[Sparse Tensors on Pytorch & Tensorflow] #Others #JupyterNotebook #CodeSnippet

SparseTensors.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sparse Tensors\n",
    "\n",
    "## Pytorch\n",
    "\n",
    "Docs: https://pytorch.org/docs/stable/sparse.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sparse matrix:\n",
      "tensor(indices=tensor([[0, 1, 0, 2, 2, 3],\n",
      "                       [0, 0, 1, 0, 1, 0]]),\n",
      "       values=tensor([1., 2., 3., 4., 5., 6.]),\n",
      "       size=(4, 2), nnz=6, layout=torch.sparse_coo)\n",
      "\n",
      "To Dense:\n",
      "tensor([[1., 3.],\n",
      "        [2., 0.],\n",
      "        [4., 5.],\n",
      "        [6., 0.]])\n"
     ]
    }
   ],
   "source": [
    "i = torch.LongTensor([[0, 1, 0, 2, 2, 3, ],\n",
    "                      [0, 0, 1, 0, 1, 0, ]]) # indices tensor\n",
    "\n",
    "v = torch.FloatTensor([1, 2, 3, 4, 5, 6]) # values values\n",
    "\n",
    "foo = torch.sparse.FloatTensor(i, v, torch.Size([4,2])) # Sparse tensor w/ values @ indices\n",
    "\n",
    "print('Sparse matrix:\\n{}\\n\\nTo Dense:\\n{}'.format(foo, foo.to_dense()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1., 1., 1., 1., 1.],\n",
       "        [1., 1., 1., 1., 1.]])"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bar = torch.ones(size=(2,5))\n",
    "bar"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[4., 4., 4., 4., 4.],\n",
       "        [2., 2., 2., 2., 2.],\n",
       "        [9., 9., 9., 9., 9.],\n",
       "        [6., 6., 6., 6., 6.]])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.sparse.mm(foo, bar)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tensorflow\n",
    "\n",
    "Docs: https://www.tensorflow.org/api_docs/python/tf/sparse/SparseTensor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "values = tf.Variable([1., 2., 3., 4., 5., 6.])\n",
    "foo = tf.sparse.SparseTensor(indices=[[0, 0], [1, 0], [0, 1], [2, 0], [2, 1], [3, 0]], values=values, dense_shape=[4, 2])\n",
    "foo_dense = tf.sparse.to_dense(foo, validate_indices=False)\n",
    "\n",
    "bar = tf.ones(shape=(2,5))\n",
    "\n",
    "mult = tf.sparse.matmul(foo, bar)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SparseTensorValue(indices=array([[0, 0],\n",
      "       [1, 0],\n",
      "       [0, 1],\n",
      "       [2, 0],\n",
      "       [2, 1],\n",
      "       [3, 0]], dtype=int64), values=array([1., 2., 3., 4., 5., 6.], dtype=float32), dense_shape=array([4, 2], dtype=int64))\n",
      "[[1. 1. 1. 1. 1.]\n",
      " [1. 1. 1. 1. 1.]]\n"
     ]
    }
   ],
   "source": [
    "with tf.Session() as sess:\n",
    "    tf.global_variables_initializer().run()\n",
    "    print(foo.eval())\n",
    "    print(bar.eval())\n",
    "    foo_dense_res, mult_res = sess.run([foo_dense, mult])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1., 3.],\n",
       "       [2., 0.],\n",
       "       [4., 5.],\n",
       "       [6., 0.]], dtype=float32)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "foo_dense_res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[4., 4., 4., 4., 4.],\n",
       "       [2., 2., 2., 2., 2.],\n",
       "       [9., 9., 9., 9., 9.],\n",
       "       [6., 6., 6., 6., 6.]], dtype=float32)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mult_res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}