search

blackfeather-wang / ISDA-for-Deep-Networks

An efficient implicit semantic augmentation method, complementary to existing non-semantic techniques.
582 stars 93 forks source link

If fc can only be nn.Linear rather nn.Conv2d? #9

Closed xiong224 closed 3 years ago

blackfeather-wang commented 3 years ago

Thank you for your attention. It can be nn.Conv2d, but some reshape operations may be essential. Please refer to our segmentation code for details.

xiong224 commented 3 years ago

Thanks very much

------------------ 原始邮件 ------------------ 发件人: "blackfeather-wang/ISDA-for-Deep-Networks" <notifications@github.com>; 发送时间: 2021年1月21日(星期四) 晚上10:53 收件人: "blackfeather-wang/ISDA-for-Deep-Networks"<ISDA-for-Deep-Networks@noreply.github.com>; 抄送: "熊晏民"<895917144@qq.com>;"Author"<author@noreply.github.com>; 主题: Re: [blackfeather-wang/ISDA-for-Deep-Networks] If fc can only be nn.Linear rather nn.Conv2d? (#9)

Thank you for your attention. It can be nn.Conv2d, but some reshape operations may be essential. Please refer to our segmentation code for details.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.