{"id":29216723,"url":"https://github.com/bytedance-seed/deepflow","last_synced_at":"2026-03-07T13:03:01.639Z","repository":{"id":301281036,"uuid":"973618239","full_name":"ByteDance-Seed/DeepFlow","owner":"ByteDance-Seed","description":"[ICCV 2025] Deeply Supervised Flow-Based Generative Models","archived":false,"fork":false,"pushed_at":"2025-06-26T03:13:42.000Z","size":739,"stargazers_count":12,"open_issues_count":0,"forks_count":2,"subscribers_count":3,"default_branch":"main","last_synced_at":"2025-06-26T04:21:41.639Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"https://deepflow-project.github.io/","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ByteDance-Seed.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","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,"zenodo":null}},"created_at":"2025-04-27T11:32:15.000Z","updated_at":"2025-06-26T03:19:48.000Z","dependencies_parsed_at":"2025-06-26T04:21:54.371Z","dependency_job_id":"9d07c4c9-7b90-401c-b5d1-98889e95878d","html_url":"https://github.com/ByteDance-Seed/DeepFlow","commit_stats":null,"previous_names":["bytedance-seed/deepflow"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/ByteDance-Seed/DeepFlow","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ByteDance-Seed%2FDeepFlow","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ByteDance-Seed%2FDeepFlow/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ByteDance-Seed%2FDeepFlow/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ByteDance-Seed%2FDeepFlow/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ByteDance-Seed","download_url":"https://codeload.github.com/ByteDance-Seed/DeepFlow/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ByteDance-Seed%2FDeepFlow/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30214618,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-07T12:15:00.571Z","status":"ssl_error","status_checked_at":"2026-03-07T12:15:00.217Z","response_time":53,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"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":"2025-07-03T01:03:30.938Z","updated_at":"2026-03-07T13:03:01.622Z","avatar_url":"https://github.com/ByteDance-Seed.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n 👋 Hi, everyone! \n    \u003cbr\u003e\n    We are \u003cb\u003eByteDance Seed team.\u003c/b\u003e\n\u003c/div\u003e\n\n\u003cp align=\"center\"\u003e\n  You can get to know us better through the following channels👇\n  \u003cbr\u003e\n  \u003ca href=\"https://team.doubao.com/\"\u003e\n    \u003cimg src=\"https://img.shields.io/badge/Website-%231e37ff?style=for-the-badge\u0026logo=bytedance\u0026logoColor=white\"\u003e\u003c/a\u003e\n  \u003ca href=\"https://github.com/user-attachments/assets/93481cda-a7f3-47f3-b333-fe6b3da86b78\"\u003e\n    \u003cimg src=\"https://img.shields.io/badge/WeChat-07C160?style=for-the-badge\u0026logo=wechat\u0026logoColor=white\"\u003e\u003c/a\u003e\n \u003ca href=\"https://www.xiaohongshu.com/user/profile/668e7e15000000000303157d?xsec_token=ABl2-aqekpytY6A8TuxjrwnZskU-6BsMRE_ufQQaSAvjc%3D\u0026xsec_source=pc_search\"\u003e\n    \u003cimg src=\"https://img.shields.io/badge/Xiaohongshu-%23FF2442?style=for-the-badge\u0026logo=xiaohongshu\u0026logoColor=white\"\u003e\u003c/a\u003e\n  \u003ca href=\"https://www.zhihu.com/org/dou-bao-da-mo-xing-tuan-dui/\"\u003e\n    \u003cimg src=\"https://img.shields.io/badge/zhihu-%230084FF?style=for-the-badge\u0026logo=zhihu\u0026logoColor=white\"\u003e\u003c/a\u003e\n\u003c/p\u003e\n\n![seed logo](https://github.com/user-attachments/assets/c42e675e-497c-4508-8bb9-093ad4d1f216)\n\n\u003c!-- 注释：以上为Seed官方信息，可直接复制使用，请注意导入“Seed WeChat”（第12行）、“Seed logo”(第20行)图片替换 --\u003e\n\n\n# Deeply Supervised Flow-Based Generative Models\n\n### [ICCV 2025]\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"https://deepflow-project.github.io/\"\u003e\n    \u003cimg src=\"https://img.shields.io/badge/DeepFlow-Project Page-yellow\"\u003e\u003c/a\u003e\n  \u003ca href=\"https://arxiv.org/abs/2503.14494\"\u003e\n    \u003cimg src=\"https://img.shields.io/badge/DeepFlow-Tech Report-red\"\u003e\u003c/a\u003e\n  \u003c!-- \u003ca href=\"https://www.apache.org/licenses/\"\u003e\n    \u003cimg src=\"https://img.shields.io/badge/License-Apache 2.0-blue\"\u003e\u003c/a\u003e --\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n    \u003ca href=\"https://dlsrbgg33.github.io/\" target=\"_blank\"\u003eInkyu\u0026nbsp;Shin\u003c/a\u003e \u0026ensp; \u003cb\u003e\u0026middot;\u003c/b\u003e \u0026ensp;\n    \u003ca href=\"https://www.chenglinyang.com/\" target=\"_blank\"\u003eChenglin\u0026nbsp;Yang\u003c/a\u003e \u0026ensp; \u003cb\u003e\u0026middot;\u003c/b\u003e \u0026ensp;\n    \u003ca href=\"http://liangchiehchen.com/\" target=\"_blank\"\u003eLiang-Chieh\u0026nbsp;Chen\u003c/a\u003e\n\u003c/p\u003e\n\n\n\nDeepFlow is a novel flow-based generation framework that enhances velocity representation through inter layer communication using deep supervision and acceleration mechanism.\n\nDeepFlow converges 8 times faster on ImageNet with equivalent performance and further reduces FID by 2.6 while halving training time compared to previous flow based models without a classifier free guidance. \n\n# News\n[2025/06/25]🎉DeepFlow is accepted to ICCV 2025.\n\n[2025/05/07]🔥We release training code, sampling code, and checkpoints of DeepFlow.\n\n# Introduction\nDeepFlow incorporates deep supervision by evenly adding velocity prediction within transformer blocks, further enhanced by the\nproposed Velocity Alignment block (VeRA).\n\u003cp\u003e\n\u003cimg src=\"assets/deepflow_teaser.png\" alt=\"teaser\" width=95% height=95%\u003e\n\u003c/p\u003e\n\n# Getting started\n## Installation\n```shell\npip install -r requirements.txt\n```\n\n## Dataset\nWe conducted our experiments using [ImageNet](https://www.kaggle.com/competitions/imagenet-object-localization-challenge/data).\nKindly refer to preprocessing guide from [REPA](https://github.com/sihyun-yu/REPA/tree/main/preprocessing) to preprocess the dataset.\n\n## Training\nFor training DeepFlow-XL/2-3T, use below training script.\n```shell\nbash script/xlarge/train.sh\n```\nwhere you can set following options:\n- `--tg-upper-bound`: time-gap between adjacent branches.\n- `--weighting`: time-step sampling during training.\n- `--df-idxs`: key transformer layers where deep supervision is applied. (currently, equally-splitted).\n- `--ssl-align`: whether to use SSL align (need to set up `--enc-type` as well).\n- `--legacy-scaling`: whether to use soft-cap for scaling factor in \"velocity_modulation\" (enable this when ssl-align is enabled for reproducibility).\n\n## Generation\nFor generative samples from DeepFlow-XL/2-3T, use below evaluation script.\n```shell\nbash script/xlarge/eval.sh\n```\nwhere you need to set specific path for checkpoint that will be evaluated.\n\nPlease follow [ADM evaluation](https://github.com/openai/guided-diffusion/tree/main/evaluations) for obtaining FID score.\n\n\n\n## Model Zoo\n\nWe release checkpoint of DeepFlow trained on ImageNet for your reference.\n\n| Dataset  | Model | SSL align | Training Epochs | Link | FID (wo/ CFG) | FID (w/ CFG) |\n| ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- |\n| ImageNet-256px | DeepFlow-XL/2-3T | X | 400 | [checkpoint](https://drive.google.com/file/d/1QXTDkeVRgfvBvyLkJcu-YJ_2pDxM73Ec/view?usp=sharing) | 7.2 | 1.97 |\n| ImageNet-256px | DeepFlow-XL/2-3T | O | 400 | [checkpoint](https://drive.google.com/file/d/1ALMz0Cdm1XSZYWTPSVEVR-sfYXsKk6Zl/view?usp=sharing) | 5.0 | 1.77 |\n\n\n# License\nThe code and model weights are licensed under CC-BY-NC. See LICENSE.txt for details.\n\n\n# Citation\nIf you find DeepFlow useful for your research and applications, feel free to give us a star ⭐ or cite us using:\n\n```bibtex\n@article{shin2025deeply,\n  title={Deeply Supervised Flow-Based Generative Models},\n  author={Shin, Inkyu and Yang, Chenglin and Chen, Liang-Chieh},\n  journal={arXiv preprint arXiv:2503.14494},\n  year={2025}\n}\n```\n\n## Acknowledgement\n\n[DiT](https://github.com/facebookresearch/DiT)\n\n[SiT](https://github.com/willisma/SiT)\n\n[REPA](https://github.com/sihyun-yu/REPA)\n\n# About [ByteDance Seed Team](https://team.doubao.com/)\n\nFounded in 2023, ByteDance Seed Team is dedicated to crafting the industry's most advanced AI foundation models. The team aspires to become a world-class research team and make significant contributions to the advancement of science and society.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbytedance-seed%2Fdeepflow","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fbytedance-seed%2Fdeepflow","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbytedance-seed%2Fdeepflow/lists"}