生信方向（DTI）模型遇见的各种问题

每个知识点下面的序号可以理解为是否容易理解的难易程度
离线安装pytorch：https://blog.csdn.net/weixin_44956153/article/details/142303905

多模态：https://blog.csdn.net/star_nwe/article/details/143416379

embedding解释：https://blog.csdn.net/qq_43426908/article/details/133046902

torch.transpose解释：https://blog.csdn.net/qq_46396470/article/details/140101469

torch.view解释:https://blog.csdn.net/york1996/article/details/81949843

torch.reshape解释:https://blog.csdn.net/qq_46396470/article/details/140101450

PyTorch 中 Tensor 的 reshape、view、transpose、permute 函数的区别：https://blog.csdn.net/qq_46396470/article/details/140103361

批量归一化解释：https://blog.csdn.net/2202_75851137/article/details/139277351

卷积神经网络CNN——卷积层、池化层、全连接层：
https://blog.csdn.net/LittleRuby/article/details/134686509
https://blog.csdn.net/qq_59702185/article/details/143744365

图卷积网络：
https://blog.csdn.net/weixin_42037511/article/details/126271782
或看B站视频理解：
（1）https://www.bilibili.com/video/BV1eeW2e6E5h?spm_id_from=333.788.videopod.episodes&vd_source=c0e9cfc36bb63a3a2d7a5bf3dba36052&p=7
（2）https://www.bilibili.com/video/BV1VV411478E/?from=search&seid=1725700777641154181&spm_id_from=333.788.comment.all.click&vd_source=c0e9cfc36bb63a3a2d7a5bf3dba36052

nn.Linear全连接层：https://blog.csdn.net/2302_79223724/article/details/142469042

权重归一化：https://blog.csdn.net/hxxjxw/article/details/124071198

注意力机制、自注意力机制、多头注意力机制：
（1）https://blog.csdn.net/weixin_60737527/article/details/127141542
（2）https://www.bilibili.com/video/BV1dt4y1J7ov/?spm_id_from=333.337.search-card.all.click&vd_source=c0e9cfc36bb63a3a2d7a5bf3dba36052
（3）https://www.bilibili.com/video/BV1PP4y1T7Yu?spm_id_from=333.788.player.switch&vd_source=c0e9cfc36bb63a3a2d7a5bf3dba36052
或
https://my.oschina.net/u/876354/blog/3061863
https://blog.csdn.net/weixin_43334693/article/details/130189238

PyTorch 张量乘法：https://blog.csdn.net/panbaoran913/article/details/125999474
https://blog.csdn.net/deephub/article/details/141156123

Adam优化器：https://blog.csdn.net/zxqq_/article/details/143820757