all_gather_into_tensor is deprecated, _gpu_gather warns on every gather

[qgallouedec]

System Info

accelerate 1.15.0, torch 2.13.0+cu129, transformers 5.18.0.dev0, 8 x H100.

Information

• The official example scripts
• My own modified scripts

Reproduction

_gpu_gather uses an API torch deprecated, so every Accelerator.gather logs a FutureWarning once per rank:

torch/distributed/c10d_logger.py:83: FutureWarning: `torch.distributed.all_gather_into_tensor` is deprecated.
Please use `torch.distributed.all_gather_single` instead.

# utils/operations.py:324
gather_op = torch.distributed.all_gather_into_tensor

It is reached from any Trainer run, Trainer._get_num_items_in_batch -> Accelerator.gather -> _gpu_gather -> _gpu_gather_one.

Expected behavior

No warning. torch.distributed.all_gather_single exists in torch 2.13, so this can be a rename behind a version check for older torch.

[lindicaphxag-tech]

PR #4271 uses all_gather_single when available and keeps the existing fallback and returned shape for older PyTorch versions. The focused regression test verifies the output-buffer layout and avoids the deprecation warning. Could a maintainer assign/review the issue?