NN-Dataset-loader-Example.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"NN-Dataset-loader-Example.ipynb","provenance":[],"authorship_tag":"ABX9TyNn/G4gWFWRRsVL54G9qC3r"},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","metadata":{"id":"Eh82F7CLDVwR"},"source":["# Required imports @author : Saurabh Pandey "]},{"cell_type":"code","metadata":{"id":"PiurZNeGC4rE"},"source":["import torch\n","import torch.nn as nn \n","from torch.optim import SGD\n","from torch.utils.data import Dataset, DataLoader"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"BM_W29wWDpkk"},"source":["# Neural Network Class "]},{"cell_type":"code","metadata":{"id":"9Mta91WfDo_C","executionInfo":{"status":"ok","timestamp":1635248037881,"user_tz":-120,"elapsed":273,"user":{"displayName":"Saurabh Pandey","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Gip9Ceu2nVlKfEZWYIPD7OOwWVzq5nUl40RCcezRg=s64","userId":"07168666420056430693"}}},"source":["class NNetwork(nn.Module):\n","  def __init__(self):\n","    super().__init__()\n","    self.input_to_hidden = nn.Linear(2,8)\n","    self.activation = nn.ReLU()\n","    self.hidden_to_output = nn.Linear(8,1)\n","  \n","  def forward(self, x):\n","    x = self.input_to_hidden(x)\n","    x = self.activation(x)\n","    x = self.hidden_to_output(x)\n","    return x"],"execution_count":6,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"B_-nxuoGDnBy"},"source":["#Dataset Class"]},{"cell_type":"code","metadata":{"id":"8WpV0pkqE2NU","executionInfo":{"status":"ok","timestamp":1635248040098,"user_tz":-120,"elapsed":198,"user":{"displayName":"Saurabh Pandey","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Gip9Ceu2nVlKfEZWYIPD7OOwWVzq5nUl40RCcezRg=s64","userId":"07168666420056430693"}}},"source":["class MyDataset(Dataset):\n","  def __init__(self, x, y):\n","    self.x = torch.tensor(x).float().to('cuda')\n","    self.y = torch.tensor(y).float().to('cuda')\n","  def __len__(self):\n","    return len(self.x)\n","  def __getitem__(self, ix):\n","    return self.x[ix], self.y[ix]"],"execution_count":7,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"P2_-5I-kF5lr"},"source":["# Training code"]},{"cell_type":"code","metadata":{"id":"MXyXlJ8BF-HF","executionInfo":{"status":"ok","timestamp":1635248087709,"user_tz":-120,"elapsed":406,"user":{"displayName":"Saurabh Pandey","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Gip9Ceu2nVlKfEZWYIPD7OOwWVzq5nUl40RCcezRg=s64","userId":"07168666420056430693"}}},"source":["x = [[1,2],[3,4],[5,6],[7,8]]\n","y = [[3],[7],[11],[15]]\n","dset = MyDataset(x, y)\n","dl = DataLoader(dset, batch_size=2, shuffle=True)\n","mynet = NNetwork().to('cuda')\n","loss_func = nn.MSELoss()\n","opt = SGD(mynet.parameters(), lr=0.001)\n","loss_history = []\n","for i in range(50):\n","  for data in dl:\n","    x, y = data \n","    opt.zero_grad()\n","    loss = loss_func(mynet(x), y)\n","    loss.backward()\n","    opt.step()\n","    loss_history.append(loss)"],"execution_count":11,"outputs":[]},{"cell_type":"code","metadata":{"id":"hHa1VQMiH65X","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1635248093390,"user_tz":-120,"elapsed":184,"user":{"displayName":"Saurabh Pandey","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Gip9Ceu2nVlKfEZWYIPD7OOwWVzq5nUl40RCcezRg=s64","userId":"07168666420056430693"}},"outputId":"1be5f518-5181-4458-9fe2-f7f508fb227d"},"source":["prediction = mynet(torch.tensor([[10,11]]).float().to('cuda'))\n","print(prediction)"],"execution_count":12,"outputs":[{"output_type":"stream","name":"stdout","text":["tensor([[20.4103]], device='cuda:0', grad_fn=<AddmmBackward>)\n"]}]}]}