Experiment (model) flow
The flow of an H2O Hydrogen Torch experiment from creation to deployment can be summarized in the following sequential steps:
- Step 1: Import an experiment's dataset
- Step 2: Train an experiment's model
- Step 3: Inspect an experiment's model
- Step 4: Deploy an experiment's model
In the below sections, each step above, in turn, is summarized.
Step 1: Import an experiment's dataset
As the first step in the experiment flow, import your experiment dataset to H2O Hydrogen Torch. Depending on the problem type the experiment aims to solve, H2O Hydrogen Torch requires you to construct your dataset in a particular format.
- To learn about all the supported problem types, see Supported problem types.
- To learn how to import your own preprocessed dataset, see Dataset connectors.
- To learn about the dataset format of a particular supported problem type, see Dataset formats.
- H2O Hydrogen Torch offers access to an array of preprocessed datasets to highlight all supported problem types. In particular, preprocessed datasets can be used as a reference to understand the format a dataset needs to follow for a specific supported problem type. To learn how to access one of the preprocessed datasets in H2O Hydrogen Torch, see Demo (preprocessed) datasets.
Step 2: Train an experiment's model
As the second step in the experiment flow, train your experiment model. H2O Hydrogen Torch offers several hyperparameter settings that you can adjust for your experiment model. H2O Hydrogen Torch also provides the ability to enable grid search to tune and experiment on several hyperparameter values.
- To learn about the settings available for each supported problem type, see Experiment settings.
- To learn how to enable grid search, see Grid search.
Step 3: Inspect an experiment's model
As the third step in the experiment flow, inspect your experiment model. H2O Hydrogen Torch allows you to inspect your experiment (model) during and after model training. Simple interactive graphs in H2O Hydrogen Torch allow you to understand the impact of selected hyperparameter values during and after model training.
- To learn about available simple and interactive graphs, see Charts.
Step 4: Deploy an experiment's model
As the fourth and final step in the experiment flow, deploy your built model (experiment). After understanding and inspecting your built model, you can deploy your model on the H2O Hydrogen Torch user interface (UI). You can also deploy a built model to any external Python environment or H2O MLOps.
- To learn how to deploy your built model on the H2O Hydrogen Torch UI, see H2O Hydrogen Torch UI.
- To learn how to deploy built models to external Python environments, see Python scoring pipeline. note
H2O Hydrogen Torch offers a standalone Python scoring pipeline that only runs on Linux-based systems.
- To learn how to deploy a built model directly to H2O MLOps from the H2O Hydrogen Torch UI, see Deploy a model to H2O MLOps through the H2O Hydrogen Torch UI.
- To learn how to deploy a built model utilizing the model's H2O MLOps pipeline, see Download an experiment's H2O MLOps pipeline.