The 3-Step Workflow in H2O Hydrogen Torch
- Predictive AI
- H2O Hydrogen Torch
- H2O MLOps
This lesson frames the entire H2O Hydrogen Torch workflow as three steps: importing a dataset formatted for your specific problem type, training a model with adjustable hyperparameters and an optional grid search across multiple values, and inspecting the trained model using interactive graphs before deploying it. It covers the three deployment paths available once training completes: through the Hydrogen Torch user interface itself, through a standalone Python scoring pipeline that runs on Linux only, or through H2O MLOps. The lesson ties these three steps together as the complete path to a deployed deep learning model.
Key takeaways
- How the three-step Hydrogen Torch workflow moves from dataset import through training to deployment
- Why the dataset must be preprocessed into the specific format required for your chosen problem type
- What grid search lets you do when training a model with more than one hyperparameter value to test
- How a trained model can be deployed through the UI, a Python scoring pipeline, or H2O MLOps
Start owning your AI