Tutorials¶
The easiest way to get started with ForML is to go through its practical tutorials presented in
this chapter. Assuming you have already installed ForML as per the installation instructions (ensure you have at least the dask
, graphviz
and rest
extra features installed) and ideally also familiarized yourself with the ForML
principles, you can now go straight through the following list of step-by-step
examples demonstrating the ForML capabilities.
Common Setup¶
The tutorials depend on the following initial environment configuration:
Assuming you have no existing feeds configured in your system yet, let’s install the Openlake feed:
$ pip install --constraints https://raw.githubusercontent.com/formlio/openlake/main/constraints.txt 'openlake[kaggle]'
Let’s now configure the local ForML platform by adding the following content to
your ~/.forml/config.toml
:
[RUNNER]
default = "compute"
[RUNNER.compute]
provider = "dask"
scheduler = "threads"
[RUNNER.visual]
provider = "graphviz"
format = "png"
[REGISTRY]
default = "tutorial"
[REGISTRY.tutorial]
provider = "posix"
path = "/tmp/forml-tutorial/registry"
[FEED]
default = ["openlake"]
[FEED.openlake]
provider = "openlake:Lite"
[SINK]
default = "print"
[SINK.print]
provider = "stdout"
[INVENTORY]
default = "tutorial"
[INVENTORY.tutorial]
provider = "posix"
path = "/tmp/forml-tutorial/inventory"
[GATEWAY]
default = "http"
[GATEWAY.http]
provider = "rest"
port = 8080
processes = 2
Important
Make sure to configure your Kaggle API access token under the ~/.kaggle/kaggle.json
as described in the Kaggle API Documentation to get
access to all of the datasets used in this tutorial.
Your local environment is now ready to perform all the runtime actions demonstrated in these tutorials.
Tutorials List¶
The list of available tutorials is:
The Titanic challenge is a complete end-to-end ML project implemented using ForML.
Pipeline demos represent a set of small snippets demonstrating the pipeline composition fundamentals.