Graphsignal is an inference observability platform that helps developers accelerate and troubleshoot AI systems. With Graphsignal, developers can:
- Trace LLM generations, communication, kernel launches, and more.
- Identify top contributors to inference latency.
- Monitor inference performance, CPU/GPU utilization, and errors.
- Track and get alerts on errors and inefficiencies.
Learn more at graphsignal.com.
Install the Graphsignal library.
pip install -U graphsignalConfigure Graphsignal tracer by specifying your API key directly or via GRAPHSIGNAL_API_KEY environment variable.
import graphsignal
graphsignal.configure(api_key='my-api-key')
# or pass the API key in GRAPHSIGNAL_API_KEY environment variableSee configure() API docs for all configuration parameters.
To get an API key, sign up for a free account at graphsignal.com. The key can then be found in your account's Settings / API Keys page.
Alternatively, you can add Graphsignal tracer from the command line, when running your module or script. Environment variables GRAPHSIGNAL_API_KEY and GRAPHSIGNAL_DEPLOYMENT must be set.
python -m graphsignal <script>python -m graphsignal -m <module>Graphsignal integrates through tracing - either via auto-instrumentation or manual setup. It automatically captures traces, errors, performance profiles, and data. All insights are available for analysis at app.graphsignal.com.
Refer to the guides below for detailed information on:
See API reference for full documentation.
Integration examples are available in examples repository.
Log in to Graphsignal to monitor and analyze your application.
Graphsignal tracer is highly lightweight. The overhead per trace is measured to be less than 100 microseconds. While profiling can introduce slight overhead, the profiling rate it is limited.
The Graphsignal tracer only establishes outbound connections to api.graphsignal.com to send data; inbound connections or commands are not possible.
Content and sensitive information, such as prompts and completions, are not recorded.
To enable debug logging, add debug_mode=True to configure(). If the debug log doesn’t provide hints for resolving the issue, report it to our support team via your account.
In case of connection issues, please make sure outgoing connections to https://api.graphsignal.com are allowed.