which all processes connect and communicate.

property main_port: int¶: An open and configured port in the main node through which all processes communicate.