AWS Bedrock CSP LLM Syntax | Text AI Analytic Functions - AWS Bedrock CSP LLM Syntax - Teradata Analytics

AWS_Bedrock_CSP_LLM_arguments
(
    ApiType('aws')
    [AUTHORIZATION([DatabaseName.]AuthorizationObjectName] 
    |{
    AcessKey('<aws access key>')
    SecretKey('<aws secret key>')|
    [SessionKey('<aws_session_key>')]
    }
    Region('<aws region>')
    ModelName('<bedrock-model-name>')
    [ModelArgs('{"Top_P":"top_p","Top_K":"top_k"}')]
)

Required Syntax Elements

ApiType

Identification of the cloud platform.

Permitted values (not case-sensitive): aws, azure, gcp, nim, litellm

AUTHORIZATION

Mutually exclusive with AccessKey, SecretKey, or SessionKey.

Authorization object containing the credentials.

Region

Name of your AWS Bedrock service region.

AccessKey

Mutually exclusive with AUTHORIZATION.

Specifies the AWS Bedrock service access key.

SecretKey

Mutually exclusive with AUTHORIZATION.

Specifies the AWS Bedrock service secret key.

ModelName

Specifies the name of the AWS Bedrock LLM to be used for text analytics tasks.

Permitted values: Case-sensitive alphanumeric string with dashes. The value is used as is for connection to AWS.

Optional Syntax Elements

ModelArgs

Specifies the JSON string representing the inference parameters for the selected LLM model.

ModelOperation

Specifies which API should be used for the model.

Permitted values: converse, invoke.

Default value: invoke. For Claude-3-7 models, default is converse.

BedrockKey

Specifies the AWS Bedrock key.