Introduction to Qadence ML Tools

Welcome to the Qadence ML Tools documentation. This submodule is designed to streamline your machine learning workflows —especially for quantum machine learning— by providing a set of robust tools for training, monitoring, and optimizing your models.

What this documentation is about

• Trainer Class Learn how to leverage the versatile Trainer class to manage your training loops, handle data loading, and integrate with experiment tracking tools like TensorBoard and MLflow. Detailed guides cover:

  • Setting up training on both GPUs and CPUs.
  • Configuring single-process, multi-processing, and distributed training setups.

• Gradient Optimization Methods Explore both gradient-based and gradient-free optimization strategies. Find examples demonstrating how to switch between these modes and how to use context managers for mixed optimization.

• Custom Loss Functions and Hooks Discover how to define custom loss functions tailored to your tasks and use hooks to insert custom behaviors at various stages of the training process.

• Callbacks for Enhanced Training Utilize built-in and custom callbacks to log metrics, save checkpoints, adjust learning rates, and more. This section explains how to integrate callbacks seamlessly into your training workflow.

• Experiment Tracking Understand how to configure experiment tracking with tools such as TensorBoard and MLflow to monitor your model’s progress and performance.

Getting Started

To dive in, explore the detailed sections below:

• Qadence Trainer Guide
• Training Configuration
• Callbacks for Trainer
• Accelerator for Distributed Training
• Training on GPU with Trainer
• Training on CPU with Trainer