Tensor in Pytorch | Learn Tech with Rookie

type

status

date

slug

summary

📝Set up

As a libarary of python, we can just import libaray “troch”:

If you want to use “gym” for developing and comparing reinforcement leanrning, please refer to

And orther supports:

If you are not familiar with tensor, please check the last note Introduction

This gives:

The answer is 0.

A scalar(torch.tensor(a natrual number)), has no dimension.

The result is 7

This shows what is in torch.tensor (relation of tensor() and tensor().item() is kind of like a set and elements in this set.

The result is tensor([6, 7])

And a matrix is similiar.

tensor([[11, 12],
        [21, 22]])

torch.Size([2, 2])

But if we look at a tensor of order 3, it is a little deceptive.

For example

tensor([[[11, 12, 13],
         [21, 22, 23],
         [31, 32, 33]]])

It looks like a 3*3 matrix, however,

torch.Size([1, 3, 3])

What is 1,3,3? We can sepearate each “[]”.

1 means there is one matrix

3 means there is 3 vector

3 means there is 3 scalar

And another example makes it clearer:

torch.Size([1, 2, 3])

1 matrix(in this tensor of order 3), 2 vectors(in each matrix), 3 scalar(in each vector).

And

shows Error:

After reading the first [11,13], torch.tensor confirms the dimension of vector being 2. So different children-dimension is forbbiden.

观点2

引用的话语

总结文章的内容

💡

有关Notion安装或者使用上的问题，欢迎您在底部评论区留言，一起交流~