关于models中LSTM参数初始化部分

[mz2sj]

nn.init.orthogonal_(getattr(self.rnn, f'weight_hh_l{i}')) nn.init.kaiming_normal_(getattr(self.rnn, f'weight_ih_l{i}')) nn.init.constant_(getattr(self.rnn, f'bias_hh_l{i}'), val=0) nn.init.constant_(getattr(self.rnn, f'bias_ih_l{i}'), val=0) getattr(self.rnn, f'bias_hh_l{i}').chunk(4)[1].fill_(1)
想请教下博主参数初始化这部分，这里对weight和bias的初始化是对LSTM中各个门和候选记忆细胞所有参数的初始化吗？最后一句getattr(self.rnn, f'bias_hh_l{i}').chunk(4)[1].fill_(1)是什么意思呢？

[mz2sj]

https://drive.google.com/file/d/1sRmt0ptcYs-6tCqiMnEcRY0Epa2xQaFC/view?usp=drivesdk

…

---原始邮件--- 发件人: "BlueSkyBubble"<notifications@github.com&gt; 发送时间: 2020年1月21日(周二) 中午12:03 收件人: "songyingxin/TextClassification-Pytorch"<TextClassification-Pytorch@noreply.github.com&gt;; 抄送: "mz2sj"<1576401831@qq.com&gt;;"Author"<author@noreply.github.com&gt;; 主题: Re: [songyingxin/TextClassification-Pytorch] 关于models中LSTM参数初始化部分 (#1) /search/hadoop02/suanfa/songyingxin/data/embedding/glove/glove.840B.300d.txt 你好，这个文件可以分箱一下吗？ — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.