<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[pythonwyq的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/pythonwyq</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; pythonwyq]]></copyright><item><title><![CDATA[最长公共子序列并输出它]]></title><link>https://blog.youkuaiyun.com/pythonwyq/article/details/125746775</link><guid>https://blog.youkuaiyun.com/pythonwyq/article/details/125746775</guid><author>pythonwyq</author><pubDate>Tue, 12 Jul 2022 16:50:50 +0800</pubDate><description><![CDATA[力扣]]></description><category></category></item><item><title><![CDATA[【论文笔记】Multi-Sample Dropout for Accelerated Training and Better Generalization]]></title><link>https://blog.youkuaiyun.com/pythonwyq/article/details/125511199</link><guid>https://blog.youkuaiyun.com/pythonwyq/article/details/125511199</guid><author>pythonwyq</author><pubDate>Tue, 28 Jun 2022 22:37:14 +0800</pubDate><description><![CDATA[论文简介：训练更快，泛化更强的Dropout：Multi-Sample Dropout
论文标题：Multi-Sample Dropout for Accelerated Training and Better Generalization
论文链接：https://arxiv.org/pdf/1905.09788.pdf
论文作者：{Hiroshi Inoue}本文阐述的也是一种 dropout 技术的变形——multi-sample dropout。传统 dropout 在每轮训练时会从输入中随机选择一]]></description><category></category></item><item><title><![CDATA[Projected Gradient Descent（PGD）对抗训练]]></title><link>https://blog.youkuaiyun.com/pythonwyq/article/details/125472030</link><guid>https://blog.youkuaiyun.com/pythonwyq/article/details/125472030</guid><author>pythonwyq</author><pubDate>Sun, 26 Jun 2022 17:38:17 +0800</pubDate><description><![CDATA[PGD对抗训练代码]]></description><category></category></item><item><title><![CDATA[FGM对抗训练]]></title><link>https://blog.youkuaiyun.com/pythonwyq/article/details/125469068</link><guid>https://blog.youkuaiyun.com/pythonwyq/article/details/125469068</guid><author>pythonwyq</author><pubDate>Sun, 26 Jun 2022 13:23:55 +0800</pubDate><description><![CDATA[​对抗训练无论是在CV领域还是在NLP领域都具有举足轻重的地位，在NLP比赛中,抗训练确确实实能够提升模型在具体任务上的泛化性能。

​]]></description><category></category></item><item><title><![CDATA[基于bert继续预训练]]></title><link>https://blog.youkuaiyun.com/pythonwyq/article/details/125464819</link><guid>https://blog.youkuaiyun.com/pythonwyq/article/details/125464819</guid><author>pythonwyq</author><pubDate>Sat, 25 Jun 2022 22:40:18 +0800</pubDate><description><![CDATA[在目前的各项NLP任务中，如果要在特定任务或者领域应用文本分类，数据分布一定是有一些差距的。这时候可以考虑进行深度预训练。进行继续预训练，有利于提升任务的性能。]]></description><category></category></item></channel></rss>