{"id":16270620,"url":"https://github.com/tongjilibo/rec4torch","last_synced_at":"2025-08-02T16:30:46.993Z","repository":{"id":60330404,"uuid":"542387630","full_name":"Tongjilibo/rec4torch","owner":"Tongjilibo","description":"推荐系统的pytorch算法实现","archived":false,"fork":false,"pushed_at":"2024-02-04T09:30:09.000Z","size":199,"stargazers_count":66,"open_issues_count":0,"forks_count":3,"subscribers_count":3,"default_branch":"master","last_synced_at":"2024-11-27T17:51:08.283Z","etag":null,"topics":["ctr","deepfm","dien","widedeep"],"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/Tongjilibo.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":"2022-09-28T03:17:07.000Z","updated_at":"2024-11-27T15:31:55.000Z","dependencies_parsed_at":"2024-10-10T18:11:05.188Z","dependency_job_id":null,"html_url":"https://github.com/Tongjilibo/rec4torch","commit_stats":{"total_commits":25,"total_committers":2,"mean_commits":12.5,"dds":0.48,"last_synced_commit":"6c3ca60e9bcc0cdfde4ec9c75430003df466cb63"},"previous_names":[],"tags_count":2,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Tongjilibo%2Frec4torch","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Tongjilibo%2Frec4torch/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Tongjilibo%2Frec4torch/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Tongjilibo%2Frec4torch/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Tongjilibo","download_url":"https://codeload.github.com/Tongjilibo/rec4torch/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":228491964,"owners_count":17928719,"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":["ctr","deepfm","dien","widedeep"],"created_at":"2024-10-10T18:11:01.417Z","updated_at":"2024-12-06T16:13:18.600Z","avatar_url":"https://github.com/Tongjilibo.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# rec4torch\n推荐系统的pytorch算法实现\n\n[![licence](https://img.shields.io/github/license/Tongjilibo/rec4torch.svg?maxAge=3600)](https://github.com/Tongjilibo/rec4torch/blob/master/LICENSE) \n[![PyPI](https://img.shields.io/pypi/v/rec4torch?label=pypi%20package)](https://pypi.org/project/rec4torch/) \n[![PyPI - Downloads](https://img.shields.io/pypi/dm/rec4torch)](https://pypistats.org/packages/rec4torch)\n[![GitHub stars](https://img.shields.io/github/stars/Tongjilibo/rec4torch?style=social)](https://github.com/Tongjilibo/rec4torch)\n[![GitHub Issues](https://img.shields.io/github/issues/Tongjilibo/rec4torch.svg)](https://github.com/Tongjilibo/rec4torch/issues)\n[![contributions welcome](https://img.shields.io/badge/contributions-welcome-brightgreen.svg?style=flat)](https://github.com/Tongjilibo/rec4torch/issues)\n\n## 1. 下载安装\n安装稳定版\n```shell\npip install rec4torch\n```\n安装最新版\n```shell\npip install git+https://www.github.com/Tongjilibo/rec4torch.git\n```\n\n\n## 2. 功能\n- **核心功能**：基于pytorch实现各类推荐算法(DeepFM, WideDeep, DCN, DIN, DIEN)\n- **主要区别**：相对于deep-ctr, 去除对tensorflow和keras的依赖，去除重复过embedding的操作，原生支持multiclass\n- **训练过程**：\n    ```text\n    2022-10-28 23:16:10 - Start Training\n    2022-10-28 23:16:10 - Epoch: 1/5\n    5000/5000 [==============================] - 13s 3ms/step - loss: 0.1351 - acc: 0.9601\n    Evaluate: 100%|██████████████████████████████████████████████████| 2500/2500 [00:03\u003c00:00, 798.09it/s] \n    test_acc: 0.98045. best_test_acc: 0.98045\n\n    2022-10-28 23:16:27 - Epoch: 2/5\n    5000/5000 [==============================] - 13s 3ms/step - loss: 0.0465 - acc: 0.9862\n    Evaluate: 100%|██████████████████████████████████████████████████| 2500/2500 [00:03\u003c00:00, 635.78it/s] \n    test_acc: 0.98280. best_test_acc: 0.98280\n\n    2022-10-28 23:16:44 - Epoch: 3/5\n    5000/5000 [==============================] - 15s 3ms/step - loss: 0.0284 - acc: 0.9915\n    Evaluate: 100%|██████████████████████████████████████████████████| 2500/2500 [00:03\u003c00:00, 673.60it/s] \n    test_acc: 0.98365. best_test_acc: 0.98365\n\n    2022-10-28 23:17:03 - Epoch: 4/5\n    5000/5000 [==============================] - 15s 3ms/step - loss: 0.0179 - acc: 0.9948\n    Evaluate: 100%|██████████████████████████████████████████████████| 2500/2500 [00:03\u003c00:00, 692.34it/s] \n    test_acc: 0.98265. best_test_acc: 0.98365\n\n    2022-10-28 23:17:21 - Epoch: 5/5\n    5000/5000 [==============================] - 14s 3ms/step - loss: 0.0129 - acc: 0.9958\n    Evaluate: 100%|██████████████████████████████████████████████████| 2500/2500 [00:03\u003c00:00, 701.77it/s] \n    test_acc: 0.98585. best_test_acc: 0.98585\n\n    2022-10-28 23:17:37 - Finish Training\n    ```\n\n\n## 3. 快速上手\n- 参考了[deepctr-torch](https://github.com/shenweichen/DeepCTR-Torch), 使用[torch4keras](https://github.com/Tongjilibo/torch4keras)中作为Trainer\n- [测试用例](https://github.com/Tongjilibo/rec4torch/tree/master/examples)\n\n\n## 4. 版本说明\n- **v0.0.2**：20240204 更新依赖项torch4keras版本\n- **v0.0.1**：20221027 dcn, deepcrossing, deepfm, din, dien, wide\u0026deep, ncf等模型，训练过程修改为传入dataloader，合并models和layers，合并简化embedding_lookup，去掉一些重复的embedding过程(提速)\n\n\n## 5. 更新：\n- **20240204**：更新依赖项torch4keras版本\n- **20221110**：增加自定义的TensorDataset和collate_fn_device，支持指定device，防止显存占用多大，用out_dim和loss来替代task参数，兼容多分类\n- **20221027**：增加deepcrossing, ncf, din, dien算法，使用torch4keras作为trainer\n- **20220930**：初版提交, 训练过程修改为传入dataloader(参考bert4torch)，合并models和layers(模型结构较简单)，合并简化embedding_lookup，去掉一些重复的embedding过程(提速)","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftongjilibo%2Frec4torch","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Ftongjilibo%2Frec4torch","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftongjilibo%2Frec4torch/lists"}