valohai-utils Toolkit

One of the Valohai core design principles is to be an unopinionated agnostic platform. The user is not required to use any specific programming language, framework or SDK.

That said, Valohai offers a Python utility library called valohai-utils to help with the everyday boilerplate.


This is an in-depth topic guide. If you are just starting out, check the Valohai fundamentals first.


pip install valohai-utils

Example Usage

import valohai

params = {
    "iterations": 10,
    "learning-rate": 0.001,

inputs = {
    "train": "",
    "test": "",

valohai.prepare(step="train", default_parameters=params, default_inputs=inputs)

for i in valohai.parameters("iterations").value:
    for path in valohai.inputs("train").paths():

What is it?

The valohai-utils is a generic utility library for the Valohai user. In general, it streamlines pipeline definition and writing Valohai-compatible code. This is achieved by offering utility functions for the most common tasks, which are all described in the sections below.

The library is designed to be used hand-in-hand with the Valohai CLI.

You can generate or update an existing valohai.yaml config file using the command vh yaml step <filename> and vh yaml pipeline <filename>.

What it isn’t?

The valohai-utils library is not a machine learning or data science library. It helps in OS-level tasks like figuring out the input file paths or parsing the command-line parameters. It does not tune hyperparameters or train a model.