train.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5Xi0rkaXwB1n"
   },
   "source": [
    "# CheXpert : A Large Chest X-Ray Dataset and Competition\n",
    "\n",
    "This competition launched by the Stanford ML group aims at finding a prediction model which could perform as well as radiologist to find different pathologies thanks to chest X-Ray. The Dataset available to train our model is composed of 223,414 chest radiographs of 65,240 patients.\n",
    "\n",
    "<img src=\"https://github.com/gaetandi/cheXpert/blob/master/view1_frontal.jpg?raw=1\" title=\"X-Ray image of the dataset\" width = 320/>\n",
    "\n",
    "The website of the competition:\n",
    "https://stanfordmlgroup.github.io/competitions/chexpert/\n",
    "\n",
    "[Publication](https://arxiv.org/pdf/1901.07031.pdf) : Irvin, Jeremy, et al. \"CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparison.\" arXiv preprint arXiv:1901.07031 (2019).\n",
    "\n",
    "Our goal is first to reproduce main results obtained in the related paper, published in January 2019.\n",
    "\n",
    "References:\n",
    "https://github.com/gaetandi/cheXpert/blob/master/cheXpert_final.ipynb\n",
    "https://pytorch.org/tutorials/beginner/data_loading_tutorial.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "4toJFrdfsxvz",
    "colab_type": "code",
    "outputId": "1788de54-bb0d-4dcd-83d8-52a4891a63cb",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 122.0
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
      "\n",
      "Enter your authorization code:\n",
      "··········\n",
      "Mounted at /content/drive\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "WpmWc4o9wHPl",
    "outputId": "bbe389ab-7347-45f3-be4f-f2117a17ec33",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 80.0
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Updating fastai...\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "# fastai colab setup\n",
    "# !curl -s https://course.fast.ai/setup/colab | bash"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "MlPuo4eFwB1p",
    "outputId": "e65dcb49-bbb2-4391-87ee-c8fe40fca04d",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34.0
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 4,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import time\n",
    "import sys\n",
    "import csv\n",
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.backends.cudnn as cudnn\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as tfunc\n",
    "from torch.utils.data import Dataset\n",
    "from torch.utils.data.dataset import random_split\n",
    "from torch.utils.data import DataLoader\n",
    "from torch.optim.lr_scheduler import ReduceLROnPlateau\n",
    "from torchvision import models\n",
    "\n",
    "import PIL.Image\n",
    "import torch.nn.functional as func\n",
    "\n",
    "from sklearn.metrics.ranking import roc_auc_score\n",
    "import sklearn.metrics as metrics\n",
    "import random\n",
    "\n",
    "use_gpu = torch.cuda.is_available()\n",
    "use_gpu"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "3DQrdeKZxQgq",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "Fro_ao-kxuCa",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "from fastai.vision import *\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "H2qRd7BCA3ZS",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "#os.chdir(\"/content/drive/My Drive/PyTorch_Hackathon_CheXpert/CheXpert-v1.0-small\")\n",
    "os.chdir(\"/content/drive/My Drive/Colab Notebooks/Challenge/data/CheXpert-v1.0-small\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "LJszyft2BAgB",
    "outputId": "d2241b11-b0c7-4b1d-ecad-1014c7367ebe",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34.0
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/content/drive/My Drive/Colab Notebooks/Challenge/data/CheXpert-v1.0-small\n"
     ]
    }
   ],
   "source": [
    "!pwd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "TOGZGehZwB1u",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "# Paths to the files with training, and validation sets.\n",
    "# Each file contains pairs (path to image, output vector)\n",
    "pathFileTrain = 'train50000.csv'\n",
    "pathFileValid = 'valid.csv'\n",
    "\n",
    "# Neural network parameters:\n",
    "nnIsTrained = False                 #pre-trained using ImageNet\n",
    "nnClassCount = 14                   #dimension of the output\n",
    "\n",
    "# Training settings: batch size, maximum number of epochs\n",
    "trBatchSize = 64\n",
    "trMaxEpoch = 3\n",
    "\n",
    "# Parameters related to image transforms: size of the down-scaled image, cropped image\n",
    "imgtransResize = (255, 255)\n",
    "imgtransCrop = 224\n",
    "\n",
    "# Class names\n",
    "class_names = ['No Finding', 'Enlarged Cardiomediastinum', 'Cardiomegaly', 'Lung Opacity', \n",
    "               'Lung Lesion', 'Edema', 'Consolidation', 'Pneumonia', 'Atelectasis', 'Pneumothorax', \n",
    "               'Pleural Effusion', 'Pleural Other', 'Fracture', 'Support Devices']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vjZztxbpwB1y"
   },
   "source": [
    "## 1. Create a Dataset\n",
    "\n",
    "In this first part, we create a loader able to load the data thanks to the csv files of the dataset.\n",
    "This loader first load labels and paths to the different images.\n",
    "\n",
    "It includes a pre-processing of labels to deal with uncertainty labels.\n",
    "Indeed, four different labels exist:\n",
    "-  **Blanks ('')**: the existence of the pathology is unknown\n",
    "-  **Ones (1)**: the pathology is detected thanks to the image\n",
    "-  **Zeros (0)**: the pathology can't be detected thanks to the image\n",
    "-  **Uncertain (-1)**: the pathology may be detected.\n",
    "\n",
    "In order to deal with uncertainty labels, different policies can be applied. The easiest ones consists in changing those **Uncertain** labels to *Blanks*, *Ones* or *Zeros*. This process is done thanks to the following class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "2NQUBySzwB1z",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "class CheXpertDataSet(Dataset):\n",
    "    def __init__(self, image_list_file, transform=None, policy=\"ones\", ttype=\"train\"):\n",
    "        \"\"\"\n",
    "        image_list_file: path to the file containing images with corresponding labels.\n",
    "        transform: optional transform to be applied on a sample.\n",
    "        Upolicy: name the policy with regard to the uncertain labels\n",
    "        \"\"\"\n",
    "        x = []\n",
    "        y = []\n",
    "\n",
    "        with open(image_list_file, \"r\") as f:\n",
    "            csvReader = csv.reader(f)\n",
    "            next(csvReader, None)\n",
    "            k=0\n",
    "            for line in csvReader:\n",
    "                k+=1\n",
    "                image_name= line[0]\n",
    "                label = line[5:]\n",
    "                \n",
    "                for i in range(14):\n",
    "                    if label[i]:\n",
    "                        a = float(label[i])\n",
    "                        if a == 1:\n",
    "                            label[i] = 1\n",
    "                        elif a == -1:\n",
    "                            if policy == \"ones\":\n",
    "                                label[i] = 1\n",
    "                            elif policy == \"zeroes\":\n",
    "                                label[i] = 0\n",
    "                            else:\n",
    "                                label[i] = 0\n",
    "                        else:\n",
    "                            label[i] = 0\n",
    "                    else:\n",
    "                        label[i] = 0\n",
    "                        \n",
    "                x.append('/content/drive/My Drive/Colab Notebooks/Challenge/data/' + image_name)\n",
    "                y.append(label)\n",
    "\n",
    "        self.x = x\n",
    "        self.y = y\n",
    "        self.transform = transform\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        \"\"\"Take the index of item and returns the image and its labels\"\"\"\n",
    "        \n",
    "        image_name = self.x[index]\n",
    "        # image = Image.open(image_name).convert('RGB')\n",
    "        image = PIL.Image.open(image_name).convert('RGB')\n",
    "        label = self.y[index]\n",
    "        if self.transform is not None:\n",
    "            image = self.transform(image)\n",
    "        return image, torch.LongTensor(label)\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "u2tyA8DFwB11"
   },
   "source": [
    "## 2. Create DataLoaders\n",
    "\n",
    "First we define a transform model to resize all images and normalize them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "bZCDvMJ9wB12",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "#TRANSFORM DATA\n",
    "\n",
    "normalize = transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "transformList = []\n",
    "transformList.append(transforms.Resize(imgtransResize))\n",
    "transformList.append(transforms.RandomResizedCrop(imgtransCrop))\n",
    "transformList.append(transforms.RandomHorizontalFlip())\n",
    "transformList.append(transforms.ToTensor())\n",
    "transformList.append(normalize)      \n",
    "transformSequence = transforms.Compose(transformList)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "VIqu36m5LbrO",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "class MyDataBunch(DataBunch):\n",
    "  def __init__(self, train_dl:DataLoader, valid_dl:DataLoader, fix_dl:DataLoader=None, test_dl:Optional[DataLoader]=None,\n",
    "                 device:torch.device=None, dl_tfms:Optional[Collection[Callable]]=None, path:PathOrStr='.',\n",
    "                 collate_fn:Callable=data_collate, no_check:bool=False, c=2):\n",
    "    DataBunch.__init__(self, train_dl, valid_dl, fix_dl=None, test_dl=None,\n",
    "                 device=None, dl_tfms=None, path='.',\n",
    "                 collate_fn=data_collate, no_check=False)\n",
    "    \n",
    "    self.c = c\n",
    "  \n",
    "  @classmethod\n",
    "  def create(cls, train_ds:Dataset, valid_ds:Dataset, test_ds:Optional[Dataset]=None, path:PathOrStr='.', bs:int=64,\n",
    "               val_bs:int=None, num_workers:int=defaults.cpus, dl_tfms:Optional[Collection[Callable]]=None,\n",
    "               device:torch.device=None, collate_fn:Callable=data_collate, no_check:bool=False, c=2, **dl_kwargs)->'DataBunch':\n",
    "        \"Create a `DataBunch` from `train_ds`, `valid_ds` and maybe `test_ds` with a batch size of `bs`. Passes `**dl_kwargs` to `DataLoader()`\"\n",
    "        datasets = cls._init_ds(train_ds, valid_ds, test_ds)\n",
    "        val_bs = ifnone(val_bs, bs)\n",
    "        dls = [DataLoader(d, b, shuffle=s, drop_last=s, num_workers=num_workers, **dl_kwargs) for d,b,s in\n",
    "               zip(datasets, (bs,val_bs,val_bs,val_bs), (True,False,False,False)) if d is not None]\n",
    "        return cls(*dls, path=path, device=device, dl_tfms=dl_tfms, collate_fn=collate_fn, no_check=no_check, c=c)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TwMb3lZJwB13"
   },
   "source": [
    "Then we build train, validation and test data loaders. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "fXHjaDpLwB14",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "#LOAD DATASET\n",
    "\n",
    "train_ds = CheXpertDataSet(pathFileTrain ,transformSequence, policy=\"ones\", ttype=\"train\")\n",
    "# test_ds, train_ds = random_split(train_ds, [500, len(train_ds) - 500])\n",
    "valid_ds = CheXpertDataSet(pathFileValid, transformSequence, ttype=\"valid\")   \n",
    "\n",
    "# Create dataloaders\n",
    "dataloaderTrain = DataLoader(dataset=train_ds, batch_size=trBatchSize, shuffle=True)\n",
    "dataloaderValid = DataLoader(dataset=valid_ds, batch_size=trBatchSize, shuffle=False)\n",
    "# dataloaderTest = DataLoader(dataset=test_ds, batch_size=trBatchSize, shuffle=False)\n",
    "\n",
    "# Create databunch\n",
    "data = MyDataBunch.create(train_ds=train_ds, valid_ds=valid_ds, bs=trBatchSize, c=nnClassCount)\n",
    "# data = MyDataBunch.create(train_ds=train_ds, valid_ds=valid_ds, test_ds=test_ds, bs=trBatchSize, c=nnClassCount)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "T_QVXd0MR8bX",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "def imshow(image, ax=None, title=None, normalize=True):\n",
    "    \"\"\"Imshow for Tensor.\"\"\"\n",
    "    if ax is None:\n",
    "        fig, ax = plt.subplots()\n",
    "    image = image.numpy().transpose((1, 2, 0))\n",
    "\n",
    "    if normalize:\n",
    "        mean = np.array([0.485, 0.456, 0.406])\n",
    "        std = np.array([0.229, 0.224, 0.225])\n",
    "        image = std * image + mean\n",
    "        image = np.clip(image, 0, 1)\n",
    "\n",
    "    ax.imshow(image)\n",
    "    ax.spines['top'].set_visible(False)\n",
    "    ax.spines['right'].set_visible(False)\n",
    "    ax.spines['left'].set_visible(False)\n",
    "    ax.spines['bottom'].set_visible(False)\n",
    "    ax.tick_params(axis='both', length=0)\n",
    "    ax.set_xticklabels('')\n",
    "    ax.set_yticklabels('')\n",
    "\n",
    "    return ax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "YeWaFe80RGmt",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "images, labels = next(iter(dataloaderValid))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "-aWBFqb9RhtT",
    "outputId": "05d586a4-200b-4600-c62d-dcd10f22549c",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 268.0
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAArAAAACjCAYAAABymS2qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nZ1fhC1iPx+PxeDwez67i/wNOf/u2bI+AUwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 864x648 with 4 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(figsize=(12,9), ncols=4)\n",
    "for ii in range(4):\n",
    "  ax = axes[ii]\n",
    "  imshow(images[ii], ax=ax, normalize=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "t-52eZqvKta-"
   },
   "source": [
    "## 3. Create Architecture and Train Model\n",
    "\n",
    "Create an architecture of resnet50, find the best learning rate, and train the model using best slice learning rates."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "1vweZ5JgQmky",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "# fastai"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "9e-5JSVAQXfX",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "arch = models.resnet50\n",
    "# arch = models.densenet121"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "GXsiFmMQQXks",
    "outputId": "2e25a339-b1f3-4202-8ef7-24c7e3bcfb7a",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 71.0
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Downloading: \"https://download.pytorch.org/models/resnet50-19c8e357.pth\" to /root/.cache/torch/checkpoints/resnet50-19c8e357.pth\n",
      "100%|██████████| 102502400/102502400 [00:00<00:00, 105529550.11it/s]\n"
     ]
    }
   ],
   "source": [
    "acc_02 = partial(accuracy_thresh, thresh=0.2)\n",
    "f_score = partial(fbeta, thresh=0.2)\n",
    "learn = cnn_learner(data, arch, metrics=[acc_02, f_score])\n",
    "learn.loss_func = BCEWithLogitsFlat()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "NuaFcU6oQXt1",
    "outputId": "d322f9ef-e2a2-41b3-8b5e-84f77d2abe1c",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34.0
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       ""
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    }
   ],
   "source": [
    "learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "nwVJWx7BQje7",
    "outputId": "37dab533-bcb2-4b05-ec76-026690e20f3c",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 283.0
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY8AAAEKCAYAAADq59mMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nxhhj/GbJwxhjjN/+H6Xr10HG1fFiAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "LcMA_MieQjmP",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "lr = 0.03"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "DMAf6AF9Qkjh",
    "outputId": "9974de24-fbba-42c4-99fa-9c581d4f0c2e",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 95.0
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='0' class='' max='5', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      0.00% [0/5 00:00<00:00]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy_thresh</th>\n",
       "      <th>fbeta</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table><p>\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='0' class='' max='536', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      0.00% [0/536 00:00<00:00]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.fit_one_cycle(5, slice(lr))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "R1gvcUN7foKi",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "# Save model before fine tuning\n",
    "learn.save('stage-1-rn50')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "SognWQhSfoKi",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "# Fine tune model\n",
    "learn.unfreeze()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "c999HYX7foKj",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "learn.lr_find()\n",
    "learn.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "fnDT958-foKk",
    "colab_type": "code",
    "outputId": "272e7e25-3888-44b4-a1c0-1838535e30e0",
    "colab": {}
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy_thresh</th>\n",
       "      <th>fbeta</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.378650</td>\n",
       "      <td>0.450442</td>\n",
       "      <td>0.788156</td>\n",
       "      <td>0.575232</td>\n",
       "      <td>03:06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.383189</td>\n",
       "      <td>0.495613</td>\n",
       "      <td>0.770146</td>\n",
       "      <td>0.552178</td>\n",
       "      <td>03:07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.376022</td>\n",
       "      <td>0.369219</td>\n",
       "      <td>0.772283</td>\n",
       "      <td>0.604469</td>\n",
       "      <td>03:07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.368444</td>\n",
       "      <td>0.371852</td>\n",
       "      <td>0.774725</td>\n",
       "      <td>0.597337</td>\n",
       "      <td>03:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.365860</td>\n",
       "      <td>0.377561</td>\n",
       "      <td>0.784799</td>\n",
       "      <td>0.594255</td>\n",
       "      <td>03:02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.fit_one_cycle(5, slice(1e-5, lr/5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "2-w_qIYKfoKl",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    ""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "rfGAZTtHfoKm",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    ""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "o8outBJtfoKm",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    ""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "4sYiz3BifoKn",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    ""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "id": "7bfFFgopfoKn",
    "colab_type": "code",
    "colab": {}
   },
   "outputs": [],
   "source": [
    ""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "THvgBNUuQlA6",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "# pytorch pre-trained model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "F4kw74uoLaa5",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "model = models.resnet50(pretrained=True)\n",
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "4Ys9UcPlgFlN",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "# Modify resnet50 for our training dataset\n",
    "for param in model.parameters():\n",
    "  param.requires_grad = False\n",
    "\n",
    "from collections import OrderedDict\n",
    "classifier = nn.Sequential(OrderedDict([\n",
    "                         ('fc', nn.Linear(1024, 256)),\n",
    "                         ('relu', nn.ReLU()),\n",
    "                         ('dropout', nn.Dropout(0.2)),\n",
    "                         ('fc2', nn.Linear(256, nnClassCount))\n",
    "                         ]))\n",
    "\n",
    "model.classifier = classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "FuR0hokLioad",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "# Training on GPU\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available else \"cpu\")\n",
    "\n",
    "criterion = nn.MultiLabelSoftMarginLoss()\n",
    "optimizer = optim.Adam(model.classifier.parameters(), lr=1e-5)\n",
    "\n",
    "model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab_type": "code",
    "id": "tluh5tfglZ8A",
    "colab": {}
   },
   "outputs": [],
   "source": [
    "epochs = 1\n",
    "for epoch in range(epochs):\n",
    "  losses = []\n",
    "  for inputs, labels in dataloaderTrain:\n",
    "      # Move input and label tensors to the default device\n",
    "      inputs, labels = inputs.to(device), labels.to(device)\n",
    "      labels.long()\n",
    "\n",
    "      optimizer.zero_grad()\n",
    "        \n",
    "      logps = model.forward(inputs)\n",
    "      loss = criterion(logps, labels)\n",
    "      loss.backward()\n",
    "      optimizer.step()\n",
    "\n",
    "      losses.append(loss.data.mean())\n",
    "  print('[%d/%d] Loss: %.3f' % (epoch+1, epochs, np.mean(losses)))"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "name": "pytorch_hackathon_cheXpert_erin-2.ipynb",
   "version": "0.3.2",
   "provenance": [],
   "collapsed_sections": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}