demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Micrograd - Arnie's version\n",
    "\n",
    "I built this project as a way to solidify my understanding of the Stanford/DeepLearningAI Coursera course.\n",
    "\n",
    "Before taking the class (June start - July end), I had watched Andrej Karpathy's Micrograd lecture (around May), but not since then.\n",
    "\n",
    "To really test my understanding, I built this out *without* re-watching Karpathy's lecture. That is, I built this with a rough understanding of autograd, and thought through first principles of differentiation, partial derivatives, chain_rule, software design, back_propagation, linear regression, optimizers, loss functions.\n",
    "\n",
    "Note:\n",
    "- I did have a working micrograd implementation, however my loss was incorrect.\n",
    "- After manually calculating the backprop and still not figuring out, I skimmed through Karpathy's micrograd code.\n",
    "- I took the opportunity to:\n",
    "-> Add in additional methods I hand't thought like __sub__, __rsub__, ....\n",
    "-> Clean-up my implementation of Neuron (notably, passing in input_feature and a single x sample, rather than the x as array itself)\n",
    "\n",
    "In the end, I realized the problem was with that I forgot to call the zero_grad() which led to the loss bouncing since the grads were compounding across iterations\n",
    "\n",
    "In principle, I don't take shortcuts, but given my urgency to progress as fast as possible in ML and since I already had a working implementation (with correct intuition), it did not make sense to spend more time debugging (I can always do that later).\n",
    "\n",
    "Extra notes: Obviously I did not use Copilot xD."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: pydantic in ./.venv/lib/python3.12/site-packages (2.8.2)\n",
      "Requirement already satisfied: annotated-types>=0.4.0 in ./.venv/lib/python3.12/site-packages (from pydantic) (0.7.0)\n",
      "Requirement already satisfied: pydantic-core==2.20.1 in ./.venv/lib/python3.12/site-packages (from pydantic) (2.20.1)\n",
      "Requirement already satisfied: typing-extensions>=4.6.1 in ./.venv/lib/python3.12/site-packages (from pydantic) (4.12.2)\n"
     ]
    }
   ],
   "source": [
    "# Install requirements\n",
    "\n",
    "!pip install pydantic"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Operation Tests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Val(value=18.0000, grad=0.0000, parents=(6.0 * 3.0))\n",
      "Val(value=2.0000, grad=0.0000, parents=(6.0 * 0.3333333333333333))\n"
     ]
    }
   ],
   "source": [
    "from micrograd.engine.value import Value\n",
    "\n",
    "a = Value(6.0)\n",
    "b = Value(3.0)\n",
    "\n",
    "c = a * b\n",
    "print(c)\n",
    "\n",
    "d = a / b\n",
    "print(d)\n",
    "\n",
    "e = Value(100.0)\n",
    "f = e.log()\n",
    "# print(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "f = f.log()\n",
    "# print(f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Simple Linear Regression Example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [100/1000], Loss: 0.0557\n",
      "Epoch [200/1000], Loss: 0.0439\n",
      "Epoch [300/1000], Loss: 0.0426\n",
      "Epoch [400/1000], Loss: 0.0425\n",
      "Epoch [500/1000], Loss: 0.0425\n",
      "Epoch [600/1000], Loss: 0.0425\n",
      "Epoch [700/1000], Loss: 0.0425\n",
      "Epoch [800/1000], Loss: 0.0425\n",
      "Epoch [900/1000], Loss: 0.0425\n",
      "Epoch [1000/1000], Loss: 0.0425\n",
      "Prediction for x=6.0 is y_pred=11.81632756353839 (Expected is 12)\n"
     ]
    }
   ],
   "source": [
    "from micrograd.nets.SimpleLinearRegression import SimpleLinearRegression\n",
    "from micrograd.loss_functions.MSE import MSE\n",
    "from micrograd.optimizers.SimpleOptimizer import SimpleOptimizer\n",
    "\n",
    "from micrograd.engine.value import Value\n",
    "\n",
    "# Set-up model, optimizers and loss function\n",
    "model = SimpleLinearRegression()\n",
    "criterion = MSE()\n",
    "\n",
    "num_epochs = 1000\n",
    "lr = 0.01\n",
    "\n",
    "optimizer = SimpleOptimizer(model.parameters(), lr)\n",
    "\n",
    "# Dataset\n",
    "x = [Value(1.0), Value(4.0), Value(9.0)]\n",
    "y = [Value(1.0), Value(8.0), Value(18.0)]\n",
    "\n",
    "for epoch in range(num_epochs):\n",
    "    # Need to applying the logistic activation!\n",
    "    y_pred = [model([xi]) for xi in x]\n",
    "\n",
    "    loss = criterion(y_pred, y)\n",
    "\n",
    "    optimizer.zero_grad()\n",
    "\n",
    "    loss.backward()\n",
    "    optimizer.step()\n",
    "\n",
    "    if (epoch + 1) % 100 == 0:\n",
    "        print(f'Epoch [{epoch+1}/{num_epochs}], Loss: {loss.value:.4f}')\n",
    "\n",
    "test_x = [Value(6.0)]\n",
    "test_y_pred = model(test_x)\n",
    "\n",
    "print(f\"Prediction for x={test_x[0].value} is y_pred={test_y_pred.value} (Expected is 12)\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Hot Dog Classifier\n",
    "Inspired by Jian-Yang from Silicon Valley, I built out a simple hot dog classifer.\n",
    "\n",
    "I downloaded a dataset from HuggingFace and used my Micrograd implementation to train and evaluate test images :)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: Pillow in ./.venv/lib/python3.12/site-packages (10.4.0)\n",
      "Requirement already satisfied: numpy in ./.venv/lib/python3.12/site-packages (2.0.0)\n"
     ]
    }
   ],
   "source": [
    "!pip install Pillow\n",
    "!pip install numpy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training samples: 200\n",
      "Validation samples: 50\n"
     ]
    }
   ],
   "source": [
    "# Load data and pre-process it\n",
    "import os\n",
    "from PIL import Image\n",
    "import numpy as np\n",
    "from micrograd.engine.value import Value\n",
    "\n",
    "IMAGE_SIZE = 225\n",
    "\n",
    "def load_and_preprocess_images(folder_path, label, image_size):\n",
    "    images = []\n",
    "    labels = []\n",
    "    for filename in os.listdir(folder_path):\n",
    "        if filename.endswith(('.jpg', '.jpeg', '.png')):\n",
    "            img_path = os.path.join(folder_path, filename)\n",
    "            with Image.open(img_path) as img:\n",
    "                img = img.resize((image_size, image_size))\n",
    "                img = img.convert('L')  # Convert to grayscale\n",
    "                img_array = np.array(img).flatten() / 255.0  # Normalize to [0, 1]\n",
    "                images.append([Value(float(pixel)) for pixel in img_array])\n",
    "                labels.append(Value(float(label)))\n",
    "    return images, labels\n",
    "\n",
    "# Load training data\n",
    "train_hotdog_path = 'dataset/hotdog_nothotdog/train/hotdog'\n",
    "train_nothotdog_path = 'dataset/hotdog_nothotdog/train/not_hotdog'\n",
    "\n",
    "x_train_hotdog, y_train_hotdog = load_and_preprocess_images(train_hotdog_path, 1, IMAGE_SIZE)\n",
    "x_train_nothotdog, y_train_nothotdog = load_and_preprocess_images(train_nothotdog_path, 0, IMAGE_SIZE)\n",
    "\n",
    "x_train = x_train_hotdog + x_train_nothotdog\n",
    "y_train = y_train_hotdog + y_train_nothotdog\n",
    "\n",
    "# Load validation data\n",
    "val_hotdog_path = 'dataset/hotdog_nothotdog/val/hotdog'\n",
    "val_nothotdog_path = 'dataset/hotdog_nothotdog/val/not_hotdog'\n",
    "\n",
    "x_valid_hotdog, y_valid_hotdog = load_and_preprocess_images(val_hotdog_path, 1, IMAGE_SIZE)\n",
    "x_valid_nothotdog, y_valid_nothotdog = load_and_preprocess_images(val_nothotdog_path, 0, IMAGE_SIZE)\n",
    "\n",
    "x_valid = x_valid_hotdog + x_valid_nothotdog\n",
    "y_valid = y_valid_hotdog + y_valid_nothotdog\n",
    "\n",
    "# Shuffle the training data\n",
    "combined = list(zip(x_train, y_train))\n",
    "np.random.shuffle(combined)\n",
    "x_train, y_train = zip(*combined)\n",
    "\n",
    "print(f\"Training samples: {len(x_train)}\")\n",
    "print(f\"Validation samples: {len(x_valid)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/1000], Loss: 19.27597753504074 - (step: 0.04)\n",
      "Epoch [2/1000], Loss: 19.194378964584704 - (step: 0.04)\n",
      "Epoch [3/1000], Loss: 19.160225360889417 - (step: 0.04)\n",
      "Epoch [4/1000], Loss: 19.12638730430335 - (step: 0.04)\n",
      "Epoch [5/1000], Loss: 19.027540792324544 - (step: 0.04)\n",
      "Epoch [6/1000], Loss: 18.949023464810374 - (step: 0.04)\n",
      "Epoch [7/1000], Loss: 18.623217836393184 - (step: 0.04)\n",
      "Epoch [8/1000], Loss: 18.419496219551196 - (step: 0.04)\n",
      "Epoch [9/1000], Loss: 18.20838775934107 - (step: 0.04)\n",
      "Epoch [10/1000], Loss: 17.962298828046166 - (step: 0.04)\n",
      "Epoch [11/1000], Loss: 17.682050020246123 - (step: 0.04)\n",
      "Epoch [12/1000], Loss: 17.17145053678899 - (step: 0.04)\n",
      "Epoch [13/1000], Loss: 16.671359444915126 - (step: 0.04)\n",
      "Epoch [14/1000], Loss: 15.523554073728265 - (step: 0.04)\n",
      "Epoch [15/1000], Loss: 13.677555737292836 - (step: 0.04)\n",
      "Epoch [16/1000], Loss: 8.458576800333601 - (step: 0.02)\n",
      "Epoch [17/1000], Loss: 4.967138607664598 - (step: 0.02)\n",
      "Epoch [18/1000], Loss: 1.666139849362127 - (step: 0.02)\n",
      "Epoch [19/1000], Loss: 6.249019037767665 - (step: 0.02)\n",
      "Epoch [20/1000], Loss: 2.545814771668153 - (step: 0.02)\n",
      "Epoch [21/1000], Loss: 1.1810002000184578 - (step: 0.02)\n",
      "Epoch [22/1000], Loss: 4.639817799604142 - (step: 0.02)\n",
      "Epoch [23/1000], Loss: 1.0242880784074566 - (step: 0.02)\n",
      "Epoch [24/1000], Loss: 5.661583432592153 - (step: 0.02)\n",
      "Epoch [25/1000], Loss: 1.419203129133773 - (step: 0.02)\n",
      "Epoch [26/1000], Loss: 6.613780065853663 - (step: 0.02)\n",
      "Epoch [27/1000], Loss: 1.9619616190452576 - (step: 0.02)\n",
      "Early stopping at epoch 28, loss: 0.7960\n"
     ]
    }
   ],
   "source": [
    "from micrograd.nets.ComplexLogisticRegression import ComplexLogisticRegression\n",
    "\n",
    "from micrograd.loss_functions.BinaryCrossEntropy import BinaryCrossEntropy\n",
    "from micrograd.optimizers.SimpleOptimizer import SimpleOptimizer\n",
    "\n",
    "import random\n",
    "\n",
    "\n",
    "model = ComplexLogisticRegression(image_size=IMAGE_SIZE)\n",
    "criterion = BinaryCrossEntropy()\n",
    "\n",
    "num_epochs = 1000\n",
    "lr = 0.02\n",
    "\n",
    "optimizer = SimpleOptimizer(model.parameters(), lr)\n",
    "\n",
    "losses = []\n",
    "\n",
    "# Train the model\n",
    "for epoch in range(num_epochs):\n",
    "\n",
    "    y_pred = [model(xi) for xi in x_train]\n",
    "\n",
    "    # for prediction in y_pred:\n",
    "        # print(f\"Prediction: {prediction}\")\n",
    "\n",
    "    loss = criterion(y_pred, y_train)\n",
    "\n",
    "    optimizer.zero_grad()\n",
    "\n",
    "    loss.backward()\n",
    "\n",
    "    # Dynamically adjust the learning rate\n",
    "    if (loss.value < 20.0):\n",
    "        optimizer.set_learn_rate(0.04)  # Reduced from 0.02\n",
    "\n",
    "    if (loss.value < 10.0):\n",
    "        optimizer.set_learn_rate(0.002)  # Increased from 0.001\n",
    "\n",
    "    # TODO: Move into optimizer.step()\n",
    "    # Add gradient clipping\n",
    "    max_grad = 1.0\n",
    "    for param in model.parameters():\n",
    "        if param.grad is not None:\n",
    "            param.grad = max(min(param.grad, max_grad), -max_grad)\n",
    "\n",
    "    # Early stopping\n",
    "    if (loss.value < 1.00):\n",
    "        print(f'Early stopping at epoch {epoch+1}, loss: {loss.value:.4f}')\n",
    "        break\n",
    "\n",
    "    optimizer.step()\n",
    "    losses.append(loss)\n",
    "\n",
    "    if (epoch + 1) % 1 == 0:\n",
    "        print(f'Epoch [{epoch+1}/{num_epochs}], Loss: {loss.value} - (step: {optimizer.get_learn_rate()})')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: matplotlib in ./.venv/lib/python3.12/site-packages (3.9.1)\n",
      "Requirement already satisfied: contourpy>=1.0.1 in ./.venv/lib/python3.12/site-packages (from matplotlib) (1.2.1)\n",
      "Requirement already satisfied: cycler>=0.10 in ./.venv/lib/python3.12/site-packages (from matplotlib) (0.12.1)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in ./.venv/lib/python3.12/site-packages (from matplotlib) (4.53.1)\n",
      "Requirement already satisfied: kiwisolver>=1.3.1 in ./.venv/lib/python3.12/site-packages (from matplotlib) (1.4.5)\n",
      "Requirement already satisfied: numpy>=1.23 in ./.venv/lib/python3.12/site-packages (from matplotlib) (2.0.0)\n",
      "Requirement already satisfied: packaging>=20.0 in ./.venv/lib/python3.12/site-packages (from matplotlib) (24.1)\n",
      "Requirement already satisfied: pillow>=8 in ./.venv/lib/python3.12/site-packages (from matplotlib) (10.4.0)\n",
      "Requirement already satisfied: pyparsing>=2.3.1 in ./.venv/lib/python3.12/site-packages (from matplotlib) (3.1.2)\n",
      "Requirement already satisfied: python-dateutil>=2.7 in ./.venv/lib/python3.12/site-packages (from matplotlib) (2.9.0.post0)\n",
      "Requirement already satisfied: six>=1.5 in ./.venv/lib/python3.12/site-packages (from python-dateutil>=2.7->matplotlib) (1.16.0)\n"
     ]
    }
   ],
   "source": [
    "!pip install matplotlib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the losses\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(range(1, len(losses) + 1), [loss.value for loss in losses])\n",
    "plt.title('Training Loss over Epochs')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')\n",
    "plt.grid(True)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Exporting model...\n",
      "{\"layers\":[{\"input_features\":225,\"output_features\":5,\"activation\":\"relu_activation\",\"neurons\":[{\"values\":[0.6989329436985595,1.249039019723437,1.1264204889661296,1.3100359981365681,0.3976737039717889,1.1862825524572798,0.74796434442675,1.340515973820429,0.44219326543351983,0.43637584369169896,1.3296454375883808,0.9582856025602015,0.8442741708829835,1.2780902294569716,0.9226763574881423,0.7603625959208007,1.328200796196609,0.9913517299498465,1.1108557867295261,0.44764068849758426,1.3072852690145393,0.6062305706649406,0.9949178130083011,1.1212015744390658,0.7459278509233918,1.098926969583689,0.700632971542126,0.6257906910608659,0.5322845620682138,1.3232608223088846,0.9737580305180237,1.2394751250788278,0.7817015187980194,1.3125640581803832,0.8258716155401092,0.3882823189517114,1.2577815716292204,0.8717329318360384,1.1667483083449393,0.8938091674976237,0.49907833793147094,0.7252543228013617,1.0338376629658506,0.5268234905222378,0.930068383611585,0.7413819901110821,0.6546542169183619,1.095229792530517,0.7382982546796812,1.171812519541777,0.8193729228985884,0.6590840767864089,0.9040481629008408,0.9849117496318214,0.7418250000292501,0.9400385219997942,0.3997464511021973,1.2251633560690995,1.1003149478793022,0.7922735923415009,0.8549613880498633,0.5560014822295225,1.3118277251061345,0.5902922424362135,1.3306486634848165,0.6663969757425655,1.1183015329259556,0.6107050240391909,1.1399953981387667,1.2171106824652165,0.9004075598252381,0.8239449795395302,0.870253209025115,0.7468827252867746,0.5535917623584986,1.1072467722636157,0.745686313948531,1.1961865554506652,0.701018184541474,1.0226625442455952,0.7344680267620284,0.6072525154943303,1.2127905137235553,0.8485385324863424,0.8730237246877568,0.5781027326015155,0.9066035937808696,0.39689528668228263,0.5146715425956523,0.5887929181741892,0.7221703022249534,0.5325055255349722,0.7876577218407034,1.0364633321947947,0.5645817945741924,0.3523464791267995,0.47069472121640316,0.3419640908017755,0.8685986976968402,0.842261075559449,0.5255803564906373,1.010095842071502,1.2310866354243184,0.879773269842818,1.2890697636935466,0.8086519266142437,1.3357049545640796,0.4165499326239079,1.2095104316968603,0.4130925563889354,1.2901962135018417,0.5758060123488369,1.1287879153428206,1.229426049368124,0.6289431742645525,1.1722691134493644,0.3715037313708072,0.4050807084414943,0.3470813169830693,0.8817820778968901,0.49637391651766094,0.8534067612235224,0.4127667930475475,0.390974325079489,0.5124765916014768,0.3785255183339838,0.8982470928098455,0.4418974356993943,0.950930815229728,0.44391397819136275,0.6210013094217556,1.0241898018272741,0.9585604821443962,0.7819775141182888,1.1920849699645817,0.8970835464864142,0.9173908813181895,1.0494832963384486,1.1610869687825947,0.9339321462106803,0.22614772582356057,0.7439681346731598,1.058850462173758,0.5592408938595226,0.5266989865193606,0.9569127231278461,0.7537656653507756,0.3156932507632462,0.12041606825464844,0.9574351741011989,0.6524730507133781,0.6022342483824734,0.10953161448773698,0.7354213062675945,0.38554390254274046,0.44981682606532,0.8588299790783326,0.7744473332588524,0.3157743132347956,0.7829672923203105,0.90772454933054,0.6561417560953334,-0.029447117699640746,0.018966929123760485,0.746066865955659,0.81512831012864,0.42990738126804784,0.6440870105492641,0.20656493764054626,0.8345153423979041,0.3982325421936519,0.021636784380987446,0.8773920863026223,0.6980108077416787,0.6261734015881767,0.46916411547487563,0.39601729664604746,0.039048798737813695,0.5948809546488121,0.6078643213114701,0.7180453586220086,0.3944561533603912,0.002612641231968145,-0.0909759263446949,0.45759397063409674,0.7979630103216004,0.37029732264919485,0.07242685530736222,0.6588475934812674,-0.1960301612359204,0.21568420888655745,0.7135394376171801,0.4446855137450024,-0.1165781803153725,0.10600327793460214,0.3672767073927528,0.3603596582990483,0.2906238629230561,0.3761023544147333,-0.16608042384577146,0.23530612702205866,0.375713864147431,0.3964068356452855,-0.010383990287163448,0.4081927425779178,0.10687509898985957,0.4000235943094152,0.6849287305716103,0.23572644665916834,-0.019027272306133703,0.5506268033585997,0.06293548207856695,-0.05642101524729323,0.4016962348286215,0.11840353472636768,0.500778061100526,-0.0691457564292008,0.5726722213661729,0.38469746996020193,0.8570001949635226,0.7818423336130008,0.3589117581999845,0.13579848903003902,0.8800939429041693,0.6119733191584564,-0.09787718185822061]},{\"values\":[0.12915037127178938,-0.13283628146188256,-0.20748244985174444,-0.5767235146870886,-0.6127172000710698,-0.2628105616014119,-0.051085321244420945,0.2044434119179593,-0.22367335109451694,-0.5220340498570132,0.1706250155215257,-0.71496243018821,-0.5249921560496635,0.009886042900188081,-0.5272809380693524,-0.5304835804000495,-0.07337230503972679,-0.40867864039443696,-0.4505359827071335,-0.5000263114650326,-0.1041826701407358,0.2725118831152149,-0.690103511968279,0.2886342791517742,-0.2790823955735526,-0.13911136105143343,0.24207127846377738,0.26738286618322393,-0.505303167910943,0.020853021351654042,0.017173172136967316,-0.2471576978103692,0.26139218643669604,-0.17407710093940706,-0.5818094789869325,-0.167596223963299,-0.6669577181343571,-0.01802224976196128,0.1154279237124441,-0.020004914268291114,-0.5261235517926033,-0.6315689789969982,-0.6370837956148264,-0.2843594640223908,0.25346810780801116,0.04628731470620398,-0.20392133294833056,-0.615013654544066,-0.433872562422971,-0.5513303476771395,-0.11835473910108767,-0.5418475576224087,-0.14416369736298393,0.18069735325074832,-0.11040803016984406,-0.4028175213514893,0.04131609506550894,-0.24885133150396102,-0.6298878731030589,-0.4647581082332349,0.012162230558020768,-0.2183418426693494,0.01758067192792698,-0.01920684233013011,-0.13079399045123943,0.25613215019998586,0.03342426049889648,-0.21370563908803983,-0.5997059021991269,0.10931777423482213,0.1863616079715653,0.10159378266162764,-0.3450427123627277,-0.5070046349834471,0.10243784906576378,0.11496085071452063,-0.1420466772384871,-0.46540299345509495,-0.28071476688698077,-0.6402194995575523,-0.41063780799703414,-0.4957677010185197,-0.20945241299368642,-0.34584380180364177,0.23709654860360993,0.11941509944273594,-0.3177200633969378,-0.5076925791088898,0.13602818436851677,-0.4904496931092588,-0.3632170867297836,0.1350030921282787,0.1340802494185715,-0.3787582674984444,-0.5550553086848368,0.05515664720524305,-0.061425171886811195,-0.11441747774145054,-0.4321605949367406,-0.5005886995542622,0.3213804364923809,-0.08503470808463282,0.07538725949226448,-0.3309385156260561,0.013463564884448307,-0.38207256911448584,0.18024621077539882,0.17930840344622928,0.012722076220456045,-0.026655779355403603,-0.17160333630544702,-0.46195413078653697,-0.39063763698036297,-0.1175577427172803,0.024330080093157656,-0.3037310361275663,-0.5228276914386965,-0.31614497277706965,-0.3812932274975306,-0.3445602316918454,0.051205400845838356,0.04180800708151723,-0.28977064076953807,0.24513996175399896,0.09908420946727217,-0.3505365916680149,-0.03145915773631751,0.04705171032925606,-0.13107952331989056,0.19147057409822074,-0.4282809034961451,-0.402423505907116,-0.22170007355094215,0.3494838281641651,0.3904554948335591,0.3675445507554238,0.10553009578767057,-0.3887719181581075,-0.21472817295980193,-0.12721783722202942,0.3248239919174418,0.1384526959998043,-0.09358501642947115,0.37662829835299916,-0.3319601916611182,-0.288418604864084,0.6059055831783489,0.6080974273962876,-0.2324082678619978,0.14059707071401228,-0.14318235393399167,-0.24389648890328555,0.16400370267490802,-0.2710022595892395,-0.20655592179045473,0.4729358480531626,-0.14266825098821287,0.05865566642484543,-0.24337782129395047,-0.05300451647736637,0.518685841084375,0.03331898203732643,0.5967323277448348,0.22230281740599167,-0.1872063600565912,0.36450387449258903,-0.12907389848281042,0.7631750644733126,-0.12752633804822008,-0.15083970741422706,0.6754872696610473,0.7071266262380922,-0.18735257765301186,0.20393568443479845,0.38568413935581236,0.6156975067547878,0.6854971244142649,0.08338612134535921,0.10318978153717204,-0.14333468410704486,0.3062797680404178,0.5927514811447057,0.42237975456086707,-0.028860398448794913,0.6827397318375569,0.3951735297445964,0.563854095175715,0.04514284193777543,0.009373878838508189,0.5194145195905864,0.28022295930899976,0.2801564573376214,0.5509949637356014,0.7819712166020005,0.255714035521253,0.9561699244120252,0.7410647022979951,0.9420898103576012,0.1166040835611128,0.08740395865659883,0.3030679463946462,0.9227528465136663,0.5103494610730948,0.17956672162629067,0.5929049757904471,0.22078646179598183,0.48328515150168183,0.6787915485555689,0.4518986825859863,0.6801270497750008,0.7321748745304567,0.4097987728285502,0.9586675526373734,0.0034775428105894117,0.31325835862218593,0.953586063642097,0.1616251309657919,0.4827043735696752,0.0800679491355163,0.22864525519061782,0.4378459835114922,0.10047863578119519,0.8751071585533556,-0.008226709848372345,0.03851055778096524,-0.11219862503069382]},{\"values\":[1.3376388946435822,1.2604849454907352,0.7340727833088562,0.9510623938896172,0.45914315945042805,1.2505994651148993,0.4580278841918961,0.7814372869830223,0.7599165504569937,0.42100397380326265,0.5487657312147398,0.5249057193311099,1.0330816641921823,1.1382884530665276,0.45228756486855526,1.0833750520282908,0.7413121505577543,0.8817335679707184,1.1287961385988154,0.9631811219988388,1.2268251412422,1.156019018181579,1.3047543718666086,0.7952139698327517,0.8870540627626607,1.1411167492763714,1.2847095409397116,0.42368371547154887,0.696152453082313,0.6175192038281283,0.6295496190518529,1.0448343831539042,0.4251809327701899,0.5394483732924302,1.034163917012507,1.1802795785976192,0.9294653848386529,0.4991683491494102,0.8315425679027689,0.7657832515207501,0.444551495836414,1.0800854541934937,0.7003458456586313,0.5973605191064407,0.5984212252613947,0.984626914198181,1.022292010199704,0.8505212579974766,1.3049016794581385,0.8680493333120881,0.6198158485798927,0.40506756942447675,0.8893058934983085,0.9175243796215474,1.2183404847140908,0.8539202770214477,0.9765353095426066,0.9692570008812152,0.7126142477454529,0.9465804587052798,1.1161327314508127,1.3422464379416008,0.7030143708969,1.3097582112941983,0.5626396281286378,1.2585961976369973,0.8539427699886665,1.2467426623243132,0.7716997807602926,0.5116130746330556,0.3011979583800777,0.5389420974732798,0.9937828765272171,0.7536453031824426,1.16702349868109,0.9721348798106926,0.2856540369751773,1.0129367051920528,0.4856435621773398,1.2253845844062767,1.1884615304473984,0.8065211563790781,1.0924689957512166,0.4868541869883269,0.3698771514165262,1.0772705554912896,0.7542126297494679,0.7354761174917293,0.6138810893171056,0.4424721970577734,0.6324652092869193,0.3652419857519523,1.1078644033986724,0.9087851058702376,0.40185633876784693,0.4100912954786048,0.9207794766216676,0.2971411965003753,0.7597312498401119,0.5900675881007904,1.133038747217431,0.9851808185432546,0.9770412310080356,0.6798757768364383,1.3530555644551008,0.7961471858722102,1.0354055640396733,1.0585421639645562,0.47409095675691937,0.42713203048487935,1.0630256779276819,0.7946276035543717,0.9813169598397512,1.1834759534769486,0.44694141394091463,0.7446837108062553,0.8246437997275425,0.30990050017444787,0.7033848743767905,1.1536159908742225,0.8754668252214858,1.2620425777089792,0.31433351351265926,0.6533187220032781,0.5267114832001011,1.0705902177614253,0.786844488548579,0.9230576804583701,0.6370844686458194,1.009498473293921,1.1215359884263907,0.46527269295837614,0.8676259087821636,0.8216187977021256,0.7186165676094508,1.233209973114213,1.200476691450291,0.5674854709728557,0.8711730817122926,0.30688584101074207,0.4666336904542068,0.5634903271883633,1.0846785872069598,0.5266730940055697,0.39214196342982993,0.28402388373371645,0.37222749230067403,0.19040726479378614,0.38409429318867533,0.5000351433368264,0.0742731460581325,0.16794504775758873,0.40460421238510974,0.1760477301139245,0.9258956359695438,0.3966153830812908,0.5204504455787337,0.30734839787305646,0.5185508189096878,0.5811268694573454,0.21268960240444199,0.9052666397246012,0.6920385091524559,0.13624708269187938,0.24471018904054562,0.8575780112612371,0.6883004110528397,0.696898362851129,0.8987616818140296,0.4388981543941003,0.15597520694877873,0.8227770102566818,0.23664960622154607,0.8707046358064225,0.28049676714468225,0.23366427619826807,0.5807317061097298,0.0023836232988781528,0.29111066405822394,-0.03252383348453493,0.716697966112411,0.31824326814800935,0.6475452275030713,0.5045633182813123,0.8494340503224529,0.5413964870795914,0.6010934759215731,0.3390243309596059,-0.10449504382861649,0.5388603786187213,0.7260482898796399,0.02066031708671977,0.006592712524608087,0.42460643061721565,0.2654894948377608,0.2722803097380025,0.42712061675078566,0.37709450650474974,0.4445009119020217,-0.17394927770523044,-0.06765577377084932,-0.21363219316288362,0.26379659209538625,0.17974782612567078,0.6112585659183439,-0.026877104926779073,0.39701554936855177,0.41666193276660674,0.46644801461311314,0.0904053224614549,0.1744176335477905,-0.018664944495069325,0.06395609553993285,0.06021945505038828,-0.16817365663554248,0.2483276738552259,0.1239184493730836,-0.004546187841126176,0.23945127404977104,0.627997803952128,0.912820723083692,0.830286085334825,0.7729436448391438,0.5222366543046882,0.446819463618149,-0.11984438278816445]},{\"values\":[0.9988801123799705,1.1523793162627445,0.9170730282816685,0.9627898466211474,1.2668103928385919,0.451852517897642,1.0287778591572525,0.5897926708220181,1.1391757284055675,0.38314116303743284,0.8731757293116799,0.9666331895807522,0.6093011923621586,1.3239207733148604,1.114721422992712,1.3207650343493145,0.5720715767918324,0.7264662970872482,1.3493685170919039,0.37752796436221947,0.9144576733921752,1.3442985351101255,0.5421357972616513,1.0721951023617209,1.05113371352527,1.079946844116939,1.0359717387667682,1.2944724530307872,0.9237482154123343,0.712529825278661,0.9153240137533609,0.6378995051135369,0.8921084626048987,0.49206309537710025,0.8114490370418256,1.2629433714146752,0.9927645108140478,1.3164330650113396,1.0131704086129019,0.8675071724368478,0.6389979227347703,0.8782599330769915,0.6346502804051994,0.6431528086681431,1.3116500733118979,0.8770018967538074,0.9841204641940596,0.5102555743720047,0.8925276141510102,0.842771696911364,0.6309877507458378,0.5199269980693476,0.7895658718970219,1.2768750612853292,1.3482843668389364,0.9838702084138937,1.0273766190428164,0.9818901564237827,1.3115954220934323,0.6643168573622604,0.5441458327515154,0.8083738623927699,0.9562090959578922,1.3551612232540289,1.1537549247350152,0.7930255758681299,0.614604902239428,0.9339330583049925,0.6040144809780428,0.7489390886758316,0.6619632777658401,0.8666006600309958,0.5318160081447949,0.3265151816402634,0.44857565498107643,0.41380173731324177,1.2335357255405843,0.6521855238832913,0.7529033451014452,0.944477406061948,0.7805756832299058,0.43587754649879196,0.3836398639353667,0.646433819953051,0.8786733226540886,0.8382640761946089,0.9255057727028182,1.0230288654069981,0.5007289830619284,1.1335605365833734,1.1352724684511555,0.895055475760088,0.6253547032582889,0.3185327900354522,1.2556736240838997,0.28178151647779454,0.7612033526472936,0.9574882488562941,0.7932082494954589,0.29383125010503897,0.9674941712752773,0.8803172140267721,0.9084284237618007,1.255834574539272,0.3846957198583525,0.4532810167804091,0.5642289785397847,0.3942170569867137,0.5874797288148221,0.588143307233913,1.2841313924185338,0.8630904046661595,1.2660421955479162,0.943700358346157,0.5838416138636763,0.3648042663867235,0.7848301321977057,1.0675800488073304,0.3340426841271711,0.615180050303554,0.6958012850930712,1.0725993192919279,0.2886012090458607,0.6898307015822732,1.2667674235769422,0.7163230397006717,0.5448430184019213,0.4759296658600434,0.5877557346676399,0.9492089981677492,1.0156113916508376,0.5874029051615719,0.5934656283981462,0.386055117032463,1.084235774157545,1.0650241215085074,0.6375958326378981,0.6395262255054593,0.6800724049210243,0.6167265384603008,0.3736465793463072,0.7370969963540629,0.6551842363648263,0.47087612945822077,0.13690358035364755,1.0165233161888687,0.6254073121703162,0.9249461926884713,0.2958427605495389,0.4399379888181976,0.12113545698584048,0.9959712988791973,1.0177732145853997,0.1084576505587473,0.39684215332976364,0.08642224034915208,0.689176568432026,0.6855585998526866,0.6668188214615476,0.06691353049954797,0.14095509319529478,0.5752554084377758,0.5264056459760803,0.49406530181046276,0.730011996056528,0.024648935084444174,0.29921722440159554,0.5037545682038607,0.2240461714875178,0.5376260733918543,0.3533489355153815,0.17339280448269984,0.7573381951506568,0.6267751642355364,0.6900535762515746,0.2757546106185449,0.27231157307783355,0.663981722115052,0.713527977280504,0.0787397200847037,0.61064649505243,0.19573826206826628,0.2752859341802351,0.2312559613314725,0.4187692756832248,0.12117345945137788,0.32171970206380485,0.29649535495708906,0.49859343221271873,0.05884070053696282,0.686371974580693,0.12034445616225871,-0.09145121549405968,0.4969728192955692,-0.23305782389034976,-0.04755303242802253,0.14853859243729464,0.20416939033701348,0.6676126805973585,0.1355802090769266,0.4161199247968954,-0.23532683722571482,-0.02117699415802574,0.14315391712636374,0.6673981948799745,0.4277117228085578,0.2529887969757269,0.450370986411845,-0.20386392892204444,0.13367771986312593,0.47683809501309554,0.3739413418238921,0.031228795158564956,0.4716376988020073,0.7651357400588652,0.1403359755906192,0.4758606702377602,0.7478972581856711,-0.1482387022025264,0.46085670296927445,0.2243049620213363,0.6799844627153679,0.3300447210021927,0.001957151224394913,-0.03053106423639284,-0.12930818126061747]},{\"values\":[0.4308303869440026,1.0578845607284835,0.98379309142267,0.8028758040048155,1.2204100278012167,0.9554225813238064,0.39143947413634805,0.5103594353029558,0.3607707080233586,1.2810335971673006,1.0229719178768701,0.8031765747917948,0.9526642023283562,0.48491518250481025,0.42162405037203593,0.7057468058446643,1.327690109599803,0.39629276023987814,1.1212558164427489,0.811791034758602,1.2038964270866304,1.2133123647395445,1.153282460593106,0.39938492691438543,1.0638881212049658,0.6098799520834124,0.6879428093542774,1.2655585743694184,0.42012538673851046,1.3091090347700856,1.1257449509558994,0.5021994080447046,0.6588255698954434,1.2570452448406761,1.1242957273670913,0.6909125341159095,0.4945844750051669,0.6452060404901364,1.1958486578600653,0.4676608165420215,1.0483554628239744,1.3413337492035522,1.006074401649311,1.1027319201206807,1.244498680124209,1.082511127977898,0.8597922843213235,1.0434972147936532,1.353045408998519,0.3914682303446752,0.6244236081787681,0.6681458098663369,1.2276783783860623,0.41348270087994593,0.6107317932425274,0.46241973710580153,1.1761569926807627,0.6656428808549342,0.6633228765527005,0.3947051777092203,0.7006129768189021,0.54228125299298,0.6498456493186664,1.043932309665939,1.2248176151754888,0.8431152412692442,0.8472681491494176,0.9189284379987788,0.6952542179792817,0.6089312723459283,0.9678586177460307,1.2893525479054262,0.4027995800282717,0.3929768613242474,0.31958672131073584,1.279396603277529,1.1565450591193376,1.0450607905410192,1.1952557314086871,0.5794070347095253,0.9376344261614152,0.6627143031754927,1.0338271708918358,0.4036673168791957,0.45277723835498923,0.8819356455996253,0.33335598319476833,1.0478960392977832,1.2601147063844842,1.1879917628055792,0.38695290052162484,1.0750071698326018,0.5510087825525486,0.3897641220514452,0.878929580797806,1.1409323911143423,1.1141047971920028,0.8347776873988076,0.6773909559022357,1.1523655553061718,1.1724004643890253,0.49246637128805515,1.162246616925141,0.8287391538750103,1.058272628485441,0.6392359197264057,0.4515889527266461,1.3393030016529313,0.5430521254666156,1.1808104315980301,0.7514414218559015,0.5958644081686632,0.3699542432345308,0.4073262922361873,1.3281685810381978,1.2232202446946896,1.0321370844149154,1.1247800158019468,0.9973752444362415,0.5717085868420492,0.340828027991917,0.9627035217705663,1.1775135151982266,0.3701331753260913,0.4130279836564781,0.9597379835695232,0.6789712092136387,0.8183318342144945,0.392712482904956,0.2906723641624834,0.41452397990787165,0.5515987640281749,0.7200842671208157,0.4303716157939223,0.9784408829860567,1.1094808449275193,1.0172021640312532,1.1785458805274747,0.9388240070805424,0.282690678194361,1.0582718664092308,0.2504431721101725,1.0274276580972563,0.35492010423825604,0.5684040003040949,0.48639171344533016,0.3809903231906007,0.29048714439767026,0.6596784620002609,0.6795088660116164,0.4808168660580935,0.7484038294410144,0.48709516689199883,0.003549457912870209,0.6227106701235123,0.528836210649946,0.036534025000147685,0.07120779641228707,0.43273680192911906,0.06259364016304304,0.31234353362915973,0.5703257170740976,0.14920639869200517,0.5524242957886825,0.3649778648353045,0.7005854073569081,0.6060573812813136,0.3055967495718754,0.5519869963949788,0.4056116381504813,0.6921892017530693,-0.006718571120892643,0.12589324954301312,0.20718041483265262,-0.049219280244284264,0.4444782538360571,0.31674519916088967,0.4783378578508317,0.3221764175525444,0.5888555390513546,0.7757246366921298,-0.0454651674002228,0.7906889121118396,0.3831428741627556,0.7685413609048828,0.8121643622603081,0.30167712202713554,0.13647164982313173,0.1594295650882566,0.31768456625263647,0.07603636491180069,0.12176216325347423,0.2534090903910003,0.49736557342000226,0.30759967567089336,-0.09279258321992703,-0.059522906176246226,0.15432102786828975,0.49668338280573626,0.46976479951874794,-0.23678814284919716,0.5555459876686415,-0.2037572214342855,0.37976353560882903,0.1623920797841313,-0.1507576631429631,0.5310835080710888,0.5569900543780778,0.39386937836019215,0.469166472259886,0.42965329572003597,-0.17143932883413118,0.4707230922476272,-0.028133723347139667,0.46317428850047976,0.4080764909533945,0.21190954209118487,0.10437148554884605,0.0437346694042104,0.7176660184988558,0.6667656821842307,-0.004240513410398324,0.8195385698136652,0.10890015278393926,0.35061355945978173,-0.13480991606694653]}]},{\"input_features\":5,\"output_features\":3,\"activation\":\"relu_activation\",\"neurons\":[{\"values\":[0.2348036272600923,-0.3299689247097655,0.23793968870139337,0.5070892228841078,0.2582554535941222,-0.0044580216026254265]},{\"values\":[0.414529829136085,0.3907555540861243,-0.14508626930087487,-0.2526643919423257,0.367120570863409,-0.006841082274194994]},{\"values\":[0.08085079988393293,0.9425247466979466,0.9732862456079628,0.48895767324099093,0.33946258863878126,-0.029212679285622514]}]},{\"input_features\":3,\"output_features\":1,\"activation\":\"sigmoid_activation\",\"neurons\":[{\"values\":[0.3494380481876106,0.2062983608303127,-0.27150807557641116,-0.07016452452704906]}]}]}\n"
     ]
    }
   ],
   "source": [
    "# Write the model to a file\n",
    "import json\n",
    "\n",
    "exported_model = model.export()\n",
    "model_json = exported_model.model_dump_json()\n",
    "print(model_json)\n",
    "\n",
    "# Write the model to a file\n",
    "with open('model_weights.json', 'w') as f:\n",
    "    f.write(model_json)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Importing model...\n",
      "Imported model!\n",
      "Model successfully imported and validated.\n"
     ]
    }
   ],
   "source": [
    "from micrograd.nets.NeuralNet import NeuralNet\n",
    "from micrograd.schemas.schemas import ModelSchema\n",
    "\n",
    "# Read the model from the file\n",
    "with open('model_weights.json', 'r') as f:\n",
    "    model_json = f.read()\n",
    "\n",
    "# Parse and validate the JSON using pydantic\n",
    "model_schema = ModelSchema.model_validate_json(model_json)\n",
    "\n",
    "# Create a new NeuralNet instance from the validated schema\n",
    "imported_model = NeuralNet.from_json(model_schema)\n",
    "\n",
    "print(\"Model successfully imported and validated.\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Val(value=0.7960, grad=1.0000, parents=(159.19965853657683 * 0.005))\n",
      "0.7959982926828841\n",
      "Loss: 0.7960\n",
      "Epoch [], Loss: 0.7960\n"
     ]
    }
   ],
   "source": [
    "print(loss)\n",
    "print(loss.value)\n",
    "print(f'Loss: {loss.value:.4f}')\n",
    "print(f'Epoch [], Loss: {loss.value:.4f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Correct guesses: 23, Wrong guesses: 27, Total: 50 - Accuracy: 46.0%\n",
      "Loss: Val(value=0.7960, grad=1.0000, parents=(159.19965853657683 * 0.005))\n"
     ]
    }
   ],
   "source": [
    "# Test the model\n",
    "n = len(x_valid)\n",
    "correct_guesses = 0\n",
    "wrong_guesses = 0\n",
    "\n",
    "y_pred = [model(xi) for xi in x_valid]\n",
    "\n",
    "predictions = []\n",
    "\n",
    "for pred in y_pred:\n",
    "    if pred.value <= 0.5:\n",
    "        predictions.append(0)\n",
    "    else:\n",
    "        predictions.append(1)\n",
    "    \n",
    "for i in range(n):\n",
    "    # print(f'Prediction: {predictions[i]} with confidence {y_pred[i].value:.2f} for actual: {y_valid[i].value}')\n",
    "\n",
    "    if predictions[i] == y_valid[i].value:\n",
    "        correct_guesses += 1\n",
    "    else:\n",
    "        wrong_guesses += 1\n",
    "\n",
    "print(f\"Correct guesses: {correct_guesses}, Wrong guesses: {wrong_guesses}, Total: {n} - Accuracy: {(correct_guesses/n)*100}%\")\n",
    "print(f\"Loss: {loss}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Correct guesses: 23, Wrong guesses: 27, Total: 50 - Accuracy: 46.0%\n",
      "Loss: Val(value=0.7960, grad=1.0000, parents=(159.19965853657683 * 0.005))\n"
     ]
    }
   ],
   "source": [
    "# Testing the imported model\n",
    "# Test the model\n",
    "n = len(x_valid)\n",
    "correct_guesses = 0\n",
    "wrong_guesses = 0\n",
    "\n",
    "y_pred = [imported_model(xi) for xi in x_valid]\n",
    "\n",
    "predictions = []\n",
    "\n",
    "for pred in y_pred:\n",
    "    if pred.value <= 0.5:\n",
    "        predictions.append(0)\n",
    "    else:\n",
    "        predictions.append(1)\n",
    "    \n",
    "for i in range(n):\n",
    "    # print(f'Prediction: {predictions[i]} with confidence {y_pred[i].value:.2f} for actual: {y_valid[i].value}')\n",
    "\n",
    "    if predictions[i] == y_valid[i].value:\n",
    "        correct_guesses += 1\n",
    "    else:\n",
    "        wrong_guesses += 1\n",
    "\n",
    "print(f\"Correct guesses: {correct_guesses}, Wrong guesses: {wrong_guesses}, Total: {n} - Accuracy: {(correct_guesses/n)*100}%\")\n",
    "print(f\"Loss: {loss}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Val(value=4.6052, grad=0.0000, parents=(2.3025850929940455 + 2.3025850929940455))\n"
     ]
    }
   ],
   "source": [
    "y_pred = Value(0.1)\n",
    "y = Value(1.0)\n",
    "\n",
    "loss_test = Value(0.0)\n",
    "loss_test += (-(y * y_pred.log()) - ((Value(1.0)-y)*(Value(1.0)-y_pred).log()))\n",
    "loss_test += (-(y * y_pred.log()) - ((Value(1.0)-y)*(Value(1.0)-y_pred).log()))\n",
    "\n",
    "\n",
    "print(loss_test)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "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": "3.12.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}