ModuleNotFoundError: No module named 'accelerate'

[barisuzar]

Hello, I am getting "ModuleNotFoundError: No module named 'accelerate'" error while performing the conversion operation ( pth_to_hf) using the following command even though 'accelerate 1.1.1' is installed.

"xtuner convert pth_to_hf pallava_domain_alignment.py domain_alignment_weight.pth domain_alignment_weight_ft"

The Python version I'm currently using is 3.10.2

You can see some of the installed packages at the bottom.

(new_xtuner_env) D:\new_xtuner_env>xtuner convert pth_to_hf pallava_domain_alignment.py domain_alignment_weight.pth domain_alignment_weight_ft
D:\new_xtuner_env\lib\site-packages\mmengine\optim\optimizer\zero_optimizer.py:11: DeprecationWarning: `TorchScript` support for functional optimizers is deprecated and will be removed in a future PyTorch release. Consider using the `torch.compile` optimizer instead.
  from torch.distributed.optim import \
[2024-11-18 12:20:06,556] [INFO] [real_accelerator.py:158:get_accelerator] Setting ds_accelerator to cuda (auto detect)
D:\new_xtuner_env\lib\site-packages\deepspeed\runtime\zero\linear.py:49: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  def forward(ctx, input, weight, bias=None):
D:\new_xtuner_env\lib\site-packages\deepspeed\runtime\zero\linear.py:67: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  def backward(ctx, grad_output):
W1118 12:20:06.763703 2816 torch\distributed\elastic\multiprocessing\redirects.py:28] NOTE: Redirects are currently not supported in Windows or MacOs.
Traceback (most recent call last):
  File "D:\new_xtuner_env\lib\site-packages\xtuner\tools\model_converters\pth_to_hf.py", line 7, in <module>
    from accelerate import init_empty_weights
ModuleNotFoundError: No module named 'accelerate'

I tried to updating numpy from 1.26.4 to different numpy versions greater than 2.0.0. This seems to have worked because I didn't get any errors regarding 'accelerate'. But this time the output was:

File "D:\new_xtuner_env\lib\site-packages\xtuner\tools\model_converters\pth_to_hf.py", line 7, in <module>
    from accelerate import init_empty_weights
  File "D:\new_xtuner_env\lib\site-packages\accelerate\__init__.py", line 16, in <module>
    from .accelerator import Accelerator
  File "D:\new_xtuner_env\lib\site-packages\accelerate\accelerator.py", line 32, in <module>
    import torch
  File "D:\new_xtuner_env\lib\site-packages\torch\__init__.py", line 2120, in <module>
    from torch._higher_order_ops import cond
  File "D:\new_xtuner_env\lib\site-packages\torch\_higher_order_ops\__init__.py", line 1, in <module>
    from .cond import cond
  File "D:\new_xtuner_env\lib\site-packages\torch\_higher_order_ops\cond.py", line 5, in <module>
    import torch._subclasses.functional_tensor
  File "D:\new_xtuner_env\lib\site-packages\torch\_subclasses\functional_tensor.py", line 42, in <module>
    class FunctionalTensor(torch.Tensor):
  File "D:\new_xtuner_env\lib\site-packages\torch\_subclasses\functional_tensor.py", line 258, in FunctionalTensor
    cpu = _conversion_method_template(device=torch.device("cpu"))
D:\new_xtuner_env\lib\site-packages\torch\_subclasses\functional_tensor.py:258: UserWarning: Failed to initialize NumPy: _ARRAY_API not found (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\utils\tensor_numpy.cpp:84.)
  cpu = _conversion_method_template(device=torch.device("cpu"))
D:\new_xtuner_env\lib\site-packages\mmengine\optim\optimizer\zero_optimizer.py:11: DeprecationWarning: `TorchScript` support for functional optimizers is deprecated and will be removed in a future PyTorch release. Consider using the `torch.compile` optimizer instead.
  from torch.distributed.optim import \
[2024-11-18 12:17:22,979] [INFO] [real_accelerator.py:158:get_accelerator] Setting ds_accelerator to cuda (auto detect)
D:\new_xtuner_env\lib\site-packages\deepspeed\runtime\zero\linear.py:49: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  def forward(ctx, input, weight, bias=None):
D:\new_xtuner_env\lib\site-packages\deepspeed\runtime\zero\linear.py:67: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  def backward(ctx, grad_output):
W1118 12:17:23.193279 7808 torch\distributed\elastic\multiprocessing\redirects.py:28] NOTE: Redirects are currently not supported in Windows or MacOs.
11/18 12:17:23 - mmengine - WARNING - WARNING: command error: 'cannot import name 'BUFSIZE' from 'numpy' (D:\new_xtuner_env\lib\site-packages\numpy\__init__.py)'!
11/18 12:17:23 - mmengine - WARNING -
    Arguments received: ['xtuner', 'convert', 'pth_to_hf', 'pallava_domain_alignment.py', 'domain_alignment_weight.pth', 'domain_alignment_weight_ft']. xtuner commands use the following syntax:                                                                            

---

Libs used:

accelerate 1.1.1
deepspeed 0.11.2+unknown
mmengine 0.10.5
numpy 1.26.4
pip 24.3.1
tokenizers 0.20.3
torch 2.4.0+cu118
torchaudio 2.4.0+cu118
torchvision 0.19.0+cu118
transformers 4.46.2
transformers-stream-generator 0.0.5
xtuner 0.1.23