Testing ConstantPadNd backward

test_backward.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch.autograd import Variable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Context:\n",
    "    pad = (1, 1, -1, -1)\n",
    "    value = 0\n",
    "    input_size = torch.Size([4, 5])\n",
    "    l_inp = 2\n",
    "    pad_tup = ((-1, -1), (1, 1))\n",
    "    l_pad = 2\n",
    "    l_diff = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[ 0.5822, -0.7529,  0.4968, -0.2158, -1.3026,  1.1133,  0.6431],\n",
       "        [-0.5976, -0.2949,  0.4010, -0.9164,  0.4081, -0.8147,  0.0278]])"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ctx = Context()\n",
    "grad_output = torch.randn(2, 7)\n",
    "grad_output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "def backward(ctx, grad_output):\n",
    "    grad_input = Variable(grad_output.data.new(ctx.input_size).zero_())\n",
    "\n",
    "    cg_input = grad_input\n",
    "    for i, p in zip(range(ctx.l_inp)[-ctx.l_pad:], ctx.pad_tup):\n",
    "        if p[0] < 0:\n",
    "            cg_input = cg_input.narrow(i, -p[0], cg_input.size(i) + p[0])\n",
    "        if p[1] < 0:\n",
    "            cg_input = cg_input.narrow(i, 0, cg_input.size(i) + p[1])\n",
    "\n",
    "    cg_output = grad_output\n",
    "    for i, p in zip(range(ctx.l_inp)[-ctx.l_pad:], ctx.pad_tup):\n",
    "        if p[0] > 0:\n",
    "            cg_output = cg_output.narrow(i, p[0], cg_output.size(i) - p[0])\n",
    "        if p[1] > 0:\n",
    "            cg_output = cg_output.narrow(i, 0, cg_output.size(i) - p[1])\n",
    "    cg_input.copy_(cg_output)\n",
    "    return grad_input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[ 0.0000,  0.0000,  0.0000,  0.0000,  0.0000],\n",
       "        [-0.7529,  0.4968, -0.2158, -1.3026,  1.1133],\n",
       "        [-0.2949,  0.4010, -0.9164,  0.4081, -0.8147],\n",
       "        [ 0.0000,  0.0000,  0.0000,  0.0000,  0.0000]])"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "backward(ctx, grad_output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "def backward_return_output(ctx, grad_output):\n",
    "    grad_input = Variable(grad_output.data.new(ctx.input_size).zero_())\n",
    "\n",
    "    cg_input = grad_input\n",
    "    for i, p in zip(range(ctx.l_inp)[-ctx.l_pad:], ctx.pad_tup):\n",
    "        if p[0] < 0:\n",
    "            cg_input = cg_input.narrow(i, -p[0], cg_input.size(i) + p[0])\n",
    "        if p[1] < 0:\n",
    "            cg_input = cg_input.narrow(i, 0, cg_input.size(i) + p[1])\n",
    "\n",
    "    cg_output = grad_output\n",
    "    for i, p in zip(range(ctx.l_inp)[-ctx.l_pad:], ctx.pad_tup):\n",
    "        if p[0] > 0:\n",
    "            cg_output = cg_output.narrow(i, p[0], cg_output.size(i) - p[0])\n",
    "        if p[1] > 0:\n",
    "            cg_output = cg_output.narrow(i, 0, cg_output.size(i) - p[1])\n",
    "    return cg_output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[-0.7529,  0.4968, -0.2158, -1.3026,  1.1133],\n",
       "        [-0.2949,  0.4010, -0.9164,  0.4081, -0.8147]])"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "backward_return_output(ctx, grad_output)"
   ]
  }
 ],
 "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}