Pytorch实现的各种知识蒸馏方法-python源码

上传者: 42125826 | 上传时间: 2021-06-30 11:04:58 | 文件大小: 48KB | 文件类型: ZIP
Pytorch实现的各种知识蒸馏方法 Knowledge-Distillation-Zoo Pytorch 实现各种知识蒸馏 (KD) 方法。 本知识库是一个简单的参考资料,主要侧重于基础知识蒸馏/转移方法。 因此没有考虑许多技巧和变化,例如逐步训练、迭代训练、教师集成、KD 方法集成、无数据、自蒸馏、量化等。 希望它对您的项目或研究有用。 我将使用新的 KD 方法定期更新此 repo。 如果我遗漏了一些基本方法,请与我联系。 Lists Name Method Paper Link Code Link Baseline basic model with softmax loss — code Logits通过回归logits模拟学习论文代码ST软目标论文代码AT注意力转移论文代码Fitnet提示薄深度网络论文代码NST神经选择性转移论文代码PKT概率知识转移论文代码 FSP 求解流程过程论文代码 FT 因子转移论文代码 RKD 关系知识蒸馏论文代码 AB 激活边界论文代码 SP 相似性保存论文代码 Sobolev sobolev/jacobian 匹配论文代码 BSS 边

文件下载

资源详情

[{"title":"( 34 个子文件 48KB ) Pytorch实现的各种知识蒸馏方法-python源码","children":[{"title":"Knowledge-Distillation-Zoo-master","children":[{"title":"train_bss.py <span style='color:#111;'> 11.05KB </span>","children":null,"spread":false},{"title":"train_dml.py <span style='color:#111;'> 11.34KB </span>","children":null,"spread":false},{"title":"kd_losses","children":[{"title":"st.py <span style='color:#111;'> 563B </span>","children":null,"spread":false},{"title":"bss.py <span style='color:#111;'> 1.94KB </span>","children":null,"spread":false},{"title":"lwm.py <span style='color:#111;'> 1.58KB </span>","children":null,"spread":false},{"title":"vid.py <span style='color:#111;'> 1.76KB </span>","children":null,"spread":false},{"title":"irg.py <span style='color:#111;'> 3.33KB </span>","children":null,"spread":false},{"title":"ft.py <span style='color:#111;'> 676B </span>","children":null,"spread":false},{"title":"__init__.py <span style='color:#111;'> 498B </span>","children":null,"spread":false},{"title":"fsp.py <span style='color:#111;'> 892B </span>","children":null,"spread":false},{"title":"dml.py <span style='color:#111;'> 548B </span>","children":null,"spread":false},{"title":"at.py <span style='color:#111;'> 821B </span>","children":null,"spread":false},{"title":"ab.py <span style='color:#111;'> 679B </span>","children":null,"spread":false},{"title":"pkt.py <span style='color:#111;'> 1.54KB </span>","children":null,"spread":false},{"title":"fitnet.py <span style='color:#111;'> 414B </span>","children":null,"spread":false},{"title":"sobolev.py <span style='color:#111;'> 1.20KB </span>","children":null,"spread":false},{"title":"logits.py <span style='color:#111;'> 463B </span>","children":null,"spread":false},{"title":"crd.py <span style='color:#111;'> 4.39KB </span>","children":null,"spread":false},{"title":"afd.py <span style='color:#111;'> 2.99KB </span>","children":null,"spread":false},{"title":"nst.py <span style='color:#111;'> 1.52KB </span>","children":null,"spread":false},{"title":"cc.py <span style='color:#111;'> 1.08KB </span>","children":null,"spread":false},{"title":"ofd.py <span style='color:#111;'> 1.47KB </span>","children":null,"spread":false},{"title":"rkd.py <span style='color:#111;'> 1.92KB </span>","children":null,"spread":false},{"title":"sp.py <span style='color:#111;'> 658B </span>","children":null,"spread":false}],"spread":false},{"title":"utils.py <span style='color:#111;'> 1.75KB </span>","children":null,"spread":false},{"title":"dataset.py <span style='color:#111;'> 4.01KB </span>","children":null,"spread":false},{"title":"train_base.py <span style='color:#111;'> 7.61KB </span>","children":null,"spread":false},{"title":"train_crd.py <span style='color:#111;'> 10.33KB </span>","children":null,"spread":false},{"title":".gitignore <span style='color:#111;'> 1.82KB </span>","children":null,"spread":false},{"title":"README.md <span style='color:#111;'> 19.58KB </span>","children":null,"spread":false},{"title":"train_ft.py <span style='color:#111;'> 12.16KB </span>","children":null,"spread":false},{"title":"train_kd.py <span style='color:#111;'> 20.55KB </span>","children":null,"spread":false},{"title":"example_train_script.sh <span style='color:#111;'> 13.86KB </span>","children":null,"spread":false},{"title":"network.py <span style='color:#111;'> 8.23KB </span>","children":null,"spread":false}],"spread":false}],"spread":true}]

评论信息

免责申明

【只为小站】的资源来自网友分享,仅供学习研究,请务必在下载后24小时内给予删除,不得用于其他任何用途,否则后果自负。基于互联网的特殊性,【只为小站】 无法对用户传输的作品、信息、内容的权属或合法性、合规性、真实性、科学性、完整权、有效性等进行实质审查;无论 【只为小站】 经营者是否已进行审查,用户均应自行承担因其传输的作品、信息、内容而可能或已经产生的侵权或权属纠纷等法律责任。
本站所有资源不代表本站的观点或立场,基于网友分享,根据中国法律《信息网络传播权保护条例》第二十二条之规定,若资源存在侵权或相关问题请联系本站客服人员,zhiweidada#qq.com,请把#换成@,本站将给予最大的支持与配合,做到及时反馈和处理。关于更多版权及免责申明参见 版权及免责申明