It is also a small summary in learning

# The following code is in jupyter notebook Running in 
import torch   
c=torch.arange(20).reshape(2,5,2)
c,c.sum(axis=0)   #(2,5,2) first 2 disappear , Two （5,2） Add up 
# Running results 
(tensor([[[ 0,  1],
          [ 2,  3],
          [ 4,  5],
          [ 6,  7],
          [ 8,  9]],
 
         [[10, 11],
          [12, 13],
          [14, 15],
          [16, 17],
          [18, 19]]]),
 tensor([[10, 12],
         [14, 16],
         [18, 20],
         [22, 24],
         [26, 28]]))

c=torch.arange(20).reshape(2,5,2)
c,c.sum(axis=1)    #(2,5,2) the second 5 disappear , Two （5,2） become （1,2） Add it up 
# Running results 
(tensor([[[ 0,  1],
          [ 2,  3],
          [ 4,  5],
          [ 6,  7],
          [ 8,  9]],
 
         [[10, 11],
          [12, 13],
          [14, 15],
          [16, 17],
          [18, 19]]]),
 tensor([[20, 25],
         [70, 75]]))

c=torch.arange(20).reshape(2,5,2)
c,c.sum(axis=2)     #(2,5,2) Third 2 disappear , become （2,5）
# Running results 
(tensor([[[ 0,  1],
          [ 2,  3],
          [ 4,  5],
          [ 6,  7],
          [ 8,  9]],
 
         [[10, 11],
          [12, 13],
          [14, 15],
          [16, 17],
          [18, 19]]]),
 tensor([[ 1,  5,  9, 13, 17],
         [21, 25, 29, 33, 37]]))

c=torch.arange(20).reshape(2,5,2)
c,c.sum(axis=1,keepdims=True).shape
# Run a screenshot 
(tensor([[[ 0,  1],
          [ 2,  3],
          [ 4,  5],
          [ 6,  7],
          [ 8,  9]],
 
         [[10, 11],
          [12, 13],
          [14, 15],
          [16, 17],
          [18, 19]]]),
 torch.Size([2, 1, 2]))

Steps of linear regression model
1. Generate data set .
2. Reading data sets .
3. Initialize model parameters .
4. Defining models
5. Loss function
6. optimization algorithm
7. Training