PyTorch-Deep-Learning/pytorch-basic-functions.ipynb

{"cells":[{"cell_type":"markdown","metadata":{"id":"wGdTfyfrTEG0"},"source":["# PyTorch Basics\n","\n","References:\n","\n","https://jovian.com/learn/deep-learning-with-pytorch-zero-to-gans\n","\n","https://pytorch.org/docs/stable/tensors.html"]},{"cell_type":"markdown","metadata":{"id":"Gii0hJK4TEG3"},"source":["## Sytem setup"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"uC494CffTEG4"},"outputs":[],"source":["# Install pytorch\n","# !conda install pytorch cpuonly -c pytorch -y"]},{"cell_type":"code","execution_count":11,"metadata":{"id":"hXMCJplpTEG4","executionInfo":{"status":"ok","timestamp":1703184024078,"user_tz":300,"elapsed":304,"user":{"displayName":"Man Ho Wong","userId":"01009376394376826107"}}},"outputs":[],"source":["import torch\n","import numpy as np"]},{"cell_type":"markdown","metadata":{"id":"-sWGfix7TEG5"},"source":["## Tensors basics"]},{"cell_type":"code","execution_count":9,"metadata":{"id":"xFb92nhzTEG5","outputId":"36e70997-582b-4c6c-9994-d35f5aee784a","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1703183901839,"user_tz":300,"elapsed":8,"user":{"displayName":"Man Ho Wong","userId":"01009376394376826107"}}},"outputs":[{"output_type":"execute_result","data":{"text/plain":["tensor([[[11., 12.],\n","         [13., 13.]],\n","\n","        [[14., 15.],\n","         [15., 16.]],\n","\n","        [[17., 17.],\n","         [18., 19.]]])"]},"metadata":{},"execution_count":9}],"source":["# Create a tensor\n","t1 = torch.tensor(4.)                # single element\n","t2 = torch.tensor([1., 2, 3, 4])     # vector\n","t3 = torch.tensor([[[11, 12, 13],\n","                    [13, 14, 15]],\n","                   [[15, 16, 17],\n","                    [17, 18, 19.]]]) # 3D array\n","t4 = torch.ones((3,2))               # 3x2 matrix filled with ones\n","t5 = torch.full((3,2), 42)           # 3x2 matrix filled with 42\n","\n","# Check dtype\n","t1.dtype # Output: torch.float32\n","\n","# Check shape\n","t1.shape # Output: torch.Size([])\n","t3.shape # Output: torch.Size([2, 2, 3])\n","\n","# Other common methods\n","t6 = torch.cat((t4, t5))  # Concatenate two tensors (shapes must be compatible)\n","t7 = torch.sin(t2)        # Compute sin of each element\n","t8 = t3.reshape(3,2,2)    # Change the shape of a tensor\n","\n","t8"]},{"cell_type":"markdown","source":["## Tensor operations and gradients"],"metadata":{"id":"DH9maEK1s-XW"}},{"cell_type":"code","source":["# Create a linear equation, y = w * x + b\n","x = torch.tensor(3.)\n","w = torch.tensor(4., requires_grad=True)\n","b = torch.tensor(5., requires_grad=True)\n","y = w * x + b  # note: y = 3 * 4 + 5 = 17\n","\n","# Compute derivatives (gradients)\n","y.backward() # derivatives of y w.r.t. to the tensors with requires_grad=True will be computed automatically\n","\n","# The values of derivatives are stored in the .grad attribute of the corresponding tensors\n","print('dy/dx:', x.grad) # Return None because requires_grad was not set to True\n","print('dy/dw:', w.grad)\n","print('dy/db:', b.grad)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"xbWzTIEEm0D1","executionInfo":{"status":"ok","timestamp":1703183230762,"user_tz":300,"elapsed":276,"user":{"displayName":"Man Ho Wong","userId":"01009376394376826107"}},"outputId":"16ff55ff-4dba-4d64-d702-f17421128a44"},"execution_count":5,"outputs":[{"output_type":"stream","name":"stdout","text":["dy/dx: None\n","dy/dw: tensor(3.)\n","dy/db: tensor(1.)\n"]}]},{"cell_type":"markdown","source":["## Working with NumPy"],"metadata":{"id":"8tXgPOQFtCql"}},{"cell_type":"code","execution_count":12,"metadata":{"id":"c9J_m7YSTEHB","executionInfo":{"status":"ok","timestamp":1703184026594,"user_tz":300,"elapsed":9,"user":{"displayName":"Man Ho Wong","userId":"01009376394376826107"}}},"outputs":[],"source":["# Convert numpy array to tensor\n","x = np.array([[1, 2], [3, 4.]])\n","y = torch.from_numpy(x) # y will have the corresponding dtype as x (e.g. torch.float64 in this example)\n","\n","# Convert tensor to numpy array\n","z = y.numpy()"]}],"metadata":{"colab":{"provenance":[{"file_id":"1mVYYe_ODzcsCCLfGzQgGGn2TiYm78bQ3","timestamp":1703181919746}]},"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":"3.7.6"}},"nbformat":4,"nbformat_minor":0}