使用Hugging Face训练AI时遭遇NoCredentialsError问题求助

我平时很少用Hugging Face，最近尝试在本地机器上结合Hugging Face框架和Amazon SageMaker训练AI，但运行训练代码时一直弹出botocore.exceptions.NoCredentialsError: Unable to locate credentials错误，附上完整报错堆栈和训练代码，求解决办法。

报错信息 #

PS C:\Users\gboss\OneDrive\Bureau\Ai training> & C:/Users/gboss/AppData/Local/Programs/Python/Python310/python.exe "c:/Users/gboss/OneDrive/Bureau/Ai training/AiTraining.py"
Traceback (most recent call last):
  File "c:\Users\gboss\OneDrive\Bureau\Ai training\AiTraining.py", line 8, in <module>
    role = iam_client.get_role(RoleName='{IAM_ROLE_WITH_SAGEMAKER_PERMISSIONS}')['Role']['Arn']
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\client.py", line 514, in _api_call
    return self._make_api_call(operation_name, kwargs)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\client.py", line 921, in _make_api_call
    http, parsed_response = self._make_request(
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\client.py", line 944, in _make_request
    return self._endpoint.make_request(operation_model, request_dict)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\endpoint.py", line 119, in make_request
    return self._send_request(request_dict, operation_model)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\endpoint.py", line 198, in _send_request
    request = self.create_request(request_dict, operation_model)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\endpoint.py", line 134, in create_request
    self._event_emitter.emit(
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\hooks.py", line 412, in emit
    return self._emitter.emit(aliased_event_name, **kwargs)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\hooks.py", line 256, in emit
    return self._emit(event_name, kwargs)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\hooks.py", line 239, in _emit
    response = handler(**kwargs)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\signers.py", line 105, in handler
    return self.sign(operation_name, request)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\signers.py", line 189, in sign
    auth.add_auth(request)
  File "C:\Users\gboss\AppData\Local\Programs\Python\Python310\lib\site-packages\botocore\auth.py", line 418, in add_auth
    raise NoCredentialsError()
botocore.exceptions.NoCredentialsError: Unable to locate credentials 

训练代码 #

import sagemaker
import boto3
from sagemaker.huggingface import HuggingFace

# 获取训练作业执行角色

iam_client = boto3.client('iam')
role = iam_client.get_role(RoleName='{IAM_ROLE_WITH_SAGEMAKER_PERMISSIONS}')['Role']['Arn']
hyperparameters = {
    'model_name_or_path':'ZipperXYZ/DialoGPT-medium-TheWorldMachineExpressive2',
    'output_dir':'/opt/ml/model'
    # 添加剩余超参数
}

# Git配置，用于下载微调脚本
git_config = {'repo': 'https://github.com/huggingface/transformers.git','branch': 'v4.17.0'}

# 创建Hugging Face estimator
huggingface_estimator = HuggingFace(
    entry_point='run_clm.py',
    source_dir='./examples/pytorch/language-modeling',
    instance_type='ml.p3.2xlarge',
    instance_count=1,
    role=role,
    git_config=git_config,
    transformers_version='4.17.0',
    pytorch_version='1.10.2',
    py_version='py38',
    hyperparameters = hyperparameters
)

# 启动训练作业
huggingface_estimator.fit()

• 配置本地AWS凭证：boto3无法找到你的AWS身份凭证，需要在本地完成配置：
  1. 安装AWS CLI后，执行aws configure命令，按提示输入你的AWS Access Key ID、Secret Access Key、默认区域（如us-east-1）和输出格式。
  2. 也可以手动在用户目录下创建.aws文件夹，新建credentials文件，内容格式如下：

     [default]
     aws_access_key_id = 你的Access Key
     aws_secret_access_key = 你的Secret Key
     

     同时新建config文件设置默认区域：

     [default]
     region = 你的AWS区域
     

• 替换占位符角色名：代码中的{IAM_ROLE_WITH_SAGEMAKER_PERMISSIONS}是占位符，必须替换为你在AWS IAM中实际创建的、拥有SageMaker权限的角色名称。
• 验证权限有效性：确保你配置的AWS账号有权限访问指定的IAM角色，且该角色已经附加了SageMaker相关权限策略（比如AmazonSageMakerFullAccess）。

内容的提问来源于stack exchange，提问作者ZipperXYZ