cage.py

import sys
import numpy as np
from neural import *

heads = 18
legs = 50

if(len(sys.argv)==3):
    heads = int(sys.argv[1])
    legs = int(sys.argv[2])

n = network("square",learning_rate=0.3)
l = layer(2,2,"liner",bias_rate=0.0)
l2 = layer(2,1,"liner",bias_rate=0.0)

# n.add_layer(l)
n.add_layer(l2)
x1 = np.array([[1,2],[1,4]]).reshape(2,-1)
y1 = np.array([[heads],[legs]]).reshape(1,-1)
# x2 = np.array([[2,4]]).reshape(2,-1)
# y2 = np.array([legs]).reshape(1,-1)

for _ in range(2500):
    # loss1 = n.learn(x2,y2)
    loss2 = n.learn(x1,y1)
    loss = (loss2)
    n.set_learning_rate(n.learning_rate*0.8)
    if(loss<0.1):
        break
    # loss = n.learn(x,y)

print loss
print "{} heads with {} legs ".format(heads,legs)
print "there are {} chicken {} rabbits,maybe, we guess ".format (l2.weights[0,0],l2.weights[0,1])