Deep_Learning_Practice_labs/lab4/plot.py at main · zhiyichin/Deep_Learning_Practice_labs · GitHub

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
64
import matplotlib.pyplot as plt
import pandas as pd


plt.rcParams["font.family"]="serif"
plt.figure()

res_list=['BLEU4-score','CrossEntropy','Gaussian-Score','KLD_weight','KLD','Teacher ratio']

# Plot KLD
filename='results/cyclic/run-train_KLD-tag-train.csv'
data=pd.read_csv(filename)
x=data['Step']
val=data['Value']
plt.plot(x,val,label='KLD')

# Plot CrossEntropy
filename='results/cyclic/run-train_CrossEntropy-tag-train.csv'
data=pd.read_csv(filename)
x=data['Step']
val=data['Value']
plt.plot(x,val,label='CrossEntropy')

plt.xlabel('500 iteration(s)')
plt.ylabel('Loss')
plt.title('Training Loss Curve')
plt.legend()
plt.savefig(f'results/cyclic/cyclic_loss.jpg')

plt.figure()

# Plot bleu
filename='results/cyclic/run-train_BLEU4-score-tag-train.csv'
data=pd.read_csv(filename)
x=data['Step']
val=data['Value']
plt.plot(x,val,'o',label='BLEU4-score',markersize=1)

# Plot KLD weight
filename='results/cyclic/run-train_KLD_weight-tag-train.csv'
data=pd.read_csv(filename)
x=data['Step']
val=data['Value']
plt.plot(x,val,'--',label='KLD_weight')

# Plot tf
filename='results/cyclic/run-train_tf-tag-train.csv'
data=pd.read_csv(filename)
x=data['Step']
val=data['Value']
plt.plot(x,val,'--',label='Teacher ratio')

# Plot Gaussian score
filename='results/cyclic/run-train_Gaussian-Score-tag-train.csv'
data=pd.read_csv(filename)
x=data['Step']
val=data['Value']
plt.plot(x,val,'o',label='Gaussian-Score',markersize=1)

plt.xlabel("500 iteration(s)")
plt.ylabel('score/weight')
plt.title('Training Ratio Curve')
plt.legend()
plt.savefig('results/cyclic/cyclic_ratio.jpg')