What H2O Hydrogen Torch Does for Deep Learning
- Predictive AI
- H2O Hydrogen Torch
- H2O MLOps
This lesson introduces H2O Hydrogen Torch, a tool for training deep learning models in computer vision, natural language processing, and audio. It covers how the tool ships with default hyperparameters and methods drawn from experienced Kaggle grandmasters' best practices, while still letting you adjust those values yourself and see the effect of each change through interactive charts. It also covers the deployment options available once a model is built: deploying to MLOps from within Hydrogen Torch's own interface, exporting to an external Python environment, or using the H2O MLOps interface directly to deploy the model package.
Key takeaways
- How H2O Hydrogen Torch's default hyperparameters are drawn from Kaggle grandmasters' best practices for deep learning models
- Why the interactive charts in Hydrogen Torch let you see how a hyperparameter change affects training before committing to it
- The three ways a Hydrogen Torch model can be deployed: within the app's own interface, an external Python environment, or H2O MLOps
- What kinds of deep learning models Hydrogen Torch is built for: computer vision, natural language processing, and audio
Start owning your AI