{"id":32639690,"url":"https://github.com/ennengyang/rankone-moe","last_synced_at":"2026-06-28T20:31:25.844Z","repository":{"id":272147361,"uuid":"889429740","full_name":"EnnengYang/RankOne-MoE","owner":"EnnengYang","description":"秩一专家混合用于多任务学习. 计算机学报, 2025.    Mixture of Rank-One Experts for Multi-Task Learning. Chinese Journal of Computers, 2025.","archived":false,"fork":false,"pushed_at":"2025-05-29T15:22:21.000Z","size":2292,"stargazers_count":2,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-10-31T02:23:54.721Z","etag":null,"topics":["model-merging","multi-task-learning"],"latest_commit_sha":null,"homepage":"","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/EnnengYang.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}},"created_at":"2024-11-16T11:11:25.000Z","updated_at":"2025-10-26T09:13:20.000Z","dependencies_parsed_at":"2025-01-12T14:36:04.347Z","dependency_job_id":"e50f1c8a-9bc9-499a-a36e-21fe2d5ae383","html_url":"https://github.com/EnnengYang/RankOne-MoE","commit_stats":null,"previous_names":["ennengyang/rankone-moe"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/EnnengYang/RankOne-MoE","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EnnengYang%2FRankOne-MoE","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EnnengYang%2FRankOne-MoE/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EnnengYang%2FRankOne-MoE/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EnnengYang%2FRankOne-MoE/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/EnnengYang","download_url":"https://codeload.github.com/EnnengYang/RankOne-MoE/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EnnengYang%2FRankOne-MoE/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34903523,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-28T02:00:05.809Z","response_time":54,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"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":["model-merging","multi-task-learning"],"created_at":"2025-10-31T02:14:23.907Z","updated_at":"2026-06-28T20:31:25.836Z","avatar_url":"https://github.com/EnnengYang.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# \"秩一专家混合用于多任务学习. 计算机学报，2025.\"\n\n## 开发环境配置\n本项目依赖于[FusionBench-v0.2.4](https://github.com/tanganke/fusion_bench)，可参考该库来配置基本环境，或者按照如下步骤创建开发环境：\n\n第一步：创建一个Conda环境\n```bash\nconda create --name rankone-moe python=3.12.4\n```\n\n第二步：激活Conda环境\n```bash\nconda activate rankone-moe\n```\n\n第三步：安装本项目的依赖的环境\n```bash\ngit clone https://github.com/EnnengYang/RankOne-MoE\ncd RankOne-MoE\n\npip install -e . # install the package in editable mode\n```\n\n\n## 运行实验\n\n\u003e 注意：本项目涉及到的所有数据集和模型权重均可在代码运行时自动下载，请确保您的网络能访问[huggingface](https://huggingface.co/)网站，您也可以考虑手动下载[相关资源](https://huggingface.co/tanganke).\n\n\u003cbr\u003e\n\n- 实验：我们的RankOne-MoE方法在CLIP-ViT-B/32, CLIP-ViT-B/16, CLIP-ViT-L/14模型下的合并性能\n```bash\nbash examples/rankone_one/rankone_moe_gridsearch.sh\n```\n\u003e ViT-B/32(8个任务)的运行结果在[examples/results/clip-vit-base-patch32](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-base-patch32)\n\n\u003e ViT-B/16(8个任务)的运行结果在[examples/results/clip-vit-base-patch16](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-base-patch16)\n\n\u003e ViT-L/14(8个任务)的运行结果在[examples/results/clip-vit-large-patch14](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-large-patch14)\n\n\u003e ViT-B/32(20个任务)的运行结果在[examples/results/clip-vit-base-patch32_20Tasks](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-base-patch32_20Tasks)\n\n\u003cbr\u003e\n\n\n- 实验：基线方法们在CLIP-ViT-B/32模型下的合并性能(8个任务)\n```bash\nbash examples/rankone_one/baseline_b32.sh\n```\n\u003e ViT-B/32(8个任务)的运行结果在[examples/results/clip-vit-base-patch32](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-base-patch32)\n\n\u003cbr\u003e\n\n- 实验：基线方法们在CLIP-ViT-B/16模型下的合并性能(8个任务)\n```bash\nbash examples/rankone_one/baseline_b16.sh\n```\n\u003e  ViT-B/16(8个任务)的运行结果在[examples/results/clip-vit-base-patch16](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-base-patch16)\n\n\u003cbr\u003e\n\n- 实验：基线方法们在CLIP-ViT-L/14模型下的合并性能(8个任务)\n```bash\nbash examples/rankone_one/baseline_l14.sh\n```\n\u003e ViT-L/14(8个任务)的运行结果在[examples/results/clip-vit-large-patch14](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-large-patch14)\n\n\u003cbr\u003e\n\n\n- 实验：基线方法们在CLIP-ViT-B/32模型下的合并性能(20个任务)\n```bash\nbash examples/rankone_one/baseline_b32_20tasks.sh\n```\n\u003e ViT-B/32(20个任务)的运行结果在[examples/results/clip-vit-base-patch32_20Tasks](https://github.com/EnnengYang/RankOne-MoE/tree/main/examples/results/clip-vit-base-patch32_20Tasks)\n\n\u003cbr\u003e\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fennengyang%2Frankone-moe","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fennengyang%2Frankone-moe","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fennengyang%2Frankone-moe/lists"}