Model training¶

The machine learning notebooks available in TranzAI are designed to be user-friendly and accessible. They can be directly instantiated and launched via the TranzAI notebook GUI. You only need to provide the parameters that will define the experiment.

TranzAI training notebok GUI

Notebook execution environment¶

Each notebook runs in a dedicated environment that is instantiated at runtime based on parameters provided. The TranzAI platform leverages cloud provider APIs to optimize cost and resource allocations.

Notebook dependencies¶

Notebook dependencies (software libraries, packages and TranzAI SDK components) are provisioned using containerization technologies (Docker). Every Docker image used by TranzAI notebooks is versioned and registered in the platform to ensure that results are accurate and reproducible.

Experiment reporting¶

Each experiment is logged and can be analyzed to identify the model version and hyper-parameter configuration that provides the most accurate and stable results.