Basic Manual RNN

In [1]:

    

import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
%matplotlib inline


    

In [2]:

    

# Number of inputs for each example
num_inputs = 2

# Number of neurons in first layer
num_neurons = 3


    

In [3]:

    

# We now need two Xs! One for each timestamp (t=0 and t=1)
x0 = tf.placeholder(tf.float32,[None, num_inputs])
x1 = tf.placeholder(tf.float32,[None, num_inputs])


    

In [4]:

    

# We'll also need a Weights variable for each x
# Notice the shape dimensions on both!
Wx = tf.Variable(tf.random_normal(shape = [num_inputs, num_neurons]))
Wy = tf.Variable(tf.random_normal(shape = [num_neurons, num_neurons]))


    

In [5]:

    

b = tf.Variable(tf.zeros([1, num_neurons]))


    

In [6]:

    

# First Activation
y0 = tf.tanh(tf.matmul(x0, Wx) + b)
y1 = tf.tanh(tf.matmul(y0, Wy) + tf.matmul(x1, Wx) + b)


    

In [7]:

    

init = tf.global_variables_initializer()


    

In [8]:

    

# BATCH 0:       example1 , example2, example 3
x0_batch = np.array([[0, 1],  [2, 3],    [4, 5]]) # DATA AT TIMESTAMP = 0

# BATCH 0:          example1 ,   example2,   example 3
x1_batch = np.array([[100, 101], [102, 103],  [104, 105]]) # DATA AT TIMESTAMP = 1


    

In [9]:

    

with tf.Session() as sess:
    sess.run(init)
    y0_output_vals , y1_output_vals  = sess.run([y0, y1],
                                                feed_dict = {x0 : x0_batch,
                                                             x1 : x1_batch})


    

In [10]:

    

# The output of values at t=0
y0_output_vals


    

    
Out[10]:





array([[ 0.02903818, -0.860023  ,  0.72572994],
       [-0.99575967, -0.997738  ,  0.8436332 ],
       [-0.9999915 , -0.9999659 ,  0.9133957 ]], dtype=float32)

In [11]:

    

# Output at t=1
y1_output_vals


    

    
Out[11]:





array([[-1., -1.,  1.],
       [-1., -1.,  1.],
       [-1., -1.,  1.]], dtype=float32)