Structured JSON Output
Output well-formatted JSON. Here we're letting the model come up with its own data, but the same prompt would work with RAG contents as well. We've incorporate a pre-filled response to the model to ensure we get json back, as described in our documentation.
User Prompt Template
{Decription of desired output}.
You MUST answer in JSON format only.
Please follow the output schema below.
Output Schema:
{
"key1": "value1",
"key2": "value2"
}
Example
Amazon Nova Lite User Prompt
Provide 5 examples of the best selling full-frame cameras in past three years.
Follow the Output Schema as described below.
Output Schema:
{
"name" : <string, the name of product>,
"brand" : <string, the name of product>,
"price" : <integer price>,
"summary": <string, the product summary>
}
Only Respond in Valid JSON, without Markdown
Amazon Nova Lite Assistant Prefilled Response
"```json"
Amazon Nova Lite Sample Response
Response
[
{
"name": "Canon EOS R5",
"brand": "Canon",
"price": "$3,899",
"summary": "The Canon EOS R5 is a high-end full-frame mirrorless camera known for its 45MP sensor, 8K video recording, and advanced autofocus capabilities."
},
{
"name": "Sony Alpha A7R IV",
"brand": "Sony",
"price": "$3,499",
"summary": "The Sony Alpha A7R IV is a top-tier full-frame mirrorless camera featuring a 61MP sensor, 8K video, and excellent low-light performance."
},
{
"name": "Nikon Z7 II",
"brand": "Nikon",
"price": "$2,999",
"summary": "The Nikon Z7 II is a professional full-frame mirrorless camera with a 45.7MP sensor, dual EXPEED 6 processors, and excellent image quality."
},
{
"name": "Canon EOS R6",
"brand": "Canon",
"price": "$2,499",
"summary": "The Canon EOS R6 is a versatile full-frame mirrorless camera with a 20.1MP sensor, capable of 20fps shooting, and 4K video recording."
},
{
"name": "Sony Alpha A7 III",
"brand": "Sony",
"price": "$1,999",
"summary": "The Sony Alpha A7 III is a popular full-frame mirrorless camera featuring a 24.2MP sensor, fast autofocus, and 10fps continuous shooting."
}
]
API Request
import boto3
import json
bedrock_runtime = boto3.client(
service_name="bedrock-runtime",
region_name="us-west-2"
)
response = bedrock_runtime.converse(
modelId='us.amazon.nova-lite-v1:0',
messages = [
{
"role": "user",
"content": [
{
"text": """Provide 5 examples of the best selling full-frame cameras in past three years.
You MUST answer in JSON format only. Please follow the output schema below.
**Output Schema:**
```json
[
{
"name": "string, the name of the camera",
"brand": "string, the brand name",
"price": "integer or string with price",
"summary": "string, brief product description"
}
]
DO NOT provide any preamble text. Return ONLY the JSON object."""
}
]
},
{
"role": "assistant",
"content": [
{
"text": "json"
}
]
}
],
inferenceConfig={
"temperature": 0.1,
"topP": .99,
"maxTokens": 1000,
"stopSequences": [""]
}
)
print(json.dumps(response, indent=2)) ```
aws bedrock-runtime converse \
--model-id "us.amazon.nova-lite-v1:0" \
--messages '[
{
"role": "user",
"content": [
{
"text": "Provide 5 examples of the best selling full-frame cameras in past three years.\nFollow the Output Schema as described below:\nOutput Schema:\n{\n\"name\" : <string, the name of product>,\n\"brand\" : <string, the name of product>,\n\"price\" : <integer price>,\n\"summary\": <string, the product summary>\n}\nOnly Respond in Valid JSON, without Markdown"
}
]
},
{
"role": "assistant",
"content": [
{
"text": "```json"
}
]
}
]' \
--inference-config '{
"temperature": 0.7,
"topP": 0.99,
"maxTokens": 512
}' \
--region us-west-2
{
"modelId": "us.amazon.nova-lite-v1:0",
"messages": [
{
"role": "user",
"content": [
{
"text": "Provide 5 examples of the best selling full-frame cameras in past three years.\nFollow the Output Schema as described below:\nOutput Schema:\n{\n\"name\" : <string, the name of product>,\n\"brand\" : <string, the name of product>,\n\"price\" : <integer price>,\n\"summary\": <string, the product summary>\n}\nOnly Respond in Valid JSON, without Markdown"
}
]
},
{
"role": "assistant",
"content": [
{
"text": "```json"
}
]
}
],
"inferenceConfig": {
"temperature": 0.7,
"topP": 0.99,
"maxTokens": 512
}
}