{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"\n\n# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\nfrom PIL import Image as Img\nimport openslide\nimport matplotlib as mpl\nimport matplotlib.pyplot as plt\nfrom matplotlib.colors import LinearSegmentedColormap\nimport seaborn as sns\nfrom IPython.display import Image, display\nimport sklearn.preprocessing\nfrom sklearn.model_selection import StratifiedKFold\nfrom sklearn.metrics import confusion_matrix\nfrom sklearn.metrics import accuracy_score\n\nfrom sklearn.model_selection import train_test_split\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\nimport cv2\nfrom tqdm.notebook import tqdm\nimport skimage.io\nfrom skimage.transform import resize, rescale\nfrom skimage.color import rgb2hed,hed2rgb\nimport os\n\nfrom fastai.callbacks import SaveModelCallback\nfrom fastai.callbacks import *\n\n# import albumentations as A\n\nimport shutil\nimport zipfile\n\nimport time\nimport cv2\nfrom copy import deepcopy\n\nimport fastai\nfrom fastai.core import *\nfrom fastai.vision import *\nfrom fastai.vision import Image as fImage\nfrom fastai.vision import pil2tensor\nfrom fastai.metrics import *\n\nimport pickle\nfrom torchvision import transforms\nfrom matplotlib.colors import ListedColormap\n# import collections\nfrom collections import defaultdict, Counter\nimport torch\nimport torch.nn as nn\nimport torch.nn.functional as F\n# from torchsummary import summary\nfrom torch.utils.data import Dataset,DataLoader\nimport torch.optim as optim\nfrom torch.optim import lr_scheduler\nfrom torch.utils.tensorboard import SummaryWriter\nimport torchvision.transforms as transforms\nimport torchvision.transforms.functional as TF\nfrom torch import autograd\n\nfrom albumentations import Compose, Normalize, HorizontalFlip, VerticalFlip\nfrom albumentations.pytorch import ToTensorV2\n\n\nfrom torch.utils.data import Dataset,DataLoader\nfrom sklearn.metrics import cohen_kappa_score\nimport scipy as sp\nimport random\n\nimport sys\n\nsys.path = [\n    '../input/efficientnet-pytorch/EfficientNet-PyTorch/EfficientNet-PyTorch-master',\n    '../input/nvidiaapex'\n] + sys.path\nfrom efficientnet_pytorch import model as enet\n\nfrom apex import amp\n\n# import torch_xla.core.xla_model as xm\n# import torch_xla.distributed.parallel_loader as pl\n# import torch_xla.distributed.xla_multiprocessing as xmp\n","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"data_path = os.path.join('/kaggle/input','prostate-cancer-grade-assessment')\n\nprint(data_path)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\n\ntrain_df = pd.read_csv(os.path.join(data_path,'train.csv'))\ntrain_df.head()\n\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_df = pd.read_csv(os.path.join(data_path,'test.csv'))\n\ntest_df.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\n\ntrain_imgs_path = []\ntrain_imgs_dir = os.path.join(data_path,'train_images')\nfor path in os.listdir(train_imgs_dir):\n#     print(train_df[train_df['image_id'] == path[:-5]]['data_provider'].values[0])\n#     if(train_df[train_df['image_id'] == path[:-5]]['data_provider'].values[0] =='radboud'):\n    train_imgs_path.append(path)\n\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\n\ntrain_labels_path = []\ntrain_labels_dir = os.path.join(data_path,'train_label_masks')\nfor path in os.listdir(train_labels_dir):\n    train_labels_path.append(path)\n\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\n\nbad_samples = []\nfor train_path in train_imgs_path:\n#     print(train_path[:-5]+\"_mask.tiff\")\n    if train_path[:-5]+\"_mask.tiff\" not in train_labels_path:\n        bad_samples.append(train_path)\n\nprint(\"Number of bad samples: \",len(bad_samples))\n\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"len(train_df)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_df = train_df.set_index('image_id')\ntrain_df = train_df.drop('ffe9bcababc858e04840669e788065a1')\ntrain_df = train_df.reset_index()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"len(train_df)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_df.info()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_imgs_path = sorted(train_imgs_path)\ntrain_labels_path = sorted(train_labels_path)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\n\ndef find_bad_samples(train_imgs_path):\n    \"\"\"\n    Looks for samples that doesn't have corresponding mask. \n    \n    param: training_imgs_path - path to directory containing images\n    \n    returns: array containing indexes of bad samples (bad = doesn't have label)\n    \"\"\"\n    bad_samples = []\n    for train_path in train_imgs_path:\n#     print(train_path[:-5]+\"_mask.tiff\")\n        if train_path[:-5]+\"_mask.tiff\" not in train_labels_path:\n            bad_samples.append(train_path)\n    return bad_samples\n\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_random_idx():\n    return train_df.sample(1)['image_id'].values[0]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_karolinska_idx():\n    return train_df[train_df['data_provider']=='karolinska'].sample(1)['image_id'].values[0]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_df.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def print_imgs_dirs():\n    count = 0\n    for file in os.listdir('/kaggle/input/preprocessingint-0-1000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-1000-2000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-2000-3000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-3000-4000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-4000-5000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-5000-6000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-6000-7000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-7000-8000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-8000-9000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-9000-10000-rgb/train_imgs'):\n        count +=1\n    for file in os.listdir('/kaggle/input/preprocessingint-10000-end-rgb/train_imgs'):\n        count +=1\n\n\n\n    print('Found ',count,' images')\nprint_imgs_dirs()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def seed_torch(seed=42):\n    random.seed(seed)\n    os.environ['PYTHONHASHSEED'] = str(seed)\n    np.random.seed(seed)\n    torch.manual_seed(seed)\n    torch.cuda.manual_seed(seed)\n    torch.backends.cudnn.deterministic = True\n\nseed_torch(seed=42)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"class PandaDataset(Dataset):\n    def __init__(self, df,idx_list, transform=None):\n        self.df = df\n        self.idx_list = idx_list\n        self.transform = transform\n        self.paths = ['/kaggle/input/preprocessingint-0-1000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-1000-2000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-2000-3000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-3000-4000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-4000-5000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-5000-6000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-6000-7000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-7000-8000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-8000-9000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-9000-10000-rgb/train_imgs',\n                     '/kaggle/input/preprocessingint-10000-end-rgb/train_imgs']\n        \n    def __len__(self):\n        return len(self.idx_list)\n\n    def __getitem__(self, idx):        \n        idx = self.idx_list[idx]\n        \n        if idx >='0005f7aaab2800f6170c399693a96917' and idx <= '1995d7f1f42ad259aae1fc004fbc6360':\n            path = self.paths[0]\n        elif idx >= '19963a5a3c42675307056e0be36851cf' and idx <= '3208b512a881073da1fad575ae4e8cf3':\n            path = self.paths[1]\n        elif idx >= '320a5a8ed48d1d03103a79659913c987' and idx <= '4b2af68b112c6842e2fce65b9f69473a':\n            path = self.paths[2]\n        elif idx >= '4b38723b387a432075f7cb3544172900' and idx <= '636430b4495809e379c2623c22f95d59':\n            path = self.paths[3]\n        elif idx >= '636498ce17020805892b1dd50e5eed64' and idx <= '7b576fbb0da9890869714dadff5226df':\n            path = self.paths[4]\n        elif idx >= '7b5b2d7f86f7725b0ea4f0ed3b4807de' and idx <= '9359e8987975c5e0d24dcb18e41270dd':\n            path = self.paths[5]\n        elif idx >= '9362e969027cc9260ceabfcbe2a7ceda' and idx <= 'aa8ee3fd85494e92cc6595f97d42dc6a':\n            path = self.paths[6]\n        elif idx >= 'aa956e452824c715d48722bbe45c2882' and idx <= 'c176b834c3a5045ac074fb70637cfa8d':\n            path = self.paths[7]\n        elif idx >= 'c176d5630828984c57bf0c94751145a8' and idx <= 'd996d408cbc9200c7254ab6955e31959':\n            path = self.paths[8]\n        elif idx >= 'd997bd54e6a545c95817a62b77fb43df' and idx <= 'f1f74c4aecfa7f9baa760bb992e4c720':\n            path = self.paths[9]\n        elif idx >= 'f1f9aea9e0bb9845704f9871346811aa':\n            path = self.paths[10]  \n        \n        final_path = os.path.join(path,idx)\n        res = []\n        \n        for tile in os.listdir(final_path):\n            tile_img = Img.open(os.path.join(final_path,tile))\n            tile_img = np.asarray(tile_img)\n            res.append(tile_img)\n        \n        rnd_perm = np.random.permutation(len(res))\n        \n        img = cv2.hconcat([\n        cv2.vconcat([res[rnd_perm[0]],res[rnd_perm[1]],res[rnd_perm[2]],res[rnd_perm[3]],res[rnd_perm[4]],res[rnd_perm[5]]]),\n        cv2.vconcat([res[rnd_perm[6]],res[rnd_perm[7]],res[rnd_perm[8]],res[rnd_perm[9]],res[rnd_perm[10]],res[rnd_perm[11]]]),\n        cv2.vconcat([res[rnd_perm[12]],res[rnd_perm[13]],res[rnd_perm[14]],res[rnd_perm[15]],res[rnd_perm[16]],res[rnd_perm[17]]]),\n        cv2.vconcat([res[rnd_perm[18]],res[rnd_perm[19]],res[rnd_perm[20]],res[rnd_perm[21]],res[rnd_perm[22]],res[rnd_perm[23]]]),\n        cv2.vconcat([res[rnd_perm[24]],res[rnd_perm[25]],res[rnd_perm[26]],res[rnd_perm[27]],res[rnd_perm[28]],res[rnd_perm[29]]]),\n        cv2.vconcat([res[rnd_perm[30]],res[rnd_perm[31]],res[rnd_perm[32]],res[rnd_perm[33]],res[rnd_perm[34]],res[rnd_perm[35]]])\n    ])\n        \n        sample =self.df.loc[idx,:]\n#         file_name = idx+\".png\"\n        isup  = sample['isup_grade']\n        \n#         img = Img.open(os.path.join(path,file_name))\n        \n#         img = np.asarray(img)\n        \n        if self.transform:\n            augmented = self.transform(image=img)\n            img = augmented['image']\n            \n            \n        label = np.zeros(5).astype(np.float32)\n        label[:int(isup)] = 1.\n#         isup = torch.tensor(isup).float()\n        \n        return img,torch.tensor(label)\n        \n\n\ndef get_transforms_type(data='train'):\n    \n    assert data in ('train', 'valid')\n    \n    if data == 'train':\n        return Compose([\n            HorizontalFlip(p=0.5),\n            VerticalFlip(p=0.5),\n            Normalize(\n                mean=[0.485, 0.456, 0.406],\n                std=[0.229, 0.224, 0.225],\n            ),\n            ToTensorV2(),\n        ])\n    \n    elif data == 'valid':\n        return Compose([\n            Normalize(\n                mean=[0.485, 0.456, 0.406],\n                std=[0.229, 0.224, 0.225],\n            ),\n            ToTensorV2(),\n        ])\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"os.listdir('/kaggle/input/efficientnet-pytorch')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"class PCAModel(nn.Module):\n    def __init__(self,output_num):\n        super().__init__()\n        self.effic = enet.EfficientNet.from_name('efficientnet-b0')\n        self.effic.load_state_dict(torch.load(os.path.join('/kaggle/input/efficientnet-pytorch','efficientnet-b0-08094119.pth')))\n        self.input_dim = self.effic._fc.in_features\n        self.head = nn.Linear(self.input_dim,output_num)\n        self.effic._fc = nn.Identity()\n    \n    def extract(self,x):\n        return self.effic(x)\n            \n\n    def forward(self,x):\n        x = self.extract(x)\n        x = self.head(x)\n        return x\n        \n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_df = pd.read_csv(os.path.join('/kaggle/input/panda-folds','folds.csv'))\ntrain_df = train_df.set_index('image_id')\ntrain_df = train_df.drop('ffe9bcababc858e04840669e788065a1')\ntrain_df = train_df.reset_index()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"fold = 1\nX_train = train_df[train_df['fold'] != fold]\nX_val = train_df[train_df['fold'] == fold]\nX_train = X_train.set_index('image_id')\nX_val = X_val.set_index('image_id')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_train.head()\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_val.head()\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_train['isup_grade'].value_counts()/len(X_train)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_val['isup_grade'].value_counts()/len(X_val)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"os.makedirs('modelw',exist_ok=True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# trainSetSample = X_train.sample(200)\n# valSetSample = X_val.sample(50)\n\ntrainSet = PandaDataset(X_train,X_train.index,transform=get_transforms_type(data='train'))\nvalSet = PandaDataset(X_val,X_val.index,transform=get_transforms_type(data='valid'))\n\nbatch_size = 5\ndataloaders = {\n    'train': DataLoader(trainSet, batch_size=batch_size, shuffle=True, num_workers=4),\n    'val': DataLoader(valSet, batch_size=batch_size, shuffle=False, num_workers=4)\n}","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Counter(X_train['isup_grade'])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Counter(X_val['isup_grade'])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"len(X_train),len(X_val)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def check_input():\n    fig,ax = plt.subplots(1,2)\n    for imgs,labels in dataloaders['val']:\n        ax[0].imshow(imgs[0].permute(1,2,0))\n        ax[0].set_title(labels[0])\n\n        ax[1].imshow(imgs[1].permute(1,2,0))\n        ax[1].set_title(labels[1])\n        break\n# check_input()    ","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"gc.collect()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def accuracy_(y_hat,y):\n    return np.mean(y_hat == y)\n\ndef quadratic_weighted_kappa(y_hat,y):\n    return cohen_kappa_score(y_hat,y,weights='quadratic')\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def train_model(model,optimizer,scheduler,start_epoch=0,end_epoch=20,save_path='modelw',apex=False,checkpoint=None,best_score=None):\n    device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n#     LOGGER.debug('using device: ',device)\n    \n    model.to(device)\n    \n    if apex:\n        print('USING APEX')\n        model, optimizer = amp.initialize(model, optimizer, opt_level=\"O1\",verbosity=0)\n        if checkpoint is not None: \n            print('Loading checkpoint - APEX')\n            model.load_state_dict(checkpoint['model'])\n            optimizer.load_state_dict(checkpoint['optimizer'])\n            amp.load_state_dict(checkpoint['amp'])\n#     best_model_wts = deepcopy(model.state_dict())\n    \n    criterion = nn.BCEWithLogitsLoss()\n    criterion2 = nn.MSELoss()\n    \n    if best_score is None:\n        best_score = -100\n    else:\n        best_score = best_score\n    best_loss = np.inf\n    \n\n    num_epochs = end_epoch - start_epoch\n    for epoch in range(start_epoch,end_epoch):\n        print('Epoch {}/{}'.format(epoch, num_epochs - 1))\n        print('-' * 10)\n        \n        since = time.time()\n        \n        for phase in ['train','val']:\n            if phase == 'train':\n            \n                model.train()\n            \n            else:\n                model.eval()\n                \n#             tk = tqdm(enumerate(dataloaders[phase]),total=len(dataloaders[phase]))\n            avg_loss = 0\n            \n            predictions = []\n            valid_labels = []\n            \n            for (imgs,labels) in dataloaders[phase]:\n\n                imgs = imgs.to(device)\n                labels = labels.to(device)\n                \n#                 print('imgs shape: ',imgs.shape)\n#                 print('lbaels shape: ',labels.shape)\n                \n                with torch.set_grad_enabled(phase=='train'):\n                \n                    preds = model(imgs)\n                    \n                \n                loss = criterion(preds,labels)\n                \n#                     print('loss ',loss)\n                        \n                avg_loss += loss.item()/ len(dataloaders[phase])\n                    \n#                     print('avg loss: ',avg_loss)\n                    \n                preds = preds.sigmoid().sum(1).detach().round()\n        \n                predictions.append(preds.to('cpu'))\n                valid_labels.append(labels.to('cpu').sum(1).detach().numpy())\n                \n                del labels,imgs\n                gc.collect()\n                    \n                if phase == 'train':\n                    \n                    if apex:\n                        with amp.scale_loss(loss, optimizer) as scaled_loss:\n                            scaled_loss.backward()\n                    else:\n                        loss.backward()\n                    optimizer.step()\n                    optimizer.zero_grad()\n                        \n                     \n            if phase =='val':\n                scheduler.step(avg_loss)\n                        \n            preds = np.concatenate(predictions)\n            valid_labels = np.concatenate(valid_labels) \n            \n                        \n            final_preds = preds\n            score = quadratic_weighted_kappa(valid_labels, final_preds)\n#                 \n            acc = accuracy_score(valid_labels,final_preds)\n                \n            time_elapsed = time.time() - since\n            print(f\"phase: {phase}, loss:{avg_loss}, kappa_score: {score}, acc: {acc}\")\n            print('Time: {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))\n                        \n                \n            if phase == 'val':\n                if score>best_score:\n                    best_score = score\n#                     best_model_wts = deepcopy(model.state_dict())\n                    print(f'  Epoch {epoch} - Save Best Score: {best_score:.4f}')\n                    if apex:\n                        checkpoint = {\n                            'model': model.state_dict(),\n                            'optimizer': optimizer.state_dict(),\n                            'amp': amp.state_dict()\n                        }\n                        torch.save(checkpoint, os.path.join(save_path,'amp_checkpoint.pt'))\n                    else:\n                        torch.save(model.state_dict(), os.path.join(save_path,f'panda_callback.pt'))\n                    \n\n                        \n#     model.load_state_dict(best_model_wts)\n#     torch.save(model,os.path.join(save_path,'fullModel.pt'))\n            \n    return model\n            \n            ","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\ngc.collect()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"lr = 1e-3\nmodel = PCAModel(5)\n# model.load_state_dict(torch.load(os.path.join('/kaggle/input/pandainterefficregr','panda_callback.pt')))\n\noptimizer = optim.Adam([\n     {'params':model.effic.parameters(),'lr':lr/10},\n    {'params':model.head.parameters()}\n] ,lr=lr, amsgrad=False)\nscheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, 'min', factor=0.5, patience=3, verbose=True, eps=1e-6)\n\ncheckpoint =  torch.load(os.path.join('/kaggle/input/eksperyment5fold1','amp_checkpoint.pt'))\n# checkpoint = None\n# scheduler = torch.optim.lr_scheduler.OneCycleLR(optimizer, max_lr=0.01, steps_per_epoch=len(dataloaders['train']), epochs=10)\n\nmodel,training_stats,coefficients =  train_model(model,optimizer,scheduler,start_epoch=14,end_epoch=20,apex=True,checkpoint=checkpoint)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#adawsaawdsassa","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}