Tensorflow Basics : Part 1

• tf.constant
• tf.ones
• tf.zeros
• tf.rank
• tf.random.normal
• tf.random.uniform
• tf.Variable
  • tf.convert_to_tensor
  • tf.argmax
  • tf.reshape
  • tf.Variable.assign
  • tf.Variable.assign_add
  • tf.Variable.assign_sub

import tensorflow as tf
print("Version , ",tf.__version__)

Version ,  2.1.0

Constant : tf.constant

tf.constant(
    value, dtype=None, shape=None, name='Const'
)

x1 = tf.constant([5, 2 ,1, 3])
print(x1)
print(x1.numpy)
print(x1.shape)
print(x1.dtype)

x2 = tf.constant([[5, 2], [1, 3]])
print(x2)

tf.Tensor([5 2 1 3], shape=(4,), dtype=int32)
<bound method _EagerTensorBase.numpy of <tf.Tensor: shape=(4,), dtype=int32, numpy=array([5, 2, 1, 3])>>
(4,)
<dtype: 'int32'>
tf.Tensor(
[[5 2]
 [1 3]], shape=(2, 2), dtype=int32)

x3 = tf.constant([5,2,1,3] , dtype=float)
print(x3)

tf.Tensor([5. 2. 1. 3.], shape=(4,), dtype=float32)

tf.ones

tf.ones(
    shape, dtype=tf.dtypes.float32, name=None
)

tf.zeros

tf.zeros(
    shape, dtype=tf.dtypes.float32, name=None
)

print(tf.ones(shape=(2, 2)))
print(tf.zeros(shape=(2, 2),dtype=tf.int16))

print(tf.ones((3, 4), tf.int32))
print(tf.zeros((3, 4), tf.int32))

tf.Tensor(
[[1. 1.]
 [1. 1.]], shape=(2, 2), dtype=float32)
tf.Tensor(
[[0 0]
 [0 0]], shape=(2, 2), dtype=int16)
tf.Tensor(
[[1 1 1 1]
 [1 1 1 1]
 [1 1 1 1]], shape=(3, 4), dtype=int32)
tf.Tensor(
[[0 0 0 0]
 [0 0 0 0]
 [0 0 0 0]], shape=(3, 4), dtype=int32)

rank : tf.rank

rank_0 = tf.Variable(7)
rank_1 = tf.Variable([7])
rank_2 = tf.Variable([[1, 2], [3, 4]])
rank_3 = tf.Variable([[[1, 2], [3, 4]], [[5, 6], [7, 8]]])
rank_4 = tf.reshape(tf.range(start=0, limit=24, delta=1), shape=[2, 3, 2, 2])

ranks = [rank_0, rank_1, rank_2, rank_3, rank_4]

for r in ranks:
    print("Rank is:", tf.rank(r).numpy())

Rank is: 0
Rank is: 1
Rank is: 2
Rank is: 3
Rank is: 4

Rank	Math entity	Example
0	Scalar (magnitude only)	[1.5]
1	Vector (magnitude and direction)	[2, 3, 5, 8 11]
2	Matrix (table of numbers)	[[1, 2], [3, 4]]
3	3-Tensor (cube of numbers)	[[[1], [2]], [[3], [4]]]
n	n-Tensor (you get the idea)	[[[1, 2], [3, 4], [5, 6]], [[7, 8], [9, 10], [11, 12]]]

Note : The rank of a tensor is not the same as the rank of a matrix. The rank of a tensor is the number of indices required to uniquely select each element of the tensor. Rank is also known as “order”, “degree”, or “ndims.”

x = tf.ones((1,3,2))
print(x)
rank = tf.rank(x).numpy()
print(rank)

tf.Tensor(
[[[1. 1.]
  [1. 1.]
  [1. 1.]]], shape=(1, 3, 2), dtype=float32)
3

normal distribution : tf.random.normal

tf.random.normal(
    shape, mean=0.0, stddev=1.0, dtype=tf.dtypes.float32, seed=None, name=None
)

from matplotlib import pyplot as plt

N = 100
tf.random.set_seed(12);
normal_dist = tf.random.normal(shape=(N,), mean=5.5, stddev=1.)
plt.plot(normal_dist)
plt.xlabel('Random number idx')
plt.ylabel('N random value')
plt.show()

uniform distribution : tf.random.uniform

tf.random.uniform(
    shape, minval=0, maxval=None, dtype=tf.dtypes.float32, seed=None, name=None
)

N = 100
tf.random.set_seed(12);
uniform_dist = tf.random.uniform(shape=(N,), minval=0, maxval=10)
plt.plot(uniform_dist)
plt.xlabel('Random number idx')
plt.ylabel('N random value')
plt.show()

Variable : tf.Variable

my_tensor = tf.constant([[1.0, 2.0], [3.0, 4.0]])
print(my_tensor)
my_variable = tf.Variable(my_tensor)
print(my_variable)

tf.Tensor(
[[1. 2.]
 [3. 4.]], shape=(2, 2), dtype=float32)
<tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy=
array([[1., 2.],
       [3., 4.]], dtype=float32)>

# Variables can be all kinds of types, just like tensors

bool_variable = tf.Variable([False, False, False, True])
complex_variable = tf.Variable([5 + 4j, 6 + 1j])

Most tensor operations work on variables as expected,
although variables cannot be reshaped

tf.convert_to_tensor

tf.argmax

tf.reshape

print("variable:", my_variable)
print("\nViewed as a tensor:", tf.convert_to_tensor(my_variable))
print("variable:", my_variable)
print("\nIndex of highest value:", tf.argmax(my_variable))

# This creates a new tensor; it does not reshape the variable.
print("\nCopying and reshaping: ", tf.reshape(my_variable, ([1,4])))

variable: <tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy=
array([[1., 2.],
       [3., 4.]], dtype=float32)>

Viewed as a tensor: tf.Tensor(
[[1. 2.]
 [3. 4.]], shape=(2, 2), dtype=float32)
variable: <tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy=
array([[1., 2.],
       [3., 4.]], dtype=float32)>

Index of highest value: tf.Tensor([1 1], shape=(2,), dtype=int64)

Copying and reshaping:  tf.Tensor([[1. 2. 3. 4.]], shape=(1, 4), dtype=float32)

can reassign the tensor using tf.Variable.assign Calling assign does not (usually) allocate a new tensor; instead, the existing tensor’s memory is reused.

new_value = tf.ones(shape=(2, 2))
print(my_variable)
my_variable.assign(new_value)
print('value of my_variable after assign new value:', my_variable)

<tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy=
array([[1., 2.],
       [3., 4.]], dtype=float32)>
value of my_variable after assign new value: <tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy=
array([[1., 1.],
       [1., 1.]], dtype=float32)>

assignment operations

tf.Variable.assign_add
tf.Variable.assign_sub

my_variable.assign_add(my_variable)
print(my_variable)

my_variable.assign_sub(my_variable*0.5)
print(my_variable)

<tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy=
array([[2., 2.],
       [2., 2.]], dtype=float32)>
<tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy=
array([[1., 1.],
       [1., 1.]], dtype=float32)>

some mathematical operations

a = tf.ones(shape=(2, 2))
b = tf.ones(shape=(2, 2))
print('a:', a.numpy().flatten())
print('b:', b.numpy().flatten())
c = a + b
print('c:', c.numpy().flatten())
d = tf.square(c)
print('square:', d.numpy().flatten())
e = tf.exp(d)
print('exp   :', e.numpy().flatten())

a: [1. 1. 1. 1.]
b: [1. 1. 1. 1.]
c: [2. 2. 2. 2.]
square: [4. 4. 4. 4.]
exp   : [54.598152 54.598152 54.598152 54.598152]