AWS Bedrock

Instrument LLM calls to AWS Bedrock via the boto3 client using OpenInference.

boto3 provides Python bindings to AWS services, including Bedrock, which provides access to a number of foundation models. Calls to these models can be instrumented using OpenInference, enabling OpenTelemetry-compliant observability of applications built using these models. Traces collected using OpenInference can be viewed in Arize.

OpenInference Traces collect telemetry data about the execution of your LLM application. Consider using this instrumentation to understand how a Bedrock-managed models are being called inside a complex system and to troubleshoot issues such as extraction and response synthesis.

Python

To get started instrumenting Bedrock calls via boto3, we need to install two components: the OpenInference instrumentation for AWS Bedrock, and an OpenTelemetry exporter used to send these traces to Phoenix.

pip install openinference-instrumentation-bedrock
pip install opentelemetry-exporter-otlp
pip install arize-otel

Instrument boto3 prior to initializing a bedrock-runtime client. All clients created after instrumentation will send traces on all calls to invoke_model.

# Import open-telemetry dependencies
from arize_otel import register_otel, Endpoints

# Setup OTEL via our convenience function
register_otel(
    endpoints = Endpoints.ARIZE,
    space_id = "your-space-id", # in app space settings page
    api_key = "your-api-key", # in app space settings page
    model_id = "your-model-id", # name this to whatever you would like
)

# Import the automatic instrumentor from OpenInference
from openinference.instrumentation.bedrock import BedrockInstrumentor

# Start the instrumentor for Bedrock
BedrockInstrumentor().instrument()

You can use the following code to test whether your tracing is working.

import boto3
session = boto3.session.Session()
client = session.client("bedrock-runtime")

# All calls to invoke_model are instrumented
prompt = (
    b'{"prompt": "Human: Hello there, how are you? Assistant:", "max_tokens_to_sample": 1024}'
)
response = client.invoke_model(modelId="anthropic.claude-v2", body=prompt)
response_body = json.loads(response.get("body").read())
print(response_body["completion"])