PyTorch_33_Weights_Initialization.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "PyTorch_33_Weights_Initialization.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyPdNcSsk4zGN2u050i2c15I",
      "include_colab_link": true
    },
    "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/github/codeplinth/pytorchbootcamp/blob/master/PyTorch_33_Weights_Initialization.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e-rJ_8t3p3WO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch\n",
        "from torch import nn,optim\n",
        "from torch.utils.data import Dataset,DataLoader\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O0NsPvMesI3c",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "fbb8aadf-5ccd-4609-895d-40abc278dd27"
      },
      "source": [
        "torch.manual_seed(0)"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<torch._C.Generator at 0x7fd5a2497198>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FamhxfdTsL7P",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#make some data\n",
        "X = torch.arange(-20,20,1).view(-1,1).type(torch.FloatTensor)\n",
        "Y = torch.zeros(X.shape[0])\n",
        "Y[(X[:, 0] > -4) & (X[:, 0] < 4)] = 1.0"
      ],
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "34wKqp5YvqdM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plotData(X,Y,model,epoch,legend = True):\n",
        "  plt.plot(X.numpy(),model(X).detach().numpy(),label = ('Epoch' + str(epoch)))\n",
        "  plt.plot(X.numpy(),Y.numpy(),'r')\n",
        "  plt.xlabel('x')\n",
        "  if legend == True:\n",
        "    plt.legend()\n",
        "  else:\n",
        "    pass"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R6CPZ3bTtBD5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class NeuralNet(nn.Module):\n",
        "  def __init__(self,D_in,H,D_out):\n",
        "    super(NeuralNet,self).__init__()\n",
        "    self.l1 = nn.Linear(D_in,H)\n",
        "    self.l2 = nn.Linear(H,D_out)\n",
        "  \n",
        "  def forward(self,x):\n",
        "    out1 = torch.sigmoid(self.l1(x))\n",
        "    y_pred = torch.sigmoid(self.l2(out1))\n",
        "    return y_pred"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "geoi7eDOyF2K",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def train_model(X,Y,model,criterion,optimzer,EPOCHS=1000):\n",
        "  cost = []\n",
        "  for epoch in range(EPOCHS):\n",
        "    total = 0\n",
        "    for x,y in zip(X,Y):\n",
        "      y_pred = model(x)\n",
        "      loss = criterion(y_pred,y)\n",
        "      optimizer.zero_grad()\n",
        "      loss.backward()\n",
        "      optimizer.step()\n",
        "      total+=loss.item()\n",
        "    cost.append(total)\n",
        "    if epoch % 300 == 0:\n",
        "      plotData(X,Y,model,epoch,legend = True)\n",
        "      plt.show()\n",
        "  return cost"
      ],
      "execution_count": 24,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rJuvN17nz1-Y",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = NeuralNet(1,2,1)"
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C9d2ath10isq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 104
        },
        "outputId": "d01400eb-ca9b-4363-b033-67f88f15095d"
      },
      "source": [
        "model.state_dict()"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "OrderedDict([('l1.weight', tensor([[-0.0075],\n",
              "                      [ 0.5364]])),\n",
              "             ('l1.bias', tensor([-0.8230, -0.7359])),\n",
              "             ('l2.weight', tensor([[-0.2723,  0.1896]])),\n",
              "             ('l2.bias', tensor([-0.0140]))])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yfl4JY5A0k5q",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.state_dict()['l1.weight'][0] = 1.0\n",
        "model.state_dict()['l1.weight'][1] = 1.0\n",
        "model.state_dict()['l1.bias'][0] = 0.0\n",
        "model.state_dict()['l1.bias'][1] = 0.0\n",
        "model.state_dict()['l2.weight'][0] = 1.0\n",
        "model.state_dict()['l2.bias'][0] = 0.0"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Iq46_2-K1i40",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 104
        },
        "outputId": "32c043c0-ccdd-4575-e356-30d0600a227f"
      },
      "source": [
        "model.state_dict()"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "OrderedDict([('l1.weight', tensor([[1.],\n",
              "                      [1.]])),\n",
              "             ('l1.bias', tensor([0., 0.])),\n",
              "             ('l2.weight', tensor([[1., 1.]])),\n",
              "             ('l2.bias', tensor([0.]))])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EEZNjoVg10X5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "e5e9357c-0268-4c07-a564-c4e1e1b2f0cf"
      },
      "source": [
        "criterion = nn.BCELoss()\n",
        "optimizer = optim.SGD(model.parameters(),lr = 0.01)\n",
        "cost = train_model(X,Y,model,criterion,optimizer)\n",
        "plt.plot(cost)\n",
        "plt.xlabel('Epochs')\n",
        "plt.show()"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fKGGA1932dEV",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 104
        },
        "outputId": "d43aa7eb-92af-469a-cb20-1ec6ef631834"
      },
      "source": [
        "model.state_dict()"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "OrderedDict([('l1.weight', tensor([[0.9841],\n",
              "                      [0.9841]])),\n",
              "             ('l1.bias', tensor([0.3722, 0.3722])),\n",
              "             ('l2.weight', tensor([[0.0997, 0.0997]])),\n",
              "             ('l2.bias', tensor([-1.6672]))])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tMiGbuyh2qfr",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 69
        },
        "outputId": "dfe30f00-7d84-46c1-a7ea-ceb9596434d8"
      },
      "source": [
        "y_pred=model(torch.tensor([[-2.0],[0.0],[2.0]]))\n",
        "y_pred"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor([[0.1633],\n",
              "        [0.1752],\n",
              "        [0.1846]], grad_fn=<SigmoidBackward>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PTGPGud82wlG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 31,
      "outputs": []
    }
  ]
}