Python predicts the model code demo of the next number through the first three numbers

Related knowledge ：

• Use scatter_ Conduct one-hot code

scatter_(self, dim, index, value) take src The data in is based on index The index in is in accordance with dim Fill in the direction of value value , This function can be converted from to onehot Code to understand

• LSTM return output、hidden and cell

output, (hidden,cell) = self.LSTM(x)

• RNN return output、hidden

output, hidden = self.RNN(x)

# -----------------------------------
#  Module import 
import numpy
import torch
from torch import nn

# -----------------------------------
#  Data preprocessing 
data_length = 30
#  Definition 30 Number , Through the first three predictions, the latter , such as ：1,2,3->4
seq_length = 3
#  It can be seen from the above that the sequence length is 3

number = [i for i in range(data_length)]
li_x = []
li_y = []
for i in range(0, data_length - seq_length):
    x = number[i: i + seq_length]
    y = number[i + seq_length]
    li_x.append(x)
    li_y.append(y)
#number: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29]
#li_x: [[0, 1, 2], [1, 2, 3], [2, 3, 4], [3, 4, 5], [4, 5, 6], [5, 6, 7], [6, 7, 8], [7, 8, 9], [8, 9, 10],
#       [9, 10, 11], [10, 11, 12], [11, 12, 13], [12, 13, 14], [13, 14, 15], [14, 15, 16], [15, 16, 17], [16, 17, 18], [17, 18, 19],
#       [18, 19, 20], [19, 20, 21], [20, 21, 22], [21, 22, 23], [22, 23, 24], [23, 24, 25], [24, 25, 26], [25, 26, 27], [26, 27, 28]]
#li_y: [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29]

#  Input data format ：seq_len, batch, input_size (27,1,3)
data_x = numpy.reshape(li_x, (len(li_x), 1, seq_length))

#  Normalize the input data 
data_x = torch.from_numpy(data_x / float(data_length)).float()

# scatter_ Function to convert onehot code 
#  Set the output data to one-hot code 
data_y = torch.zeros(len(li_y), data_length).scatter_(1, torch.tensor(li_y).unsqueeze_(dim=1), 1).float()


# -----------------------------------
#  Define the network model 
class net(nn.Module):
    #  Model structure ：LSTM +  Full connection  + Softmax
    def __init__(self, input_size, hidden_size, output_size, num_layer):
        super(net, self).__init__()
        # LSTM return output、hidden and cell
        # RNN return output、hidden
        self.layer1 = nn.LSTM(input_size, hidden_size, num_layer)
        self.layer2 = nn.Linear(hidden_size, output_size)
        self.layer3 = nn.Softmax()

    def forward(self, x):
        x, _ = self.layer1(x)
        #  Format ：[27, 1, 32], Represents the number of samples ,batch Size and hidden layer size 
        sample, batch, hidden = x.size()
        x = x.reshape(-1, hidden)
        #  Convert it into a two-dimensional matrix and calculate it with full connection 
        x = self.layer2(x)
        x = self.layer3(x)
        return x


model = net(seq_length, 32, data_length, 4)

# -----------------------------------
#  Define the loss function and optimizer 
loss_fun = nn.MSELoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.01)

# -----------------------------------
#  Training models 

#  Before training, you can first look at the difference between the predicted results of the initialized parameters 
# result = model(data_x)
# for target, pred in zip(data_y, result):
#     print("{} -> {}".format(target.argmax().data, pred.argmax().data))

#  Start training 1000 round 
for _ in range(1000):
    output = model(data_x)
    loss = loss_fun(data_y, output)
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

    if (_ + 1) % 50 == 0:
        print('Epoch: {}, Loss: {}'.format(_, loss.data))

# -----------------------------------
#  Predicted results 
result = model(data_x)
for target, pred in zip(data_y, result):
    print(" Correct result ：{}, forecast ：{}".format(target.argmax().data, pred.argmax().data))

precision = (result.argmax(dim=1).data==data_y.argmax(dim=1).data)


    #  result ：
    #  Correct result ：3, forecast ：3
    #  Correct result ：4, forecast ：4
    #  Correct result ：5, forecast ：5
    #  Correct result ：6, forecast ：6
    #  Correct result ：7, forecast ：7
    #  Correct result ：8, forecast ：8
    #  Correct result ：9, forecast ：9
    #  Correct result ：10, forecast ：10
    #  Correct result ：11, forecast ：11
    #  Correct result ：12, forecast ：12
    #  Correct result ：13, forecast ：13
    #  Correct result ：14, forecast ：14
    #  Correct result ：15, forecast ：15
    #  Correct result ：16, forecast ：16
    #  Correct result ：17, forecast ：21
    #  Correct result ：18, forecast ：18
    #  Correct result ：19, forecast ：27
    #  Correct result ：20, forecast ：21
    #  Correct result ：21, forecast ：21
    #  Correct result ：22, forecast ：21
    #  Correct result ：23, forecast ：21
    #  Correct result ：24, forecast ：24
    #  Correct result ：25, forecast ：25
    #  Correct result ：26, forecast ：26
    #  Correct result ：27, forecast ：27
    #  Correct result ：28, forecast ：28
    #  Correct result ：29, forecast ：29