{"id":17031392,"url":"https://github.com/breandan/hgan","last_synced_at":"2026-05-19T02:04:20.647Z","repository":{"id":146966541,"uuid":"127544203","full_name":"breandan/hGAN","owner":"breandan","description":"Hyper volume maximization for GAN training","archived":false,"fork":false,"pushed_at":"2018-03-31T15:34:29.000Z","size":110,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":3,"default_branch":"master","last_synced_at":"2025-06-04T19:18:41.438Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":false,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/breandan.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2018-03-31T15:21:03.000Z","updated_at":"2018-05-06T04:32:23.000Z","dependencies_parsed_at":"2023-05-05T22:30:39.230Z","dependency_job_id":null,"html_url":"https://github.com/breandan/hGAN","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/breandan/hGAN","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/breandan%2FhGAN","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/breandan%2FhGAN/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/breandan%2FhGAN/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/breandan%2FhGAN/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/breandan","download_url":"https://codeload.github.com/breandan/hGAN/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/breandan%2FhGAN/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":265249557,"owners_count":23734478,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-10-14T08:23:59.120Z","updated_at":"2026-05-19T02:04:20.619Z","avatar_url":"https://github.com/breandan.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Hyper Volume Generative Adversarial Network - hGAN\n\nReplication of [Stabilizing GAN Training with Multiple Random Projections](https://arxiv.org/abs/1705.07831) and extension including training with multi-objective training via hyper volume maximization\n\n## To run\n\nDownload the [cropped and aligned version of CelebA](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html) and unzip it\n\n```\npython train.py --ndiscriminators 12\n```\n\n```\noptional arguments:\n  -h, --help            show this help message and exit\n  --batch-size N        input batch size for training (default: 64)\n  --epochs N            number of epochs to train (default: 50)\n  --lr LR               learning rate (default: 0.0002)\n  --beta1 lambda        Adam beta param (default: 0.5)\n  --beta2 lambda        Adam beta param (default: 0.999)\n  --ndiscriminators NDISCRIMINATORS\n                        Number of discriminators. Default=8\n  --checkpoint-epoch N  epoch to load for checkpointing. If None, training\n                        starts from scratch\n  --checkpoint-path Path\n                        Path for checkpointing\n  --data-path Path      Path to data\n  --workers WORKERS     number of data loading workers\n  --seed S              random seed (default: 1)\n  --save-every N        how many epochs to wait before logging training\n                        status. Default is 5\n  --hyper-mode          enables training with hypervolume maximization\n  --nadir-factor nadir  Factor of the max disc loss to initialize nadir point\n                        (default: 50.0)\n  --no-cuda             Disables GPU use\n```\n\n## Tested with\n\n\n- Python 3.6\n- Pytorch 0.3.0\n\n## To do\n\n- Scheduler for the nadir point\n\nCollaborators: Isabela Albuquerque, Breandan Considine\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbreandan%2Fhgan","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fbreandan%2Fhgan","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbreandan%2Fhgan/lists"}