{"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_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# 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)\n\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\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2023-08-25T06:29:19.086039Z","iopub.execute_input":"2023-08-25T06:29:19.086447Z","iopub.status.idle":"2023-08-25T06:29:19.131835Z","shell.execute_reply.started":"2023-08-25T06:29:19.086417Z","shell.execute_reply":"2023-08-25T06:29:19.130951Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import matplotlib.pyplot as plt\nimport numpy as np\nimport pandas as pd\nimport torch\nimport torch.nn as nn\nfrom torch.utils.data import DataLoader, TensorDataset, Dataset\nimport torchvision.transforms as transforms\nimport torchvision.datasets as datasets\nfrom sklearn.model_selection import train_test_split\nfrom tqdm import tqdm\n%matplotlib inline","metadata":{"execution":{"iopub.status.busy":"2023-08-25T07:12:08.435020Z","iopub.execute_input":"2023-08-25T07:12:08.435536Z","iopub.status.idle":"2023-08-25T07:12:08.445784Z","shell.execute_reply.started":"2023-08-25T07:12:08.435489Z","shell.execute_reply":"2023-08-25T07:12:08.444584Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_all_df = pd.read_csv('/kaggle/input/digit-recognizer/train.csv')\ntest_df = pd.read_csv('/kaggle/input/digit-recognizer/test.csv')\ntrain_all_df.head()","metadata":{"execution":{"iopub.status.busy":"2023-08-25T06:29:42.614778Z","iopub.execute_input":"2023-08-25T06:29:42.615317Z","iopub.status.idle":"2023-08-25T06:29:48.078558Z","shell.execute_reply.started":"2023-08-25T06:29:42.615273Z","shell.execute_reply":"2023-08-25T06:29:48.077078Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_all_labels = train_all_df['label'].copy()\ntrain_all_features = train_all_df.drop('label', axis=1)","metadata":{"execution":{"iopub.status.busy":"2023-08-24T08:03:06.385637Z","iopub.execute_input":"2023-08-24T08:03:06.386224Z","iopub.status.idle":"2023-08-24T08:03:06.523523Z","shell.execute_reply.started":"2023-08-24T08:03:06.386180Z","shell.execute_reply":"2023-08-24T08:03:06.521946Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# データの準備\n","metadata":{}},{"cell_type":"markdown","source":"## train_test_split","metadata":{}},{"cell_type":"code","source":"# datasetsに乗っけるために、torch.utils.data.TensorDataset()を使う\n# 特徴量とラベルを指定することで、組ごとにdatasetで呼び出せるようになる\n\n# train_test_split\ntrain_df, valid_df = train_test_split(train_all_df, test_size=0.2, random_state=42, shuffle=True)\ntrain_labels = train_df['label'].values\ntrain_features = train_df.drop('label', axis=1).values\nvalid_labels = valid_df['label'].values\nvalid_features = valid_df.drop('label', axis=1).values\n","metadata":{"execution":{"iopub.status.busy":"2023-08-25T06:30:04.414573Z","iopub.execute_input":"2023-08-25T06:30:04.415039Z","iopub.status.idle":"2023-08-25T06:30:04.791689Z","shell.execute_reply.started":"2023-08-25T06:30:04.414983Z","shell.execute_reply":"2023-08-25T06:30:04.790675Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## transformsとDataloader\ndatasets は今回はdataframe形式でデータが提供されているので Dataset.Imagefolder() のような形でdatasetを定義できない?  \nかわりに、 torch.utils.data.TensorDataset()を使う  ","metadata":{}},{"cell_type":"code","source":"train_features","metadata":{"execution":{"iopub.status.busy":"2023-08-25T07:58:34.733034Z","iopub.execute_input":"2023-08-25T07:58:34.733461Z","iopub.status.idle":"2023-08-25T07:58:34.743161Z","shell.execute_reply.started":"2023-08-25T07:58:34.733429Z","shell.execute_reply":"2023-08-25T07:58:34.741941Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# transforms\n\n# toTensor書こうとしたけど、 TensorDatasetを書くときはTensor型じゃないと行けないので、かかないでおく\ntrain_transform = transforms.Compose([\n    transforms.Normalize(0.5, 0.5)\n\n])\n\ntest_transform = transforms.Compose([\n    transforms.Normalize(0.5, 0.5)\n])\n\n# def train_transform(tensor):\n#     tensor = tensor.view(-1, 28, 28)\n#     return tensor\n\n\n\n\n# dataset定義\ntrain_features = torch.Tensor(train_features)\ntrain_labels = torch.Tensor(train_labels).long() # longにしないとloss計算ができない loss計算はint\nvalid_features = torch.Tensor(valid_features)\nvalid_labels = torch.Tensor(valid_labels).long() # longにしないとloss計算ができない loss計算はint\n\n## dataset.imagefolderで本来はデータを指定するが、今回は１回テンソルでdataframeで与えられてるので、以下のTensorDatasetを使う。\n## ただし、TensorDatasetではtransformsが使えないので、その場合はカスタムで作ってあげないといけない\n# train_dataset = TensorDataset(train_features, train_labels)\n# valid_dataset = TensorDataset(valid_features, valid_labels)\n\n# カスタムDataset https://stackoverflow.com/questions/55588201/pytorch-transforms-on-tensordataset\nclass CustomTensorDataset(Dataset):\n    \"\"\"TensorDataset with support of transforms.\n    \"\"\"\n    def __init__(self, tensors, transform=None):\n        assert all(tensors[0].size(0) == tensor.size(0) for tensor in tensors)\n        self.tensors = tensors\n        self.transform = transform\n\n    def __getitem__(self, index):\n        x = self.tensors[0][index]\n\n        if self.transform:\n            x = self.transform(x)\n\n        y = self.tensors[1][index]\n\n        return x, y\n\n    def __len__(self):\n        return self.tensors[0].size(0)\n\ntrain_dataset = CustomTensorDataset(tensors=(train_features, train_labels), transform=transform) \nvalid_dataset = CustomTensorDataset(tensors=(valid_features, valid_labels), transform=transform)\n\n    \n# dataloader 定義\nbatch_size = 100\ntrain_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\nvalid_loader = DataLoader(valid_dataset, batch_size=batch_size, shuffle=False)\n\n","metadata":{"execution":{"iopub.status.busy":"2023-08-25T08:05:31.460844Z","iopub.execute_input":"2023-08-25T08:05:31.461333Z","iopub.status.idle":"2023-08-25T08:05:31.476649Z","shell.execute_reply.started":"2023-08-25T08:05:31.461297Z","shell.execute_reply":"2023-08-25T08:05:31.475287Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# model定義","metadata":{}},{"cell_type":"code","source":"class MyCNN(nn.Module):\n    def __init__(self):\n        super().__init__()\n        \n        self.conv1 = nn.Conv2d(1, 16, kernel_size=5, stride=1, padding=0)\n        self.conv2 = nn.Conv2d(16, 32, kernel_size=5, stride=1, padding=0)\n        self.relu = nn.ReLU()\n        self.maxpool = nn.MaxPool2d(2)\n        \n        self.l1 = nn.LazyLinear(128)\n        self.l2 = nn.Linear(128, 10)\n        \n        self.features = nn.Sequential(\n            self.conv1,\n            self.relu,\n            self.maxpool,\n            self.conv2,\n            self.relu,\n            self.maxpool\n        )\n        \n        self.classifier = nn.Sequential(\n            self.l1,\n            self.relu,\n            self.l2\n        )\n        \n        self.flatten = nn.Flatten()\n        \n    def forward(self, x):\n        x1 = self.features(x)\n        x2 = self.flatten(x1)\n        x3 = self.classifier(x2)\n        \n        return x3","metadata":{"execution":{"iopub.status.busy":"2023-08-25T08:05:32.608380Z","iopub.execute_input":"2023-08-25T08:05:32.609222Z","iopub.status.idle":"2023-08-25T08:05:32.623162Z","shell.execute_reply.started":"2023-08-25T08:05:32.609182Z","shell.execute_reply":"2023-08-25T08:05:32.619441Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"n_iters = 2500 # epochとbatchsize で計算回したい回数\nnum_epoch = int(n_iters / (len(train_df) / batch_size))\n\nmodel = MyCNN()\nerror = nn.CrossEntropyLoss()\nlr = 0.1\noptimizer = torch.optim.SGD(model.parameters(), lr=lr, momentum=0.9)\n\nhistory = np.ones((0, 5))","metadata":{"execution":{"iopub.status.busy":"2023-08-25T08:05:32.825576Z","iopub.execute_input":"2023-08-25T08:05:32.826005Z","iopub.status.idle":"2023-08-25T08:05:32.838503Z","shell.execute_reply.started":"2023-08-25T08:05:32.825971Z","shell.execute_reply":"2023-08-25T08:05:32.837177Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"for epoch in range(num_epoch):\n\n    for images, labels in tqdm(train_loader):\n        break\n    ","metadata":{"execution":{"iopub.status.busy":"2023-08-25T08:05:33.418200Z","iopub.execute_input":"2023-08-25T08:05:33.419020Z","iopub.status.idle":"2023-08-25T08:05:33.611983Z","shell.execute_reply.started":"2023-08-25T08:05:33.418972Z","shell.execute_reply":"2023-08-25T08:05:33.610261Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}