TorchRunner

TorchRunner: Open-Source ML Experiment Management Tool

Track hyperparameters, metrics, code versions and more with TorchRunner, an open-source tool for managing machine learning experiments.

What is TorchRunner?

TorchRunner is an open-source experiment management tool designed specifically for machine learning projects. It integrates with PyTorch to automatically track metrics, parameters, and other metadata during model training.

Key features of TorchRunner include:

• Automatic tracking of hyperparameters, metrics, dataset versions, and code versions without having to change training code.
• Web UI and APIs to visualize experiments, compare runs, and search through experiment history.
• Tools for hyperparameter optimization like grid search and random search.
• Model registry to version and organize saved models.
• Integrations with tools like TensorBoard, MLflow, Weights & Biases, and more.
• Designed to accelerate research by enabling better organization and reproducibility.

TorchRunner helps machine learning teams standardize and scale their training pipelines. By automatically tracking experiments and making data readily accessible, it saves time and effort while encouraging best practices like experiment tracking. The open source tools provide flexibility to customize and integrate with existing infrastructure.

TorchRunner Features

Official Links