RoterDL/katib

Repository for hyperparameter tuning

Katib is a Kubernetes Native System for Hyperparameter Tuning and Neural Architecture Search. The system is inspired by Google vizier and supports multiple ML/DL frameworks (e.g. TensorFlow, MXNet, and PyTorch).

Table of Contents generated with DocToc

• Name
• Concepts in Katib
  • Experiment
  • Trial
  • Job
  • Suggestion
• Components in Katib
  • v1alpha1
  • v1alpha2
• Getting Started
• Web UI
• API Documentation
• Quickstart to run tfjob and pytorch operator jobs in Katib
  • TFjob operator
  • Pytorch operator
  • Katib
  • Running examples
  • Cleanups
• CONTRIBUTING

Name

Katib stands for secretary in Arabic. As Vizier stands for a high official or a prime minister in Arabic, this project Katib is named in the honor of Vizier.

Concepts in Katib

Katib has the concepts of Experiment, Trial, Job and Suggestion.

Experiment

Experiment represents a single optimization run over a feasible space. Each Experiment contains a configuration describing the feasible space, as well as a set of Trials. It is assumed that objective function f(x) does not change in the course of a Experiment.

In v1alpha1, Experiment is defined as a CRD StudyJob in Kubernetes. In v1alpha2, Experiment is defined as a CRD Experiment.

Trial

A Trial is a list of parameter values, x, that will lead to a single evaluation of f(x). A Trial can be “Completed”, which means that it has been evaluated and the objective value f(x) has been assigned to it, otherwise it is “Pending”.

In v1alpha1, Trial is just a concept inside Katib and not exposed to users. In v1alpha2, Trial is defined as a CRD Trial in Kubernetes.

Job

A Job refers to a process responsible for evaluating a Pending Trial and calculating its objective value.

The job kind can be Kubernetes Job, Kubeflow TFJob or Kubeflow PyTorchJob. Thus Katib supports multiple frameworks with the help of different job kinds.

Suggestion

A Suggestion is an algorithm to construct a parameter set according to the Experiment. Then Trial will be created to evaluate the parameter set.

Currently Katib supports the following exploration algorithms in v1alpha1:

• random search
• grid search
• hyperband
• bayesian optimization
• NAS based on reinforcement learning
• NAS based on EnvelopeNets

And Katib supports the following exploration algorithms in v1alpha2:

• random search

Components in Katib

v1alpha1

Katib consists of several components as shown below. Each component is running on k8s as a deployment. Each component communicates with others via GRPC and the API is defined at pkg/api/v1alpha1/api.proto.

• vizier: main components.
  • vizier-core: GRPC API server of vizier.
  • vizier-core-rest: REST API server of vizier.
  • vizier-db: Data storage backend of vizier.
  • suggestion: implementation of each exploration algorithm.
    • suggestion-random
    • suggestion-grid
    • suggestion-hyperband
    • suggestion-bayesianoptimization
    • suggestion-nasrl
    • suggestion-nasenvelopenets
• studyjob-controller: Controller for StudyJob CRD in Kubernetes.
• modeldb : WebUI
  • modeldb-frontend
  • modeldb-backend
  • modeldb-db

v1alpha2

Katib consists of several components as shown below. Each component is running on k8s as a deployment. Each component communicates with others via GRPC and the API is defined at pkg/api/v1alpha2/api.proto.

• katib: main components.
  • katib-manager: GRPC API server of katib.
  • katib-manager-rest: REST API server of katib.
  • katib-db: Data storage backend of katib.
  • katib-ui: User interface of katib.
  • suggestion: implementation of each exploration algorithm.
    • suggestion-random
• katib-controller: Controller for katib CRDs in Kubernetes.
• experiment-controller: Controller for Experiment CRD in Kubernetes.
• trial-controller: Controller for Trial CRD in Kubernetes.

Getting Started

Please see here for more details about katib v1alpha1.
Please see here for more details about katib v1alpha2.

Web UI

Katib provides a Web UI. You can visualize general trend of Hyper parameter space and each training history. You can use random-example or other examples to generate a similar UI.

API Documentation

Please refer to api.md.

Quickstart to run tfjob and pytorch operator jobs in Katib

For running tfjob and pytorch operator jobs in Katib, you have to install their packages.

In your Ksonnet app root, run the following

export KF_ENV=default
ks env set ${KF_ENV} --namespace=kubeflow
ks registry add kubeflow github.com/kubeflow/kubeflow/tree/master/kubeflow

TFjob operator

For installing tfjob operator, run the following

ks pkg install kubeflow/tf-training
ks pkg install kubeflow/common
ks generate tf-job-operator tf-job-operator
ks apply ${KF_ENV} -c tf-job-operator

Pytorch operator

For installing pytorch operator, run the following

ks pkg install kubeflow/pytorch-job
ks generate pytorch-operator pytorch-operator
ks apply ${KF_ENV} -c pytorch-operator

Katib

Finally, you can install Katib

ks pkg install kubeflow/katib
ks generate katib katib
ks apply ${KF_ENV} -c katib

If you want to use Katib not in GKE and you don't have StorageClass for dynamic volume provisioning at your cluster, you have to create persistent volume to bound your persistent volume claim.

This is yaml file for persistent volume

apiVersion: v1
kind: PersistentVolume
metadata:
  name: katib-mysql
  labels:
    type: local
    app: katib
spec:
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteOnce
  hostPath:
    path: /data/katib

Create this pv after deploying Katib package

kubectl create -f https://raw.githubusercontent.com/kubeflow/katib/master/manifests/v1alpha1/pv/pv.yaml

Running examples

After deploy everything, you can run examples.

To run tfjob operator example, you have to install volume for it.

If you are using GKE and default StorageClass, you have to create this pvc

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: tfevent-volume
  namespace: kubeflow
  labels:
    type: local
    app: tfjob
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi

If you are not using GKE and you don't have StorageClass for dynamic volume provisioning at your cluster, you have to create pvc and pv

kubectl create -f https://raw.githubusercontent.com/kubeflow/katib/master/examples/v1alpha1/tfevent-volume/tfevent-pvc.yaml

kubectl create -f https://raw.githubusercontent.com/kubeflow/katib/master/examples/v1alpha1/tfevent-volume/tfevent-pv.yaml

This is example for tfjob operator

kubectl create -f https://raw.githubusercontent.com/kubeflow/katib/master/examples/v1alpha1/tfjob-example.yaml

This is example for pytorch operator

kubectl create -f https://raw.githubusercontent.com/kubeflow/katib/master/examples/v1alpha1/pytorchjob-example.yaml

You can check status of StudyJob

$ kubectl describe studyjob pytorchjob-example -n kubeflow

Name:         pytorchjob-example
Namespace:    kubeflow
Labels:       controller-tools.k8s.io=1.0
Annotations:  <none>
API Version:  kubeflow.org/v1alpha1
Kind:         StudyJob
Metadata:
  Cluster Name:
  Creation Timestamp:  2019-01-15T18:35:20Z
  Generation:          1
  Resource Version:    1058135
  Self Link:           /apis/kubeflow.org/v1alpha1/namespaces/kubeflow/studyjobs/pytorchjob-example
  UID:                 4fc7ad83-18f4-11e9-a6de-42010a8e0225
Spec:
  Metricsnames:
    accuracy
  Objectivevaluename:  accuracy
  Optimizationgoal:    0.99
  Optimizationtype:    maximize
  Owner:               crd
  Parameterconfigs:
    Feasible:
      Max:          0.05
      Min:          0.01
    Name:           --lr
    Parametertype:  double
    Feasible:
      Max:          0.9
      Min:          0.5
    Name:           --momentum
    Parametertype:  double
  Requestcount:     4
  Study Name:       pytorchjob-example
  Suggestion Spec:
    Request Number:        3
    Suggestion Algorithm:  random
    Suggestion Parameters:
      Name:   SuggestionCount
      Value:  0
  Worker Spec:
    Go Template:
      Raw Template:  apiVersion: "kubeflow.org/v1"
kind: PyTorchJob
metadata:
  name: {{.WorkerID}}
  namespace: kubeflow
spec:
 pytorchReplicaSpecs:
  Master:
    replicas: 1
    restartPolicy: OnFailure
    template:
      spec:
        containers:
          - name: pytorch
            image: gcr.io/kubeflow-ci/pytorch-mnist-with-summary:1.0
            imagePullPolicy: Always
            command:
              - "python"
              - "/opt/pytorch_dist_mnist/dist_mnist_with_summary.py"
              {{- with .HyperParameters}}
              {{- range .}}
              - "{{.Name}}={{.Value}}"
              {{- end}}
              {{- end}}
  Worker:
    replicas: 2
    restartPolicy: OnFailure
    template:
      spec:
        containers:
          - name: pytorch
            image: gcr.io/kubeflow-ci/pytorch-mnist-with-summary:1.0
            imagePullPolicy: Always
            command:
              - "python"
              - "/opt/pytorch_dist_mnist/dist_mnist_with_summary.py"
              {{- with .HyperParameters}}
              {{- range .}}
              - "{{.Name}}={{.Value}}"
              {{- end}}
              {{- end}}
    Retain:  true
Status:
  Conditon:                     Running
  Early Stopping Parameter Id:
  Last Reconcile Time:          2019-01-15T18:35:20Z
  Start Time:                   2019-01-15T18:35:20Z
  Studyid:                      k291b444a0b68631
  Suggestion Count:             1
  Suggestion Parameter Id:      n6f17dd9ff466a2b
  Trials:
    Trialid:  o104235328003ad9
    Workeridlist:
      Completion Time:  <nil>
      Conditon:         Running
      Kind:             PyTorchJob
      Start Time:       2019-01-15T18:35:20Z
      Workerid:         b3b371c89144727f
    Trialid:            ca207b2432231de3
    Workeridlist:
      Completion Time:  <nil>
      Conditon:         Running
      Kind:             PyTorchJob
      Start Time:       2019-01-15T18:35:20Z
      Workerid:         f291b04fb27ece3c
    Trialid:            ddff69212e826432
    Workeridlist:
      Completion Time:  <nil>
      Conditon:         Running
      Kind:             PyTorchJob
      Start Time:       2019-01-15T18:35:20Z
      Workerid:         ncbed67bbcd4a8ed
Events:                 <none>

When the spec.Status.Condition becomes Completed, the StudyJob is finished.

You can monitor your results in Katib UI. For accessing to Katib UI, you have to install Ambassador.

In your Ksonnet app root, run the following

ks generate ambassador ambassador
ks apply ${KF_ENV} -c ambassador

After this, you have to port-forward Ambassador service

kubectl port-forward svc/ambassador -n kubeflow 8080:80

Finally, you can access to Katib UI using this URL: http://localhost:8080/katib/.

Cleanups

Delete installed components

ks delete ${KF_ENV} -c katib
ks delete ${KF_ENV} -c pytorch-operator
ks delete ${KF_ENV} -c tf-job-operator

If you create pv for Katib, delete it

kubectl delete -f https://raw.githubusercontent.com/kubeflow/katib/master/manifests/v1alpha1/pv/pv.yaml

If you deploy Ambassador, delete it

ks delete ${KF_ENV} -c ambassador

CONTRIBUTING

Please feel free to test the system! developer-guide.md is a good starting point for developers.