Feature/examples (#576)

* Update examples, fix small errors * Update adversarial example notebook * Update svm example * Update GAN example
pytorchbearer · Jun 21, 2019 · fb55c8d · fb55c8d
1 parent 2c6294f
commit fb55c8d
Show file tree

Hide file tree

Showing 9 changed files with 380 additions and 213 deletions.
diff --git a/.gitignore b/.gitignore
@@ -108,4 +108,5 @@ ENV/
 /docs/_static/examples/images
 /docs/_static/examples/logs
 /docs/_static/notebooks/data
-/docs/_static/notebooks/*.png
+/docs/_static/notebooks/*.png
+/docs/_static/notebooks/images
diff --git a/docs/_static/notebooks/adversarial.ipynb b/docs/_static/notebooks/adversarial.ipynb
diff --git a/docs/_static/notebooks/amsgrad.ipynb b/docs/_static/notebooks/amsgrad.ipynb
diff --git a/docs/_static/notebooks/basic_opt.ipynb b/docs/_static/notebooks/basic_opt.ipynb
@@ -17,7 +17,7 @@
     "\n",
     "> **Runtime**   →   **Change runtime type**   →   **Hardware Accelerator: GPU**\n",
     "\n",
-    "## Dependencies\n",
+    "## Install Torchbearer\n",
     "\n",
     "First we install torchbearer if needed. "
    ]

diff --git a/docs/_static/notebooks/callbacks.ipynb b/docs/_static/notebooks/callbacks.ipynb
@@ -10,7 +10,7 @@
     "\n",
     "**Note**: The easiest way to use this tutorial is as a colab notebook, which allows you to dive in with no setup. You won't need a GPU for these examples.\n",
     "\n",
-    "## Dependencies\n",
+    "## Install Torchbearer\n",
     "\n",
     "First we install torchbearer if needed. "
    ]

diff --git a/docs/_static/notebooks/gan.ipynb b/docs/_static/notebooks/gan.ipynb
@@ -7,45 +7,40 @@
     "id": "mpfN9z4hrrYx"
    },
    "source": [
-    "Training a GAN\n",
-    "====================================\n",
+    "# Training a GAN\n",
     "\n",
     "We shall try to implement something more complicated using torchbearer - a Generative Adverserial Network (GAN). This tutorial is a modified version of the [GAN](https://github.com/eriklindernoren/PyTorch-GAN#gan) from the brilliant collection of GAN implementations [PyTorch_GAN](https://github.com/eriklindernoren/PyTorch-GAN) by eriklindernoren on github.\n",
     "\n",
+    "**Note**: The easiest way to use this tutorial is as a colab notebook, which allows you to dive in with no setup. We recommend you enable a free GPU with\n",
     "\n",
-    "First lets import all the necessary packages. "
+    "> **Runtime**   →   **Change runtime type**   →   **Hardware Accelerator: GPU**\n",
+    "\n",
+    "## Install Torchbearer\n",
+    "\n",
+    "First we install torchbearer if needed. "
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "colab": {},
-    "colab_type": "code",
-    "id": "vcMl2qlkr8UM"
-   },
-   "outputs": [],
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "0.3.2\n"
+     ]
+    }
+   ],
    "source": [
-    "import os\n",
-    "\n",
-    "import numpy as np\n",
-    "import torch\n",
-    "import torch.nn as nn\n",
-    "import torchvision.transforms as transforms\n",
-    "from torch.autograd import Variable\n",
-    "from torch.utils.data import DataLoader\n",
-    "from torchvision import datasets\n",
-    "from torchvision.utils import save_image\n",
-    "\n",
     "try:\n",
-    "  import torchbearer\n",
+    "    import torchbearer\n",
     "except:\n",
-    "  !pip install torchbearer\n",
-    "  import torchbearer\n",
-    "  \n",
-    "import torchbearer.callbacks as callbacks\n",
-    "from torchbearer import state_key\n",
-    "from torchbearer.bases import base_closure"
+    "    !pip install -q torchbearer\n",
+    "    import torchbearer\n",
+    "    \n",
+    "print(torchbearer.__version__)"
    ]
   },
   {
@@ -63,24 +58,26 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 2,
    "metadata": {
     "colab": {},
     "colab_type": "code",
     "id": "4uuz1gTYsGLp"
    },
    "outputs": [],
    "source": [
+    "import torch\n",
+    "from torchbearer import state_key\n",
+    "\n",
     "# Define constants\n",
-    "epochs = 200\n",
-    "batch_size = 64\n",
-    "lr = 0.0002\n",
-    "nworkers = 8\n",
+    "train_steps = 50000\n",
+    "batch_size = 128\n",
+    "lr = 0.002\n",
     "latent_dim = 100\n",
     "sample_interval = 400\n",
     "img_shape = (1, 28, 28)\n",
     "adversarial_loss = torch.nn.BCELoss()\n",
-    "device = 'cuda'\n",
+    "device = 'cuda' if torch.cuda.is_available() else 'cpu'\n",
     "valid = torch.ones(batch_size, 1, device=device)\n",
     "fake = torch.zeros(batch_size, 1, device=device)\n",
     "batch = torch.randn(25, latent_dim).to(device)\n",
@@ -112,14 +109,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 3,
    "metadata": {
     "colab": {},
     "colab_type": "code",
     "id": "GBFLYyH5sP1C"
    },
    "outputs": [],
    "source": [
+    "from torchvision import datasets, transforms\n",
+    "\n",
     "transform = transforms.Compose([\n",
     "                        transforms.ToTensor(),\n",
     "                   ])\n",
@@ -142,14 +141,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 4,
    "metadata": {
     "colab": {},
     "colab_type": "code",
     "id": "y2YUr7jWsXTX"
    },
    "outputs": [],
    "source": [
+    "import torch.nn as nn\n",
+    "import numpy as np\n",
+    "\n",
     "class Generator(nn.Module):\n",
     "    def __init__(self):\n",
     "        super(Generator, self).__init__()\n",
@@ -165,13 +167,12 @@
     "            *block(latent_dim, 128, normalize=False),\n",
     "            *block(128, 256),\n",
     "            *block(256, 512),\n",
-    "            *block(512, 1024),\n",
-    "            nn.Linear(1024, int(np.prod(img_shape))),\n",
-    "            nn.Tanh()\n",
+    "            nn.Linear(512, int(np.prod(img_shape))),\n",
+    "            nn.Sigmoid()\n",
     "        )\n",
     "\n",
     "    def forward(self, real_imgs, state):\n",
-    "        z = Variable(torch.Tensor(np.random.normal(0, 1, (real_imgs.shape[0], latent_dim)))).to(state[torchbearer.DEVICE])\n",
+    "        z = torch.randn(real_imgs.shape[0], latent_dim, device = state[torchbearer.DEVICE])\n",
     "        img = self.model(z)\n",
     "        img = img.view(img.size(0), *img_shape)\n",
     "        return img\n",
@@ -209,7 +210,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 5,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -240,7 +241,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 6,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -278,7 +279,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 7,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -298,13 +299,12 @@
    },
    "source": [
     "\n",
-    "Closures\n",
-    "------------------------------------\n",
+    "## Closures\n",
     "The training loop of a GAN is a bit different to a standard model training loop.\n",
     "GANs require separate forward and backward passes for the generator and discriminator.\n",
     "To achieve this in torchbearer we can write a new closure.\n",
     "Since the individual training loops for the generator and discriminator are the same as a\n",
-    "standard training loop we can use a :func:`~torchbearer.bases.base_closure`.\n",
+    "standard training loop we can use a [`~torchbearer.bases.base_closure`](https://torchbearer.readthedocs.io/en/latest/code/main.html#torchbearer.bases.base_closure).\n",
     "The base closure takes state keys for required objects (data, model, optimiser, etc.) and returns a standard closure consisting of:\n",
     "\n",
     "1. Zero gradients\n",
@@ -317,7 +317,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 8,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -342,7 +342,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 9,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -373,14 +373,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 10,
    "metadata": {
     "colab": {},
     "colab_type": "code",
     "id": "FRmFEk8EtklV"
    },
    "outputs": [],
    "source": [
+    "from torchvision.utils import save_image\n",
+    "from torchbearer import callbacks\n",
+    "import os\n",
     "os.makedirs('images', exist_ok=True)\n",
     "\n",
     "@callbacks.on_step_training\n",
@@ -407,7 +410,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 11,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -416,7 +419,7 @@
    "outputs": [],
    "source": [
     "trial = torchbearer.Trial(generator, None, criterion=gen_crit, metrics=metrics, callbacks=[saver_callback])\n",
-    "trial.with_train_generator(dataloader, steps=200000)\n",
+    "trial.with_train_generator(dataloader, steps=train_steps)\n",
     "_ = trial.to(device)"
    ]
   },
@@ -432,7 +435,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 12,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/",
@@ -456,7 +459,45 @@
     "id": "6G-pLJLMtOwo",
     "outputId": "6a8e2b33-0d19-42b7-c250-da8a80572168"
    },
-   "outputs": [],
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "6325852d4ae743a8be2e8c79655578f2",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "HBox(children=(IntProgress(value=0, description='0/1(t)', max=50000), HTML(value='')))"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "[((50000, None),\n",
+       "  {'running_loss': 0.32024282217025757,\n",
+       "   'running_d_loss': 0.32024282217025757,\n",
+       "   'running_g_loss': 2.202518939971924,\n",
+       "   'loss': 0.4093643128871918,\n",
+       "   'd_loss': 0.4093643128871918,\n",
+       "   'g_loss': 1.8098881244659424})]"
+      ]
+     },
+     "execution_count": 12,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
    "source": [
     "new_keys = {DISC_MODEL: discriminator.to(device), DISC_OPT: optimizer_D, GEN_OPT: optimizer_G, DISC_CRIT: disc_crit}\n",
     "trial.state.update(new_keys)\n",
@@ -501,7 +542,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.8"
+   "version": "3.6.7"
   }
  },
  "nbformat": 4,

diff --git a/docs/_static/notebooks/quickstart.ipynb b/docs/_static/notebooks/quickstart.ipynb
@@ -15,7 +15,7 @@
     "\n",
     "> **Runtime**   →   **Change runtime type**   →   **Hardware Accelerator: GPU**\n",
     "\n",
-    "## Dependencies\n",
+    "## Install Torchbearer\n",
     "\n",
     "First we install torchbearer if needed. \n",
     "\n"