Setting Verta Config for the Python client

The Client config file allows you to define settings that are easily repeatable and sharable. This can be either a JSON or a YAML file.

When the Client is instantiated, it will look for a file called verta_config.json or verta_config.yaml in the following locations—in order—using the first one it finds:

1. the current directory

2. parent directories, recursively

3. $HOME/.verta/

Example

Defining a config file like this:

verta_config.yaml

email: "[e-mail]"
dev_key: "XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX"
host: "app.verta.ai"
workspace: "My Workspace"
project: "My Project"
experiment: "My Experiment"

allows you go directly from instantiating a client to working with your ExperimentRun without needing to explicitly set the project or experiment in your code:

workflow.py

from verta import Client

client = Client()
run = client.set_experiment_run("New Run")
# setting email from config file
# setting dev_key from config file
# setting host from config file
# connection successfully established
# setting experiment from config file
# setting project from config file
# set existing Project: My Project from workspace: My Workspace
# set existing Experiment: My Experiment
# created new ExperimentRun: New Run

Supported fields

For the fields and structure of the config file itself, please refer to the specification here.