.. note:: :class: sphx-glr-download-link-note Click :ref:`here ` to download the full example code .. rst-class:: sphx-glr-example-title .. _sphx_glr_beginner_basics_buildmodel_tutorial.py: `Learn the Basics `_ || `Quickstart `_ || `Tensors `_ || `Datasets & DataLoaders `_ || `Transforms `_ || **Build Model** || `Autograd `_ || `Optimization `_ || `Save & Load Model `_ Build the Neural Network =================== Neural networks comprise of layers/modules that perform operations on data. The `torch.nn `_ namespace provides all the building blocks you need to build your own neural network. Every module in PyTorch subclasses the `nn.Module `_. A neural network is a module itself that consists of other modules (layers). This nested structure allows for building and managing complex architectures easily. In the following sections, we'll build a neural network to classify images in the FashionMNIST dataset. .. code-block:: default import os import torch from torch import nn from torch.utils.data import DataLoader from torchvision import datasets, transforms Get Device for Training ----------------------- We want to be able to train our model on a hardware accelerator like the GPU, if it is available. Let's check to see if `torch.cuda `_ is available, else we continue to use the CPU. .. code-block:: default device = "cuda" if torch.cuda.is_available() else "cpu" print(f"Using {device} device") .. rst-class:: sphx-glr-script-out Out: .. code-block:: none Using cuda device Define the Class ------------------------- We define our neural network by subclassing ``nn.Module``, and initialize the neural network layers in ``__init__``. Every ``nn.Module`` subclass implements the operations on input data in the ``forward`` method. .. code-block:: default class NeuralNetwork(nn.Module): def __init__(self): super(NeuralNetwork, self).__init__() self.flatten = nn.Flatten() self.linear_relu_stack = nn.Sequential( nn.Linear(28*28, 512), nn.ReLU(), nn.Linear(512, 512), nn.ReLU(), nn.Linear(512, 10), ) def forward(self, x): x = self.flatten(x) logits = self.linear_relu_stack(x) return logits We create an instance of ``NeuralNetwork``, and move it to the ``device``, and print its structure. .. code-block:: default model = NeuralNetwork().to(device) print(model) .. rst-class:: sphx-glr-script-out Out: .. code-block:: none NeuralNetwork( (flatten): Flatten(start_dim=1, end_dim=-1) (linear_relu_stack): Sequential( (0): Linear(in_features=784, out_features=512, bias=True) (1): ReLU() (2): Linear(in_features=512, out_features=512, bias=True) (3): ReLU() (4): Linear(in_features=512, out_features=10, bias=True) ) ) To use the model, we pass it the input data. This executes the model's ``forward``, along with some `background operations `_. Do not call ``model.forward()`` directly! Calling the model on the input returns a 10-dimensional tensor with raw predicted values for each class. We get the prediction probabilities by passing it through an instance of the ``nn.Softmax`` module. .. code-block:: default X = torch.rand(1, 28, 28, device=device) logits = model(X) pred_probab = nn.Softmax(dim=1)(logits) y_pred = pred_probab.argmax(1) print(f"Predicted class: {y_pred}") .. rst-class:: sphx-glr-script-out Out: .. code-block:: none Predicted class: tensor([8], device='cuda:0') -------------- Model Layers ------------------------- Let's break down the layers in the FashionMNIST model. To illustrate it, we will take a sample minibatch of 3 images of size 28x28 and see what happens to it as we pass it through the network. .. code-block:: default input_image = torch.rand(3,28,28) print(input_image.size()) .. rst-class:: sphx-glr-script-out Out: .. code-block:: none torch.Size([3, 28, 28]) nn.Flatten ^^^^^^^^^^^^^^^^^^^^^^ We initialize the `nn.Flatten `_ layer to convert each 2D 28x28 image into a contiguous array of 784 pixel values ( the minibatch dimension (at dim=0) is maintained). .. code-block:: default flatten = nn.Flatten() flat_image = flatten(input_image) print(flat_image.size()) .. rst-class:: sphx-glr-script-out Out: .. code-block:: none torch.Size([3, 784]) nn.Linear ^^^^^^^^^^^^^^^^^^^^^^ The `linear layer `_ is a module that applies a linear transformation on the input using its stored weights and biases. .. code-block:: default layer1 = nn.Linear(in_features=28*28, out_features=20) hidden1 = layer1(flat_image) print(hidden1.size()) .. rst-class:: sphx-glr-script-out Out: .. code-block:: none torch.Size([3, 20]) nn.ReLU ^^^^^^^^^^^^^^^^^^^^^^ Non-linear activations are what create the complex mappings between the model's inputs and outputs. They are applied after linear transformations to introduce *nonlinearity*, helping neural networks learn a wide variety of phenomena. In this model, we use `nn.ReLU `_ between our linear layers, but there's other activations to introduce non-linearity in your model. .. code-block:: default print(f"Before ReLU: {hidden1}\n\n") hidden1 = nn.ReLU()(hidden1) print(f"After ReLU: {hidden1}") .. rst-class:: sphx-glr-script-out Out: .. code-block:: none Before ReLU: tensor([[ 0.3980, 0.1355, -0.2899, 0.0650, 0.1646, -0.1050, -0.3243, -0.2629, 0.0117, -0.4639, -0.3012, -0.3025, 0.3991, 0.2786, -0.1655, 0.0047, -0.4105, 0.3910, -0.3500, -0.0958], [ 0.2369, -0.1007, -0.3273, 0.4325, -0.0558, 0.2747, -0.4837, -0.0559, -0.1406, -0.2160, -0.3125, -0.3584, -0.2178, 0.2848, -0.0265, 0.3975, -0.3476, -0.1200, -0.2461, -0.2075], [ 0.2284, -0.2248, 0.0620, 0.1601, 0.4152, 0.3929, -0.3526, 0.1674, 0.0494, 0.0103, -0.6525, -0.4557, 0.0912, 0.3204, -0.2013, 0.0638, -0.0363, 0.0150, -0.0437, -0.1595]], grad_fn=) After ReLU: tensor([[0.3980, 0.1355, 0.0000, 0.0650, 0.1646, 0.0000, 0.0000, 0.0000, 0.0117, 0.0000, 0.0000, 0.0000, 0.3991, 0.2786, 0.0000, 0.0047, 0.0000, 0.3910, 0.0000, 0.0000], [0.2369, 0.0000, 0.0000, 0.4325, 0.0000, 0.2747, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.2848, 0.0000, 0.3975, 0.0000, 0.0000, 0.0000, 0.0000], [0.2284, 0.0000, 0.0620, 0.1601, 0.4152, 0.3929, 0.0000, 0.1674, 0.0494, 0.0103, 0.0000, 0.0000, 0.0912, 0.3204, 0.0000, 0.0638, 0.0000, 0.0150, 0.0000, 0.0000]], grad_fn=) nn.Sequential ^^^^^^^^^^^^^^^^^^^^^^ `nn.Sequential `_ is an ordered container of modules. The data is passed through all the modules in the same order as defined. You can use sequential containers to put together a quick network like ``seq_modules``. .. code-block:: default seq_modules = nn.Sequential( flatten, layer1, nn.ReLU(), nn.Linear(20, 10) ) input_image = torch.rand(3,28,28) logits = seq_modules(input_image) nn.Softmax ^^^^^^^^^^^^^^^^^^^^^^ The last linear layer of the neural network returns `logits` - raw values in [-\infty, \infty] - which are passed to the `nn.Softmax `_ module. The logits are scaled to values [0, 1] representing the model's predicted probabilities for each class. ``dim`` parameter indicates the dimension along which the values must sum to 1. .. code-block:: default softmax = nn.Softmax(dim=1) pred_probab = softmax(logits) Model Parameters ------------------------- Many layers inside a neural network are *parameterized*, i.e. have associated weights and biases that are optimized during training. Subclassing ``nn.Module`` automatically tracks all fields defined inside your model object, and makes all parameters accessible using your model's ``parameters()`` or ``named_parameters()`` methods. In this example, we iterate over each parameter, and print its size and a preview of its values. .. code-block:: default print(f"Model structure: {model}\n\n") for name, param in model.named_parameters(): print(f"Layer: {name} | Size: {param.size()} | Values : {param[:2]} \n") .. rst-class:: sphx-glr-script-out Out: .. code-block:: none Model structure: NeuralNetwork( (flatten): Flatten(start_dim=1, end_dim=-1) (linear_relu_stack): Sequential( (0): Linear(in_features=784, out_features=512, bias=True) (1): ReLU() (2): Linear(in_features=512, out_features=512, bias=True) (3): ReLU() (4): Linear(in_features=512, out_features=10, bias=True) ) ) Layer: linear_relu_stack.0.weight | Size: torch.Size([512, 784]) | Values : tensor([[-0.0142, 0.0180, -0.0141, ..., -0.0041, 0.0017, -0.0218], [ 0.0197, 0.0216, 0.0001, ..., -0.0165, -0.0150, -0.0318]], device='cuda:0', grad_fn=) Layer: linear_relu_stack.0.bias | Size: torch.Size([512]) | Values : tensor([ 0.0173, -0.0246], device='cuda:0', grad_fn=) Layer: linear_relu_stack.2.weight | Size: torch.Size([512, 512]) | Values : tensor([[-0.0036, -0.0268, 0.0280, ..., 0.0238, 0.0004, 0.0399], [-0.0287, -0.0391, -0.0414, ..., 0.0267, -0.0159, -0.0017]], device='cuda:0', grad_fn=) Layer: linear_relu_stack.2.bias | Size: torch.Size([512]) | Values : tensor([-0.0299, 0.0030], device='cuda:0', grad_fn=) Layer: linear_relu_stack.4.weight | Size: torch.Size([10, 512]) | Values : tensor([[ 0.0351, 0.0246, 0.0390, ..., -0.0363, 0.0398, -0.0236], [ 0.0317, -0.0276, 0.0116, ..., 0.0070, 0.0116, 0.0197]], device='cuda:0', grad_fn=) Layer: linear_relu_stack.4.bias | Size: torch.Size([10]) | Values : tensor([0.0266, 0.0033], device='cuda:0', grad_fn=) -------------- Further Reading -------------- - `torch.nn API `_ .. rst-class:: sphx-glr-timing **Total running time of the script:** ( 0 minutes 0.080 seconds) .. _sphx_glr_download_beginner_basics_buildmodel_tutorial.py: .. only :: html .. container:: sphx-glr-footer :class: sphx-glr-footer-example .. container:: sphx-glr-download :download:`Download Python source code: buildmodel_tutorial.py ` .. container:: sphx-glr-download :download:`Download Jupyter notebook: buildmodel_tutorial.ipynb ` .. only:: html .. rst-class:: sphx-glr-signature `Gallery generated by Sphinx-Gallery `_