345 lecturas
345 lecturas

No hay TensorFlow sin Tensores

Demasiado Largo; Para Leer

Los tensores son arreglos multidimensionales en el núcleo de TensorFlow, lo que permite una representación y manipulación de datos eficientes.Esta guía cubre la creación de tensores, operaciones y conceptos avanzados como la transmisión y los tensores de ragged, proporcionando una comprensión integral para los practicantes del aprendizaje automático.
featured image - No hay TensorFlow sin Tensores
Tensor Flow - [Technical Documentation] HackerNoon profile picture
0-item

Contenido Overview

  • básicos
  • Sobre las formas
  • Índice
  • Manipulación de formas
  • Más sobre DTypes
  • Transmisión
  • tf.convert_to a tensor
  • Tensores Tensores
  • Tensores de string
  • Tensores de ahorro

Tensores son arreglos multidimensionales con un tipo uniforme (llamadodtypePuedes ver todos los apoyadosdtypesentf.dtypes.

Si usted está familiarizado conNúmeroLos tensores son comonp.arrays.

Todos los tensores son inmutables como números y cadenas de Python: nunca se puede actualizar el contenido de un tensor, solo se puede crear un nuevo.

import tensorflow as tf
import numpy as np

2024-08-15 03:05:18.327501: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2024-08-15 03:05:18.348450: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:8454] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2024-08-15 03:05:18.354825: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1452] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered

básicos

Primero, crear algunos tensores básicos.

Aquí está un tensor "escalar" o "rank-0".Un escalar contiene un único valor, y no hay "ejes".

# This will be an int32 tensor by default; see "dtypes" below.
rank_0_tensor = tf.constant(4)
print(rank_0_tensor)

tf.Tensor(4, shape=(), dtype=int32)
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
I0000 00:00:1723691120.932442  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.936343  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.940040  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.943264  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.954872  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.958376  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.961894  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.964843  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.967730  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.971300  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.974711  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691120.977717  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.208679  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.210786  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.212791  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.214776  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.216798  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.218734  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.220650  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.222554  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.224486  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.226429  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.228329  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.230251  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.269036  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.271069  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.273006  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.274956  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.276917  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.278854  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.280754  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.282664  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.284613  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.287058  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.289508  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1723691122.291891  176945 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355


Un tensor "vector" o "rank-1" es como una lista de valores.

# Let's make this a float tensor.
rank_1_tensor = tf.constant([2.0, 3.0, 4.0])
print(rank_1_tensor)
tf.Tensor([2. 3. 4.], shape=(3,), dtype=float32)


A "matrix" or "rank-2" tensor has two axes:

# If you want to be specific, you can set the dtype (see below) at creation time
rank_2_tensor = tf.constant([[1, 2],
                             [3, 4],
                             [5, 6]], dtype=tf.float16)
print(rank_2_tensor)
tf.Tensor(
[[1. 2.]
 [3. 4.]
 [5. 6.]], shape=(3, 2), dtype=float16)

A scalar, shape: []

A vector, shape: [3]

A matrix, shape: [3, 2]

A scalar, the number 4

The line with 3 sections, each one containing a number.

A 3x2 grid, with each cell containing a number.

A scalar, the number 4

The line with 3 sections, each one containing a number.

A 3x2 grid, with each cell containing a number.

Tensors may have more axes; here is a tensor with three axes:

# There can be an arbitrary number of
# axes (sometimes called "dimensions")
rank_3_tensor = tf.constant([
  [[0, 1, 2, 3, 4],
   [5, 6, 7, 8, 9]],
  [[10, 11, 12, 13, 14],
   [15, 16, 17, 18, 19]],
  [[20, 21, 22, 23, 24],
   [25, 26, 27, 28, 29]],])

print(rank_3_tensor)
tf.Tensor(
[[[ 0  1  2  3  4]
  [ 5  6  7  8  9]]

 [[10 11 12 13 14]
  [15 16 17 18 19]]

 [[20 21 22 23 24]
  [25 26 27 28 29]]], shape=(3, 2, 5), dtype=int32)

Hay muchas maneras de visualizar un tensor con más de dos ejes.

A 3-axis tensor, shape: [3, 2, 5]



Puede convertir un tensor a una matriz NumPy usandonp.arrayO eltensor.numpyEl método:

np.array(rank_2_tensor)
array([[1., 2.],
       [3., 4.],
       [5., 6.]], dtype=float16)
rank_2_tensor.numpy()
array([[1., 2.],
       [3., 4.],
       [5., 6.]], dtype=float16)

Los sensores a menudo contienen flotadores e incs, pero tienen muchos otros tipos, incluyendo:

  • Números complejos
  • Strings

La basetf.TensorLa clase requiere que los tensores sean "rectangulares"---es decir, a lo largo de cada eje, cada elemento es del mismo tamaño.

  • Tensores Tensores
  • Tensores de ahorro

Puede hacer matemáticas básicas en tensores, incluyendo la suma, la multiplicación en forma de elementos y la multiplicación de matriz.

a = tf.constant([[1, 2],
                 [3, 4]])
b = tf.constant([[1, 1],
                 [1, 1]]) # Could have also said `tf.ones([2,2], dtype=tf.int32)`

print(tf.add(a, b), "\n")
print(tf.multiply(a, b), "\n")
print(tf.matmul(a, b), "\n")
tf.Tensor(
[[2 3]
 [4 5]], shape=(2, 2), dtype=int32) 

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

tf.Tensor(
[[3 3]
 [7 7]], shape=(2, 2), dtype=int32)
print(a + b, "\n") # element-wise addition
print(a * b, "\n") # element-wise multiplication
print(a @ b, "\n") # matrix multiplication

tf.Tensor(
[[2 3]
 [4 5]], shape=(2, 2), dtype=int32) 

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

tf.Tensor(
[[3 3]
 [7 7]], shape=(2, 2), dtype=int32)

Los sensores se utilizan en todo tipo de operaciones (o "Ops").

c = tf.constant([[4.0, 5.0], [10.0, 1.0]])

# Find the largest value
print(tf.reduce_max(c))
# Find the index of the largest value
print(tf.math.argmax(c))
# Compute the softmax
print(tf.nn.softmax(c))
tf.Tensor(10.0, shape=(), dtype=float32)
tf.Tensor([1 0], shape=(2,), dtype=int64)
tf.Tensor(
[[2.6894143e-01 7.3105854e-01]
 [9.9987662e-01 1.2339458e-04]], shape=(2, 2), dtype=float32)

Nota: Por lo general, donde una función TensorFlow espera un Tensor como entrada, la función también aceptará cualquier cosa que pueda ser convertida a un Tensor usando tf.convert_to_tensor.

Nota: Por lo general, donde una función TensorFlow espera un Tensor como entrada, la función también aceptará cualquier cosa que pueda ser convertida a un Tensor usando tf.convert_to_tensor.

tf.convert_to_tensor([1,2,3])
<tf.Tensor: shape=(3,), dtype=int32, numpy=array([1, 2, 3], dtype=int32)>
tf.reduce_max([1,2,3])
<tf.Tensor: shape=(), dtype=int32, numpy=3>
tf.reduce_max(np.array([1,2,3]))
<tf.Tensor: shape=(), dtype=int64, numpy=3>


Sobre las formas

Tensores tienen formas. Algunos vocabulario:

  • Forma: La longitud (número de elementos) de cada uno de los ejes de un tensor.
  • Rango: Número de ejes tensores.Un escalar tiene rango 0, un vector tiene rango 1, una matriz es rango 2.
  • Eje o Dimensión: Una dimensión particular de un tensor.
  • Tamaño: El número total de elementos en el tensor, el producto de los elementos del vector de forma.

Nota: Aunque puede ver la referencia a un "tensor de dos dimensiones", un tensor de rango 2 no suele describir un espacio 2D.

Note:Aunque se puede ver la referencia a un "tensor de dos dimensiones", un tensor de rango 2 no suele describir un espacio 2D.

Tensores ytf.TensorShapeLos objetos tienen propiedades convenientes para acceder a estos:

rank_4_tensor = tf.zeros([3, 2, 4, 5])


A rank-4 tensor, shape: [3, 2, 4, 5]

A tensor shape is like a vector.

A 4-axis tensor

A tensor shape is like a vector.

A 4-axis tensor

print("Type of every element:", rank_4_tensor.dtype)
print("Number of axes:", rank_4_tensor.ndim)
print("Shape of tensor:", rank_4_tensor.shape)
print("Elements along axis 0 of tensor:", rank_4_tensor.shape[0])
print("Elements along the last axis of tensor:", rank_4_tensor.shape[-1])
print("Total number of elements (3*2*4*5): ", tf.size(rank_4_tensor).numpy())
Type of every element: <dtype: 'float32'>
Number of axes: 4
Shape of tensor: (3, 2, 4, 5)
Elements along axis 0 of tensor: 3
Elements along the last axis of tensor: 5
Total number of elements (3*2*4*5):  120

Sin embargo, tenga en cuenta que laTensor.ndimyTensor.shapeLos atributos no vuelvenTensorobjetos. si usted necesita unTensorUtilice eltf.rankotf.shapeEsta diferencia es sutil, pero puede ser importante cuando se construyen gráficos (más tarde).

tf.rank(rank_4_tensor)
<tf.Tensor: shape=(), dtype=int32, numpy=4>
tf.shape(rank_4_tensor)
<tf.Tensor: shape=(4,), dtype=int32, numpy=array([3, 2, 4, 5], dtype=int32)>

A menudo, los ejes se ordenan de global a local: el eje de lote primero, seguido por las dimensiones espaciales, y las características para cada ubicación por último.

Typical axis order

Keep track of what each axis is. A 4-axis tensor might be: Batch, Width, Height, Features

Keep track of what each axis is. A 4-axis tensor might be: Batch, Width, Height, Features


Índice

Índice de un solo eje

TensorFlow sigue las reglas de indexación estándar de Python, similares aCómo indexar una lista o cadena en Python, y las reglas básicas para la indexación NumPy.

  • Los índices comienzan en 0
  • los índices negativos cuentan hacia atrás desde el final
  • Las columnas, :, se utilizan para las flechas: start:stop:step


rank_1_tensor = tf.constant([0, 1, 1, 2, 3, 5, 8, 13, 21, 34])
print(rank_1_tensor.numpy())
[ 0  1  1  2  3  5  8 13 21 34]

La indexación con un escalar elimina el eje:

print("First:", rank_1_tensor[0].numpy())
print("Second:", rank_1_tensor[1].numpy())
print("Last:", rank_1_tensor[-1].numpy())
First: 0
Second: 1
Last: 34

Índice con a:El eje mantiene el eje:

print("Everything:", rank_1_tensor[:].numpy())
print("Before 4:", rank_1_tensor[:4].numpy())
print("From 4 to the end:", rank_1_tensor[4:].numpy())
print("From 2, before 7:", rank_1_tensor[2:7].numpy())
print("Every other item:", rank_1_tensor[::2].numpy())
print("Reversed:", rank_1_tensor[::-1].numpy())
Everything: [ 0  1  1  2  3  5  8 13 21 34]
Before 4: [0 1 1 2]
From 4 to the end: [ 3  5  8 13 21 34]
From 2, before 7: [1 2 3 5 8]
Every other item: [ 0  1  3  8 21]
Reversed: [34 21 13  8  5  3  2  1  1  0]

Indicaciones multi-axis

Los tensores de rango más alto se indexan pasando varios índices.

Las mismas reglas que en el caso de un solo eje se aplican a cada eje de forma independiente.

print(rank_2_tensor.numpy())
[[1. 2.]
 [3. 4.]
 [5. 6.]]

Pasando un número entero para cada índice, el resultado es un escalar.

# Pull out a single value from a 2-rank tensor
print(rank_2_tensor[1, 1].numpy())

You can index using any combination of integers and slices:

# Get row and column tensors
print("Second row:", rank_2_tensor[1, :].numpy())
print("Second column:", rank_2_tensor[:, 1].numpy())
print("Last row:", rank_2_tensor[-1, :].numpy())
print("First item in last column:", rank_2_tensor[0, -1].numpy())
print("Skip the first row:")
print(rank_2_tensor[1:, :].numpy(), "\n")
Second row: [3. 4.]
Second column: [2. 4. 6.]
Last row: [5. 6.]
First item in last column: 2.0
Skip the first row:
[[3. 4.]
 [5. 6.]]

Here is an example with a 3-axis tensor:

print(rank_3_tensor[:, :, 4])
tf.Tensor(
[[ 4  9]
 [14 19]
 [24 29]], shape=(3, 2), dtype=int32)


Selecting the last feature across all locations in each example in the batch

A 3x2x5 tensor with all the values at the index-4 of the last axis selected.

The selected values packed into a 2-axis tensor.

A 3x2x5 tensor with all the values at the index-4 of the last axis selected.

The selected values packed into a 2-axis tensor.

Leer elGuía de cortes de tensiónpara aprender cómo puede aplicar la indexación para manipular elementos individuales en sus tensores.

Manipulación de formas

Recuperar un tensor es de gran utilidad.

# Shape returns a `TensorShape` object that shows the size along each axis
x = tf.constant([[1], [2], [3]])
print(x.shape)
(3, 1)
# You can convert this object into a Python list, too
print(x.shape.as_list())
[3, 1]

Usted puede transformar un tensor en una nueva forma.tf.reshapeLa operación es rápida y barata ya que los datos subyacentes no necesitan ser duplicados.

# You can reshape a tensor to a new shape.
# Note that you're passing in a list
reshaped = tf.reshape(x, [1, 3])
print(x.shape)
print(reshaped.shape)
(3, 1)
(1, 3)

Los datos mantienen su diseño en la memoria y se crea un nuevo tensor, con la forma solicitada, apuntando a los mismos datos. TensorFlow utiliza el ordenamiento de la memoria "row-major" de estilo C, donde incrementar el índice más derecho corresponde a un solo paso en la memoria.

print(rank_3_tensor)
tf.Tensor(
[[[ 0  1  2  3  4]
  [ 5  6  7  8  9]]

 [[10 11 12 13 14]
  [15 16 17 18 19]]

 [[20 21 22 23 24]
  [25 26 27 28 29]]], shape=(3, 2, 5), dtype=int32)

Si planea un tensor, puede ver qué orden está colocado en la memoria.

# A `-1` passed in the `shape` argument says "Whatever fits".
print(tf.reshape(rank_3_tensor, [-1]))
tf.Tensor(
[ 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 24 25 26 27 28 29], shape=(30,), dtype=int32)

El único uso razonable detf.reshapees combinar o dividir los ejes adyacentes (o agregar / eliminar1de la s).

Para este tensor 3x2x5, la reforma a (3x2)x5 o 3x(2x5) son dos cosas razonables que hacer, ya que las flechas no se mezclan:

print(tf.reshape(rank_3_tensor, [3*2, 5]), "\n")
print(tf.reshape(rank_3_tensor, [3, -1]))
tf.Tensor(
[[ 0  1  2  3  4]
 [ 5  6  7  8  9]
 [10 11 12 13 14]
 [15 16 17 18 19]
 [20 21 22 23 24]
 [25 26 27 28 29]], shape=(6, 5), dtype=int32) 

tf.Tensor(
[[ 0  1  2  3  4  5  6  7  8  9]
 [10 11 12 13 14 15 16 17 18 19]
 [20 21 22 23 24 25 26 27 28 29]], shape=(3, 10), dtype=int32)



Some good reshapes.

A 3x2x5 tensor

The same data reshaped to (3x2)x5

The same data reshaped to 3x(2x5)

A 3x2x5 tensor

The same data reshaped to (3x2)x5

The same data reshaped to 3x(2x5)

La remodelación "trabajará" para cualquier nueva forma con el mismo número total de elementos, pero no hará nada útil si no respeta el orden de los ejes.

Cambiar los ejes entf.reshapeNo funciona, es necesariotf.transposePara eso.

# Bad examples: don't do this

# You can't reorder axes with reshape.
print(tf.reshape(rank_3_tensor, [2, 3, 5]), "\n") 

# This is a mess
print(tf.reshape(rank_3_tensor, [5, 6]), "\n")

# This doesn't work at all
try:
  tf.reshape(rank_3_tensor, [7, -1])
except Exception as e:
  print(f"{type(e).__name__}: {e}")


tf.Tensor(
[[[ 0  1  2  3  4]
  [ 5  6  7  8  9]
  [10 11 12 13 14]]

 [[15 16 17 18 19]
  [20 21 22 23 24]
  [25 26 27 28 29]]], shape=(2, 3, 5), dtype=int32) 

tf.Tensor(
[[ 0  1  2  3  4  5]
 [ 6  7  8  9 10 11]
 [12 13 14 15 16 17]
 [18 19 20 21 22 23]
 [24 25 26 27 28 29]], shape=(5, 6), dtype=int32) 

InvalidArgumentError: { {function_node __wrapped__Reshape_device_/job:localhost/replica:0/task:0/device:GPU:0} } Input to reshape is a tensor with 30 values, but the requested shape requires a multiple of 7 [Op:Reshape]



Some bad reshapes.

You can't reorder axes, use tf.transpose for that

Anything that mixes the slices of data together is probably wrong.

The new shape must fit exactly.

You can't reorder axes, use tf.transpose for that

Anything that mixes the slices of data together is probably wrong.

The new shape must fit exactly.

Usted puede correr a través de formas no especificadas por completo. O bien la forma contiene unaNone(una longitud de eje es desconocida) o la forma entera esNone(El rango del tensor es desconocido).

A excepción de tf.RaggedTensor, tales formas sólo ocurrirán en el contexto de las APIs simbólicas de construcción de gráficos de TensorFlow:

  • F. Funciones
  • La fuente de fuego funcional.


Más enTipos

Tipos

Para inspeccionar atf.TensorEl tipo de datos utiliza elTensor.dtypede propiedad.

Cuando se crea atf.TensorDesde un objeto de Python, puede especificar opcionalmente el tipo de datos.

Si no lo hace, TensorFlow elige un tipo de datos que puede representar sus datos.tf.int32Python y los números de puntos flotantestf.float32De lo contrario, TensorFlow utiliza las mismas reglas que NumPy utiliza cuando se convierten en matrizes.

Puede ser de tipo a tipo.

the_f64_tensor = tf.constant([2.2, 3.3, 4.4], dtype=tf.float64)
the_f16_tensor = tf.cast(the_f64_tensor, dtype=tf.float16)
# Now, cast to an uint8 and lose the decimal precision
the_u8_tensor = tf.cast(the_f16_tensor, dtype=tf.uint8)
print(the_u8_tensor)
tf.Tensor([2 3 4], shape=(3,), dtype=uint8)


Transmisión

La transmisión es un concepto tomado de laCaracterísticas equivalentes en NumPyEn resumen, bajo ciertas condiciones, los tensores más pequeños se "estiran" automáticamente para encajar los tensores más grandes cuando se ejecutan operaciones combinadas sobre ellos.

El caso más simple y más común es cuando se intenta multiplicar o agregar un tensor a un escalar.

x = tf.constant([1, 2, 3])

y = tf.constant(2)
z = tf.constant([2, 2, 2])
# All of these are the same computation
print(tf.multiply(x, 2))
print(x * y)
print(x * z)
tf.Tensor([2 4 6], shape=(3,), dtype=int32)
tf.Tensor([2 4 6], shape=(3,), dtype=int32)
tf.Tensor([2 4 6], shape=(3,), dtype=int32)

Asimismo, los ejes de longitud 1 se pueden estirar para coincidir con los otros argumentos.

En este caso, una matriz 3x1 es elementalmente multiplicada por una matriz 1x4 para producir una matriz 3x4.[4].

# These are the same computations
x = tf.reshape(x,[3,1])
y = tf.range(1, 5)
print(x, "\n")
print(y, "\n")
print(tf.multiply(x, y))
tf.Tensor(
[[1]
 [2]
 [3]], shape=(3, 1), dtype=int32) 

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

tf.Tensor(
[[ 1  2  3  4]
 [ 2  4  6  8]
 [ 3  6  9 12]], shape=(3, 4), dtype=int32)

A broadcasted add: a [3, 1] times a [1, 4] gives a [3,4]

Adding a 3x1 matrix to a 4x1 matrix results in a 3x4 matrix

Adding a 3x1 matrix to a 4x1 matrix results in a 3x4 matrix

Aquí está la misma operación sin transmisión:

x_stretch = tf.constant([[1, 1, 1, 1],
                         [2, 2, 2, 2],
                         [3, 3, 3, 3]])

y_stretch = tf.constant([[1, 2, 3, 4],
                         [1, 2, 3, 4],
                         [1, 2, 3, 4]])

print(x_stretch * y_stretch)  # Again, operator overloading
tf.Tensor(
[[ 1  2  3  4]
 [ 2  4  6  8]
 [ 3  6  9 12]], shape=(3, 4), dtype=int32)

La mayor parte del tiempo, la transmisión es eficiente tanto en el tiempo como en el espacio, ya que la operación de transmisión nunca materializa los tensores expandidos en la memoria.

Vea cómo se utiliza la transmisióntf.broadcast_to.

print(tf.broadcast_to(tf.constant([1, 2, 3]), [3, 3]))
tf.Tensor(
[[1 2 3]
 [1 2 3]
 [1 2 3]], shape=(3, 3), dtype=int32)

A diferencia de las matemáticas, por ejemplo,broadcast_toNo hace nada especial para salvar la memoria. Aquí, usted está materializando el tensor.

Puede hacerse aún más complicado.Esta secciónEl libro de Jake VanderPlasPython Manual de Ciencia de Datosmuestra más trucos de transmisión (de nuevo en NumPy).


tf.convert_to a tensor

La mayoría de las operaciones, comotf.matmulytf.reshapeArgumentos de clasetf.TensorSin embargo, notará en el caso anterior, los objetos de Python con la forma de tensores son aceptados.

La mayoría, pero no todos, de las llamadas de OSconvert_to_tensorExiste un registro de conversiones, y la mayoría de las clases de objetos como NumPyndarray, deTensorShapeLas listas de Python, ytf.VariableTodos se convertirán automáticamente.

estetf.register_tensor_conversion_functionpara más detalles, y si tiene su propio tipo que le gustaría convertir automáticamente a un tensor.


Tensores Tensores

Un tensor con números variables de elementos a lo largo de algún eje se llama "ragged".tf.ragged.RaggedTensorpor los datos racionalizados.

Por ejemplo, esto no puede ser representado como un tensor regular:

tf.RaggedTensor, shape: [4, None]

A 2-axis ragged tensor, each row can have a different length.

A 2-axis ragged tensor, each row can have a different length.

ragged_list = [
    [0, 1, 2, 3],
    [4, 5],
    [6, 7, 8],
    [9]]
try:
  tensor = tf.constant(ragged_list)
except Exception as e:
  print(f"{type(e).__name__}: {e}")
ValueError: Can't convert non-rectangular Python sequence to Tensor.

En lugar de crear atf.RaggedTensorUsotf.ragged.constant:

ragged_tensor = tf.ragged.constant(ragged_list)
print(ragged_tensor)
<tf.RaggedTensor [[0, 1, 2, 3], [4, 5], [6, 7, 8], [9]]>

La forma de unatf.RaggedTensorcontendrá algunos ejes con longitudes desconocidas:

print(ragged_tensor.shape)
(4, None)


Tensores de string

tf.stringes adtype, que es decir que puede representar datos como cadenas (arrayes de bytes de longitud variable) en tensores.

Las cadenas son atómicas y no pueden ser indexadas como las cadenas de Python. La longitud de la cadena no es uno de los ejes del tensor.tf.stringsfunciones para manipularlos.

Aquí está un tensor de cuerdas escalares:

# Tensors can be strings, too here is a scalar string.
scalar_string_tensor = tf.constant("Gray wolf")
print(scalar_string_tensor)
tf.Tensor(b'Gray wolf', shape=(), dtype=string)

Un vector de cuerdas:

A vector of strings, shape: [3,]

The string length is not one of the tensor's axes.

The string length is not one of the tensor's axes.


# If you have three string tensors of different lengths, this is OK.
tensor_of_strings = tf.constant(["Gray wolf",
                                 "Quick brown fox",
                                 "Lazy dog"])
# Note that the shape is (3,). The string length is not included.
print(tensor_of_strings)
tf.Tensor([b'Gray wolf' b'Quick brown fox' b'Lazy dog'], shape=(3,), dtype=string)

En el artículo anterior se imprime elbEl prefijo indica quetf.stringdtype no es una cadena de código unicode, sino una cadena de byte.Unicode Tutorialpara más información sobre cómo trabajar con texto unicode en TensorFlow.

Si pasas los caracteres de unicode, son UTF-8 codificados.

tf.constant("🥳👍")
<tf.Tensor: shape=(), dtype=string, numpy=b'\xf0\x9f\xa5\xb3\xf0\x9f\x91\x8d'>

Algunas funciones básicas con cuerdas se pueden encontrar entf.stringsIncluyendotf.strings.split.

# You can use split to split a string into a set of tensors
print(tf.strings.split(scalar_string_tensor, sep=" "))
tf.Tensor([b'Gray' b'wolf'], shape=(2,), dtype=string)
# ...but it turns into a `RaggedTensor` if you split up a tensor of strings,
# as each string might be split into a different number of parts.
print(tf.strings.split(tensor_of_strings))
<tf.RaggedTensor [[b'Gray', b'wolf'], [b'Quick', b'brown', b'fox'], [b'Lazy', b'dog']]>

Three strings split, shape: [3, None]

Splitting multiple strings returns a tf.RaggedTensor

Splitting multiple strings returns a tf.RaggedTensor

ytf.strings.to_number:

text = tf.constant("1 10 100")
print(tf.strings.to_number(tf.strings.split(text, " ")))
tf.Tensor([  1.  10. 100.], shape=(3,), dtype=float32)

Aunque no se puede utilizartf.castPara convertir un tensor de cuerdas en números, puede convertirlo en bytes y luego en números.

byte_strings = tf.strings.bytes_split(tf.constant("Duck"))
byte_ints = tf.io.decode_raw(tf.constant("Duck"), tf.uint8)
print("Byte strings:", byte_strings)
print("Bytes:", byte_ints)
Byte strings: tf.Tensor([b'D' b'u' b'c' b'k'], shape=(4,), dtype=string)
Bytes: tf.Tensor([ 68 117  99 107], shape=(4,), dtype=uint8)
# Or split it up as unicode and then decode it
unicode_bytes = tf.constant("アヒル 🦆")
unicode_char_bytes = tf.strings.unicode_split(unicode_bytes, "UTF-8")
unicode_values = tf.strings.unicode_decode(unicode_bytes, "UTF-8")

print("\nUnicode bytes:", unicode_bytes)
print("\nUnicode chars:", unicode_char_bytes)
print("\nUnicode values:", unicode_values)
Unicode bytes: tf.Tensor(b'\xe3\x82\xa2\xe3\x83\x92\xe3\x83\xab \xf0\x9f\xa6\x86', shape=(), dtype=string)

Unicode chars: tf.Tensor([b'\xe3\x82\xa2' b'\xe3\x83\x92' b'\xe3\x83\xab' b' ' b'\xf0\x9f\xa6\x86'], shape=(5,), dtype=string)

Unicode values: tf.Tensor([ 12450  12498  12523     32 129414], shape=(5,), dtype=int32)

Eltf.stringdtype se utiliza para todos los datos de bytes crudos en TensorFlow.tf.ioEl módulo contiene funciones para la conversión de datos a y desde bytes, incluyendo la decodificación de imágenes y el análisis de csv.

Tensores de ahorro

A veces, sus datos son escasos, como un espacio de embalaje muy amplio.tf.sparse.SparseTensory operaciones relacionadas para almacenar datos escasos de manera eficiente.

tf.SparseTensor, shape: [3, 4]

An 3x4 grid, with values in only two of the cells.

An 3x4 grid, with values in only two of the cells.

# Sparse tensors store values by index in a memory-efficient manner
sparse_tensor = tf.sparse.SparseTensor(indices=[[0, 0], [1, 2]],
                                       values=[1, 2],
                                       dense_shape=[3, 4])
print(sparse_tensor, "\n")

# You can convert sparse tensors to dense
print(tf.sparse.to_dense(sparse_tensor))
SparseTensor(indices=tf.Tensor(
[[0 0]
 [1 2]], shape=(2, 2), dtype=int64), values=tf.Tensor([1 2], shape=(2,), dtype=int32), dense_shape=tf.Tensor([3 4], shape=(2,), dtype=int64)) 

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

Originalmente publicado en el sitio web de TensorFlow, este artículo aparece aquí bajo un nuevo título y está licenciado bajo CC BY 4.0.

Originally published on the TensorFlow website,este artículo aparece aquí bajo un nuevo título y está licenciado bajoCC BY 4.0.Las muestras de código compartidas bajo laApache 2.0 License.

Sitio web de TensorFlow


Trending Topics

blockchaincryptocurrencyhackernoon-top-storyprogrammingsoftware-developmenttechnologystartuphackernoon-booksBitcoinbooks