soumith/dcgan.torch

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1488 stars · 410 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A Torch (Lua) implementation of the DCGAN paper for training deep convolutional generative adversarial networks that generate images. It supports training on datasets like Celeb-A, LSUN, and ImageNet, plus generating samples and latent-space vector arithmetic with pre-trained models.

Use cases

• train a GAN to generate faces from the Celeb-A dataset
• generate images of bedrooms or churches with DCGAN on LSUN
• generate large artsy images from a pre-trained generator
• interpolate between samples by walking the latent space
• do vector arithmetic on images in latent space
• train an image generator on my own image dataset

When to choose

• you specifically need the classic DCGAN architecture in Torch7
• you want to reproduce the DCGAN paper experiments
• you have legacy Torch7 infrastructure and an NVIDIA GPU

When to avoid

• you want a maintained framework - Torch7 is deprecated
• you prefer PyTorch or TensorFlow implementations of GANs
• you only have a CPU and need fast training

Facets

library · maturity abandoned

deep-learning machine-learning image-processing gpu-computing deep-learning machine-learning image-processing artificial-intelligence lua dcgan gan torch7 image-generation generative-models celeba lsun linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="soumith/dcgan.torch")