road_map_lr_finder.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Finds an initial learning as described [here](https://arxiv.org/abs/1506.01186)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "from argparse import Namespace\n",
    "\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torchvision\n",
    "import torch.nn.functional as F\n",
    "from pytorch_lightning import Trainer\n",
    "\n",
    "from data_helper import LabeledDataset\n",
    "from helper import collate_fn\n",
    "from road_map_construction import RoadMapNetwork\n",
    "\n",
    "matplotlib.rcParams['figure.figsize'] = [5, 5]\n",
    "matplotlib.rcParams['figure.dpi'] = 200\n",
    "\n",
    "random.seed(0)\n",
    "np.random.seed(0)\n",
    "torch.manual_seed(0)\n",
    "\n",
    "args = Namespace(\n",
    "    NUM_LAYERS=3,\n",
    "    FEATURES_START=256,\n",
    "    DROPOUT=0.5,\n",
    "    LOSS=\"dice_loss\",\n",
    "    BATCH_SIZE=32,\n",
    "    EPOCHS=10,\n",
    "    LEARNING_RATE=0.1,\n",
    "    L2_PENALTY=5e-4\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:lightning:GPU available: True, used: True\n",
      "INFO:lightning:CUDA_VISIBLE_DEVICES: [0]\n",
      "INFO:lightning:Set SLURM handle signals.\n",
      "INFO:lightning:\n",
      "   | Name                            | Type                 | Params\n",
      "---------------------------------------------------------------------\n",
      "0  | feature_extractor               | DenoisingAutoencoder | 788 K \n",
      "1  | feature_extractor.encoder       | Sequential           | 394 K \n",
      "2  | feature_extractor.encoder.0     | Conv2d               | 7 K   \n",
      "3  | feature_extractor.encoder.1     | LeakyReLU            | 0     \n",
      "4  | feature_extractor.encoder.2     | Conv2d               | 295 K \n",
      "5  | feature_extractor.encoder.3     | LeakyReLU            | 0     \n",
      "6  | feature_extractor.encoder.4     | Conv2d               | 73 K  \n",
      "7  | feature_extractor.encoder.5     | LeakyReLU            | 0     \n",
      "8  | feature_extractor.encoder.6     | Conv2d               | 18 K  \n",
      "9  | feature_extractor.encoder.7     | LeakyReLU            | 0     \n",
      "10 | feature_extractor.decoder       | Sequential           | 394 K \n",
      "11 | feature_extractor.decoder.0     | ConvTranspose2d      | 18 K  \n",
      "12 | feature_extractor.decoder.1     | LeakyReLU            | 0     \n",
      "13 | feature_extractor.decoder.2     | ConvTranspose2d      | 73 K  \n",
      "14 | feature_extractor.decoder.3     | LeakyReLU            | 0     \n",
      "15 | feature_extractor.decoder.4     | ConvTranspose2d      | 295 K \n",
      "16 | feature_extractor.decoder.5     | LeakyReLU            | 0     \n",
      "17 | feature_extractor.decoder.6     | ConvTranspose2d      | 6 K   \n",
      "18 | feature_extractor.decoder.7     | Tanh                 | 0     \n",
      "19 | classifier                      | UNet                 | 30 M  \n",
      "20 | classifier.layers               | ModuleList           | 30 M  \n",
      "21 | classifier.layers.0             | DoubleConv           | 1 M   \n",
      "22 | classifier.layers.0.net         | Sequential           | 1 M   \n",
      "23 | classifier.layers.0.net.0       | Conv2d               | 442 K \n",
      "24 | classifier.layers.0.net.1       | BatchNorm2d          | 512   \n",
      "25 | classifier.layers.0.net.2       | ReLU                 | 0     \n",
      "26 | classifier.layers.0.net.3       | Conv2d               | 590 K \n",
      "27 | classifier.layers.0.net.4       | BatchNorm2d          | 512   \n",
      "28 | classifier.layers.0.net.5       | ReLU                 | 0     \n",
      "29 | classifier.layers.1             | Down                 | 3 M   \n",
      "30 | classifier.layers.1.net         | Sequential           | 3 M   \n",
      "31 | classifier.layers.1.net.0       | MaxPool2d            | 0     \n",
      "32 | classifier.layers.1.net.1       | DoubleConv           | 3 M   \n",
      "33 | classifier.layers.1.net.1.net   | Sequential           | 3 M   \n",
      "34 | classifier.layers.1.net.1.net.0 | Conv2d               | 1 M   \n",
      "35 | classifier.layers.1.net.1.net.1 | BatchNorm2d          | 1 K   \n",
      "36 | classifier.layers.1.net.1.net.2 | ReLU                 | 0     \n",
      "37 | classifier.layers.1.net.1.net.3 | Conv2d               | 2 M   \n",
      "38 | classifier.layers.1.net.1.net.4 | BatchNorm2d          | 1 K   \n",
      "39 | classifier.layers.1.net.1.net.5 | ReLU                 | 0     \n",
      "40 | classifier.layers.1.net.dropout | Dropout2d            | 0     \n",
      "41 | classifier.layers.2             | Down                 | 14 M  \n",
      "42 | classifier.layers.2.net         | Sequential           | 14 M  \n",
      "43 | classifier.layers.2.net.0       | MaxPool2d            | 0     \n",
      "44 | classifier.layers.2.net.1       | DoubleConv           | 14 M  \n",
      "45 | classifier.layers.2.net.1.net   | Sequential           | 14 M  \n",
      "46 | classifier.layers.2.net.1.net.0 | Conv2d               | 4 M   \n",
      "47 | classifier.layers.2.net.1.net.1 | BatchNorm2d          | 2 K   \n",
      "48 | classifier.layers.2.net.1.net.2 | ReLU                 | 0     \n",
      "49 | classifier.layers.2.net.1.net.3 | Conv2d               | 9 M   \n",
      "50 | classifier.layers.2.net.1.net.4 | BatchNorm2d          | 2 K   \n",
      "51 | classifier.layers.2.net.1.net.5 | ReLU                 | 0     \n",
      "52 | classifier.layers.2.net.dropout | Dropout2d            | 0     \n",
      "53 | classifier.layers.3             | Up                   | 9 M   \n",
      "54 | classifier.layers.3.upsample    | ConvTranspose2d      | 2 M   \n",
      "55 | classifier.layers.3.conv        | DoubleConv           | 7 M   \n",
      "56 | classifier.layers.3.conv.net    | Sequential           | 7 M   \n",
      "57 | classifier.layers.3.conv.net.0  | Conv2d               | 4 M   \n",
      "58 | classifier.layers.3.conv.net.1  | BatchNorm2d          | 1 K   \n",
      "59 | classifier.layers.3.conv.net.2  | ReLU                 | 0     \n",
      "60 | classifier.layers.3.conv.net.3  | Conv2d               | 2 M   \n",
      "61 | classifier.layers.3.conv.net.4  | BatchNorm2d          | 1 K   \n",
      "62 | classifier.layers.3.conv.net.5  | ReLU                 | 0     \n",
      "63 | classifier.layers.4             | Up                   | 2 M   \n",
      "64 | classifier.layers.4.upsample    | ConvTranspose2d      | 524 K \n",
      "65 | classifier.layers.4.conv        | DoubleConv           | 1 M   \n",
      "66 | classifier.layers.4.conv.net    | Sequential           | 1 M   \n",
      "67 | classifier.layers.4.conv.net.0  | Conv2d               | 1 M   \n",
      "68 | classifier.layers.4.conv.net.1  | BatchNorm2d          | 512   \n",
      "69 | classifier.layers.4.conv.net.2  | ReLU                 | 0     \n",
      "70 | classifier.layers.4.conv.net.3  | Conv2d               | 590 K \n",
      "71 | classifier.layers.4.conv.net.4  | BatchNorm2d          | 512   \n",
      "72 | classifier.layers.4.conv.net.5  | ReLU                 | 0     \n",
      "73 | classifier.layers.5             | Conv2d               | 257   \n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e3554e34ba87449182b46fd7896c9f70",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Finding best initial lr', style=ProgressStyle(description…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = RoadMapNetwork(hparams=args)\n",
    "trainer = Trainer(gpus=1)\n",
    "\n",
    "# Run learning rate finder\n",
    "lr_finder = trainer.lr_find(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot with\n",
    "fig = lr_finder.plot(suggest=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1e-08"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pick point based on plot, or get suggestion\n",
    "lr_finder.suggestion()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:root] *",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}