Pytorch Numpy Array Observations

Pytorch Numpy Array Observations.ipynb

{
    "nbformat": 4,
    "nbformat_minor": 0,
    "metadata": {
        "colab": {
            "name": "Pytorch Numpy Array Observations.ipynb",
            "provenance": [],
            "authorship_tag": "ABX9TyOn+Xm2YMMuyKnkTydwQFpr"
        },
        "kernelspec": {
            "name": "python3",
            "display_name": "Python 3"
        }
    },
    "cells": [
        {
            "cell_type": "markdown",
            "metadata": {
                "id": "view-in-github",
                "colab_type": "text"
            },
            "source": [
                "<a href=\"https://colab.research.google.com/gist/kalmufti/b5609b6930eb99befeba663b06a9aa89/pytorch-numpy-array-observations.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
            ]
        },
        {
            "cell_type": "markdown",
            "metadata": {
                "id": "Q6mbBhmUxzBY",
                "colab_type": "text"
            },
            "source": [
                "Observation/differences between Numpy and Pytorch arrays and data types."
            ]
        },
        {
            "cell_type": "code",
            "metadata": {
                "id": "5EvwykMEq0g5",
                "colab_type": "code",
                "colab": {}
            },
            "source": [
                "import torch\n",
                "import numpy as np"
            ],
            "execution_count": 0,
            "outputs": []
        },
        {
            "cell_type": "code",
            "metadata": {
                "id": "Vb1IS_OdrJy3",
                "colab_type": "code",
                "colab": {}
            },
            "source": [
                "n = np.zeros(3)\n",
                "t = torch.zeros(3)"
            ],
            "execution_count": 0,
            "outputs": []
        },
        {
            "cell_type": "markdown",
            "metadata": {
                "id": "NHQBBouZslEI",
                "colab_type": "text"
            },
            "source": [
                "Numpy array is a Python like list of type numpy.ndarrays (number of dimensions arrays). While Pytorch array is of type tensor that can be used on a GPU. Both are multidimensional homogeneous data type arrays."
            ]
        },
        {
            "cell_type": "code",
            "metadata": {
                "id": "SWs1XLbnr1YP",
                "colab_type": "code",
                "colab": {
                    "base_uri": "https://localhost:8080/",
                    "height": 85
                },
                "outputId": "9d4db3ea-a90d-4d33-d891-bbf1a60f1c71",
                "executionInfo": {
                    "status": "ok",
                    "timestamp": 1580849138140,
                    "user_tz": 360,
                    "elapsed": 624,
                    "user": {
                        "displayName": "Khalid Almufti",
                        "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDKIcxSgguIJDPHq2fm_5baTlkB8BZ3_JqIjgAi=s64",
                        "userId": "04536491061392437351"
                    }
                }
            },
            "source": [
                "print(n)\n",
                "print(t)\n",
                "print(type(n))\n",
                "print(type(t))"
            ],
            "execution_count": 3,
            "outputs": [
                {
                    "output_type": "stream",
                    "text": [
                        "[0. 0. 0.]\n",
                        "tensor([0., 0., 0.])\n",
                        "<class 'numpy.ndarray'>\n",
                        "<class 'torch.Tensor'>\n"
                    ],
                    "name": "stdout"
                }
            ]
        },
        {
            "cell_type": "markdown",
            "metadata": {
                "id": "z3QIh2nxuGfd",
                "colab_type": "text"
            },
            "source": [
                "Numpy shape attribute return a tuple, while Pytorch return the same, shape in Pytorch is an alias for .size() see next."
            ]
        },
        {
            "cell_type": "code",
            "metadata": {
                "id": "j1gHf8vVrOTL",
                "colab_type": "code",
                "colab": {
                    "base_uri": "https://localhost:8080/",
                    "height": 85
                },
                "outputId": "859c979b-75b4-4823-b2a3-e2e06f2320f4",
                "executionInfo": {
                    "status": "ok",
                    "timestamp": 1580849138141,
                    "user_tz": 360,
                    "elapsed": 617,
                    "user": {
                        "displayName": "Khalid Almufti",
                        "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDKIcxSgguIJDPHq2fm_5baTlkB8BZ3_JqIjgAi=s64",
                        "userId": "04536491061392437351"
                    }
                }
            },
            "source": [
                "print(n.shape)\n",
                "print(t.shape)\n",
                "print(type(n.shape))\n",
                "print(type(t.shape))"
            ],
            "execution_count": 4,
            "outputs": [
                {
                    "output_type": "stream",
                    "text": [
                        "(3,)\n",
                        "torch.Size([3])\n",
                        "<class 'tuple'>\n",
                        "<class 'torch.Size'>\n"
                    ],
                    "name": "stdout"
                }
            ]
        },
        {
            "cell_type": "markdown",
            "metadata": {
                "id": "kG1syYKr0MU4",
                "colab_type": "text"
            },
            "source": [
                "Numpy size is an attributes which return an integer. While Pytorch is a method that returns a tuple."
            ]
        },
        {
            "cell_type": "code",
            "metadata": {
                "id": "iTbDXonmzR5U",
                "colab_type": "code",
                "colab": {
                    "base_uri": "https://localhost:8080/",
                    "height": 68
                },
                "outputId": "a0a3796b-ce2b-4616-d9d3-4b1955d68a65",
                "executionInfo": {
                    "status": "ok",
                    "timestamp": 1580849138142,
                    "user_tz": 360,
                    "elapsed": 611,
                    "user": {
                        "displayName": "Khalid Almufti",
                        "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDKIcxSgguIJDPHq2fm_5baTlkB8BZ3_JqIjgAi=s64",
                        "userId": "04536491061392437351"
                    }
                }
            },
            "source": [
                "print(n.size)\n",
                "print(t.size)\n",
                "print(t.size())"
            ],
            "execution_count": 5,
            "outputs": [
                {
                    "output_type": "stream",
                    "text": [
                        "3\n",
                        "<built-in method size of Tensor object at 0x7f22ff86daf8>\n",
                        "torch.Size([3])\n"
                    ],
                    "name": "stdout"
                }
            ]
        },
        {
            "cell_type": "code",
            "metadata": {
                "id": "1DYDa-oj0JVA",
                "colab_type": "code",
                "colab": {
                    "base_uri": "https://localhost:8080/",
                    "height": 255
                },
                "outputId": "411c8086-823e-4483-a6cb-127d1bc1011c",
                "executionInfo": {
                    "status": "ok",
                    "timestamp": 1580849138142,
                    "user_tz": 360,
                    "elapsed": 603,
                    "user": {
                        "displayName": "Khalid Almufti",
                        "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDKIcxSgguIJDPHq2fm_5baTlkB8BZ3_JqIjgAi=s64",
                        "userId": "04536491061392437351"
                    }
                }
            },
            "source": [
                "print(help(t.size))"
            ],
            "execution_count": 6,
            "outputs": [
                {
                    "output_type": "stream",
                    "text": [
                        "Help on built-in function size:\n",
                        "\n",
                        "size(...) method of torch.Tensor instance\n",
                        "    size() -> torch.Size\n",
                        "    \n",
                        "    Returns the size of the :attr:`self` tensor. The returned value is a subclass of\n",
                        "    :class:`tuple`.\n",
                        "    \n",
                        "    Example::\n",
                        "    \n",
                        "        >>> torch.empty(3, 4, 5).size()\n",
                        "        torch.Size([3, 4, 5])\n",
                        "\n",
                        "None\n"
                    ],
                    "name": "stdout"
                }
            ]
        },
        {
            "cell_type": "markdown",
            "metadata": {
                "id": "wWWUwiMqwkxG",
                "colab_type": "text"
            },
            "source": [
                " Numpy data type defaults to float64. While torch tensor is float32 data type."
            ]
        },
        {
            "cell_type": "code",
            "metadata": {
                "id": "opzq2tziwHiL",
                "colab_type": "code",
                "colab": {
                    "base_uri": "https://localhost:8080/",
                    "height": 85
                },
                "outputId": "6504efc2-0095-4042-9c16-1dc340ded79b",
                "executionInfo": {
                    "status": "ok",
                    "timestamp": 1580849138143,
                    "user_tz": 360,
                    "elapsed": 594,
                    "user": {
                        "displayName": "Khalid Almufti",
                        "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDKIcxSgguIJDPHq2fm_5baTlkB8BZ3_JqIjgAi=s64",
                        "userId": "04536491061392437351"
                    }
                }
            },
            "source": [
                "print(type(n[0]))\n",
                "print(type(t[0]))\n",
                "\n",
                "print(n.dtype)\n",
                "print(t.dtype)"
            ],
            "execution_count": 7,
            "outputs": [
                {
                    "output_type": "stream",
                    "text": [
                        "<class 'numpy.float64'>\n",
                        "<class 'torch.Tensor'>\n",
                        "float64\n",
                        "torch.float32\n"
                    ],
                    "name": "stdout"
                }
            ]
        }
    ]
}