Define Valohai parameters¶

See also

This how-to is a part of our Bring your existing projects to Valohai series.

Parameters are passed as command-line arguments to your code.

Using parameters allows you to:

Easily keep track and compare executions with different parameters
Rerun an execution with a different set of parameters
Run multiple executions with different parameter sets (parameter sweeps, hyperparameter optimization etc.)

valohai-utils (Python)

import valohai

# Define parameters in a dictionary
default_parameters = {
    'iterations': 10,
}

# Create a step 'Train Model' in valohai.yaml with a set of parameters
valohai.prepare(step="Train Model", default_parameters=default_parameters)

# Access the parameters in your code
for i in range(valohai.parameters('iterations').value):
    print("Iteration %s" % i)

Generate or update your existing YAML file by running

vh yaml step myfile.py

The generated valohai.yaml configuration file looks like:

Python

import argparse

def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument('--iterations', type=int, default=10)
return parser.parse_args()

args = parse_args()

print(args.iterations)

Create a valohai.yaml configuration file and define your step in it:

R

library(optparse)

option_list <- list(
    make_option("--iterations", default = 10),
)

parser <- OptionParser(option_list = option_list)
args <- parse_args(parser)

iterations <- args$iterations
write(iterations, stdout())

Create a valohai.yaml configuration file and define your step in it:

- step:
    name: Train model
    image: tensorflow/tensorflow:1.13.1
    command: python myfile.py {parameters}
    parameters:
      - name: iterations
        type: integer
        default: 100