main.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This file shows a different usage from using command line to run main.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-07-23 18:17:35.556793: I tensorflow/core/util/port.cc:111] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n",
      "2024-07-23 18:17:35.619120: E tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:9342] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
      "2024-07-23 18:17:35.619153: E tensorflow/compiler/xla/stream_executor/cuda/cuda_fft.cc:609] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
      "2024-07-23 18:17:35.619188: E tensorflow/compiler/xla/stream_executor/cuda/cuda_blas.cc:1518] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
      "2024-07-23 18:17:35.633109: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 AVX512F AVX512_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2024-07-23 18:17:36.881081: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import display_tools\n",
    "from models import VitAug\n",
    "from data_utilis import spilt_data\n",
    "from process import Processor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prepare Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "(images_train, labels_train), (images_test, labels_test) = tf.keras.datasets.cifar10.load_data()    \n",
    "images = np.concatenate((images_train, images_test), axis=0)\n",
    "labels = np.concatenate((labels_train,labels_test), axis=0)\n",
    "data_train, data_val, data_test = spilt_data((images, labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = spilt_data((images_train, labels_train))\n",
    "dataset_dev = (dataset[0],dataset[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-07-23 18:17:42.182265: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1886] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 30570 MB memory:  -> device: 0, name: NVIDIA GeForce RTX 4080 SUPER, pci bus id: 0000:31:00.0, compute capability: 8.9\n"
     ]
    },
    {
     "data": {
      "image/jpeg": "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",
      "image/png": "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",
      "text/plain": [
       "<PIL.Image.Image image mode=RGB size=128x128>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_names = ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']\n",
    "display_tools.visualise_mixup(images_train[0:16], (4,4), 'beta', alpha = 1.0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train the Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## model paramters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# two model configurations\n",
    "model_param1 = {\n",
    "    'sampling_method': 'beta',\n",
    "    'image_size': 32,\n",
    "    'patch_size': 4,\n",
    "    'num_classes': 10,\n",
    "    'dim': 256,\n",
    "    'depth': 8,\n",
    "    'num_heads': 8,\n",
    "    'mlp_dim': 512,\n",
    "    'dropout': 0.2,\n",
    "    'alpha': 1.0\n",
    "}\n",
    "\n",
    "model_param2 = {\n",
    "    'sampling_method': 'uniform',\n",
    "    'image_size': 32,\n",
    "    'patch_size': 4,\n",
    "    'num_classes': 10,\n",
    "    'dim': 256,\n",
    "    'depth': 8,\n",
    "    'num_heads': 8,\n",
    "    'mlp_dim': 512,\n",
    "    'dropout': 0.2,\n",
    "    'uniform_range': (0.0,1.0)\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## training paramters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_epochs = 10\n",
    "batch_size = 128\n",
    "learning_rate = 0.001"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## augmentation method 1 (beta)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = VitAug(**model_param1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "processor = Processor(\n",
    "    model=model,\n",
    "    data_train=data_train,\n",
    "    data_test=data_test,\n",
    "    data_val=data_val,\n",
    "    learning_rate=learning_rate,\n",
    "    batch_size=batch_size,\n",
    "    num_epochs=num_epochs\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "start: training, with sampling method beta\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-07-23 18:17:46.449819: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x124fe3f0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:\n",
      "2024-07-23 18:17:46.449836: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): NVIDIA GeForce RTX 4080 SUPER, Compute Capability 8.9\n",
      "2024-07-23 18:17:46.454458: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.\n",
      "2024-07-23 18:17:46.467771: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:442] Loaded cuDNN version 8600\n",
      "2024-07-23 18:17:46.594473: I ./tensorflow/compiler/jit/device_compiler.h:186] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1, train:: time: 124.319s, loss: 2.277, aug_accuracy: 10.710%                     --- test:: time: 3.050s, loss: 2.085, accuracy: 12.716%\n",
      "Epoch 2, train:: time: 105.019s, loss: 2.128, aug_accuracy: 13.914%                     --- test:: time: 2.017s, loss: 2.002, accuracy: 15.078%\n",
      "Epoch 3, train:: time: 104.396s, loss: 2.090, aug_accuracy: 15.694%                     --- test:: time: 2.109s, loss: 1.987, accuracy: 16.413%\n",
      "Epoch 4, train:: time: 105.291s, loss: 2.062, aug_accuracy: 16.879%                     --- test:: time: 1.841s, loss: 1.919, accuracy: 17.520%\n",
      "Epoch 5, train:: time: 102.923s, loss: 2.029, aug_accuracy: 17.892%                     --- test:: time: 1.910s, loss: 1.825, accuracy: 18.454%\n",
      "Epoch 6, train:: time: 106.308s, loss: 2.005, aug_accuracy: 18.816%                     --- test:: time: 1.977s, loss: 1.785, accuracy: 19.283%\n",
      "Epoch 7, train:: time: 105.118s, loss: 1.991, aug_accuracy: 19.524%                     --- test:: time: 2.033s, loss: 1.808, accuracy: 19.862%\n",
      "Epoch 8, train:: time: 105.714s, loss: 2.004, aug_accuracy: 20.021%                     --- test:: time: 1.827s, loss: 1.756, accuracy: 20.296%\n",
      "Epoch 9, train:: time: 102.420s, loss: 1.979, aug_accuracy: 20.478%                     --- test:: time: 1.948s, loss: 1.755, accuracy: 20.733%\n",
      "Epoch 10, train:: time: 101.963s, loss: 2.015, aug_accuracy: 20.824%                     --- test:: time: 2.153s, loss: 1.800, accuracy: 20.957%\n",
      "finish: training, with sampling method beta\n"
     ]
    }
   ],
   "source": [
    "processor.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "********************summary********************\n",
      "\ton train set:\n",
      "            lowest loss: 1.979 at epoch 9\n",
      "            final loss: 2.015\n",
      "            higest aug_accuracy: 20.824% at epoch 10\n",
      "            final aug_accuracy: 20.824%\n",
      "            average time: 106.347 s/epoch\n",
      "        \n",
      "\ton validation set:\n",
      "            lowest loss: 1.755 at epoch 9\n",
      "            final loss: 1.800\n",
      "            higest accuracy: 20.957% at epoch 10\n",
      "            final accuracy: 20.957%\n",
      "            average time: 2.086 s/epoch\n",
      "        \n"
     ]
    }
   ],
   "source": [
    "print('*'*20 + 'summary' + '*' * 20)\n",
    "display_tools.report_summary(record=(processor.losses_train, processor.accuracies_train, processor.times_train), dataset_type='train')\n",
    "display_tools.report_summary(record=(processor.losses_val, processor.accuracies_val, processor.times_val), dataset_type='validation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "start: test\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "finish: test. Accuracy is 21.126%\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.21125973761081696"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "processor.test()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## augmentation method 2 (uniform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = VitAug(**model_param2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "processor.reset_model(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "start: training, with sampling method uniform\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1, train:: time: 116.796s, loss: 2.263, aug_accuracy: 11.032%                     --- test:: time: 2.879s, loss: 2.067, accuracy: 13.414%\n",
      "Epoch 2, train:: time: 102.768s, loss: 2.110, aug_accuracy: 14.648%                     --- test:: time: 2.036s, loss: 1.920, accuracy: 16.127%\n",
      "Epoch 3, train:: time: 102.318s, loss: 2.045, aug_accuracy: 17.021%                     --- test:: time: 2.085s, loss: 1.876, accuracy: 18.118%\n",
      "Epoch 4, train:: time: 101.921s, loss: 2.000, aug_accuracy: 18.738%                     --- test:: time: 2.170s, loss: 1.781, accuracy: 19.470%\n",
      "Epoch 5, train:: time: 96.406s, loss: 1.994, aug_accuracy: 19.796%                     --- test:: time: 2.188s, loss: 1.789, accuracy: 20.315%\n",
      "Epoch 6, train:: time: 98.111s, loss: 1.990, aug_accuracy: 20.591%                     --- test:: time: 2.124s, loss: 1.793, accuracy: 20.908%\n",
      "Epoch 7, train:: time: 98.536s, loss: 1.968, aug_accuracy: 21.099%                     --- test:: time: 2.172s, loss: 1.774, accuracy: 21.402%\n",
      "Epoch 8, train:: time: 99.218s, loss: 1.968, aug_accuracy: 21.560%                     --- test:: time: 2.174s, loss: 1.709, accuracy: 21.833%\n",
      "Epoch 9, train:: time: 96.692s, loss: 1.941, aug_accuracy: 22.003%                     --- test:: time: 2.172s, loss: 1.693, accuracy: 22.254%\n",
      "Epoch 10, train:: time: 99.588s, loss: 1.950, aug_accuracy: 22.389%                     --- test:: time: 2.192s, loss: 1.735, accuracy: 22.590%\n",
      "finish: training, with sampling method uniform\n"
     ]
    }
   ],
   "source": [
    "processor.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "********************summary********************\n",
      "\ton train set:\n",
      "            lowest loss: 1.941 at epoch 9\n",
      "            final loss: 1.950\n",
      "            higest aug_accuracy: 22.389% at epoch 10\n",
      "            final aug_accuracy: 22.389%\n",
      "            average time: 101.236 s/epoch\n",
      "        \n",
      "\ton validation set:\n",
      "            lowest loss: 1.693 at epoch 9\n",
      "            final loss: 1.735\n",
      "            higest accuracy: 22.590% at epoch 10\n",
      "            final accuracy: 22.590%\n",
      "            average time: 2.219 s/epoch\n",
      "        \n"
     ]
    }
   ],
   "source": [
    "print('*'*20 + 'summary' + '*' * 20)\n",
    "display_tools.report_summary(record=(processor.losses_train, processor.accuracies_train, processor.times_train), dataset_type='train')\n",
    "display_tools.report_summary(record=(processor.losses_val, processor.accuracies_val, processor.times_val), dataset_type='validation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "start: test\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "finish: test. Accuracy is 22.808%\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.22808009386062622"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "processor.test()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "comp0197-cw1-pt",
   "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": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}