1. Data Parallelism Training steps: The master device loads the model and copies the model parameters to every worker device The master device splits the training data along the batch dimension and passes each batch to every worker device Each worker device performs training The master device aggregates the gradients from each worker device and updates the model parameters The master device broadcasts the model parameters to every worker device, preparing for the next batch of training Core idea:

1. The Advantages of Multi-Cluster Tekton Builds Thanks to Kubernetes, Tekton already has good elasticity and can support large-scale builds. At the same time, developing Tasks mainly uses Yaml and Shell, which widens the range of scenarios Tekton can adapt to. The figure above is a sketch of Tekton across multiple clusters.