Project_Notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "fb8a552a",
   "metadata": {},
   "source": [
    "# Bandgap Prediction"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb6e6739",
   "metadata": {},
   "source": [
    "Using QM9 data set to predict the bandgap of molecules using LSTM and GNN's"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "673c6786",
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch_geometric.datasets import QM9\n",
    "from torch_geometric.loader import DataLoader as GraphDataLoader\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from torch.utils.data import Dataset\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm import tqdm\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "from rdkit import RDLogger\n",
    "from torch_geometric.nn import GCNConv, global_mean_pool\n",
    "import torch.nn.functional as F"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "fdbb1c28",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              SMILES                                        Coordinates  \\\n",
      "0  [H]C([H])([H])[H]  [[tensor(-0.0127), tensor(1.0858), tensor(0.00...   \n",
      "1       [H]N([H])[H]  [[tensor(-0.0404), tensor(1.0241), tensor(0.06...   \n",
      "2            [H]O[H]  [[tensor(-0.0344), tensor(0.9775), tensor(0.00...   \n",
      "3          [H]C#C[H]  [[tensor(0.5995), tensor(0.), tensor(1.)], [te...   \n",
      "4             [H]C#N  [[tensor(-0.0133), tensor(1.1325), tensor(0.00...   \n",
      "\n",
      "        HOMO      LUMO       Egap  \n",
      "0 -10.549854  3.186453  13.736308  \n",
      "1  -6.993326  2.255824   9.249149  \n",
      "2  -7.967494  1.869422   9.836916  \n",
      "3  -7.741639  1.376896   9.118535  \n",
      "4  -9.806983  0.519737  10.329442  \n"
     ]
    }
   ],
   "source": [
    "def load_qm9(path=\"./QM9\"):\n",
    "    return QM9(path)\n",
    "\n",
    "qm9 = load_qm9()\n",
    "smiles = []\n",
    "coords = []\n",
    "homo_lumo_gaps = []\n",
    "stock_gap = []\n",
    "homo = []\n",
    "lumo = []\n",
    "\n",
    "for entry in qm9:\n",
    "    smiles.append(entry.smiles)  # Get SMILES notation\n",
    "    coords.append(entry.pos)     # Get atomic coordinates\n",
    "    homo_energy = entry.y[0, 2].item()  # Get HOMO energy from the tensor\n",
    "    homo.append(homo_energy)\n",
    "    lumo_energy = entry.y[0, 3].item()  # Get LUMO energy from the tensor\n",
    "    lumo.append(lumo_energy)\n",
    "    gap_stock = entry.y[0, 4].item() \n",
    "    gap = lumo_energy - homo_energy  # Calculate the bandgap\n",
    "    stock_gap.append(gap_stock)\n",
    "    homo_lumo_gaps.append(gap)  # Store the bandgap\n",
    "# Creating a DataFrame\n",
    "qm9_data = pd.DataFrame({\n",
    "    \"SMILES\": smiles,\n",
    "    \"Coordinates\": coords,\n",
    "    \"HOMO\" : homo,\n",
    "    \"LUMO\" : lumo,\n",
    "    \"Egap\" : stock_gap\n",
    "})\n",
    "\n",
    "print(qm9_data.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "b5d4b280",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'HOMO-LUMO gap')"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(np.arange(len(qm9_data[\"Egap\"])), qm9_data[\"Egap\"])\n",
    "plt.title(\"Visualization of HOMO-LUMO Gap Values\")\n",
    "plt.xlabel(\"Index\")\n",
    "plt.ylabel(\"HOMO-LUMO gap\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "5acfdfcf",
   "metadata": {},
   "outputs": [],
   "source": [
    "def pad_start_end_token(smiles):\n",
    "    \"\"\"\n",
    "    Pad a list of SMILES with \"SOS\" and \"EOS\" token\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    smiles: list of str\n",
    "        A list containing SMILES strings to pad\n",
    "    \n",
    "    Returns\n",
    "    ------\n",
    "    padded: list of list of str\n",
    "        A list containing padded SMILES strings. Example: [['SOS', 'C', 'EOS'], ...]\n",
    "    \"\"\"\n",
    "    padded = []\n",
    "    for smile in smiles:\n",
    "        padded.append(['SOS'] + list(smile) + ['EOS'])\n",
    "    \n",
    "    return padded"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b510a1f0",
   "metadata": {},
   "outputs": [],
   "source": [
    "padded_smiles = pad_start_end_token(qm9_data[\"SMILES\"])\n",
    "vocab = np.unique(np.concatenate(padded_smiles))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "a74099d3",
   "metadata": {},
   "outputs": [],
   "source": [
    "class SmilesDataset(Dataset):\n",
    "    def __init__(self, smiles, vocab, y):\n",
    "        \n",
    "        self.vocab = np.array(vocab, dtype=str).reshape(-1, 1)\n",
    "        \n",
    "        self.encoder = OneHotEncoder()\n",
    "        # fit the encoder\n",
    "        self.encoder.fit(self.vocab)\n",
    "\n",
    "        self.y = torch.tensor(y)\n",
    "        \n",
    "        # one-hot encoding\n",
    "        self.data = [\n",
    "            torch.tensor(\n",
    "                self.encoder.transform(np.array(s).reshape(-1, 1)).toarray(),\n",
    "                dtype=torch.float\n",
    "            ) for s in smiles\n",
    "        ]\n",
    "        \n",
    "        self.data = nn.utils.rnn.pad_sequence(self.data, batch_first=True)\n",
    "        self.X = self.data[:, :-1, :]\n",
    "        \n",
    "    def __len__(self):\n",
    "        return int(self.data.shape[0])\n",
    "    \n",
    "    def get_data(self):\n",
    "        return self.X\n",
    "\n",
    "    def get_labels(self):\n",
    "        return self.y\n",
    "    \n",
    "    def __getitem__(self, idx):\n",
    "        return self.X[idx], self.y[idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ac98ebd2",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_homo = SmilesDataset(padded_smiles, vocab, qm9_data[\"HOMO\"])\n",
    "data_homo_lite = SmilesDataset(padded_smiles[:50000], vocab, qm9_data[\"HOMO\"][:50000])\n",
    "\n",
    "data_lumo = SmilesDataset(padded_smiles, vocab, qm9_data[\"LUMO\"])\n",
    "data_lumo_lite = SmilesDataset(padded_smiles[:50000], vocab, qm9_data[\"LUMO\"][:50000])\n",
    "input_size = data_homo.vocab.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "9647fd4e",
   "metadata": {},
   "outputs": [],
   "source": [
    "val_data_homo = SmilesDataset(padded_smiles[50000:60000], vocab, qm9_data[\"HOMO\"].values[50000:60000])\n",
    "val_data_lumo = SmilesDataset(padded_smiles[50000:60000], vocab, qm9_data[\"LUMO\"].values[50000:60000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "3ad64ecf",
   "metadata": {},
   "outputs": [],
   "source": [
    "RDLogger.DisableLog(\"rdApp.*\")\n",
    "\n",
    "class Trainer:\n",
    "    def __init__(self, model, opt_method, learning_rate, batch_size, epoch, l2):\n",
    "        self.model = model\n",
    "        if opt_method == \"sgdm\":\n",
    "            self.optimizer = torch.optim.SGD(model.parameters(), learning_rate, momentum=0.9)\n",
    "        elif opt_method == \"adam\":\n",
    "            self.optimizer = torch.optim.Adam(model.parameters(), learning_rate, weight_decay=l2)\n",
    "        else:\n",
    "            raise NotImplementedError(\"This optimization is not supported\")\n",
    "\n",
    "        self.epoch = epoch\n",
    "        self.batch_size = batch_size\n",
    "\n",
    "    def train(self, train_data, draw_curve=True):\n",
    "        self.encoder = train_data.encoder\n",
    "\n",
    "        train_loader = DataLoader(train_data, batch_size=self.batch_size, shuffle=True)\n",
    "        train_loss_list, train_acc_list = [], []\n",
    "\n",
    "        loss_func = nn.L1Loss()\n",
    "        for n in tqdm(range(self.epoch), leave=False):\n",
    "            self.model.train()\n",
    "            epoch_loss, epoch_err = 0.0, 0.0\n",
    "            for X_batch, y_batch in train_loader:\n",
    "                batch_importance = y_batch.shape[0] / len(train_data)\n",
    "                y_pred, _ = self.model(X_batch)\n",
    "                batch_loss = loss_func(y_pred.flatten(), y_batch.float())\n",
    "                self.optimizer.zero_grad()\n",
    "                batch_loss.backward()\n",
    "                self.optimizer.step()\n",
    "\n",
    "                batch_err = torch.mean(y_batch.float() ** 2 - y_pred.flatten() ** 2)\n",
    "\n",
    "                epoch_err += batch_err.detach().cpu().item() * batch_importance\n",
    "                epoch_loss += batch_loss.detach().cpu().item() * batch_importance\n",
    "\n",
    "            train_acc_list.append(epoch_err)\n",
    "            train_loss_list.append(epoch_loss)\n",
    "\n",
    "        if draw_curve:\n",
    "            x_axis = np.arange(self.epoch)\n",
    "            fig, axes = plt.subplots(1, 2, figsize=(10, 4))\n",
    "            axes[0].plot(x_axis, train_loss_list, label=\"Train\")\n",
    "            axes[0].set_title(\"Loss\")\n",
    "            axes[0].legend()\n",
    "            axes[1].plot(x_axis, train_acc_list, label='Train')\n",
    "            axes[1].set_title(\"Error\")\n",
    "            axes[1].legend()\n",
    "            return {\n",
    "                \"train_accs\": train_acc_list,\n",
    "                \"train_losses\": train_loss_list\n",
    "            }\n",
    "        else:\n",
    "            return {\n",
    "                \"train_accs\": train_acc_list,\n",
    "                \"train_losses\": train_loss_list\n",
    "            }\n",
    "\n",
    "    def sample(self, num_seq=10):\n",
    "        self.model.eval()\n",
    "        seqs = []\n",
    "        with torch.no_grad():\n",
    "            for _ in tqdm(range(num_seq), leave=False):\n",
    "                chars = ['SOS']\n",
    "                hidden = self.model.init_hidden(1)\n",
    "                while chars[-1] != 'EOS':\n",
    "                    input_encoding = self.encoder.transform(np.array([chars[-1]]).reshape(-1, 1)).toarray()\n",
    "                    input_encoding = torch.tensor(input_encoding, dtype=torch.float).reshape(1, 1, -1)\n",
    "                    out, hidden = self.model(input_encoding, hidden)\n",
    "\n",
    "                    prob = out.detach().numpy().flatten()\n",
    "                    prob /= np.sum(prob)\n",
    "\n",
    "                    index = np.random.choice(self.model.input_size, p=prob)\n",
    "                    out_encoding = np.zeros((1, self.model.input_size))\n",
    "                    out_encoding[0, index] = 1.0\n",
    "                    char = data.encoder.inverse_transform(out_encoding).flatten().tolist()[0]\n",
    "                    chars.append(char)\n",
    "                seqs.append(''.join(chars[1:-1]))\n",
    "        return seqs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "05161df0",
   "metadata": {},
   "outputs": [],
   "source": [
    "class LSTM(nn.Module):\n",
    "    def __init__(self, input_size, output_size, hidden_size, num_layers=1):\n",
    "        super().__init__()\n",
    "\n",
    "        self.input_size = input_size\n",
    "        self.hidden_size = hidden_size\n",
    "        self.output_size = output_size\n",
    "        self.num_layers = num_layers\n",
    "\n",
    "        self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True)\n",
    "        self.fc = nn.Linear(hidden_size, output_size)\n",
    "\n",
    "    def forward(self, x, h=None):\n",
    "        out, h = self.lstm(x, h)\n",
    "        out = out[:, -1, :]\n",
    "\n",
    "        out = self.fc(out)\n",
    "        return out, h\n",
    "\n",
    "    def predict(self, x):\n",
    "        # Check if the input tensor has the right number of dimensions\n",
    "        if x.ndim != 3:\n",
    "            raise ValueError(f\"Expected input to be a 3D tensor, got {x.ndim}D tensor instead.\")\n",
    "\n",
    "        self.eval()  \n",
    "        with torch.no_grad():  \n",
    "            h = self.init_hidden(x.size(0))\n",
    "            out, _ = self.forward(x, h)\n",
    "            out = out.squeeze()  # Remove the batch dimension\n",
    "\n",
    "            return out\n",
    "\n",
    "    def init_hidden(self, batch_size):\n",
    "        return (torch.zeros(self.num_layers, batch_size, self.hidden_size), \n",
    "                torch.zeros(self.num_layers, batch_size, self.hidden_size))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "49911950",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained HOMO model with adam optimizer, hidden size of 16, 0.001 learning rate, 128 batch size, and 5 epochs\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained LUMO model with adam optimizer, hidden size of 16, 0.001 learning rate, 128 batch size, and 5 epochs\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                              \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained HOMO model with adam optimizer, hidden size of 32, 1e-05 learning rate, 64 batch size, and 5 epochs\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained LUMO model with adam optimizer, hidden size of 32, 1e-05 learning rate, 64 batch size, and 5 epochs\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained HOMO model with sgdm optimizer, hidden size of 64, 0.001 learning rate, 128 batch size, and 5 epochs\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained LUMO model with sgdm optimizer, hidden size of 64, 0.001 learning rate, 128 batch size, and 5 epochs\n"
     ]
    }
   ],
   "source": [
    "model_homo = LSTM(input_size, 1, 32, 2)\n",
    "model_lumo = LSTM(input_size, 1, 32, 2)\n",
    "trainer_homo = Trainer(model_homo, \"adam\", 1e-3, 128, 5, 1e-5)\n",
    "trainer_lumo = Trainer(model_lumo, \"adam\", 1e-3, 128, 5, 1e-5)\n",
    "\n",
    "training_args = [\n",
    "    {\n",
    "        \"hidden_size\": 16,\n",
    "        \"optim_meth\": \"adam\",\n",
    "        \"learn_rate\": 1e-3,\n",
    "        \"batch_size\": 128,\n",
    "        \"epochs\": 5\n",
    "    },\n",
    "    {\n",
    "        \"hidden_size\": 32,\n",
    "        \"optim_meth\": \"adam\",\n",
    "        \"learn_rate\": 1e-5,\n",
    "        \"batch_size\": 64,\n",
    "        \"epochs\": 5\n",
    "    },\n",
    "    {\n",
    "        \"hidden_size\": 64,\n",
    "        \"optim_meth\": \"sgdm\",\n",
    "        \"learn_rate\": 1e-3,\n",
    "        \"batch_size\": 128,\n",
    "        \"epochs\": 5\n",
    "    }\n",
    "]\n",
    "\n",
    "train_losses_homo = []\n",
    "train_losses_lumo = []\n",
    "train_accs_homo = []\n",
    "train_accs_lumo = []\n",
    "val_gap_errs = []\n",
    "\n",
    "for trial in range(3):\n",
    "    curr_args = training_args[trial]\n",
    "    curr_model_homo = LSTM(input_size, 1, curr_args[\"hidden_size\"], 2)\n",
    "    curr_model_lumo = LSTM(input_size, 1, curr_args[\"hidden_size\"], 2)\n",
    "\n",
    "    curr_trainer_homo = Trainer(curr_model_homo, curr_args[\"optim_meth\"], \n",
    "                                curr_args[\"learn_rate\"], curr_args[\"batch_size\"], \n",
    "                                curr_args[\"epochs\"], 1e-5)\n",
    "    curr_trainer_lumo = Trainer(curr_model_lumo, curr_args[\"optim_meth\"], \n",
    "                                curr_args[\"learn_rate\"], curr_args[\"batch_size\"], \n",
    "                                curr_args[\"epochs\"], 1e-5)\n",
    "    \n",
    "    homo_out = curr_trainer_homo.train(data_homo_lite, False)\n",
    "    train_losses_homo.append(homo_out[\"train_losses\"])\n",
    "    train_accs_homo.append(homo_out[\"train_accs\"])\n",
    "\n",
    "    print(\"Trained HOMO model with {} optimizer, hidden size of {}, {} learning rate, {} batch size, and {} epochs\".format(\n",
    "        curr_args[\"optim_meth\"], curr_args[\"hidden_size\"], \n",
    "        curr_args[\"learn_rate\"], curr_args[\"batch_size\"], \n",
    "        curr_args[\"epochs\"]))\n",
    "\n",
    "    lumo_out = curr_trainer_lumo.train(data_lumo_lite, False)\n",
    "    train_losses_lumo.append(lumo_out[\"train_losses\"])\n",
    "    train_accs_lumo.append(lumo_out[\"train_accs\"])\n",
    "\n",
    "    print(\"Trained LUMO model with {} optimizer, hidden size of {}, {} learning rate, {} batch size, and {} epochs\".format(\n",
    "        curr_args[\"optim_meth\"], curr_args[\"hidden_size\"], \n",
    "        curr_args[\"learn_rate\"], curr_args[\"batch_size\"], \n",
    "        curr_args[\"epochs\"]))\n",
    "    \n",
    "    homo_preds = curr_model_homo.predict(val_data_homo.get_data())\n",
    "    lumo_preds = curr_model_lumo.predict(val_data_lumo.get_data())\n",
    "\n",
    "    targets = torch.sub(val_data_homo.get_labels(), val_data_lumo.get_labels())\n",
    "    preds = torch.sub(homo_preds, lumo_preds)\n",
    "\n",
    "    error = torch.mean((preds - targets) ** 2)\n",
    "    val_gap_errs.append(error)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "962bb0c3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'hidden_size': 16, 'optim_meth': 'adam', 'learn_rate': 0.001, 'batch_size': 128, 'epochs': 5}\n",
      "validation error: 0.8942833421693475\n",
      "\n",
      "\n",
      "{'hidden_size': 32, 'optim_meth': 'adam', 'learn_rate': 1e-05, 'batch_size': 64, 'epochs': 5}\n",
      "validation error: 9.554098086468183\n",
      "\n",
      "\n",
      "{'hidden_size': 64, 'optim_meth': 'sgdm', 'learn_rate': 0.001, 'batch_size': 128, 'epochs': 5}\n",
      "validation error: 1.3018904304633623\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for trial in range(3):\n",
    "    print(training_args[trial])\n",
    "    print(\"validation error:\", val_gap_errs[trial].item())\n",
    "    print(\"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "ce00b48a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Training Loss')"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.arange(5), train_accs_lumo[1])\n",
    "plt.title(\"Training Error VS Epoch HOMO\")\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Training Loss\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96f00862",
   "metadata": {},
   "source": [
    "We can see from the above results an LSTM with a hidden size of 16 trained using ADAM for 5 epochs with a batch size of 64 performs the best, achieveing the lowest error."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5bd1fa0",
   "metadata": {},
   "source": [
    "# GNN"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "302ab9ad",
   "metadata": {},
   "source": [
    "We got the code to process the SMILE strings into a graph from here: https://www.blopig.com/blog/2022/02/how-to-turn-a-smiles-string-into-a-molecular-graph-for-pytorch-geometric/\\"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "03ab348c",
   "metadata": {},
   "outputs": [],
   "source": [
    "gnn_data = []\n",
    "\n",
    "target_idx = 4\n",
    "for data in qm9:\n",
    "    data.y = data.y[:, target_idx]\n",
    "    gnn_data.append(data)\n",
    "\n",
    "\n",
    "# Split the dataset into training and testing sets\n",
    "train_dataset = gnn_data[:50000]\n",
    "test_dataset = gnn_data[50000:60000]\n",
    "\n",
    "train_loader = GraphDataLoader(train_dataset, batch_size=64, shuffle=True)\n",
    "test_loader = GraphDataLoader(test_dataset, batch_size=64, shuffle=False)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "8b1f43a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "class GNN(torch.nn.Module):\n",
    "    def __init__(self, hidden_channels):\n",
    "        super(GNN, self).__init__()\n",
    "        self.conv1 = GCNConv(qm9.num_node_features, hidden_channels)\n",
    "        self.conv2 = GCNConv(hidden_channels, hidden_channels)\n",
    "        self.lin = torch.nn.Linear(hidden_channels, 1)  # Outputs one value per graph\n",
    "\n",
    "    def forward(self, data):\n",
    "        x, edge_index, batch = data.x, data.edge_index, data.batch\n",
    "        x = F.relu(self.conv1(x, edge_index))\n",
    "        x = F.relu(self.conv2(x, edge_index))\n",
    "        x = global_mean_pool(x, batch)  # Global pooling\n",
    "        x = self.lin(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "270e8d7a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1, Loss: 1.1845\n",
      "Epoch: 2, Loss: 0.8455\n",
      "Epoch: 3, Loss: 0.8143\n",
      "Epoch: 4, Loss: 0.7945\n",
      "Epoch: 5, Loss: 0.7855\n",
      "Epoch: 6, Loss: 0.7806\n",
      "Epoch: 7, Loss: 0.7790\n",
      "Epoch: 8, Loss: 0.7732\n",
      "Epoch: 9, Loss: 0.7683\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = GNN(hidden_channels=64)\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.001)\n",
    "loss_fn = torch.nn.L1Loss()\n",
    "\n",
    "def train():\n",
    "    model.train()\n",
    "    total_loss = 0\n",
    "    for data in train_loader:\n",
    "        optimizer.zero_grad()\n",
    "        pred = model(data)\n",
    "        target = data.y.view(-1, 1).float()\n",
    "        loss = loss_fn(pred, target)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        total_loss += loss.item()\n",
    "    return total_loss / len(train_loader)\n",
    "\n",
    "losses = []\n",
    "\n",
    "# Training the model\n",
    "for epoch in range(1, 10):\n",
    "    loss = train()\n",
    "    losses.append(loss)\n",
    "    print(f'Epoch: {epoch}, Loss: {loss:.4f}')\n",
    "\n",
    "plt.plot(np.arange(9), losses)\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.title(\"Training curve\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "26b72014",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(1.1115)\n"
     ]
    }
   ],
   "source": [
    "def eval():\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        tot_error = 0.0\n",
    "        for data in test_loader:\n",
    "            pred = model(data)\n",
    "            target = data.y.view(-1, 1).float()\n",
    "            error = (pred - target) ** 2\n",
    "            tot_error += torch.mean(error)\n",
    "        return tot_error / len(test_loader)\n",
    "    \n",
    "error = eval()\n",
    "print(error)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "c142",
   "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": "3.11.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}