Pipeline Step

Pipeline step is a task in a pipeline. A pipeline step can be one of two types:

  • Data Science Job: the OCID of an existing Data Science Job must be provided.

  • Custom Script: the artifact of the Python script and the execution configuration must be specified.

This section shows how you can use the ADS Pipeline APIs to create pipeline steps.

Data Science Job Step

Create a Data Science Job step with the OCID of an existing Job.

  • Python
  • Python (Alternative)
  • YAML
from ads.pipeline import PipelineStep

pipeline_step = PipelineStep(
    name="<pipeline_step_name>",
    description="<pipeline_step_description>",
    job_id="<job_id>"
)

from ads.pipeline import PipelineStep

pipeline_step = (
    PipelineStep("<pipeline_step_name>")
    .with_description("<pipeline_step_description>")
    .with_job_id("<job_id>")
)

kind: pipeline
spec:
  ...
  stepDetails:
  ...
  - kind: dataScienceJob
    spec:
      description: <pipeline_step_description>
      jobId: ocid1.datasciencejob..<unique_id>
      name: <pipeline_step_name>
  ...
type: pipeline

Custom Script Step

To create a Custom Script step, infrastructure and runtime must be specified.

The Custom Script step infrastructure is defined by a CustomScriptStep instance. When constructing a Custom Scrip step infrastructure, you specify the Compute shape, Block Storage size in the CustomScriptStep instance. For example:

  • Python
  • Python (Alternative)
from ads.pipeline import CustomScriptStep

infrastructure = CustomScriptStep(
    block_storage_size=200,
    shape_name="VM.Standard3.Flex",
    shape_config_details={"ocpus": 4, "memory_in_gbs": 32},
)

from ads.pipeline import CustomScriptStep

infrastructure = (
  CustomScriptStep()
  .with_block_storage_size(200)
  .with_shape_name("VM.Standard3.Flex")
  .with_shape_config_details(ocpus=4, memory_in_gbs=32)
)

A Custom Script step can have different types of runtime depending on the source code you run:

  • GitPythonRuntime allows you to run source code from a Git repository, see Run from Git.

  • NotebookRuntime allows you to run a JupyterLab Python notebook, see Run a Notebook.

  • PythonRuntime allows you to run Python code with additional options, including setting a working directory, adding Python paths, and copying output files, see Run a ZIP file or folder.

  • ScriptRuntime allows you to run Python, Bash, and Java scripts from a single source file (.zip or .tar.gz) or code directory, see Run a Script and Run a ZIP file or folder.

All of these runtime options allow you to configure a Data Science Conda Environment for running your code.

To define a Custom Script step with GitPythonRuntime you can use:

  • Python
  • Python (Alternative)
from ads.pipeline import GitPythonRuntime

runtime = GitPythonRuntime(
      env={"GREETINGS": "Welcome to OCI Data Science"}
      conda={"type": "service", "slug": "pytorch19_p37_gpu_v1"}
      url="https://github.com/pytorch/tutorials.git",
      entrypoint="beginner_source/examples_nn/polynomial_nn.py",
      output_dir="~/Code/tutorials/beginner_source/examples_nn",
      outputURI="oci://<bucket_name>@<namespace>/<prefix>",
    )

from ads.pipeline import GitPythonRuntime

runtime = (
    GitPythonRuntime()
    .with_environment_variable(GREETINGS="Welcome to OCI Data Science")
    .with_service_conda("pytorch19_p37_gpu_v1")
    .with_source("https://github.com/pytorch/tutorials.git")
    .with_entrypoint("beginner_source/examples_nn/polynomial_nn.py")
    .with_output(
    output_dir="~/Code/tutorials/beginner_source/examples_nn",
    output_uri="oci://<bucket_name>@<namespace>/<prefix>"
  )
)

To define a Custom Script step with NotebookRuntime you can use:

  • Python
  • Python (Alternative)
from ads.pipeline import NotebookRuntime

runtime = NotebookRuntime(
    notebook_path_uri="https://raw.githubusercontent.com/tensorflow/docs/master/site/en/tutorials/customization/basics.ipynb",
    notebook_encoding="utf-8",
    conda={"type": "service", "slug": "tensorflow26_p37_cpu_v2"}
    env={"GREETINGS": "Welcome to OCI Data Science"}
    outputURI="oci://<bucket_name>@<namespace>/<prefix>",
)

from ads.pipeline import NotebookRuntime

runtime = (
    NotebookRuntime()
      .with_notebook(
          path="https://raw.githubusercontent.com/tensorflow/docs/master/site/en/tutorials/customization/basics.ipynb",
          encoding='utf-8'
      )
      .with_service_conda("tensorflow26_p37_cpu_v2")
      .with_environment_variable(GREETINGS="Welcome to OCI Data Science")
      .with_output("oci://bucket_name@namespace/path/to/dir")
)

To define a Custom Script step with PythonRuntime you can use:

  • Python
  • Python (Alternative)
from ads.pipeline import PythonRuntime

runtime = PythonRuntime(
    script_path_uri="local/path/to/zip_or_dir",
    entrypoint="zip_or_dir/my_package/entry.py",
    working_dir="zip_or_dir",
    python_path=["my_python_packages"],
    output_uri="oci://<bucket_name>@<namespace>/<prefix>",
    conda={"type": "service", "slug": "pytorch19_p37_cpu_v1"}
)

from ads.pipeline import PythonRuntime

runtime = (
    PythonRuntime()
    .with_service_conda("pytorch19_p37_cpu_v1")
    # The job artifact directory is named "zip_or_dir"
    .with_source("local/path/to/zip_or_dir", entrypoint="zip_or_dir/my_package/entry.py")
    # Change the working directory to be inside the job artifact directory
    # Working directory a relative path from the parent of the job artifact directory
    # Working directory is also added to Python paths
    .with_working_dir("zip_or_dir")
    # Add an additional Python path
    # The "my_python_packages" folder is under "zip_or_dir" (working directory)
    .with_python_path("my_python_packages")
    # Files in "output" directory will be copied to OCI object storage once the job finishes
    # Here we assume "output" is a folder under "zip_or_dir" (working directory)
    .with_output("output", "oci://bucket_name@namespace/path/to/dir")
)

To define a Custom Script step with ScriptRuntime you can use:

  • Python
  • Python (Alternative)
from ads.pipeline import ScriptRuntime

runtime = ScriptRuntime(
    script_path_uri="oci://<bucket_name>@<namespace>/<prefix>/<script.py>",
    conda={"type": "service", "slug": "tensorflow26_p37_cpu_v2"}
)

from ads.pipeline import ScriptRuntime

runtime = (
    ScriptRuntime()
    .with_source("oci://<bucket_name>@<namespace>/<prefix>/<script.py>")
    .with_service_conda("tensorflow26_p37_cpu_v2")
)

With Infrastructure and runtime provided, create a pipeline step of the Custom Script type.

  • Python
  • Python (Alternative)
  • YAML
from ads.pipeline import PipelineStep

pipeline_step = PipelineStep(
    name="<pipeline_step_name>",
    description="<pipeline_step_description>",
    infrastructure=infrastructure,
    runtime=runtime,
)

from ads.pipeline import PipelineStep

pipeline_step = (
    PipelineStep("<pipeline_step_name>")
    .with_description("<pipeline_step_description>")
    .with_infrastructure(infrastructure)
    .with_runtime(runtime)
)

kind: pipeline
spec:
  ...
  stepDetails:
  ...
  - kind: customScript
    spec:
      description: <pipeline_step_description>
      infrastructure:
        kind: infrastructure
        spec:
          blockStorageSize: 200
          shapeConfigDetails:
            memoryInGBs: 32
            ocpus: 4
          shapeName: VM.Standard3.Flex
      name: <pipeline_step_name>
      runtime:
        kind: runtime
        spec:
          conda:
            slug: <slug>
            type: service
          scriptPathURI: oci://<bucket_name>@<namespace>/<prefix>/<script.py>
        type: script
  ...
type: pipeline