reconfigured talk

Author: sethjuarez

data.ipynb

@@ -1,173 +1 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import torch\n",
-    "from torch.utils.data import DataLoader\n",
-    "from torch.utils.data.dataset import Dataset\n",
-    "from torchvision import datasets, transforms"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "# Squares"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "class SquareDataset(Dataset):\n",
-    "    def __init__(self, size):\n",
-    "        self.size = size\n",
-    "        self.X = torch.randint(255, (size, 9), dtype=torch.float)\n",
-    "\n",
-    "        real_w = torch.tensor([[1,1,1,0,0,0,0,0,0],\n",
-    "                               [0,0,0,1,1,1,0,0,0],\n",
-    "                               [0,0,0,0,0,0,1,1,1]], \n",
-    "                               dtype=torch.float)\n",
-    "\n",
-    "        y = torch.argmax(self.X.mm(real_w.t()), 1)\n",
-    "        \n",
-    "        self.Y = torch.zeros(size, 3, dtype=torch.float) \\\n",
-    "                      .scatter_(1, y.view(-1, 1), 1)\n",
-    "\n",
-    "    def __getitem__(self, index):\n",
-    "        return (self.X[index], self.Y[index])\n",
-    "\n",
-    "    def __len__(self):\n",
-    "        return self.size"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "squares = SquareDataset(256)\n",
-    "print(squares[34])\n",
-    "print(squares[254])\n",
-    "print(squares[25])"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "scrolled": false
-   },
-   "outputs": [],
-   "source": [
-    "dataloader = DataLoader(squares, batch_size=5)\n",
-    "\n",
-    "for batch, (X, Y) in enumerate(dataloader):\n",
-    "    print(X, '\\n\\n', Y)\n",
-    "    break"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "# Digits"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "digits = datasets.MNIST('data', train=True, download=True,\n",
-    "                    transform=transforms.Compose([\n",
-    "                        transforms.ToTensor(),\n",
-    "                        transforms.Lambda(lambda x: x.view(28*28))\n",
-    "                    ]),\n",
-    "                    target_transform=transforms.Compose([\n",
-    "                        transforms.Lambda(lambda y: \n",
-    "                                          torch.zeros(10, dtype=torch.float).scatter_(0, torch.tensor(y), value=1))\n",
-    "                    ])\n",
-    "                 )"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "dataloader = DataLoader(digits, batch_size=10, shuffle=True)\n",
-    "\n",
-    "for batch, (X, Y) in enumerate(dataloader):\n",
-    "    print(X, '\\n\\n', Y)\n",
-    "    break"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  }
- ],
- "metadata": {
-  "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": "3.7.3"
-  },
-  "varInspector": {
-   "cols": {
-    "lenName": 16,
-    "lenType": 16,
-    "lenVar": 40
-   },
-   "kernels_config": {
-    "python": {
-     "delete_cmd_postfix": "",
-     "delete_cmd_prefix": "del ",
-     "library": "var_list.py",
-     "varRefreshCmd": "print(var_dic_list())"
-    },
-    "r": {
-     "delete_cmd_postfix": ") ",
-     "delete_cmd_prefix": "rm(",
-     "library": "var_list.r",
-     "varRefreshCmd": "cat(var_dic_list()) "
-    }
-   },
-   "types_to_exclude": [
-    "module",
-    "function",
-    "builtin_function_or_method",
-    "instance",
-    "_Feature"
-   ],
-   "window_display": false
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2
-}
+{"cells":[{"cell_type":"code","execution_count":3,"metadata":{},"outputs":[],"source":"import torch\nfrom torch.utils.data import DataLoader\nfrom torch.utils.data.dataset import Dataset\nfrom torchvision import datasets, transforms"},{"cell_type":"markdown","metadata":{},"outputs":[],"source":"# Dataset"},{"cell_type":"code","execution_count":6,"metadata":{},"outputs":[],"source":"class SquareDataset(Dataset):\n    def __init__(self, size):\n        self.size = size\n        self.X = torch.randint(255, (size, 9), dtype=torch.float)\n\n        real_w = torch.tensor([[1,1,1,0,0,0,0,0,0],\n                               [0,0,0,1,1,1,0,0,0],\n                               [0,0,0,0,0,0,1,1,1]], \n                               dtype=torch.float)\n\n        y = torch.argmax(self.X.mm(real_w.t()), 1)\n        \n        self.Y = torch.zeros(size, 3, dtype=torch.float) \\\n                      .scatter_(1, y.view(-1, 1), 1)\n\n    def __getitem__(self, index):\n        return (self.X[index], self.Y[index])\n\n    def __len__(self):\n        return self.size"},{"cell_type":"code","execution_count":7,"metadata":{},"outputs":[],"source":"squares = SquareDataset(256)\nprint(squares[34])\nprint(squares[254])\nprint(squares[25])"},{"cell_type":"code","execution_count":8,"metadata":{},"outputs":[],"source":"dataloader = DataLoader(squares, batch_size=5)\n\nfor batch, (X, Y) in enumerate(dataloader):\n    print(X, '\\n\\n', Y)\n    break"},{"cell_type":"markdown","metadata":{},"outputs":[],"source":"# Digits\nTransforms!"},{"cell_type":"code","execution_count":4,"metadata":{},"outputs":[],"source":"digits = datasets.MNIST('data', train=True, download=True,\n                    transform=transforms.Compose([\n                        transforms.ToTensor(),\n                        transforms.Lambda(lambda x: x.view(28*28))\n                    ]),\n                    target_transform=transforms.Compose([\n                        transforms.Lambda(lambda y: \n                                          torch.zeros(10, dtype=torch.float).scatter_(0, torch.tensor(y), value=1))\n                    ])\n                 )"},{"cell_type":"code","execution_count":5,"metadata":{},"outputs":[],"source":"dataloader = DataLoader(digits, batch_size=10, shuffle=True)\n\nfor batch, (X, Y) in enumerate(dataloader):\n    print(X, '\\n\\n', Y)\n    break"}],"nbformat":4,"nbformat_minor":2,"metadata":{"language_info":{"name":"python","codemirror_mode":{"name":"ipython","version":3}},"orig_nbformat":2,"file_extension":".py","mimetype":"text/x-python","name":"python","npconvert_exporter":"python","pygments_lexer":"ipython3","version":3}}

digits.ipynb

@@ -1 +1 @@
-{"cells":[{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"import torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nfrom utils.draw import draw_digits\nfrom torch.utils.data import DataLoader\nfrom torch.optim.lr_scheduler import StepLR\nfrom torchvision import datasets, transforms"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"class CNN(nn.Module):\n    def __init__(self):\n        super(CNN, self).__init__()\n        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)\n        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)\n        self.conv2_drop = nn.Dropout2d()\n        self.fc1 = nn.Linear(320, 50)\n        self.fc2 = nn.Linear(50, 10)\n\n    def forward(self, x):\n        x = x.view(-1, 1, 28, 28)\n        x = F.relu(F.max_pool2d(self.conv1(x), 2))\n        x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))\n        x = x.view(-1, 320)\n        x = F.relu(self.fc1(x))\n        x = F.dropout(x, training=self.training)\n        x = self.fc2(x)\n        return F.softmax(x, dim=1)"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"digits = datasets.MNIST('data', download=True,\n                        transform=transforms.Compose([\n                            transforms.ToTensor(),\n                            transforms.Lambda(lambda x: x.reshape(28*28))\n                        ]),\n                        target_transform=transforms.Compose([\n                            transforms.Lambda(lambda y: \n                                              torch.zeros(10, dtype=torch.float).scatter_(0, torch.tensor(y), value=1))\n                        ])\n                     )"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"draw_digits(digits)"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"torch.cuda.is_available()"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n\n# Use the nn package to define our model and loss function.\nmodel = CNN()\nmodel = model.to(device)\n\ncost = torch.nn.BCELoss()\n\n# optimizer which Tensors it should update.\nlearning_rate = 1e-3\noptimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)\nscheduler = StepLR(optimizer, 3)\n\n# dataset!\ndataloader = DataLoader(digits, batch_size=64, num_workers=0, pin_memory=True)\n\nepochs = 10"},{"cell_type":"code","execution_count":null,"metadata":{"scrolled":false},"outputs":[],"source":"for t in range(epochs):\n    print('\\nepoch {}, (lr: {:>.1e})'.format(t, scheduler.get_lr()[0]))\n    print('-------------------------------')\n    for batch, (X, Y) in enumerate(dataloader):\n        X, Y = X.to(device), Y.to(device)\n        optimizer.zero_grad()\n        pred = model(X)\n        loss = cost(pred, Y)\n        loss.backward()\n        optimizer.step()\n        \n        if batch % 100 == 0:\n            print('loss: {:>10f}  [{:>5d}/{:>5d}]'.format(loss.item(), batch * len(X), len(dataloader.dataset)))\n            \n    # step scheduler\n    scheduler.step()"},{"cell_type":"markdown","metadata":{},"source":["# Does it work???"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"test_data = datasets.MNIST('data', train=False, download=True,\n                        transform=transforms.Compose([\n                            transforms.ToTensor(),\n                            transforms.Lambda(lambda x: x.reshape(28*28))\n                        ]),\n                        target_transform=transforms.Compose([\n                            transforms.Lambda(lambda y: \n                                              torch.zeros(10, dtype=torch.float).scatter_(0, torch.tensor(y), value=1))\n                        ])\n                     )\ntest_loader = DataLoader(digits, batch_size=64, num_workers=0, pin_memory=True)"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"model.eval()\ntest_loss = 0\ncorrect = 0\nwith torch.no_grad():\n    for batch, (X, Y) in enumerate(test_loader):\n        X, Y = X.to(device), Y.to(device)\n        pred = model(X)\n\n        test_loss += cost(pred, Y).item()\n        correct += (pred.argmax(1) == Y.argmax(1)).type(torch.float).sum().item()\n\ntest_loss /= len(dataloader.dataset)\ncorrect /= len(dataloader.dataset)\nprint('Test Error:')\nprint('acc: {:>0.1f}%, avg loss: {:>8f}'.format(100*correct, test_loss))"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"len(test_data)"},{"cell_type":"markdown","metadata":{},"source":["# Saving Things!"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":"import torch.onnx as onnx\n\n# create dummy variable to traverse graph\nx = torch.randint(255, (1, 28*28), dtype=torch.float).to(device) / 255\nonnx.export(model, x, 'superfile.onnx')\nprint('Saved onnx model to \"superfile.onnx\"')"}],"nbformat":4,"nbformat_minor":2,"metadata":{"language_info":{"name":"python","codemirror_mode":{"name":"ipython","version":3}},"orig_nbformat":2,"file_extension":".py","mimetype":"text/x-python","name":"python","npconvert_exporter":"python","pygments_lexer":"ipython3","version":3}}
+{"cells":[{"cell_type":"code","execution_count":1,"metadata":{},"outputs":[],"source":"import torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nfrom utils.draw import draw_digits\nfrom torch.utils.data import DataLoader\nfrom torch.optim.lr_scheduler import StepLR\nfrom torchvision import datasets, transforms"},{"cell_type":"code","execution_count":2,"metadata":{},"outputs":[],"source":"class CNN(nn.Module):\n    def __init__(self):\n        super(CNN, self).__init__()\n        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)\n        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)\n        self.conv2_drop = nn.Dropout2d()\n        self.fc1 = nn.Linear(320, 50)\n        self.fc2 = nn.Linear(50, 10)\n\n    def forward(self, x):\n        x = x.view(-1, 1, 28, 28)\n        x = F.relu(F.max_pool2d(self.conv1(x), 2))\n        x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))\n        x = x.view(-1, 320)\n        x = F.relu(self.fc1(x))\n        x = F.dropout(x, training=self.training)\n        x = self.fc2(x)\n        return F.softmax(x, dim=1)"},{"cell_type":"code","execution_count":3,"metadata":{},"outputs":[],"source":"digits = datasets.MNIST('data', download=True,\n                        transform=transforms.Compose([\n                            transforms.ToTensor(),\n                            transforms.Lambda(lambda x: x.reshape(28*28))\n                        ]),\n                        target_transform=transforms.Compose([\n                            transforms.Lambda(lambda y: \n                                              torch.zeros(10, dtype=torch.float).scatter_(0, torch.tensor(y), value=1))\n                        ])\n                     )"},{"cell_type":"code","execution_count":4,"metadata":{},"outputs":[],"source":"draw_digits(digits)"},{"cell_type":"code","execution_count":5,"metadata":{},"outputs":[],"source":"torch.cuda.is_available()"},{"cell_type":"code","execution_count":6,"metadata":{},"outputs":[],"source":"device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n\n# Use the nn package to define our model and loss function.\nmodel = CNN()\nmodel = model.to(device)\n\ncost = torch.nn.BCELoss()\n\n# optimizer which Tensors it should update.\nlearning_rate = 1e-3\noptimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)\nscheduler = StepLR(optimizer, 3)\n\n# dataset!\ndataloader = DataLoader(digits, batch_size=64, num_workers=0, pin_memory=True)\n\nepochs = 10"},{"cell_type":"code","execution_count":7,"metadata":{},"outputs":[],"source":"for t in range(epochs):\n    print('\\nepoch {}, (lr: {:>.1e})'.format(t, scheduler.get_lr()[0]))\n    print('-------------------------------')\n    for batch, (X, Y) in enumerate(dataloader):\n        X, Y = X.to(device), Y.to(device)\n        optimizer.zero_grad()\n        pred = model(X)\n        loss = cost(pred, Y)\n        loss.backward()\n        optimizer.step()\n        \n        if batch % 100 == 0:\n            print('loss: {:>10f}  [{:>5d}/{:>5d}]'.format(loss.item(), batch * len(X), len(dataloader.dataset)))\n            \n    # step scheduler\n    scheduler.step()"},{"cell_type":"markdown","metadata":{},"outputs":[],"source":"# Does it work???"},{"cell_type":"code","execution_count":8,"metadata":{},"outputs":[],"source":"test_data = datasets.MNIST('data', train=False, download=True,\n                        transform=transforms.Compose([\n                            transforms.ToTensor(),\n                            transforms.Lambda(lambda x: x.reshape(28*28))\n                        ]),\n                        target_transform=transforms.Compose([\n                            transforms.Lambda(lambda y: \n                                              torch.zeros(10, dtype=torch.float).scatter_(0, torch.tensor(y), value=1))\n                        ])\n                     )\ntest_loader = DataLoader(digits, batch_size=64, num_workers=0, pin_memory=True)"},{"cell_type":"code","execution_count":9,"metadata":{},"outputs":[],"source":"model.eval()\ntest_loss = 0\ncorrect = 0\nwith torch.no_grad():\n    for batch, (X, Y) in enumerate(test_loader):\n        X, Y = X.to(device), Y.to(device)\n        pred = model(X)\n\n        test_loss += cost(pred, Y).item()\n        correct += (pred.argmax(1) == Y.argmax(1)).type(torch.float).sum().item()\n\ntest_loss /= len(dataloader.dataset)\ncorrect /= len(dataloader.dataset)\nprint('Test Error:')\nprint('acc: {:>0.1f}%, avg loss: {:>8f}'.format(100*correct, test_loss))"},{"cell_type":"code","execution_count":10,"metadata":{},"outputs":[],"source":"len(test_data)"},{"cell_type":"markdown","metadata":{},"outputs":[],"source":"# Saving Things!"},{"cell_type":"code","execution_count":11,"metadata":{},"outputs":[],"source":"import torch.onnx as onnx\n\n# create dummy variable to traverse graph\nx = torch.randint(255, (1, 28*28), dtype=torch.float).to(device) / 255\nonnx.export(model, x, 'superfile.onnx')\nprint('Saved onnx model to \"superfile.onnx\"')"},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":""}],"nbformat":4,"nbformat_minor":2,"metadata":{"language_info":{"name":"python","codemirror_mode":{"name":"ipython","version":3}},"orig_nbformat":2,"file_extension":".py","mimetype":"text/x-python","name":"python","npconvert_exporter":"python","pygments_lexer":"ipython3","version":3}}