qnn_mnist.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Imports and Setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/qnn/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import classiq\n",
    "\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torchinfo import summary\n",
    "\n",
    "from models.leqm3 import linear_entanglement_r3_quantum_model\n",
    "from models.qnn import execute_fn, post_process_fn, QNN\n",
    "\n",
    "from scripts.helper import create_writer, write_train_results\n",
    "from scripts.data_setup import create_mnist_dataloaders\n",
    "from scripts.data_transforms import input_transform, target_transform\n",
    "from scripts.train import train\n",
    "from scripts.test import test\n",
    "from scripts.save_model import save_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "The current version of 'classiq' has been deprecated, and will not be supported as of 2024-01-18. Please run \"pip install -U classiq\" to upgrade the classiq SDK to the latest version.\n",
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/qnn/lib/python3.11/site-packages/classiq/_internals/authentication/token_manager.py:82: UserWarning: Device is already registered.\n",
      "Generating a new refresh token should only be done if the current refresh token is compromised.\n",
      "To do so, set the overwrite parameter to true\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "## Authenticate Classiq\n",
    "classiq.authenticate()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'cpu'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## For setting up device agnostic code\n",
    "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
    "device = 'cpu'\n",
    "device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Clear Output Files\n",
    "post_process_output_file = open(\"post_process_output.txt\", \"w\")\n",
    "print(\"-----------------------------------------------------------------------------------------------------------------\", file=post_process_output_file)\n",
    "print(\"--------------------------------------------POST PROCESS OUTPUT--------------------------------------------------\", file=post_process_output_file)\n",
    "print(\"-----------------------------------------------------------------------------------------------------------------\", file=post_process_output_file)\n",
    "post_process_output_file.close()\n",
    "\n",
    "test_loop_output_file = open(\"test_loop_output.txt\", \"w\")\n",
    "print(\"-----------------------------------------------------------------------------------------------------------------\", file=test_loop_output_file)\n",
    "print(\"-----------------------------------------------TEST LOOP OUTPUT--------------------------------------------------\", file=test_loop_output_file)\n",
    "print(\"-----------------------------------------------------------------------------------------------------------------\", file=test_loop_output_file)\n",
    "test_loop_output_file.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# HYPER PARAMETERS\n",
    "_LEARNING_RATE = 1.0\n",
    "BATCH_SIZE = 64\n",
    "EPOCHS = 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Quantum Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#  Create a Linear Entanglement Quantum Model for MNIST Data Classification with three linear entanglement layers of RXX, RYY, and RZZ.\n",
    "quantum_model = linear_entanglement_r3_quantum_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "quantum_program = classiq.synthesize(quantum_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Opening: https://platform.classiq.io/circuit/5ff23bf6-d6a0-45d2-a46a-fc742af01b37?version=0.33.0\n"
     ]
    }
   ],
   "source": [
    "# View Quantum Program on Classiq Platform\n",
    "classiq.show(quantum_program)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Quantum Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "gio: https://platform.classiq.io/circuit/5ff23bf6-d6a0-45d2-a46a-fc742af01b37?version=0.33.0: Operation not supported\n"
     ]
    }
   ],
   "source": [
    "qnn = QNN(\n",
    "    quantum_program=quantum_program,\n",
    "    execute=execute_fn,\n",
    "    post_process=post_process_fn,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# summary(model=qnn, input_size=(32, 16), verbose=0, col_names=[\"input_size\", \"output_size\", \"num_params\", \"trainable\"], col_width=20, row_settings=[\"var_names\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# choosing our loss function\n",
    "loss_fn = nn.L1Loss()\n",
    "# choosing our optimizer\n",
    "optimizer = optim.SGD(qnn.parameters(), lr=_LEARNING_RATE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preparing Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataloader, test_dataloader, class_names = create_mnist_dataloaders(\n",
    "    root=\"data\",\n",
    "    transform=input_transform,\n",
    "    target_transform=target_transform,\n",
    "    batch_size=BATCH_SIZE,\n",
    "    create_subset=True,\n",
    "    subset_size=64\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataloaders: (<torch.utils.data.dataloader.DataLoader object at 0x7ff8066d1590>, <torch.utils.data.dataloader.DataLoader object at 0x7ff7fff44710>)\n",
      "Length of train dataloader: 1 batches of 64\n",
      "Length of test dataloader: 1 batches of 64\n",
      "Our Dataset have following classes: ['0 - zero', '1 - one', '2 - two', '3 - three', '4 - four', '5 - five', '6 - six', '7 - seven', '8 - eight', '9 - nine']\n"
     ]
    }
   ],
   "source": [
    "# Let's check out what we've created\n",
    "print(f\"Dataloaders: {train_dataloader, test_dataloader}\") \n",
    "print(f\"Length of train dataloader: {len(train_dataloader)} batches of {BATCH_SIZE}\")\n",
    "print(f\"Length of test dataloader: {len(test_dataloader)} batches of {BATCH_SIZE}\")\n",
    "print(f\"Our Dataset have following classes: {class_names}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/qnn/lib/python3.11/site-packages/torchvision/transforms/functional.py:1603: UserWarning: The default value of the antialias parameter of all the resizing transforms (Resize(), RandomResizedCrop(), etc.) will change from None to True in v0.17, in order to be consistent across the PIL and Tensor backends. To suppress this warning, directly pass antialias=True (recommended, future default), antialias=None (current default, which means False for Tensors and True for PIL), or antialias=False (only works on Tensors - PIL will still use antialiasing). This also applies if you are using the inference transforms from the models weights: update the call to weights.transforms(antialias=True).\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Image shape: torch.Size([64, 16]) -> [batch_size, pixel_angle]\n",
      "Label shape: torch.Size([64, 10]) -> [batch_size, label_value]\n"
     ]
    }
   ],
   "source": [
    "data, label = next(iter(train_dataloader))\n",
    "\n",
    "print(f\"Image shape: {data.shape} -> [batch_size, pixel_angle]\")\n",
    "print(f\"Label shape: {label.shape} -> [batch_size, label_value]\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Run Experiment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### 01. Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[INFO] Created SummaryWriter, saving to: runs/2024-01-04/data_0.1_percent/linear_entanglement_r3/2_epochs...\n"
     ]
    }
   ],
   "source": [
    "# Create a writer for tracking our experiment\n",
    "writer = create_writer(experiment_name=\"data_0.1_percent\", model_name=\"linear_entanglement_r3\", extra=f\"{EPOCHS}_epochs\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/2 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0\n",
      "----------\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/qnn/lib/python3.11/site-packages/torchvision/transforms/functional.py:1603: UserWarning: The default value of the antialias parameter of all the resizing transforms (Resize(), RandomResizedCrop(), etc.) will change from None to True in v0.17, in order to be consistent across the PIL and Tensor backends. To suppress this warning, directly pass antialias=True (recommended, future default), antialias=None (current default, which means False for Tensors and True for PIL), or antialias=False (only works on Tensors - PIL will still use antialiasing). This also applies if you are using the inference transforms from the models weights: update the call to weights.transforms(antialias=True).\n",
      "  warnings.warn(\n",
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/models/qnn.py:42: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
      "  return torch.tensor(pred_probs)\n",
      " 50%|█████     | 1/2 [00:53<00:53, 53.47s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1 | Train loss: 0.19275\n",
      "Epoch: 1\n",
      "----------\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/qnn/lib/python3.11/site-packages/torchvision/transforms/functional.py:1603: UserWarning: The default value of the antialias parameter of all the resizing transforms (Resize(), RandomResizedCrop(), etc.) will change from None to True in v0.17, in order to be consistent across the PIL and Tensor backends. To suppress this warning, directly pass antialias=True (recommended, future default), antialias=None (current default, which means False for Tensors and True for PIL), or antialias=False (only works on Tensors - PIL will still use antialiasing). This also applies if you are using the inference transforms from the models weights: update the call to weights.transforms(antialias=True).\n",
      "  warnings.warn(\n",
      "100%|██████████| 2/2 [01:46<00:00, 53.35s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 2 | Train loss: 0.38348\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "train_results = train(\n",
    "    model = qnn, \n",
    "    data_loader = train_dataloader, \n",
    "    loss_fn = loss_fn, \n",
    "    optimizer = optimizer, \n",
    "    writer = writer, \n",
    "    epochs = EPOCHS,\n",
    "    device = device\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'train_loss': [0.19274857640266418, 0.38347625732421875]}\n"
     ]
    }
   ],
   "source": [
    "# Check out the model results\n",
    "print(train_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "write_train_results(experiment_name=\"data_0.1_percent\", model_name=\"linear_entanglement_r3\", epochs=EPOCHS, results=train_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %load_ext tensorboard\n",
    "# %tensorboard --logdir runs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### 02. Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/qnn/lib/python3.11/site-packages/torchvision/transforms/functional.py:1603: UserWarning: The default value of the antialias parameter of all the resizing transforms (Resize(), RandomResizedCrop(), etc.) will change from None to True in v0.17, in order to be consistent across the PIL and Tensor backends. To suppress this warning, directly pass antialias=True (recommended, future default), antialias=None (current default, which means False for Tensors and True for PIL), or antialias=False (only works on Tensors - PIL will still use antialiasing). This also applies if you are using the inference transforms from the models weights: update the call to weights.transforms(antialias=True).\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy of the model: 9.38%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/devilkillerag/document/python_pro/QNN-MNIST-Classification/models/qnn.py:42: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
      "  return torch.tensor(pred_probs)\n"
     ]
    }
   ],
   "source": [
    "test_results = test(\n",
    "    model = qnn, \n",
    "    data_loader = test_dataloader,\n",
    "    device = device\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9.375\n"
     ]
    }
   ],
   "source": [
    "print(test_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Save the trained model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[INFO] Saving model to: outputs/saved_models/leqmr3_subset64.pt\n"
     ]
    }
   ],
   "source": [
    "save_model(\n",
    "    model=qnn,\n",
    "    target_dir='outputs/saved_models',\n",
    "    model_name='leqmr3_subset64.pt'\n",
    ")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "qnn",
   "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": "3.11.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}