Modularize your build function

envd allows you to easily create reusable components, so that the building file can be more organized.

Defining your own package

envd support most Python syntax, including string format, for-loop, if-statement, function definition.

For example, you can define a function for tensorflow package like this

def tensorflow(version):
    if version.startswith("1"):
        # execute command to install tf1.x
        run(["pip install tensorflow-gpu=={}".format(version)])
    else:
        # execute command to install tf2.x
        run(["pip install tensorflow=={}".format(version)])

The full list of language spec can be found at https://github.com/google/starlark-go/blob/master/doc/spec.md.

TIP

More examples can be found at https://github.com/tensorchord/envdlib/tree/main/src.

Using predefined package by envdlib

envd provided a set of predefined libraries which are commonly used in machine learning tasks. You can find them at https://github.com/tensorchord/envdlib.

To use it you just need one line in your envd file

# import envdlib packages
envdlib = include("https://github.com/tensorchord/envdlib")

# use it in your build function
def build():
    base(dev=True)
    install.conda()
    install.python()
    envdlib.tensorboard(host_port=8888)

And now you'll have tensorboard on your 8888 port.

TIP

You can also build your own package such as for internal or domain-specific tools following envdlib and share it with others.

Contribute to envdlib

We're extending our package coverage of envdlib. Please don't hesitate to file a pull request or raise an issue if you have any need. Your contribution is welcomed.