Remove the deprecated AllGatherGrad class

src/pytorch_lightning/CHANGELOG.md

@@ -35,6 +35,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
   * Removed the `Trainer(ipus=...)` argument
   * Removed the `Trainer(num_processes=...)` argument
 
+- Removed the deprecated `pytorch_lightning.utilities.AllGatherGrad` class ([#16360](https://github.com/Lightning-AI/lightning/pull/16360))
+
 - Removed the deprecated `resume_from_checkpoint` Trainer argument ([#16167](https://github.com/Lightning-AI/lightning/pull/16167))
 
 - Removed the deprecated automatic GPU selection ([#16184](https://github.com/Lightning-AI/lightning/pull/16184))

src/pytorch_lightning/utilities/__init__.py

@@ -17,7 +17,6 @@
 
 from lightning_fabric.utilities import LightningEnum  # noqa: F401
 from lightning_fabric.utilities import move_data_to_device  # noqa: F401
-from pytorch_lightning.utilities.distributed import AllGatherGrad  # noqa: F401
 from pytorch_lightning.utilities.enums import GradClipAlgorithmType  # noqa: F401
 from pytorch_lightning.utilities.grads import grad_norm  # noqa: F401
 from pytorch_lightning.utilities.imports import (  # noqa: F401

src/pytorch_lightning/utilities/distributed.py

@@ -12,10 +12,9 @@
 # limitations under the License.
 """Utilities that can be used with distributed training."""
 
-from typing import Any, Callable, Dict, Optional, Tuple
+from typing import Any, Callable, Dict, Optional
 
 import torch
-from torch import Tensor
 from torch.nn.parallel.distributed import DistributedDataParallel
 
 from lightning_fabric.utilities.distributed import _all_gather_ddp_if_available as new_all_gather_ddp_if_available
@@ -177,40 +176,6 @@ def gather_all_tensors(*args: Any, **kwargs: Any) -> Any:
     return new_gather_all_tensors(*args, **kwargs)
 
 
-class AllGatherGrad(torch.autograd.Function):
-    """Gathers tensors from the whole group and stacks them.
-
-    This implementation is copied from PyTorch.
-
-    .. deprecated:: v1.8.0
-        This function has been deprecated in v1.8.0 in favor of :func:`torch.distributed.nn.functional.all_gather` and
-        will be removed in v2.0.0.
-    """
-
-    @staticmethod
-    def forward(  # type: ignore[override]
-        ctx: Any,
-        tensor: Tensor,
-        group: Optional["torch.distributed.ProcessGroup"] = None,
-    ) -> Tensor:
-        rank_zero_deprecation(
-            "`AllGatherGrad` has been deprecated in v1.8.0 and will be removed in v2.0.0."
-            " Use `torch.distributed.nn.functional.all_gather` instead.",
-            stacklevel=6,
-        )
-        ctx.group = group
-        gathered_tensor = [torch.zeros_like(tensor) for _ in range(torch.distributed.get_world_size())]
-        torch.distributed.all_gather(gathered_tensor, tensor, group=group)
-        gathered_tensor = torch.stack(gathered_tensor, dim=0)
-        return gathered_tensor
-
-    @staticmethod
-    def backward(ctx: Any, *grad_output: Tensor) -> Tuple[Tensor, None]:
-        grad_output = torch.cat(grad_output)
-        torch.distributed.all_reduce(grad_output, op=torch.distributed.ReduceOp.SUM, async_op=False, group=ctx.group)
-        return grad_output[torch.distributed.get_rank()], None
-
-
 def get_default_process_group_backend_for_device(*args: Any, **kwargs: Any) -> Any:
     rank_zero_deprecation(
         "`pytorch_lightning.utilities.distributed.get_default_process_group_backend_for_device` has been deprecated"

tests/tests_pytorch/deprecated_api/test_remove_2-0.py

@@ -303,16 +303,6 @@ def test_tuning_trainer_property():
         trainer.tuning = True
 
 
-@RunIf(skip_windows=True)
-def test_v1_8_0_deprecated_all_gather_grad():
-    tensor1 = torch.ones(1, requires_grad=True)
-    with mock.patch("torch.distributed.all_gather"), mock.patch("torch.distributed.get_world_size", return_value=1):
-        from pytorch_lightning.utilities import AllGatherGrad
-
-        with pytest.deprecated_call(match="`AllGatherGrad` has been deprecated in v1.8"):
-            AllGatherGrad.apply(tensor1)
-
-
 def test_v1_8_1_deprecated_rank_zero_only():
     from pytorch_lightning.utilities.distributed import rank_zero_only