v0.13.2 Cannot from accelerate import Accelerator in Windows since module 'signal' has no attribute 'SIGKILL'

[T-Atlas]

System Info

v0.13.2 (working nomally when in v0.12.0)
window 11
py 3.9.13
torch 1.13.0+gpu

Information

• The official example scripts
• My own modified scripts

Tasks

• One of the scripts in the examples/ folder of Accelerate or an officially supported no_trainer script in the examples folder of the transformers repo (such as run_no_trainer_glue.py)
• My own task or dataset (give details below)

Reproduction

from accelerate import Accelerator

Expected behavior

(d2l) PS C:\Users\LianJunhong\Desktop\eat_pytorch_in_20_days> accelerate env
NOTE: Redirects are currently not supported in Windows or MacOs.
Traceback (most recent call last):
  File "D:\Miniconda3\envs\d2l\lib\runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "D:\Miniconda3\envs\d2l\lib\runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "D:\Miniconda3\envs\d2l\Scripts\accelerate.exe\__main__.py", line 4, in <module>
  File "D:\Miniconda3\envs\d2l\lib\site-packages\accelerate\__init__.py", line 7, in <module>
    from .accelerator import Accelerator
  File "D:\Miniconda3\envs\d2l\lib\site-packages\accelerate\accelerator.py", line 27, in <module>
    from .checkpointing import load_accelerator_state, load_custom_state, save_accelerator_state, save_custom_state
  File "D:\Miniconda3\envs\d2l\lib\site-packages\accelerate\checkpointing.py", line 24, in <module>
    from .utils import (
  File "D:\Miniconda3\envs\d2l\lib\site-packages\accelerate\utils\__init__.py", line 96, in <module>
    from .launch import PrepareForLaunch, _filter_args, get_launch_prefix
  File "D:\Miniconda3\envs\d2l\lib\site-packages\accelerate\utils\launch.py", line 25, in <module>
    import torch.distributed.run as distrib_run
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\run.py", line 386, in <module>
    from torch.distributed.launcher.api import LaunchConfig, elastic_launch
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\launcher\__init__.py", line 10, in <module>
    from torch.distributed.launcher.api import (  # noqa: F401
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\launcher\api.py", line 15, in <module>
    from torch.distributed.elastic.agent.server.api import WorkerSpec
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\elastic\agent\server\__init__.py", line 40, in <module>
    from .local_elastic_agent import TORCHELASTIC_ENABLE_FILE_TIMER, TORCHELASTIC_TIMER_FILE
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\elastic\agent\server\local_elastic_agent.py", line 19, in <module>
    import torch.distributed.elastic.timer as timer
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\elastic\timer\__init__.py", line 44, in <module>
    from .file_based_local_timer import FileTimerClient, FileTimerServer, FileTimerRequest  # noqa: F401
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\elastic\timer\file_based_local_timer.py", line 63, in <module>
    class FileTimerClient(TimerClient):
  File "D:\Miniconda3\envs\d2l\lib\site-packages\torch\distributed\elastic\timer\file_based_local_timer.py", line 81, in FileTimerClient
    def __init__(self, file_path: str, signal=signal.SIGKILL) -> None:
AttributeError: module 'signal' has no attribute 'SIGKILL'

[sgugger]

This comes from PyTorch directly, you can follow up the issue and its resolution here. In the meantime, you should downgrade your PyTorch to a version < 1.13.

[T-Atlas]

This comes from PyTorch directly, you can follow up the issue and its resolution here. In the meantime, you should downgrade your PyTorch to a version < 1.13.

Thanks! I will follow up that issue

[versus666jzx]

Same issue

• python3.10
• PyTorch 1.13.0+gpu
• Win 11

File "C:\Users\Me\venv\lib\site-packages\torch\distributed\elastic\timer\file_based_local_timer.py", line 81, in FileTimerClient
    def __init__(self, file_path: str, signal=signal.SIGKILL) -> None:
AttributeError: module 'signal' has no attribute 'SIGKILL'. Did you mean: 'SIGILL'?

Downgrade torch to 1.12.1 solve this.

[T-Atlas]

Same issue

• python3.10
• PyTorch 1.13.0+gpu
• Win 11

File "C:\Users\Me\venv\lib\site-packages\torch\distributed\elastic\timer\file_based_local_timer.py", line 81, in FileTimerClient
    def __init__(self, file_path: str, signal=signal.SIGKILL) -> None:
AttributeError: module 'signal' has no attribute 'SIGKILL'. Did you mean: 'SIGILL'?

Downgrade torch to 1.12.1 solve this.

I think so, but if you need to try PyTorch version==1.13.0, you can use accelerate v0.12.0.

[muellerzr]

@T-Atlas or @versus666jzx can you verify that retrying and installing accelerate with pip install git+https://github.com/huggingface/accelerate fixes your issues? Thanks!

[T-Atlas]

@muellerzr Bug fixed with accelerate-0.14.0.dev0. Thanks!

[versus666jzx]

@muellerzr I confirm that the bug has been fixed. Thanks a lot!

• torch - 1.13.0+cu117
• accelerate - 0.14.0.dev0