{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7fb27b941602401d91542211134fc71a",
   "metadata": {
    "colab_type": "text",
    "id": "view-in-github"
   },
   "source": [
    "# MNIST Classification with blox\n",
    "\n",
    "This tutorial demonstrates how to train a Convolutional Neural Network (CNN) on MNIST using **blox**.\n",
    "\n",
    "We will strictly follow a probabilistic approach:\n",
    "1.  **Model**: Defines a conditional distribution $P(Y | X)$.\n",
    "2.  **Objective**: Maximize the likelihood of the data (minimize Negative Log Likelihood).\n",
    "\n",
    "We use **Distrax** to handle the probability distributions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "acae54e37e7d407bbb7b55eff062a284",
   "metadata": {
    "id": "install-deps"
   },
   "outputs": [],
   "source": [
    "# Setup: ensure blox is importable.\n",
    "import sys\n",
    "\n",
    "sys.path.insert(0, \"../src\")\n",
    "\n",
    "# Install necessary packages.\n",
    "!pip install -q optax tensorflow tensorflow-datasets matplotlib distrax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "9a63283cbaf04dbcab1f6479b197f3a8",
   "metadata": {
    "id": "imports"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/python/3.12.1/lib/python3.12/site-packages/keras/src/export/tf2onnx_lib.py:8: FutureWarning: In the future `np.object` will be defined as the corresponding NumPy scalar.\n",
      "  if not hasattr(np, \"object\"):\n"
     ]
    }
   ],
   "source": [
    "import blox as bx\n",
    "import distrax\n",
    "import jax\n",
    "import jax.numpy as jnp\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import optax\n",
    "import tensorflow as tf\n",
    "import tensorflow_datasets as tfds\n",
    "\n",
    "# Set random seed for reproducibility.\n",
    "tf.random.set_seed(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "8dd0d8092fe74a7c96281538738b07e2",
   "metadata": {
    "id": "load-data"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2025-12-26 18:10:15.582305: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303)\n",
      "2025-12-26 18:10:18.369270: W tensorflow/core/kernels/data/cache_dataset_ops.cc:917] The calling iterator did not fully read the dataset being cached. In order to avoid unexpected truncation of the dataset, the partially cached contents of the dataset  will be discarded. This can happen if you have an input pipeline similar to `dataset.cache().take(k).repeat()`. You should use `dataset.take(k).cache().repeat()` instead.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Batch shape: (32, 28, 28, 1)\n"
     ]
    }
   ],
   "source": [
    "batch_size = 32\n",
    "train_steps = 1200\n",
    "eval_every = 200\n",
    "\n",
    "\n",
    "def get_datasets(batch_size: int, train_steps: int):\n",
    "  \"\"\"Load MNIST datasets as tf.data Datasets.\"\"\"\n",
    "\n",
    "  def normalize(sample):\n",
    "    image = tf.cast(sample['image'], tf.float32) / 255.0\n",
    "    return {'image': image, 'label': sample['label']}\n",
    "\n",
    "  train_ds = tfds.load('mnist', split='train')\n",
    "  train_ds = train_ds.map(normalize)\n",
    "  train_ds = train_ds.repeat().shuffle(1024)\n",
    "  train_ds = train_ds.batch(batch_size, drop_remainder=True)\n",
    "  train_ds = train_ds.take(train_steps).prefetch(-1)\n",
    "\n",
    "  test_ds = tfds.load('mnist', split='test')\n",
    "  test_ds = test_ds.map(normalize)\n",
    "  test_ds = test_ds.batch(batch_size, drop_remainder=True).prefetch(-1)\n",
    "\n",
    "  return train_ds, test_ds\n",
    "\n",
    "\n",
    "train_ds, test_ds = get_datasets(batch_size, train_steps)\n",
    "sample_batch = next(iter(train_ds.as_numpy_iterator()))\n",
    "print(f'Batch shape: {sample_batch[\"image\"].shape}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "72eea5119410473aa328ad9291626812",
   "metadata": {
    "id": "define-model"
   },
   "outputs": [],
   "source": "class CNN(bx.Module):\n  \"\"\"A probabilistic CNN classifier.\"\"\"\n\n  def __init__(self, graph: bx.Graph, num_classes: int, rng: bx.Rng):\n    super().__init__(graph)\n    self.conv1 = bx.Conv(\n        graph.child('conv1'), output_channels=32, kernel_size=(3, 3), rng=rng\n    )\n    self.conv2 = bx.Conv(\n        graph.child('conv2'), output_channels=64, kernel_size=(3, 3), rng=rng\n    )\n    self.linear1 = bx.Linear(graph.child('linear1'), output_size=256, rng=rng)\n    self.linear2 = bx.Linear(\n        graph.child('linear2'), output_size=num_classes, rng=rng\n    )\n    self.dropout = bx.Dropout(graph.child('dropout'), rate=0.5, rng=rng)\n\n  def __call__(self, params: bx.Params, x: jax.Array, is_training: bool = True):\n    # Convolutional feature extraction.\n    x, params = self.conv1(params, x)\n    x = jax.nn.relu(x)\n    x = bx.max_pool(x, window_shape=(2, 2), strides=(2, 2))\n\n    x, params = self.conv2(params, x)\n    x = jax.nn.relu(x)\n    x = bx.max_pool(x, window_shape=(2, 2), strides=(2, 2))\n\n    # Flatten and dense layers.\n    x = x.reshape((x.shape[0], -1))\n    x, params = self.linear1(params, x)\n    x = jax.nn.relu(x)\n    x, params = self.dropout(params, x, is_training=is_training)\n\n    # Output logits for the Categorical distribution.\n    logits, params = self.linear2(params, x)\n    return logits, params"
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8edb47106e1a46a883d545849b8ab81b",
   "metadata": {
    "id": "init-model"
   },
   "outputs": [],
   "source": "# Create model components.\ngraph = bx.Graph('mnist')\nrng = bx.Rng(graph.child('rng'))\nmodel = CNN(graph.child('cnn'), num_classes=10, rng=rng)\n\n# Initialize params with a sample batch item.\nparams = rng.seed(bx.Params(), seed=42)\n_, params = model(params, sample_batch['image'][:1], is_training=False)\nparams = params.locked()\n\n# Create a separate eval key for evaluation.\neval_key = jax.random.key(1)\n\nbx.display(model.graph, params)"
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "10185d26023b46108eb7d9f57d49d2b3",
   "metadata": {
    "id": "train-step"
   },
   "outputs": [],
   "source": [
    "@jax.jit(static_argnames=['optimizer'], donate_argnames=['params', 'opt_state'])\n",
    "def train_step(params, opt_state, batch_images, batch_labels, optimizer):\n",
    "  trainable, non_trainable = params.split()\n",
    "\n",
    "  def loss_fn(trainable):\n",
    "    params = trainable.merge(non_trainable)\n",
    "    logits, new_params = model(params, batch_images, is_training=True)\n",
    "\n",
    "    # Probabilistic Loss: Negative Log Likelihood.\n",
    "    dist = distrax.Categorical(logits=logits)\n",
    "    loss = -dist.log_prob(batch_labels).mean()\n",
    "    accuracy = jnp.mean(dist.mode() == batch_labels)\n",
    "\n",
    "    _, new_non_trainable = new_params.split()\n",
    "    return loss, ({'loss': loss, 'accuracy': accuracy}, new_non_trainable)\n",
    "\n",
    "  grads, (metrics, new_non_trainable) = jax.grad(loss_fn, has_aux=True)(\n",
    "      trainable\n",
    "  )\n",
    "  updates, new_opt_state = optimizer.update(grads, opt_state, trainable)\n",
    "  new_trainable = optax.apply_updates(trainable, updates)\n",
    "\n",
    "  return new_trainable.merge(new_non_trainable), new_opt_state, metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8763a12b2bbd4a93a75aff182afb95dc",
   "metadata": {
    "id": "eval-step"
   },
   "outputs": [],
   "source": "@jax.jit\ndef eval_step(params, batch_images, batch_labels):\n  \"\"\"Evaluate a batch. Does not return params to avoid memory waste.\"\"\"\n  logits, _ = model(params, batch_images, is_training=False)\n  dist = distrax.Categorical(logits=logits)\n  loss = -dist.log_prob(batch_labels).mean()\n  accuracy = jnp.mean(dist.mode() == batch_labels)\n  return {'loss': loss, 'accuracy': accuracy}"
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7623eae2785240b9bd12b16a66d81610",
   "metadata": {
    "id": "training-loop"
   },
   "outputs": [],
   "source": [
    "def train_model(params):\n",
    "  optimizer = optax.adamw(1e-3)\n",
    "  trainable_params, _ = params.split()\n",
    "  opt_state = optimizer.init(trainable_params)\n",
    "\n",
    "  # Track eval counter for unique RNG keys during evaluation.\n",
    "  eval_counter = 0\n",
    "\n",
    "  history = {\n",
    "      'train_loss': [],\n",
    "      'train_acc': [],\n",
    "      'test_loss': [],\n",
    "      'test_acc': [],\n",
    "  }\n",
    "\n",
    "  for step, batch in enumerate(train_ds.as_numpy_iterator()):\n",
    "    params, opt_state, metrics = train_step(\n",
    "        params, opt_state, batch['image'], batch['label'], optimizer\n",
    "    )\n",
    "    history['train_loss'].append(float(metrics['loss']))\n",
    "    history['train_acc'].append(float(metrics['accuracy']))\n",
    "\n",
    "    # Evaluate periodically.\n",
    "    if (step + 1) % eval_every == 0:\n",
    "      test_metrics = {'loss': [], 'accuracy': []}\n",
    "      for test_batch in test_ds.as_numpy_iterator():\n",
    "        # Fold in counter to eval_key for unique RNG each batch.\n",
    "        eval_params = rng.seed(\n",
    "            params, seed=jax.random.fold_in(eval_key, eval_counter)\n",
    "        )\n",
    "        eval_counter += 1\n",
    "\n",
    "        metrics = eval_step(\n",
    "            eval_params, test_batch['image'], test_batch['label']\n",
    "        )\n",
    "        test_metrics['loss'].append(metrics['loss'])\n",
    "        test_metrics['accuracy'].append(metrics['accuracy'])\n",
    "\n",
    "      test_loss = np.mean(test_metrics['loss'])\n",
    "      test_acc = np.mean(test_metrics['accuracy'])\n",
    "      history['test_loss'].append(test_loss)\n",
    "      history['test_acc'].append(test_acc)\n",
    "\n",
    "      train_loss = np.mean(history['train_loss'][-eval_every:])\n",
    "      train_acc = np.mean(history['train_acc'][-eval_every:])\n",
    "      print(\n",
    "          f'Step {step + 1}: '\n",
    "          f'train_loss={train_loss:.4f}, train_acc={train_acc:.4f}, '\n",
    "          f'test_loss={test_loss:.4f}, test_acc={test_acc:.4f}'\n",
    "      )\n",
    "\n",
    "  return params, history\n",
    "\n",
    "\n",
    "trained_params, history = train_model(params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "7cdc8c89c7104fffa095e18ddfef8986",
   "metadata": {
    "id": "viz"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x300 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize Training Progress.\n",
    "eval_steps = [eval_every * (i + 1) for i in range(len(history['test_acc']))]\n",
    "\n",
    "plt.figure(figsize=(12, 4))\n",
    "\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(history['train_loss'], alpha=0.3, label='Train (per step)')\n",
    "plt.plot(eval_steps, history['test_loss'], 'o-', label='Test')\n",
    "plt.xlabel('Step')\n",
    "plt.ylabel('Loss (NLL)')\n",
    "plt.legend()\n",
    "\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(history['train_acc'], alpha=0.3, label='Train (per step)')\n",
    "plt.plot(eval_steps, history['test_acc'], 'o-', label='Test')\n",
    "plt.xlabel('Step')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()\n",
    "\n",
    "\n",
    "# Visualize Predictions.\n",
    "def show_predictions(params, count=5):\n",
    "  test_batch = next(iter(test_ds.as_numpy_iterator()))\n",
    "  images = test_batch['image'][:count]\n",
    "  labels = test_batch['label'][:count]\n",
    "  logits, _ = model(params, images, is_training=False)\n",
    "\n",
    "  dist = distrax.Categorical(logits=logits)\n",
    "  preds = dist.mode()\n",
    "\n",
    "  fig, axes = plt.subplots(1, count, figsize=(15, 3))\n",
    "  for i, ax in enumerate(axes):\n",
    "    ax.imshow(images[i].squeeze(), cmap='gray')\n",
    "    ax.set_title(f'True: {labels[i]}, Pred: {preds[i]}')\n",
    "    ax.axis('off')\n",
    "  plt.show()\n",
    "\n",
    "\n",
    "show_predictions(trained_params)"
   ]
  }
 ],
 "metadata": {
  "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": "3.12.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}