retrieveAndGenerate 语法错误：未知参数生成配置或检索配置（Claude-v3、Amazon Bedrock）

Question

我正在尝试使用 claude-v3 模型从知识库检索并生成响应。为此，我遵循了boto3文档和亚马逊上的博客文章并编写了这个方法：

def retrieveAndGenerate(input, kbId, modelArn=None):
    response = boto_runtime.retrieve_and_generate(
        input={
            'text': input
        },
        retrieveAndGenerateConfiguration={
            'knowledgeBaseConfiguration': {
                'generationConfiguration': {
                    'promptTemplate': {
                        'textPromptTemplate': promptTemplate
                    }
                },
                'knowledgeBaseId': kbId,
                'modelArn': modelArn,
                "retrievalConfiguration": {
                    'vectorSearchConfiguration': {
                        'numberOfResults': 5
                    }
                }
            },
            'type': 'KNOWLEDGE_BASE'
        }
    )
    
    return response

但是它给了我这个错误：

ParamValidationError: Parameter validation failed: Unknown parameter in retrieveAndGenerateConfiguration.knowledgeBaseConfiguration: "generationConfiguration", must be one of: knowledgeBaseId, modelArn
Unknown parameter in retrieveAndGenerateConfiguration.knowledgeBaseConfiguration: "retrievalConfiguration", must be of one: knowledgeBaseId, modelArn

我将该函数称为：

anthropicModelArns = [arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0']
response = retrieveAndGenerate(input='Felsefe nedir?', kbId='VPY6GXXXXX', modelArn=anthropicModelArns[0])

即使上述字段之一也会引发相同的错误。

异常的完整跟踪

ParamValidationError                      Traceback (most recent call last)
Cell In[45], line 1
----> 1 response = retrieveAndGenerate(input='Felsefe nedir?', kbId='VPY6GXXXX', modelArn=anthropicModelArns[0])

Cell In[44], line 2
      1 def retrieveAndGenerate(input, kbId, modelArn=None):
----> 2     response = boto_runtime.retrieve_and_generate(
      3         input={
      4             'text': input
      5         },
      6         retrieveAndGenerateConfiguration={
      7             'knowledgeBaseConfiguration': {
      8                 'generationConfiguration': {
      9                     'promptTemplate': {
     10                         'textPromptTemplate': promptTemplate
     11                     }
     12                 },
     13                 'knowledgeBaseId': kbId,
     14                 'modelArn': modelArn,
     15                 "retrievalConfiguration": {
     16                     'vectorSearchConfiguration': {
     17                         'numberOfResults': 5
     18                     }
     19                 }
     20             },
     21             'type': 'KNOWLEDGE_BASE'
     22         }
     23     )
     25     return response

File /usr/local/lib/python3.12/site-packages/botocore/client.py:553, in ClientCreator._create_api_method.<locals>._api_call(self, *args, **kwargs)
    549     raise TypeError(
    550         f"{py_operation_name}() only accepts keyword arguments."
    551     )
    552 # The "self" in this scope is referring to the BaseClient.
--> 553 return self._make_api_call(operation_name, kwargs)

File /usr/local/lib/python3.12/site-packages/botocore/client.py:962, in BaseClient._make_api_call(self, operation_name, api_params)
    958 if properties:
    959     # Pass arbitrary endpoint info with the Request
    960     # for use during construction.
    961     request_context['endpoint_properties'] = properties
--> 962 request_dict = self._convert_to_request_dict(
    963     api_params=api_params,
    964     operation_model=operation_model,
    965     endpoint_url=endpoint_url,
    966     context=request_context,
    967     headers=additional_headers,
    968 )
    969 resolve_checksum_context(request_dict, operation_model, api_params)
    971 service_id = self._service_model.service_id.hyphenize()

File /usr/local/lib/python3.12/site-packages/botocore/client.py:1036, in BaseClient._convert_to_request_dict(self, api_params, operation_model, endpoint_url, context, headers, set_user_agent_header)
   1027 def _convert_to_request_dict(
   1028     self,
   1029     api_params,
   (...)
   1034     set_user_agent_header=True,
   1035 ):
-> 1036     request_dict = self._serializer.serialize_to_request(
   1037         api_params, operation_model
   1038     )
   1039     if not self._client_config.inject_host_prefix:
   1040         request_dict.pop('host_prefix', None)

File /usr/local/lib/python3.12/site-packages/botocore/validate.py:381, in ParamValidationDecorator.serialize_to_request(self, parameters, operation_model)
    377     report = self._param_validator.validate(
    378         parameters, operation_model.input_shape
    379     )
    380     if report.has_errors():
--> 381         raise ParamValidationError(report=report.generate_report())
    382 return self._serializer.serialize_to_request(
    383     parameters, operation_model
    384 )

ParamValidationError: Parameter validation failed:
Unknown parameter in retrieveAndGenerateConfiguration.knowledgeBaseConfiguration: "generationConfiguration", must be one of: knowledgeBaseId, modelArn
Unknown parameter in retrieveAndGenerateConfiguration.knowledgeBaseConfiguration: "retrievalConfiguration", must be one of: knowledgeBaseId, modelArn

我尝试将

generationConfiguration

和

retrievalConfiguration

排除在

knowledgeBaseConfiguration

之外，但这些情况也引发了相同的错误。

它仅适用于最少的必填字段，如下所示：

def retrieveAndGenerate(input, kbId, modelArn=None):
    response = boto_runtime.retrieve_and_generate(
        input={
            'text': input
        },
        retrieveAndGenerateConfiguration={
            'knowledgeBaseConfiguration': {
                'knowledgeBaseId': kbId,
                'modelArn': modelArn
            },
            'type': 'KNOWLEDGE_BASE'
        }
    )
    
    return response

我的错误是什么以及如何解决？感谢您的回复。

Answer 1

我遇到了同样的问题，但在 lambda 上更新 boto3 版本后解决了。

您可以直接在 Lambda 代码中或使用 Lambda 层更新 boto3。有关在 AWS Lambda 中更新 boto3 的指南，您可以参考此 Stack Overflow 讨论：AWS Lambda 控制台 - 升级 boto3 版本。