TorchGAN: A Flexible Framework for GAN Training and Evaluation

Abstract

TorchGAN is a PyTorch based framework for writing succinct and comprehensible code for training and evaluation of Generative Adversarial Networks. The framework's modular design allows effortless customization of the model architecture, loss functions, training paradigms, and evaluation metrics. The key features of TorchGAN are its extensibility, built-in support for a large number of popular models, losses and evaluation metrics, and zero overhead compared to vanilla PyTorch. By using the framework to implement several popular GAN models, we demonstrate its extensibility and ease of use. We also benchmark the training time of our framework for said models against the corresponding baseline PyTorch implementations and observe that TorchGAN's features bear almost zero overhead.

References in corpus (6)

• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Self-Attention Generative Adversarial Networks
• Energy-based Generative Adversarial Network
• Real-valued (Medical) Time Series Generation with Recurrent Conditional GANs
• Adversarial Feature Matching for Text Generation
• Training Generative Adversarial Networks with Binary Neurons by End-to-end Backpropagation

Cited by in corpus (4)

• medigan: a Python library of pretrained generative models for medical image synthesis
• Generating transient noise artifacts in gravitational-wave detector data with generative adversarial networks
• On Characterizing GAN Convergence Through Proximal Duality Gap
• Ward2ICU: A Vital Signs Dataset of Inpatients from the General Ward