forked from epochventures/PyTorch
-
Notifications
You must be signed in to change notification settings - Fork 0
/
XOR.py
65 lines (52 loc) · 1.73 KB
/
XOR.py
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
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torch.autograd import Variable
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.fc1 = nn.Linear(2, 50) # 2 Input noses, 50 in middle layers
self.fc2 = nn.Linear(50, 1) # 50 middle layer, 1 output nodes
self.rl1 = nn.ReLU()
self.rl2 = nn.ReLU()
def forward(self, x):
x = self.fc1(x)
x = self.rl1(x)
x = self.fc2(x)
x = self.rl2(x)
return x
if __name__ == "__main__":
## Create Network
net = Net()
#print net
## Optimization and Loss
#criterion = nn.CrossEntropyLoss() # use a Classification Cross-Entropy loss
criterion = nn.MSELoss()
#criterion = nn.L1Loss()
#criterion = nn.NLLLoss()
#criterion = nn.BCELoss()
optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.1)
#optimizer = optim.Adam(net.parameters(), lr=0.01)
trainingdataX = [[[0.01, 0.01], [0.01, 0.90], [0.90, 0.01], [0.95, 0.95]], [[0.02, 0.03], [0.04, 0.95], [0.97, 0.02], [0.96, 0.95]]]
trainingdataY = [[[0.01], [0.90], [0.90], [0.01]], [[0.04], [0.97], [0.98], [0.1]]]
NumEpoches = 20000
for epoch in range(NumEpoches):
running_loss = 0.0
for i, data in enumerate(trainingdataX, 0):
inputs = data
labels = trainingdataY[i]
inputs = Variable(torch.FloatTensor(inputs))
labels = Variable(torch.FloatTensor(labels))
optimizer.zero_grad()
outputs = net(inputs)
loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
running_loss += loss.data[0]
if i % 1000 == 0:
print "loss: ", running_loss
running_loss = 0.0
print "Finished training..."
print net(Variable(torch.FloatTensor(trainingdataX[0])))