部署报错提示 module 'torch' has no attribute 'float32'

严格按照 README 教程操作的，环境是 ananconda + python3.8，显卡是 A100 80G 版

依赖也是仓库中 requirements.txt 中的版本，但是运行到第三步的时候报错

Python 3.8.16 | packaged by conda-forge | (default, Feb  1 2023, 16:01:55)
[GCC 11.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from transformers import AutoTokenizer, AutoModelForCausalLM
>>> tokenizer = AutoTokenizer.from_pretrained("fnlp/moss-moon-003-sft", trust_remote_code=True)
Explicitly passing a `revision` is encouraged when loading a model with custom code to ensure no malicious code has been contributed in a newer revision.
>>> model = AutoModelForCausalLM.from_pretrained("fnlp/moss-moon-003-sft", trust_remote_code=True).half().cuda()
Explicitly passing a `revision` is encouraged when loading a configuration with custom code to ensure no malicious code has been contributed in a newer revision.
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/mnt/conda-envs/moss/lib/python3.8/site-packages/transformers/models/auto/auto_factory.py", line 441, in from_pretrained
    config, kwargs = AutoConfig.from_pretrained(
  File "/mnt/conda-envs/moss/lib/python3.8/site-packages/transformers/models/auto/configuration_auto.py", line 935, in from_pretrained
    return config_class.from_pretrained(pretrained_model_name_or_path, **kwargs)
  File "/mnt/conda-envs/moss/lib/python3.8/site-packages/transformers/configuration_utils.py", line 553, in from_pretrained
    return cls.from_dict(config_dict, **kwargs)
  File "/mnt/conda-envs/moss/lib/python3.8/site-packages/transformers/configuration_utils.py", line 696, in from_dict
    config = cls(**config_dict)
  File "/home/almalinux/.cache/huggingface/modules/transformers_modules/moss-moon-003-sft/configuration_moss.py", line 116, in __init__
    super().__init__(
  File "/mnt/conda-envs/moss/lib/python3.8/site-packages/transformers/configuration_utils.py", line 338, in __init__
    self.torch_dtype = getattr(torch, self.torch_dtype)
AttributeError: module 'torch' has no attribute 'float32'

Start with the README's deployment tutorial and requirements.txt, reproducing the error at the third step in the reported Python 3.8 environment. Inspect the torch and transformers versions involved, then follow the traceback into configuration_moss.py. Done means the documented setup can load the model without the reported AttributeError, with the compatible dependency guidance captured in the relevant setup documentation.