Inference notebook

Roboflow Inference Servers come equipped with a built in Jupyterlab environment. This environment is the fastest way to get up and running with inference for development and testing. To use it, first start an inference server.

The easiest way to start an inference server is with the inference CLI. Install it via pip:

pip install inference-cli

Now run the inference sever start command. Be sure to specify the --dev flag so that the notebook environment is enabled (it is disabled by default).

inference server start --dev

Now visit localhost:9001 in your browser to see the inference landing page. This page contains links to resources and examples related to inference. It also contains a link to the built in Jupyterlab environment.

From the landing page, select the button labeled "Jump Into an Inference Enabled Notebook" to open a new tab for the Jupyterlab environment.

This Jupyterlab environment comes preloaded with several example notebooks and all of the dependencies needed to run inference.