Permute 3.5.6 Multilingual macOS. Video, audio and image files come in many different kinds and shapes, but sometimes you need a specific format since your iPad or DVD player won’t play that video. That is what Permute is for – easily convert your media files to various different formats. Permute 3 was started from scratch - completely new project, everything written from the ground up again. It includes many new features and the entire workflow has been improved with many per-conversion customizations. February 17, 2019. Permute offers easy-to-use drag-and-drop video conversion. Features: Easy to Use – built from the ground up, Permute is a perfect example of what a Mac app should be. With a gorgeous interface and drag & drop simplicity no need for complicated options.
1 先看看官方中英文doc:
Hi, I am trying to permute the feature map dimension in a tensor. As a very simplified case, If I have a tensor of size (5, 4, 3, 6) I want to rearrange the above tensor along its dimension 1 (i.e. 4) from 0,1,2,3 to 0,2,1,3 One possible way I found was to do a indexselect followed by cat. Swackett x v1 7. Permute 3.1.1 Multilingual macOS 52 mbVideo, audio and image files come in many different kinds and shapes, but sometimes you need a specific format since your iPad or DVD player won't play that.
1.1 permute(dims)
将tensor的维度换位。
参数: - __dims__ (int .*) - 换位顺序
例:
1.2 permute(*dims) → Tensor
Permute the dimensions of this tensor.
Parameters: *dims (int..) – The desired ordering of dimensions
Example:
2 pytorch permute的使用
permute函数功能还是比较简单的,下面主要介绍几个细节点:
2.1 transpose与permute的异同
Tensor.permute(a,b,c,d, ..):permute函数可以对任意高维矩阵进行转置,但没有 torch.permute() 这个调用方式, 只能 Tensor.permute():
torch.transpose(Tensor, a,b):transpose只能操作2D矩阵的转置,有两种调用方式;
另:连续使用transpose也可实现permute的效果:
从以上操作中可知,permute相当于可以同时操作于tensor的若干维度,transpose只能同时作用于tensor的两个维度;
Vmware fusion pro 10 0 1. 2.2 permute函数与contiguous、view函数之关联
contiguous:view只能作用在contiguous的variable上,如果在view之前调用了transpose、permute等,就需要调用contiguous()来返回一个contiguous copy;
一种可能的解释是:有些tensor并不是占用一整块内存,而是由不同的数据块组成,而tensor的view()操作依赖于内存是整块的,这时只需要执行contiguous()这个函数,把tensor变成在内存中连续分布的形式;
判断ternsor是否为contiguous,可以调用torch.Tensor.is_contiguous()函数:
另:在pytorch的最新版本0.4版本中,增加了torch.reshape(),与 numpy.reshape() 的功能类似,大致相当于 tensor.contiguous().view(),这样就省去了对tensor做view()变换前,调用contiguous()的麻烦;
Permute 3 1 1/3
3 permute与view函数功能demo
利用函数 permute(2,0,1) 可以把 Tensor([[[1,2,3],[4,5,6]]]) 转换成:
如果使用view(1,3,2) 可以得到:
5 参考
Permute 3 1 1/4
发布于 2019-08-09