In [2]:
from matplotlib import pyplot as plt
import numpy as np
In [43]:
# Load stats:
"""
# Rows: [generations]
# Cols:
0: evaluations (one evaluation of one individual -- may be multiple rollouts)
1: median fitness
2: maximum fitness
3: top fitness (best fitness ever)
4: peak fitness (fitness earned with best single weight value by max individual)
5: median # of nodes in population
6: median # of connections in population]
"""
stats = np.loadtxt('log/sample_stats.out', delimiter=',')
runTime = np.loadtxt('log/sample_runTime.out', delimiter=',')
In [159]:
# Plot fitness progress:
plt.figure(figsize=(12,4))
plt.grid(linestyle='--', linewidth=0.1)
plt.plot(stats[:,1], label="Median Fitness")
plt.plot(stats[:,2], label="Max Fitness")
plt.plot(stats[:,3], label="Top Fitness")
plt.plot(stats[:,4], label="Peak Fitness")
plt.legend()
plt.grid(linestyle=':', linewidth=2)
plt.xlabel('Generations')
plt.ylabel('Fitness')
# Approximate time on 8 cores
genKey = np.array([32,128,256,384,640,896],dtype=int)
for i in range(len(genKey)):
gen = int(genKey[i]/8)-1
time = runTime[gen]/3600
plt.annotate(str(round(time,1))+' hours', xy=(gen*8,0), xytext=(gen*8,-200),
arrowprops=dict(facecolor='black', shrink=0.05), horizontalalignment='center')
plt.title('Sample Run of WANN Cart-Pole Swing-Up on 8 Core Laptop')
Out[159]: