Yuan TangandGitHub
5a767fcf1c
* feat: Support customizable deployment strategy for RawDeployment mode Signed-off-by: Yuan Tang <[email protected]> * regen Signed-off-by: Yuan Tang <[email protected]> * lint Signed-off-by: Yuan Tang <[email protected]> * Correctly apply rollingupdate Signed-off-by: Yuan Tang <[email protected]> * address comments Signed-off-by: Yuan Tang <[email protected]> * Add validation Signed-off-by: Yuan Tang <[email protected]> --------- Signed-off-by: Yuan Tang <[email protected]>
Deploy to KServe from Kubeflow Pipelines
Kubeflow Pipelines KServe component
The following examples illustrate how to use the Kubeflow Pipelines component for KServe using the v1beta1 API. These assume your cluster has a KServe version >= v0.7.0.
- Deploy a custom model.
- Deploy a TensorFlow model. There is also a notebook which illustrates this.
Additional usage instructions can be found in the component README. To dive into the source behind the KServe Kubeflow Pipelines Component, please look into the YAML for the KServe Component and the source code.
Note: For those still using an older version of KFServing less than v0.5.0, an older version of the KFServing Pipelines component must be used as demonstrated in this notebook. The source code for this version of the component can be found here.
End to end pipeline example using KServe
Deploy a sample MNIST model end to end using Kubeflow Pipelines with Tekton. The notebook demonstrates how to compile and execute an End to End Machine Learning workflow that uses Katib, TFJob, KServe, and Tekton pipeline. This pipeline contains 5 steps, it finds the best hyperparameter using Katib, creates PVC for storing models, processes the hyperparameter results, distributedly trains the model on TFJob with the best hyperparameter using more iterations, and finally serves the model using KServe. You can visit this medium blog for more details on this pipeline.
