{"id":13994009,"url":"https://github.com/ZhangYuanhan-AI/NOAH","last_synced_at":"2025-07-22T18:32:57.236Z","repository":{"id":37634051,"uuid":"500317631","full_name":"ZhangYuanhan-AI/NOAH","owner":"ZhangYuanhan-AI","description":"[TPAMI] Searching prompt modules for parameter-efficient transfer learning.","archived":false,"fork":false,"pushed_at":"2023-12-08T03:25:30.000Z","size":45268,"stargazers_count":233,"open_issues_count":9,"forks_count":12,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-07-11T09:01:29.951Z","etag":null,"topics":["deep-learning","domain-generalization","pre-trained-model","prompt-tuning","pytorch","transfer-learning","visual-prompting"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ZhangYuanhan-AI.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}},"created_at":"2022-06-06T06:29:20.000Z","updated_at":"2025-07-02T22:42:53.000Z","dependencies_parsed_at":"2023-12-08T04:26:25.658Z","dependency_job_id":"d9fa9778-ed07-4f78-88da-f01e8148580a","html_url":"https://github.com/ZhangYuanhan-AI/NOAH","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/ZhangYuanhan-AI/NOAH","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ZhangYuanhan-AI%2FNOAH","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ZhangYuanhan-AI%2FNOAH/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ZhangYuanhan-AI%2FNOAH/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ZhangYuanhan-AI%2FNOAH/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ZhangYuanhan-AI","download_url":"https://codeload.github.com/ZhangYuanhan-AI/NOAH/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ZhangYuanhan-AI%2FNOAH/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":265642167,"owners_count":23803820,"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":["deep-learning","domain-generalization","pre-trained-model","prompt-tuning","pytorch","transfer-learning","visual-prompting"],"created_at":"2024-08-09T14:02:39.949Z","updated_at":"2025-07-22T18:32:52.223Z","avatar_url":"https://github.com/ZhangYuanhan-AI.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n\u003ch1\u003eNeural Prompt Search\u003c/h1\u003e\n\n\u003cdiv\u003e\n    \u003ca href='https://zhangyuanhan-ai.github.io/' target='_blank'\u003eYuanhan Zhang\u003c/a\u003e\u0026emsp;\n    \u003ca href='https://kaiyangzhou.github.io/' target='_blank'\u003eKaiyang Zhou\u003c/a\u003e\u0026emsp;\n    \u003ca href='https://liuziwei7.github.io/' target='_blank'\u003eZiwei Liu\u003c/a\u003e\n\u003c/div\u003e\n\u003cdiv\u003e\n    S-Lab, Nanyang Technological University\n\u003c/div\u003e\n\n\n\u003cimg src=\"figures/motivation.png\"\u003e\n\n\u003ch3\u003eTL;DR\u003c/h3\u003e\n    \nThe idea is simple: we view existing parameter-efficient tuning modules, including [Adapter](https://arxiv.org/abs/1902.00751), [LoRA](https://arxiv.org/abs/2106.09685) and [VPT](https://arxiv.org/abs/2203.12119), as prompt modules and propose to search the optimal configuration via neural architecture search. Our approach is named **NOAH** (Neural prOmpt seArcH).\n\n---\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"https://arxiv.org/abs/2206.04673\" target='_blank'\u003e[arXiv]\u003c/a\u003e •\n  \u003ca href=\"https://zhangyuanhan-ai.github.io/NOAH\" target='_blank'\u003e[project page]\u003c/a\u003e\n\u003c/p\u003e\n\n\u003c/div\u003e\n\n\n\n## Updatas\n[05/2022] [arXiv](https://arxiv.org/abs/2206.04673) paper has been **released**.\n\n## Environment Setup\n```\nconda create -n NOAH python=3.8\nconda activate NOAH\npip install -r requirements.txt\n```\n\n## Data Preparation\n\n### 1. Visual Task Adaptation Benchmark (VTAB)\n```\ncd data/vtab-source\npython get_vtab1k.py\n```\n\n### 2. Few-Shot and Domain Generation\n\n- Images\n\n    Please refer to [DATASETS.md](https://github.com/KaiyangZhou/CoOp/blob/main/DATASETS.md) to download the datasets.\n\n- Train/Val/Test splits\n\n    Please refer to files under `data/XXX/XXX/annotations` for the detail information.\n\n\n## Quick Start For NOAH\nWe use the VTAB experiments as examples.\n\n### 1. Downloading the Pre-trained Model\n| Model | Link |\n|-------|------|\n|ViT B/16 | [link](https://storage.googleapis.com/vit_models/imagenet21k/ViT-B_16.npz)|\n\n### 2. Supernet Training\n```\nsh configs/NOAH/VTAB/supernet/slurm_train_vtab.sh PATH-TO-YOUR-PRETRAINED-MODEL\n```\n\n### 3. Subnet Search\n```\nsh configs/NOAH/VTAB/search/slurm_search_vtab.sh PARAMETERS-LIMITES\n```\n### 4. Subnet Retraining\n```\nsh configs/NOAH/VTAB/subnet/slurm_retrain_vtab.sh PATH-TO-YOUR-PRETRAINED-MODEL\n```\nWe add the optimal subnet architecture of each dataset in the ``experiments/NOAH/subnet/VTAB``.\n\n### 5. Performance\n![fig1](figures/table1.jpg)\n\n## Citation\nIf you use this code in your research, please kindly cite this work.\n```\n@misc{zhang2022neural,\n      title={Neural Prompt Search}, \n      author={Yuanhan Zhang and Kaiyang Zhou and Ziwei Liu},\n      year={2022},\n      eprint={2206.04673},\n      archivePrefix={arXiv},\n      primaryClass={cs.CV}\n}\n```\n\n## Acknoledgments\nPart of the code is borrowed from [CoOp](https://github.com/KaiyangZhou/CoOp), [AutoFormer](https://github.com/microsoft/Cream/tree/main/AutoFormer), [timm](https://github.com/rwightman/pytorch-image-models) and [mmcv](https://github.com/open-mmlab/mmcv).\n\nThanks to Chong Zhou (https://chongzhou96.github.io/) for the code of downloading the VTAB-1k.\n\n\u003cdiv align=\"center\"\u003e\n\n[![Hits](https://hits.seeyoufarm.com/api/count/incr/badge.svg?url=https%3A%2F%2Fgithub.com%2FZhangYuanhan-AI%2FNOAH\u0026count_bg=%2379C83D\u0026title_bg=%23555555\u0026icon=\u0026icon_color=%23E7E7E7\u0026title=visitors\u0026edge_flat=false)](https://hits.seeyoufarm.com)\n\n\u003c/div\u003e\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FZhangYuanhan-AI%2FNOAH","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FZhangYuanhan-AI%2FNOAH","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FZhangYuanhan-AI%2FNOAH/lists"}