{"id":21923094,"url":"https://github.com/26hzhang/coteaching","last_synced_at":"2025-08-11T06:40:42.672Z","repository":{"id":92547414,"uuid":"189856187","full_name":"26hzhang/CoTeaching","owner":"26hzhang","description":"A TensorFlow implementation of \"Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels\"","archived":false,"fork":false,"pushed_at":"2019-06-14T05:10:12.000Z","size":38,"stargazers_count":7,"open_issues_count":0,"forks_count":4,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-04-19T18:22:14.939Z","etag":null,"topics":["deep-learning","noisy-label-learning","noisy-labels","pytorch","tensorflow"],"latest_commit_sha":null,"homepage":null,"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/26hzhang.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":"2019-06-02T14:25:56.000Z","updated_at":"2024-11-29T07:31:29.000Z","dependencies_parsed_at":"2023-03-07T02:15:18.467Z","dependency_job_id":null,"html_url":"https://github.com/26hzhang/CoTeaching","commit_stats":null,"previous_names":["26hzhang/coteaching","isaacchanghau/coteaching"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/26hzhang/CoTeaching","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/26hzhang%2FCoTeaching","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/26hzhang%2FCoTeaching/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/26hzhang%2FCoTeaching/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/26hzhang%2FCoTeaching/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/26hzhang","download_url":"https://codeload.github.com/26hzhang/CoTeaching/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/26hzhang%2FCoTeaching/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":269842894,"owners_count":24484107,"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","status":"online","status_checked_at":"2025-08-11T02:00:10.019Z","response_time":75,"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":["deep-learning","noisy-label-learning","noisy-labels","pytorch","tensorflow"],"created_at":"2024-11-28T21:09:05.495Z","updated_at":"2025-08-11T06:40:42.622Z","avatar_url":"https://github.com/26hzhang.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# CoTeaching\n\nThis repository reproduces the NeurIPS'18 paper\n[Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels](\nhttps://papers.nips.cc/paper/8072-co-teaching-robust-training-of-deep-neural-networks-with-extremely-noisy-labels.pdf) \nby TensorFlow.\n\n- TensorFlow implementation, see all `*_tf.py` files.\n- Adapt original co-teaching PyTorch implementation to PyTorch `1.1.0`, see all `*_th.py` files. The original PyTorch\nimplementation is provided by the author \"[Bo Han](https://bhanml.github.io)\" as: [[bhanML/Co-teaching]](\nhttps://github.com/bhanML/Co-teaching).\n\n## Requirements\nThe codes are developed and tested on MacOS (`python==3.7.x`, CPU) and Ubuntu 18.04 (`python==3.6.x`, NVIDIA GeForce GTX \n1080 Ti GPU with `CUDA==10.0`) with following environment:\n- tensorflow==1.13.1 (\u003e=1.8.0)\n- pytorch==1.1.0 (\u003e=0.4.1)\n- numpy==1.14.6 (\u003e=1.14.2)\n\n## Setups\n**On MacOS**\n\nInstall TensorFlow via:\n```bash\n$ pip3 install tensorflow==1.13.1\n```\nInstall PyTorch via:\n```bash\n$ pip3 install torch torchvision\n```\n**On Ubuntu**\n\nInstall TensorFlow via:\n```bash\n$ pip3 install tensorflow==1.13.1  # CPU version\n$ pip3 install tensorflow-gpu==1.13.1  # GPU version\n```\nInstall PyTorch via:\n```bash\n# CPU version\n$ pip3 install https://download.pytorch.org/whl/cpu/torch-1.1.0-cp36-cp36m-linux_x86_64.whl  \n$ pip3 install https://download.pytorch.org/whl/cpu/torchvision-0.3.0-cp36-cp36m-linux_x86_64.whl\n# GPU version\n$ pip3 install https://download.pytorch.org/whl/cu100/torch-1.1.0-cp36-cp36m-linux_x86_64.whl\n$ pip3 install https://download.pytorch.org/whl/cu100/torchvision-0.3.0-cp36-cp36m-linux_x86_64.whl\n```\n\n## Usage\nHere is an example for TensorFlow:\n```bash\n$ python3 main_tf.py --dataset cifar10 --noise_type symmetric --noise_rate 0.5\n```\nHere is an example for PyTorch: \n```bash\n$ python3 main_th.py --dataset cifar10 --noise_type symmetric --noise_rate 0.5\n```\n\n## Performance\nPerformance on benchmark datasets reported by the Author:\n\n| (Flipping, Rate) | MNIST  | CIFAR-10 | CIFAR-100 |\n| ---------------: | -----: | -------: | --------: |\n| (Pair, 45%)      | 87.58% | 72.85%   | 34.40%    |\n| (Symmetry, 50%)  | 91.68% | 74.49%   | 41.23%    |\n| (Symmetry, 20%)  | 97.71% | 82.18%   | 54.36%    |\n\nPerformance on benchmark datasets derived by the codes in this repository:\n\u003e `th` means PyTorch while `tf` means TensorFlow. \n\n| (Flipping, Rate) | MNIST (th -- tf) | CIFAR-10 (th -- tf) | CIFAR-100 (th -- tf) |\n| ---------------: | ---------------: | ------------------: | -------------------: |\n| (Pair, 45%)      | 88.63% -- 94.16% | 72.88% -- 76.04%    | 34.05% -- 35.24%     |\n| (Symmetry, 50%)  | 92.34% -- 98.05% | 74.56% -- 79.64%    | 41.17% -- 49.09%     |\n| (Symmetry, 20%)  | 97.84% -- 99.16% | 82.87% -- 87.02%    | 54.11% -- 59.55%     |\n\n\u003e The model structure and parameters setting of TensorFlow version are almost same as those of PyTorch version, \nbut the performance of TensorFlow version is generally better than the PyTorch version, I think it maybe caused by the \ninternal implementation of some functions are different between these two frameworks.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2F26hzhang%2Fcoteaching","html_url":"https://awesome.ecosyste.ms/projects/github.com%2F26hzhang%2Fcoteaching","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2F26hzhang%2Fcoteaching/lists"}