AWS bedrock documentation change
Summary
Added support for inline input requests in batch inference jobs and updated examples
Security assessment
Introduces new functionality for inline data processing but doesn't address security vulnerabilities or document security-specific features. Focuses on usability enhancements.
Diff
diff --git a/bedrock/latest/userguide/batch-inference-data.md b/bedrock/latest/userguide/batch-inference-data.md index 136a7bd6b..595e31708 100644 --- a//bedrock/latest/userguide/batch-inference-data.md +++ b//bedrock/latest/userguide/batch-inference-data.md @@ -7 +7 @@ -Example text input for Anthropic Claude 3 HaikuExample video input for Amazon NovaExample Converse input +Example text inputExample video input for Amazon NovaExample Converse inputExample inline input @@ -9 +9 @@ Example text input for Anthropic Claude 3 HaikuExample video input for Amazon No -# Format and upload your batch inference data +# Prepare your batch inference data @@ -11 +11 @@ Example text input for Anthropic Claude 3 HaikuExample video input for Amazon No -You must add your batch inference data to an S3 location that you'll choose or specify when submitting a model invocation job. The S3 location must contain the following items: +You must add your batch inference data to an S3 location that you'll choose or specify when submitting a model invocation job. Alternatively, you can pass inference requests inline in the API call using the `inputInlineRequests` field. If you use S3 input, the S3 location must contain the following items: @@ -55 +55 @@ To better understand how to set up your batch inference inputs, see the followin -## Example text input for Anthropic Claude 3 Haiku +## Example text input @@ -57 +57 @@ To better understand how to set up your batch inference inputs, see the followin -If you plan to run batch inference using the [Messages API](./model-parameters-anthropic-claude-messages.html) format for the Anthropic Claude 3 Haiku model, you might provide a JSONL file containing the following JSON object as one of the lines: +If you plan to run batch inference using the [Messages API](./model-parameters-anthropic-claude-messages.html) format for an Anthropic Claude model, you might provide a JSONL file containing the following JSON object as one of the lines: @@ -158,0 +159,56 @@ For the full list of fields supported in the Converse request body, see [Convers +## Example inline input + +Instead of uploading a JSONL file to Amazon S3, you can pass inference requests directly in the `CreateModelInvocationJob` API call using the `inputInlineRequests` field. Each item in the list uses the same record format as a line in the input JSONL file. + +The following example shows a full `CreateModelInvocationJob` request body with inline requests for an Anthropic Claude model: + + + { + "jobName": "my-inline-batch-job", + "roleArn": "arn:aws:iam::123456789012:role/BatchServiceRole", + "modelId": "anthropic.claude-sonnet-4-20250514", + "inputInlineRequests": [ + { + "recordId": "CALL0000001", + "modelInput": { + "anthropic_version": "bedrock-2023-05-31", + "max_tokens": 1024, + "messages": [ + { + "role": "user", + "content": [ + { + "type": "text", + "text": "Summarize the following call transcript: ..." + } + ] + } + ] + } + }, + { + "recordId": "CALL0000002", + "modelInput": { + "anthropic_version": "bedrock-2023-05-31", + "max_tokens": 1024, + "messages": [ + { + "role": "user", + "content": [ + { + "type": "text", + "text": "Extract the key action items from this meeting: ..." + } + ] + } + ] + } + } + ], + "outputDataConfig": { + "s3OutputDataConfig": { + "s3Uri": "s3://amzn-s3-demo-bucket/output/" + } + } + } +