Forward pass of a Simple RNN

The Softmax function is also known as the normalized exponential function. It normalizes the input into a probability distribution that sums to 1.

raise SystemExit("Stop right there!");

An exception has occurred, use %tb to see the full traceback.

SystemExit: Stop right there!

Importing libraries and packages

# System
import os

# Mathematical operations and data manipulation
import numpy as np

# Modelling
import tensorflow as tf

from IPython.display import display, HTML

display(HTML("<style>.container {width:80% !important;}</style>"))

os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2"

Forward pass

np.random.seed(0)
tf.random.set_seed(0)

# Number of inputs (2) and the number of neurons in the hidden layer (3)
num_inputs = 2
num_neurons = 3

# Feedforward weights and recurrent weights variables for the weight matrices
wf = tf.Variable(tf.random.normal(shape=[num_inputs, num_neurons]))
wr = tf.Variable(tf.random.normal(shape=[num_neurons, num_neurons]))

# Bias variable (as many values as the number of neurons in the hidden layer)
bias = tf.Variable(tf.zeros([1, num_neurons]))

# Creating data
xt0_batch = np.array([[0, 1], [2, 3], [4, 5]]).astype(np.float32)
xt1_batch = np.array([[100, 101], [102, 103], [104, 105]]).astype(np.float32)

def forward_pass(xt0, xt1):
    yt0 = tf.tanh(tf.matmul(xt0, wf) + bias)
    yt1 = tf.tanh(tf.matmul(yt0, wr) + tf.matmul(xt1, wf) + bias)
    return yt0, yt1


# Forward pass
yt0_output, yt1_output = forward_pass(xt0_batch, xt1_batch)

# Printing the output values yt0 and yt1
tf.print(yt0_output)
tf.print(yt1_output)

[[-0.776318431 -0.844548464 0.438419849]
 [-0.0857750699 -0.993522227 0.516408086]
 [0.6983459 -0.999749422 0.586677969]]
[[1 -1 0.99999851]
 [1 -1 0.999998331]
 [1 -1 0.999997377]]