tiny img + data aug

Author: slegroux

config/data/image/tiny_imagenet_aug.yaml

@@ -4,7 +4,6 @@ data_dir: '../data/image'
 train_val_split: [0.8, 0.2]
 batch_size: 512
 num_workers: 0
-size: 64
 pin_memory: True
 persistent_workers: False
 transforms: 
@@ -15,11 +14,10 @@ transforms:
       mean: [0.4822, 0.4495, 0.3985]
       std: [0.2771, 0.2690, 0.2826]
     - _target_: torchvision.transforms.Resize
-      size: [${..size},${..size}]
+      size: 64
     - _target_: torchvision.transforms.RandomCrop
-      size: [${..size},${..size}]
+      size: 64
     - _target_: torchvision.transforms.RandomHorizontalFlip
     - _target_: torchvision.transforms.RandomVerticalFlip
-    - _target_: torchvision.transforms.RandomRotation
 
 

tutorials/super_resolution.ipynb

@@ -28,25 +28,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 16,
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "Seed set to 42\n",
-      "/Users/slegroux/miniforge3/envs/nimrod/lib/python3.11/site-packages/torchvision/transforms/v2/_deprecated.py:42: UserWarning: The transform `ToTensor()` is deprecated and will be removed in a future release. Instead, please use `v2.Compose([v2.ToImage(), v2.ToDtype(torch.float32, scale=True)])`.Output is equivalent up to float precision.\n",
-      "  warnings.warn(\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
+    "import torch.nn as nn\n",
+    "import torch\n",
+    "\n",
     "from nimrod.image.datasets import ImageDataset, ImageDataModule\n",
     "from nimrod.models.core import lr_finder, train_one_cycle\n",
+    "from nimrod.models.resnet import ResBlock\n",
+    "\n",
     "from hydra.utils import instantiate\n",
     "from omegaconf import OmegaConf\n",
-    "from rich import print"
+    "from rich import print\n",
+    "from typing import Optional, Type"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## tiny imagenet"
    ]
   },
   {
@@ -71,13 +74,6 @@
     ")"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  },
   {
    "cell_type": "code",
    "execution_count": 5,
@@ -148,6 +144,79 @@
     "print(dm.dim)"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#| export\n",
+    "class UpBlock(nn.Module):\n",
+    "    def __init__(\n",
+    "        self,\n",
+    "        in_channels:int,\n",
+    "        out_channels:int,\n",
+    "        kernel_size:int=3,\n",
+    "        activation:Optional[Type[nn.Module]]=nn.ReLU\n",
+    "    ):\n",
+    "        super().__init__()\n",
+    "        layers = []\n",
+    "        # upsample receptive field\n",
+    "        layers.append(nn.UpsamplingNearest2d(scale_factor=2))\n",
+    "        # resnet block increase channels\n",
+    "        layers.append(ResBlock(in_channels, out_channels, kernel_size=kernel_size, activation=activation))\n",
+    "        self.nnet = nn.Sequential(*layers)\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        return self.nnet(x)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 31,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"color: #800080; text-decoration-color: #800080; font-weight: bold\">torch.Size</span><span style=\"font-weight: bold\">([</span><span style=\"color: #008080; text-decoration-color: #008080; font-weight: bold\">1</span>, <span style=\"color: #008080; text-decoration-color: #008080; font-weight: bold\">8</span>, <span style=\"color: #008080; text-decoration-color: #008080; font-weight: bold\">128</span>, <span style=\"color: #008080; text-decoration-color: #008080; font-weight: bold\">128</span><span style=\"font-weight: bold\">])</span>\n",
+       "</pre>\n"
+      ],
+      "text/plain": [
+       "\u001b[1;35mtorch.Size\u001b[0m\u001b[1m(\u001b[0m\u001b[1m[\u001b[0m\u001b[1;36m1\u001b[0m, \u001b[1;36m8\u001b[0m, \u001b[1;36m128\u001b[0m, \u001b[1;36m128\u001b[0m\u001b[1m]\u001b[0m\u001b[1m)\u001b[0m\n"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "m = UpBlock(3, 8)\n",
+    "x = torch.randn(1, 3, 64, 64)\n",
+    "y = m(x)\n",
+    "print(y.shape)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 29,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "torch.Size([1, 3, 128, 128])"
+      ]
+     },
+     "execution_count": 29,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "m.nnet[0](x).shape"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,