restruture old ch02 into appendix A

appendix-A/01_optional-python-setup-preferences/README.md

@@ -0,0 +1,78 @@
+# Python Setup Tips
+
+
+
+There are several different ways you can install Python and set up your computing environment. Here, I am illustrating my personal preference. 
+
+(I am using computers running macOS, but this workflow is similar for Linux machines and may work for other operating systems as well.)
+
+
+
+## 1. Download and install Miniforge
+
+Download miniforge from the GitHub repository [here](https://github.com/conda-forge/miniforge).
+
+<img src="figures/download.png" alt="download" style="zoom:33%;" />
+
+Depending on your operating system, this should download either an `.sh` (macOS, Linux) or `.exe` file (Windows). 
+
+For the `.sh` file, open your command line terminal and execute the following command
+
+```bash
+sh ~/Desktop/Miniforge3-MacOSX-arm64.sh
+```
+
+where `Desktop/` is the folder where the Miniforge installer was downloaded to. On your computer, you may have to replace it with `Downloads/`.
+
+<img src="figures/miniforge-install.png" alt="miniforge-install" style="zoom:33%;" />
+
+Next, step through the download instructions, confirming with "Enter".
+
+## 2. Create a new virtual environment
+
+After the installation was successfully completed, I recommend creating a new virtual environment called `dl-fundamentals`, which you can do by executing
+
+```bash
+conda create -n LLMs python=3.10
+```
+
+<img src="figures/new-env.png" alt="new-env" style="zoom:33%;" />
+
+Next, activate your new virtual environment (you have to do it every time you open a new terminal window or tab):
+
+```bash
+conda activate dl-workshop
+```
+
+<img src="figures/activate-env.png" alt="activate-env" style="zoom:33%;" />
+
+## Optional: styling your terminal
+
+If you want to style your terminal similar to mine so that you can see which virtual environment is active,  check out the [Oh My Zsh](https://github.com/ohmyzsh/ohmyzsh) project.
+
+
+
+# 3. Install new Python libraries
+
+
+
+To install new Python libraries, you can now use the `conda` package installer. For example, you can install [JupyterLab](https://jupyter.org/install) and [watermark](https://github.com/rasbt/watermark) as follows:
+
+```bash
+conda install jupyterlab watermark
+```
+
+<img src="figures/conda-install.png" alt="conda-install" style="zoom:33%;" />
+
+
+
+You can also still use `pip` to install libraries. By default, `pip` should be linked to your new `LLms` conda environment:
+
+<img src="figures/check-pip.png" alt="check-pip" style="zoom:33%;" />
+
+---
+
+
+
+
+Any questions? Please feel free to reach out in the [Discussion Forum](https://github.com/rasbt/LLMs-from-scratch/discussions).

appendix-A/02_installing-python-libraries/README.md

@@ -0,0 +1,46 @@
+# Libraries Used In This Workshop
+
+We will be using the following libraries in this workshop, and I highly recommend installing them before attending the event:
+
+- numpy >= 1.24.3 (The fundamental package for scientific computing with Python)
+- scipy >= 1.10.1 (Additional functions for NumPy)
+- pandas >= 2.0.2 (A data frame library)
+- matplotlib >= 3.7.1 (A plotting library)
+- jupyterlab >= 4.0 (An application for running Jupyter notebooks)
+- ipywidgets >= 8.0.6 (Fixes progress bar issues in Jupyter Lab)
+- scikit-learn >= 1.2.2 (A general machine learning library)
+- watermark >= 2.4.2 (An IPython/Jupyter extension for printing package information)
+- torch >= 2.0.1 (The PyTorch deep learning library)
+- torchvision >= 0.15.2 (PyTorch utilities for computer vision)
+- torchmetrics >= 0.11.4 (Metrics for PyTorch)
+- transformers >= 4.30.2 (Language transformers and LLMs for PyTorch)
+- lightning >= 2.0.3 (A library for advanced PyTorch features: multi-GPU, mixed-precision etc.)
+
+To install these requirements most conveniently, you can use the `requirements.txt` file:
+
+```
+pip install -r requirements.txt
+```
+
+![install-requirements](figures/install-requirements.png)
+
+Then, after completing the installation, please check if all the packages are installed and are up to date using
+
+```
+python_environment_check.py
+```
+
+![check_1](figures/check_1.png)
+
+It's also recommended to check the versions in JupyterLab by running the `jupyter_environment_check.ipynb` in this directory. Ideally, it should look like as follows:
+
+![check_1](figures/check_2.png)
+
+If you see the following issues, it's likely that your JupyterLab instance is connected to wrong conda environment:
+
+![jupyter-issues](figures/jupyter-issues.png)
+
+
+In this case, you may want to use `watermark` to check if you opened the JupyterLab instance in the right conda environment using the `--conda` flag:
+
+![watermark](figures/watermark.png)

appendix-A/02_installing-python-libraries/jupyter_environment_check.ipynb

@@ -0,0 +1,123 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "18d54544-92d0-412c-8e28-f9083b2bab6f",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[OK] Your Python version is 3.10.12\n"
+     ]
+    }
+   ],
+   "source": [
+    "from python_environment_check import check_packages, get_requirements_dict"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "60e03297-4337-4181-b8eb-f483f406954a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "d = get_requirements_dict()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "d982ddf9-c167-4ed2-9fce-e271f2b1e1de",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[OK] numpy 1.25.1\n",
+      "[OK] scipy 1.11.1\n",
+      "[OK] pandas 2.0.3\n",
+      "[OK] matplotlib 3.7.2\n",
+      "[OK] jupyterlab 4.0.3\n",
+      "[OK] ipywidgets 8.0.7\n",
+      "[OK] watermark 2.4.3\n",
+      "[OK] torch 2.0.1\n"
+     ]
+    }
+   ],
+   "source": [
+    "check_packages(d)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e0bdd547-333c-42a9-92f3-4e552f206cf3",
+   "metadata": {},
+   "source": [
+    "Same checks as above but using watermark:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "9d696044-9272-4b96-8305-34602807bb94",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%load_ext watermark"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "ce321731-a15a-4579-b33b-035730371eb3",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "numpy     : 1.25.1\n",
+      "scipy     : 1.11.1\n",
+      "pandas    : 2.0.3\n",
+      "matplotlib: 3.7.2\n",
+      "sklearn   : 1.3.0\n",
+      "watermark : 2.4.3\n",
+      "torch     : 2.0.1\n",
+      "\n",
+      "conda environment: LLMs\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "%watermark --conda -p numpy,scipy,pandas,matplotlib,sklearn,watermark,torch"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.10.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

appendix-A/02_installing-python-libraries/python_environment_check.py

@@ -0,0 +1,62 @@
+# Sebastian Raschka, 2023
+
+from os.path import dirname, join, realpath
+from packaging.version import parse as version_parse
+import platform
+import sys
+
+if version_parse(platform.python_version()) < version_parse('3.9'):
+    print('[FAIL] We recommend Python 3.9 or newer but'
+          ' found version %s' % (sys.version))
+else:
+    print('[OK] Your Python version is %s' % (platform.python_version()))
+
+
+def get_packages(pkgs):
+    versions = []
+    for p in pkgs:
+        try:
+            imported = __import__(p)
+            try:
+                versions.append(imported.__version__)
+            except AttributeError:
+                try:
+                    versions.append(imported.version)
+                except AttributeError:
+                    try:
+                        versions.append(imported.version_info)
+                    except AttributeError:
+                        versions.append('0.0')
+        except ImportError:
+            print(f'[FAIL]: {p} is not installed and/or cannot be imported.')
+            versions.append('N/A')
+    return versions
+
+
+def get_requirements_dict():
+    PROJECT_ROOT = dirname(realpath(__file__))
+    REQUIREMENTS_FILE = join(PROJECT_ROOT, "requirements.txt")
+    d = {}
+    with open(REQUIREMENTS_FILE) as f:
+        for line in f:
+            line = line.split(" ")
+            d[line[0]] = line[-1]
+    return d
+
+
+def check_packages(d):
+    versions = get_packages(d.keys())
+
+    for (pkg_name, suggested_ver), actual_ver in zip(d.items(), versions):
+        if actual_ver == 'N/A':
+            continue
+        actual_ver, suggested_ver = version_parse(actual_ver), version_parse(suggested_ver)
+        if actual_ver < suggested_ver:
+            print(f'[FAIL] {pkg_name} {actual_ver}, please upgrade to >= {suggested_ver}')
+        else:
+            print(f'[OK] {pkg_name} {actual_ver}')
+
+
+if __name__ == '__main__':
+    d = get_requirements_dict()
+    check_packages(d)

appendix-A/02_installing-python-libraries/requirements.txt

@@ -0,0 +1,8 @@
+numpy >= 1.24.3
+scipy >= 1.10.1
+pandas >= 2.0.2
+matplotlib >= 3.7.1
+jupyterlab >= 4.0
+ipywidgets >= 8.0.6
+watermark >= 2.4.2
+torch >= 2.0.1

appendix-A/03_main-chapter-code/DDP-script.py

@@ -0,0 +1,178 @@
+# Appendix A: Introduction to PyTorch (Part 3)
+
+import torch
+import torch.nn.functional as F
+from torch.utils.data import Dataset, DataLoader
+
+# NEW imports:
+import os
+import torch.multiprocessing as mp
+from torch.utils.data.distributed import DistributedSampler
+from torch.nn.parallel import DistributedDataParallel as DDP
+from torch.distributed import init_process_group, destroy_process_group
+
+
+# NEW: function to initialize a distributed process group (1 process / GPU)
+# this allows communication among processes
+def ddp_setup(rank, world_size):
+    """
+    Arguments:
+        rank: a unique process ID
+        world_size: total number of processes in the group
+    """
+    # rank of machine running rank:0 process
+    # here, we assume all GPUs are on the same machine
+    os.environ["MASTER_ADDR"] = "localhost"
+    # any free port on the machine
+    os.environ["MASTER_PORT"] = "12345"
+
+    # initialize process group
+    # Windows users may have to use "gloo" instead of "nccl" as backend
+    # nccl: NVIDIA Collective Communication Library
+    init_process_group(backend="nccl", rank=rank, world_size=world_size)
+    torch.cuda.set_device(rank)
+
+
+class ToyDataset(Dataset):
+    def __init__(self, X, y):
+        self.features = X
+        self.labels = y
+
+    def __getitem__(self, index):
+        one_x = self.features[index]
+        one_y = self.labels[index]
+        return one_x, one_y
+
+    def __len__(self):
+        return self.labels.shape[0]
+
+
+class NeuralNetwork(torch.nn.Module):
+    def __init__(self, num_inputs, num_outputs):
+        super().__init__()
+
+        self.layers = torch.nn.Sequential(
+            # 1st hidden layer
+            torch.nn.Linear(num_inputs, 30),
+            torch.nn.ReLU(),
+
+            # 2nd hidden layer
+            torch.nn.Linear(30, 20),
+            torch.nn.ReLU(),
+
+            # output layer
+            torch.nn.Linear(20, num_outputs),
+        )
+
+    def forward(self, x):
+        logits = self.layers(x)
+        return logits
+
+
+def prepare_dataset():
+    X_train = torch.tensor([
+        [-1.2, 3.1],
+        [-0.9, 2.9],
+        [-0.5, 2.6],
+        [2.3, -1.1],
+        [2.7, -1.5]
+    ])
+    y_train = torch.tensor([0, 0, 0, 1, 1])
+
+    X_test = torch.tensor([
+        [-0.8, 2.8],
+        [2.6, -1.6],
+    ])
+    y_test = torch.tensor([0, 1])
+
+    train_ds = ToyDataset(X_train, y_train)
+    test_ds = ToyDataset(X_test, y_test)
+
+    train_loader = DataLoader(
+        dataset=train_ds,
+        batch_size=2,
+        shuffle=False, # NEW: False because of DistributedSampler below
+        pin_memory=True,
+        drop_last=True,
+        # NEW: chunk batches across GPUs without overlapping samples:
+        sampler=DistributedSampler(train_ds) # NEW
+    )
+    test_loader = DataLoader(
+        dataset=test_ds,
+        batch_size=2,
+        shuffle=False,
+    )
+    return train_loader, test_loader
+
+
+# NEW: wrapper
+def main(rank, world_size, num_epochs):
+
+    ddp_setup(rank, world_size) # NEW: initialize process groups
+
+    train_loader, test_loader = prepare_dataset()
+    model = NeuralNetwork(num_inputs=2, num_outputs=2)
+    model.to(rank)
+    optimizer = torch.optim.SGD(model.parameters(), lr=0.5)
+
+    model = DDP(model, device_ids=[rank]) # NEW: wrap model with DDP
+    # the core model is now accessible as model.module
+    
+    for epoch in range(num_epochs):
+    
+        model.train()
+        for features, labels in enumerate(train_loader):
+    
+            features, labels = features.to(rank), labels.to(rank) # New: use rank
+            logits = model(features)
+            loss = F.cross_entropy(logits, labels) # Loss function
+    
+            optimizer.zero_grad()
+            loss.backward()
+            optimizer.step()
+    
+            ### LOGGING
+            print(f"[GPU{rank}] Epoch: {epoch+1:03d}/{num_epochs:03d}"
+                  f" | Batchsize {labels.shape[0]:03d}"
+                  f" | Train/Val Loss: {loss:.2f}")
+    
+    model.eval()
+    train_acc = compute_accuracy(model, train_loader, device=rank)
+    print(f"[GPU{rank}] Training accuracy", train_acc)
+    test_acc = compute_accuracy(model, test_loader, device=rank)
+    print(f"[GPU{rank}] Test accuracy", test_acc)
+
+    destroy_process_group() # NEW: cleanly exit distributed mode
+
+
+def compute_accuracy(model, dataloader, device):
+    model = model.eval()
+    correct = 0.0
+    total_examples = 0
+
+    for idx, (features, labels) in enumerate(dataloader):
+        features, labels = features.to(device), labels.to(device)
+
+        with torch.no_grad():
+            logits = model(features)
+        predictions = torch.argmax(logits, dim=1)
+        compare = labels == predictions
+        correct += torch.sum(compare)
+        total_examples += len(compare)
+    return (correct / total_examples).item()
+
+
+if __name__ == "__main__":
+    print("PyTorch version:", torch.__version__)
+    print("CUDA available:", torch.cuda.is_available())
+    print("Number of GPUs available:", torch.cuda.device_count())
+
+    torch.manual_seed(123)
+
+    # NEW: spawn new processes
+    # note that spawn will automatically pass the rank
+    num_epochs = 3
+    world_size = torch.cuda.device_count()
+    mp.spawn(main, args=(world_size, num_epochs), nprocs=world_size)
+    # nprocs=world_size spawns one process per GPU
+

appendix-A/03_main-chapter-code/code-part2.ipynb

@@ -0,0 +1,452 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "O9i6kzBsZVaZ"
+   },
+   "source": [
+    "# Appendix A: Introduction to PyTorch (Part 2)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "ppbG5d-NZezH"
+   },
+   "source": [
+    "## A.9 Optimizing training performance with GPUs"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "6jH0J_DPZhbn"
+   },
+   "source": [
+    "### A.9.1 PyTorch computations on GPU devices"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "RM7kGhwMF_nO",
+    "outputId": "ac60b048-b81f-4bb0-90fa-1ca474f04e9a"
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "2.0.1+cu118\n"
+     ]
+    }
+   ],
+   "source": [
+    "import torch\n",
+    "\n",
+    "print(torch.__version__)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "OXLCKXhiUkZt",
+    "outputId": "39fe5366-287e-47eb-cc34-3508d616c4f9"
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "True\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(torch.cuda.is_available())"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "MTTlfh53Va-T",
+    "outputId": "f31d8bbe-577f-4db4-9939-02e66b9f96d1"
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "tensor([5., 7., 9.])"
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "tensor_1 = torch.tensor([1., 2., 3.])\n",
+    "tensor_2 = torch.tensor([4., 5., 6.])\n",
+    "\n",
+    "print(tensor_1 + tensor_2)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "Z4LwTNw7Vmmb",
+    "outputId": "1c025c6a-e3ed-4c7c-f5fd-86c14607036e"
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tensor([5., 7., 9.], device='cuda:0')\n"
+     ]
+    }
+   ],
+   "source": [
+    "tensor_1 = tensor_1.to(\"cuda\")\n",
+    "tensor_2 = tensor_2.to(\"cuda\")\n",
+    "\n",
+    "print(tensor_1 + tensor_2)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/",
+     "height": 184
+    },
+    "id": "tKT6URN1Vuft",
+    "outputId": "e6f01e7f-d9cf-44cb-cc6d-46fc7907d5c0"
+   },
+   "outputs": [
+    {
+     "ename": "RuntimeError",
+     "evalue": "ignored",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
+      "\u001b[0;32m<ipython-input-7-4ff3c4d20fc3>\u001b[0m in \u001b[0;36m<cell line: 2>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mtensor_1\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtensor_1\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"cpu\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtensor_1\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mtensor_2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
+      "\u001b[0;31mRuntimeError\u001b[0m: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!"
+     ]
+    }
+   ],
+   "source": [
+    "tensor_1 = tensor_1.to(\"cpu\")\n",
+    "print(tensor_1 + tensor_2)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "c8j1cWDcWAMf"
+   },
+   "source": [
+    "## A.9.2 Single-GPU training"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {
+    "id": "GyY59cjieitv"
+   },
+   "outputs": [],
+   "source": [
+    "X_train = torch.tensor([\n",
+    "    [-1.2, 3.1],\n",
+    "    [-0.9, 2.9],\n",
+    "    [-0.5, 2.6],\n",
+    "    [2.3, -1.1],\n",
+    "    [2.7, -1.5]\n",
+    "])\n",
+    "\n",
+    "y_train = torch.tensor([0, 0, 0, 1, 1])\n",
+    "\n",
+    "X_test = torch.tensor([\n",
+    "    [-0.8, 2.8],\n",
+    "    [2.6, -1.6],\n",
+    "])\n",
+    "\n",
+    "y_test = torch.tensor([0, 1])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {
+    "id": "v41gKqEJempa"
+   },
+   "outputs": [],
+   "source": [
+    "from torch.utils.data import Dataset\n",
+    "\n",
+    "\n",
+    "class ToyDataset(Dataset):\n",
+    "    def __init__(self, X, y):\n",
+    "        self.features = X\n",
+    "        self.labels = y\n",
+    "\n",
+    "    def __getitem__(self, index):\n",
+    "        one_x = self.features[index]\n",
+    "        one_y = self.labels[index]\n",
+    "        return one_x, one_y\n",
+    "\n",
+    "    def __len__(self):\n",
+    "        return self.labels.shape[0]\n",
+    "\n",
+    "train_ds = ToyDataset(X_train, y_train)\n",
+    "test_ds = ToyDataset(X_test, y_test)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {
+    "id": "UPGVRuylep8Y"
+   },
+   "outputs": [],
+   "source": [
+    "from torch.utils.data import DataLoader\n",
+    "\n",
+    "torch.manual_seed(123)\n",
+    "\n",
+    "train_loader = DataLoader(\n",
+    "    dataset=train_ds,\n",
+    "    batch_size=2,\n",
+    "    shuffle=True,\n",
+    "    num_workers=1,\n",
+    "    drop_last=True\n",
+    ")\n",
+    "\n",
+    "test_loader = DataLoader(\n",
+    "    dataset=test_ds,\n",
+    "    batch_size=2,\n",
+    "    shuffle=False,\n",
+    "    num_workers=1\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {
+    "id": "drhg6IXofAXh"
+   },
+   "outputs": [],
+   "source": [
+    "class NeuralNetwork(torch.nn.Module):\n",
+    "    def __init__(self, num_inputs, num_outputs):\n",
+    "        super().__init__()\n",
+    "\n",
+    "        self.layers = torch.nn.Sequential(\n",
+    "\n",
+    "            # 1st hidden layer\n",
+    "            torch.nn.Linear(num_inputs, 30),\n",
+    "            torch.nn.ReLU(),\n",
+    "\n",
+    "            # 2nd hidden layer\n",
+    "            torch.nn.Linear(30, 20),\n",
+    "            torch.nn.ReLU(),\n",
+    "\n",
+    "            # output layer\n",
+    "            torch.nn.Linear(20, num_outputs),\n",
+    "        )\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        logits = self.layers(x)\n",
+    "        return logits"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "7jaS5sqPWCY0",
+    "outputId": "84c74615-38f2-48b8-eeda-b5912fed1d3a"
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch: 001/003 | Batch 000/002 | Train/Val Loss: 0.75\n",
+      "Epoch: 001/003 | Batch 001/002 | Train/Val Loss: 0.65\n",
+      "Epoch: 002/003 | Batch 000/002 | Train/Val Loss: 0.44\n",
+      "Epoch: 002/003 | Batch 001/002 | Train/Val Loss: 0.13\n",
+      "Epoch: 003/003 | Batch 000/002 | Train/Val Loss: 0.03\n",
+      "Epoch: 003/003 | Batch 001/002 | Train/Val Loss: 0.00\n"
+     ]
+    }
+   ],
+   "source": [
+    "import torch.nn.functional as F\n",
+    "\n",
+    "\n",
+    "torch.manual_seed(123)\n",
+    "model = NeuralNetwork(num_inputs=2, num_outputs=2)\n",
+    "\n",
+    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\") # NEW\n",
+    "model = model.to(device) # NEW\n",
+    "\n",
+    "optimizer = torch.optim.SGD(model.parameters(), lr=0.5)\n",
+    "\n",
+    "num_epochs = 3\n",
+    "\n",
+    "for epoch in range(num_epochs):\n",
+    "\n",
+    "    model.train()\n",
+    "    for batch_idx, (features, labels) in enumerate(train_loader):\n",
+    "\n",
+    "        features, labels = features.to(device), labels.to(device) # NEW\n",
+    "        logits = model(features)\n",
+    "        loss = F.cross_entropy(logits, labels) # Loss function\n",
+    "\n",
+    "        optimizer.zero_grad()\n",
+    "        loss.backward()\n",
+    "        optimizer.step()\n",
+    "\n",
+    "        ### LOGGING\n",
+    "        print(f\"Epoch: {epoch+1:03d}/{num_epochs:03d}\"\n",
+    "              f\" | Batch {batch_idx:03d}/{len(train_loader):03d}\"\n",
+    "              f\" | Train/Val Loss: {loss:.2f}\")\n",
+    "\n",
+    "    model.eval()\n",
+    "    # Optional model evaluation"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "metadata": {
+    "id": "4qrlmnPPe7FO"
+   },
+   "outputs": [],
+   "source": [
+    "def compute_accuracy(model, dataloader, device):\n",
+    "\n",
+    "    model = model.eval()\n",
+    "    correct = 0.0\n",
+    "    total_examples = 0\n",
+    "\n",
+    "    for idx, (features, labels) in enumerate(dataloader):\n",
+    "\n",
+    "        features, labels = features.to(device), labels.to(device) # New\n",
+    "\n",
+    "        with torch.no_grad():\n",
+    "            logits = model(features)\n",
+    "\n",
+    "        predictions = torch.argmax(logits, dim=1)\n",
+    "        compare = labels == predictions\n",
+    "        correct += torch.sum(compare)\n",
+    "        total_examples += len(compare)\n",
+    "\n",
+    "    return (correct / total_examples).item()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "1_-BfkfEf4HX",
+    "outputId": "473bf21d-5880-4de3-fc8a-051d75315b94"
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "1.0"
+      ]
+     },
+     "execution_count": 27,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "compute_accuracy(model, train_loader, device=device)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "iYtXKBGEgKss",
+    "outputId": "508edd84-3fb7-4d04-cb23-9df0c3d24170"
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "1.0"
+      ]
+     },
+     "execution_count": 21,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "compute_accuracy(model, test_loader, device=device)"
+   ]
+  }
+ ],
+ "metadata": {
+  "accelerator": "GPU",
+  "colab": {
+   "gpuType": "T4",
+   "provenance": []
+  },
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.10.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}