Training Models

Train your models on a scalable infrastructure using Qwak.

Train models on Qwak

Qwak let's you easily train models on a scalable infrastructure, either with CPU or GPU-based instances. Easily choose the resources you'd like to use and train your models with several easy commands.

Training takes place during the build phase on the model. Your training code should be placed inside the build() method, which is called only once during the build phase.

Training data can be ingested from the Feature store, in an uploaded CSV file or directly from S3 or Parquet files.

In this section we will show you how to train your models using various examples.

Train models with GPUs

GPU instances provide high-performance resources to accelerate the model training process.

Learn how to train models with GPUs →