7_9_resnext50_bs32.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Kaggle Competition:  Iceberg or Ship"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# resnext50 architecture \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total 5380\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu 1405706 Jan 15 01:35 1b_getdata-csv.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  625023 Jan 14 18:47 1_getdata.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  245619 Jan 14 18:36 2_model_resnet34.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu   13742 Jan 15 01:35 3_model_resnext50.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu   14220 Jan 13 21:14 4_model_nasnet.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu   64965 Jan 14 19:40 4_resnext_resize.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  147078 Jan 14 19:57 5_resnet18_resize.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  138629 Jan 14 19:52 6_resnext50_resize.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  205624 Jan 20 20:02 7_1_resnext50_more_epoch_submitted.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  210001 Jan 15 18:51 7_2_resnext50_more_epoch.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  254485 Jan 20 19:51 7_5_resnext50_more_epoch_update_sz.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  253300 Jan 20 20:00 7_6_resnext50_more_epoch_update_sz.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  255888 Jan 20 22:14 7_7_resnext50_more_epoch_update_sz_zm15.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  326004 Jan 20 21:44 7_8_resnext50_more_epoch_update_sz_zm16_dropout.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu   15701 Jan 21 00:31 7_9_resnext50_bs32.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  179449 Jan 14 21:28 7_resnext50_sz_100.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  201400 Jan 15 02:17 8_resnext50_more_epoch_sz300.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  137540 Jan 14 18:36 baseline.ipynb\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu    3299 Jan 15 01:35 helper.py\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu  767392 Jan 14 18:36 other_groverpr.ipynb\r\n",
      "drwxrwxr-x 2 ubuntu ubuntu    4096 Jan 15 01:42 __pycache__\r\n",
      "-rw-rw-r-- 1 ubuntu ubuntu    2170 Jan 20 21:49 README.md\r\n"
     ]
    }
   ],
   "source": [
    "!ls -l"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Put these at the top of every notebook, to get automatic reloading and inline plotting\n",
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai.imports import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This file contains all the main external libs we'll use\n",
    "from fastai.imports import *\n",
    "from fastai.transforms import *\n",
    "from fastai.conv_learner import *\n",
    "from fastai.model import *\n",
    "from fastai.dataset import *\n",
    "from fastai.sgdr import *\n",
    "from fastai.plots import *\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import shutil\n",
    "from IPython.display import FileLink, FileLinks\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "import helper"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#PATH = \"/home/nidhin/data/iceberg/\"\n",
    "PATH = os.path.expanduser(\"~/data/iceberg/\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!ls -l {PATH}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Set Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "arch=resnext50\n",
    "sz=128\n",
    "bs=32\n",
    "lr = 0.0075"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "614a8795e19a479685a252eb150ddbec",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=6), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "def get_data(sz, bs):\n",
    "    tfms = tfms_from_model(arch, sz, aug_tfms=transforms_top_down, max_zoom=1.05)\n",
    "    data = ImageClassifierData.from_paths(PATH, tfms=tfms, bs=bs,test_name='test')\n",
    "    return data\n",
    "\n",
    "data = get_data(sz, bs)\n",
    "data = data.resize(int(sz*1.0), 'tmp')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#tfms = tfms_from_model(arch, sz, aug_tfms=transforms_side_on, max_zoom=1.1)\n",
    "#data = ImageClassifierData.from_paths(PATH, tfms=tfms, bs=bs,test_name='test',)\n",
    "\n",
    "learn = ConvLearner.pretrained(arch, data, precompute=True, ps=0.55)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning Rate Finder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fbf7099bd7a94b6ab10e7a45331565b8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=1), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 61%|██████    | 25/41 [00:00<00:00, 26.40it/s, loss=13]   \n",
      "                                                        \r"
     ]
    }
   ],
   "source": [
    "lrf = learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbfd98b6be0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.sched.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbfd98b6940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.sched.plot() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4c36a1c04005488684f042729bcba07d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=9), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.      0.58914 0.38332 0.82954]                          \n",
      "[1.      0.49857 0.35798 0.84091]                          \n",
      "[2.      0.49615 0.37098 0.82954]                          \n",
      "[3.      0.45108 0.38589 0.83523]                          \n",
      "[4.      0.42321 0.36436 0.83239]                          \n",
      "[5.      0.40413 0.34622 0.86079]                          \n",
      "[6.      0.39234 0.3788  0.83807]                          \n",
      "[7.      0.34126 0.36086 0.82386]                          \n",
      "[8.      0.32163 0.40159 0.80966]                          \n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn.fit(lr, 9)\n",
    "learn.precompute=False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Stochastic Gradient Descent with Restart"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "lr = 0.0075"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d22e6be9d4014d1c85d771ea11fe1e8b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=31), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.      0.49519 0.34976 0.83807]                          \n",
      "[1.      0.52507 0.4109  0.73579]                          \n",
      "[2.      0.50282 0.38799 0.83807]                          \n",
      "[3.      0.50093 0.39744 0.82102]                          \n",
      "[4.      0.49359 0.35228 0.85511]                          \n",
      "[5.      0.45569 0.34807 0.86648]                          \n",
      "[6.      0.44707 0.33876 0.84943]                          \n",
      "[7.      0.44135 0.37259 0.83523]                          \n",
      "[8.      0.46557 0.35429 0.84375]                          \n",
      "[9.      0.46279 0.36772 0.84659]                          \n",
      "[10.       0.4748   0.34667  0.83807]                      \n",
      "[11.       0.45682  0.35541  0.84659]                      \n",
      "[12.       0.43367  0.35683  0.84943]                      \n",
      "[13.       0.43849  0.35987  0.83807]                      \n",
      "[14.       0.42141  0.35925  0.82954]                      \n",
      "[15.       0.43512  0.35571  0.83239]                      \n",
      "[16.       0.43175  0.35148  0.83239]                      \n",
      "[17.       0.46023  0.36394  0.84091]                      \n",
      "[18.       0.47645  0.33597  0.84659]                      \n",
      "[19.       0.45793  0.3491   0.83807]                      \n",
      "[20.       0.45832  0.3354   0.83239]                      \n",
      "[21.       0.44504  0.35342  0.84375]                      \n",
      "[22.       0.45346  0.33515  0.84659]                      \n",
      "[23.       0.46197  0.34799  0.84943]                      \n",
      "[24.       0.43829  0.33671  0.85511]                      \n",
      "[25.       0.44923  0.34409  0.84091]                      \n",
      "[26.       0.42269  0.33563  0.84943]                      \n",
      "[27.       0.40611  0.33201  0.84943]                      \n",
      "[28.       0.39033  0.33294  0.85227]                      \n",
      "[29.       0.42099  0.33231  0.84659]                      \n",
      "[30.       0.4236   0.33621  0.85796]                      \n",
      "\n"
     ]
    }
   ],
   "source": [
    "#learn.fit(lr, 4, cycle_len=2)\n",
    "learn.fit(lr, 5, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbf9843e5f8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.sched.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbf98090d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.333332394520013, 0.8411214953271028)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# lr/10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# lr/9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# lr/3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# lr/4"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Unfreeze Layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# unfreeze layers\n",
    "\n",
    "learn.unfreeze()\n",
    "#lrs = np.array([lr/15,lr/12,lr/9])\n",
    "#lrs = np.array([lr/9,lr/3,lr/1])\n",
    "\n",
    "lrs = np.array([lr/11,lr/5,lr/2])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Run Learner Again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e458eae552a64ee18f46f689ec25d017",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=63), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.      0.5125  0.35994 0.85227]                          \n",
      "[1.      0.4873  0.29016 0.86364]                          \n",
      "[2.      0.40126 0.31746 0.88352]                          \n",
      "[3.      0.34912 0.27811 0.86932]                          \n",
      "[4.      0.34752 0.25278 0.88636]                          \n",
      "[5.      0.31504 0.28099 0.86932]                          \n",
      "[6.      0.27475 0.2769  0.86648]                          \n",
      "[7.      0.2791  0.26938 0.88921]                          \n",
      "[8.      0.30772 0.26277 0.86932]                          \n",
      "[9.      0.31938 0.2383  0.89773]                          \n",
      "[10.       0.32929  0.25089  0.88636]                      \n",
      "[11.       0.32302  0.26279  0.88636]                      \n",
      "[12.       0.28451  0.27441  0.87784]                      \n",
      "[13.       0.25382  0.23607  0.89773]                      \n",
      "[14.       0.24284  0.263    0.88921]                      \n",
      "[15.       0.26591  0.34303  0.84091]                      \n",
      "[16.       0.31529  0.33585  0.85511]                      \n",
      "[17.       0.31004  0.36745  0.84091]                      \n",
      "[18.       0.31     0.27708  0.86648]                      \n",
      "[19.       0.3273   0.33432  0.86932]                      \n",
      "[20.       0.29274  0.3535   0.86648]                      \n",
      "[21.       0.282    0.23202  0.88921]                      \n",
      "[22.       0.27662  0.25852  0.87784]                      \n",
      "[23.       0.24266  0.2406   0.89204]                      \n",
      "[24.       0.22303  0.24664  0.88921]                      \n",
      "[25.       0.20969  0.25598  0.87784]                      \n",
      "[26.       0.20739  0.25315  0.88068]                      \n",
      "[27.       0.22146  0.22365  0.90341]                      \n",
      "[28.       0.21011  0.31121  0.88068]                      \n",
      "[29.       0.21411  0.2674   0.88636]                      \n",
      "[30.       0.25078  0.27494  0.88921]                      \n",
      "[31.       0.28343  0.29865  0.875  ]                      \n",
      "[32.       0.32911  0.23919  0.89489]                      \n",
      "[33.       0.34214  0.36736  0.85796]                      \n",
      "[34.       0.34371  0.24471  0.88352]                      \n",
      "[35.       0.36206  0.41033  0.84659]                      \n",
      "[36.       0.31564  0.31259  0.875  ]                      \n",
      "[37.       0.31439  0.33086  0.86648]                      \n",
      "[38.       0.29205  0.23101  0.89773]                      \n",
      "[39.       0.26514  0.25737  0.88352]                      \n",
      "[40.       0.23247  0.23578  0.89489]                      \n",
      "[41.       0.23461  0.26832  0.89204]                      \n",
      "[42.       0.26883  0.27565  0.89204]                      \n",
      "[43.       0.28195  0.29177  0.86079]                      \n",
      "[44.       0.2246   0.28898  0.87216]                      \n",
      "[45.       0.23623  0.27536  0.88636]                      \n",
      "[46.       0.23205  0.23353  0.89489]                      \n",
      "[47.       0.20394  0.30007  0.88921]                      \n",
      "[48.       0.19011  0.30107  0.89204]                      \n",
      "[49.       0.19163  0.26857  0.90341]                      \n",
      "[50.       0.17314  0.27137  0.89204]                      \n",
      "[51.       0.15789  0.25387  0.89773]                      \n",
      "[52.       0.15534  0.25101  0.90625]                      \n",
      "[53.       0.14899  0.281    0.89773]                      \n",
      "[54.       0.14303  0.28411  0.90341]                      \n",
      "[55.       0.15562  0.28343  0.90625]                      \n",
      "[56.       0.1397   0.23035  0.91761]                      \n",
      "[57.       0.13164  0.27475  0.90057]                      \n",
      "[58.       0.11998  0.27829  0.89773]                      \n",
      "[59.       0.16301  0.25888  0.91477]                      \n",
      "[60.       0.13676  0.25525  0.90341]                      \n",
      "[61.       0.13701  0.26686  0.90625]                      \n",
      "[62.       0.12712  0.31445  0.88352]                      \n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn.fit(lrs, 6, cycle_len=1, cycle_mult=2)\n",
    "#learn.fit(lrs, 5, cycle_len=1, cycle_mult=2,use_wd_sched=True,wds=wds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbf9966b0f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.sched.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbf980351d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Log Loss | Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.22724770442431763, 0.897196261682243)"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('resnext50_sz75_zm15_sequ_4')\n",
    "learn.load('resnext50_sz75_zm15_sequ_4')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2e3f9dde1b884237a2078f023ef96e9c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=31), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.      0.15277 0.2538  0.90909]                          \n",
      "[1.      0.22474 0.31864 0.875  ]                          \n",
      "[2.      0.16597 0.24882 0.90625]                          \n",
      "[3.      0.1705  0.29324 0.89489]                          \n",
      "[4.      0.15465 0.29524 0.87216]                          \n",
      "[5.      0.15689 0.2943  0.89204]                          \n",
      "[6.      0.14741 0.2499  0.90909]                          \n",
      "[7.      0.14434 0.26739 0.89489]                          \n",
      "[8.      0.1664  0.27573 0.88352]                          \n",
      "[9.      0.17475 0.21598 0.91477]                          \n",
      "[10.       0.1468   0.2635   0.89773]                      \n",
      "[11.       0.12381  0.2555   0.90341]                      \n",
      "[12.       0.12278  0.25627  0.91477]                      \n",
      "[13.       0.15629  0.28315  0.90341]                      \n",
      "[14.       0.11612  0.25585  0.91193]                      \n",
      "[15.       0.10963  0.28706  0.88636]                      \n",
      "[16.       0.09801  0.39648  0.86932]                       \n",
      "[17.       0.10111  0.32353  0.89204]                       \n",
      "[18.       0.15559  0.6089   0.80398]                      \n",
      "[19.       0.31464  0.29457  0.88352]                      \n",
      "[20.       0.26932  0.29557  0.88068]                      \n",
      "[21.       0.2277   0.25074  0.88921]                      \n",
      "[22.       0.27436  0.2333   0.91193]                      \n",
      "[23.       0.20607  0.28225  0.88636]                      \n",
      "[24.       0.16791  0.24639  0.90625]                      \n",
      "[25.       0.15601  0.27951  0.89204]                      \n",
      "[26.       0.18618  0.26971  0.90057]                      \n",
      "[27.       0.15603  0.26465  0.90341]                      \n",
      "[28.       0.1334   0.30864  0.89489]                      \n",
      "[29.       0.17631  0.27247  0.90909]                      \n",
      "[30.       0.14434  0.29834  0.89204]                      \n",
      "\n",
      "                                             \r"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.2085947903473867, 0.9065420560747663)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbf8e7c56d8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Run Learner again\n",
    "\n",
    "learn.fit(lrs, 5, cycle_len=1, cycle_mult=2)\n",
    "\n",
    "learn.sched.plot_loss()\n",
    "\n",
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('resnext50_sz75_zm15_sequ_5')\n",
    "learn.load('resnext50_sz75_zm15_sequ_5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    }
   ],
   "source": [
    "helper.make_submission(learn=learn, data=data, output_file_name = f'{PATH}sub/resnext50_sz75_zm15_sequ_5.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ce6134bd2a7248a886f6c5fab653f879",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=31), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.      0.12886 0.26851 0.90909]                          \n",
      "[1.      0.13513 0.26437 0.89773]                          \n",
      "[2.      0.17852 0.30053 0.89204]                          \n",
      "[3.      0.17632 0.32158 0.88636]                          \n",
      "[4.      0.18563 0.30407 0.87784]                          \n",
      "[5.      0.14389 0.28103 0.89773]                          \n",
      "[6.      0.14204 0.3219  0.87784]                          \n",
      "[7.      0.23589 0.22447 0.90909]                          \n",
      "[8.      0.25183 0.21184 0.91193]                          \n",
      "[9.      0.20777 0.22018 0.90909]                          \n",
      "[10.       0.18591  0.2622   0.90625]                      \n",
      "[11.       0.14641  0.25867  0.90909]                      \n",
      "[12.       0.14831  0.25211  0.92046]                      \n",
      "[13.       0.1443   0.25274  0.91193]                      \n",
      "[14.       0.11526  0.23744  0.91193]                      \n",
      "[15.       0.12268  0.26277  0.91477]                      \n",
      "[16.       0.12457  0.3015   0.91193]                      \n",
      "[17.       0.15982  0.42422  0.89204]                      \n",
      "[18.       0.15655  0.2942   0.89773]                      \n",
      "[19.       0.14317  0.22529  0.92329]                      \n",
      "[20.       0.10827  0.26758  0.90057]                      \n",
      "[21.       0.09365  0.32936  0.89773]                       \n",
      "[22.       0.11531  0.3729   0.89204]                       \n",
      "[23.       0.10638  0.27948  0.90341]                      \n",
      "[24.       0.10517  0.26335  0.92046]                      \n",
      "[25.       0.08263  0.29391  0.91477]                       \n",
      "[26.       0.08973  0.28632  0.91477]                       \n",
      "[27.       0.08738  0.30903  0.91761]                       \n",
      "[28.       0.07474  0.28013  0.92046]                       \n",
      "[29.       0.07108  0.28627  0.90909]                       \n",
      "[30.       0.07115  0.27455  0.90909]                       \n",
      "\n",
      "                                             \r"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.22784101516728522, 0.9190031152647975)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fbf98493470>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Run Learner again\n",
    "\n",
    "learn.fit(lrs, 5, cycle_len=1, cycle_mult=2)\n",
    "\n",
    "learn.sched.plot_loss()\n",
    "\n",
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('resnext50_sz75_zm15_sequ_6')\n",
    "learn.load('resnext50_sz75_zm15_sequ_6')\n",
    "#helper.make_submission(learn=learn, data=data, output_file_name = f'{PATH}sub/resnext50_sz75_zm15_sequ_4.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Resize the images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_data(sz, bs):\n",
    "    tfms = tfms_from_model(arch, sz, aug_tfms=transforms_top_down, max_zoom=1.10)\n",
    "    data = ImageClassifierData.from_paths(PATH, tfms=tfms, bs=bs,test_name='test')\n",
    "    return data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Image Size of 140"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#train for image size\n",
    "learn.set_data(get_data(140, 16))\n",
    "\n",
    "learn.freeze()\n",
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.unfreeze()\n",
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.sched.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Log Loss | Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('resnext50_sz75_zm15_sequ_4')\n",
    "learn.load('resnext50_sz75_zm15_sequ_4')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#helper.make_submission(learn=learn, data=data, output_file_name = f'{PATH}sub/resnext50_sz75_zm15_sequ_4.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Image Size of 150"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#train for image size \n",
    "learn.set_data(get_data(150, 16))\n",
    "\n",
    "learn.freeze()\n",
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)\n",
    "\n",
    "learn.unfreeze()\n",
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.sched.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Log Loss | Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Save learner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('resnext50_sz75_sequ_5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.load('resnext50_sz75_sequ_5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Output predictions to csv file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#helper.make_submission(learn=learn, data=data, output_file_name = f'{PATH}sub/resnext50_sz75_sequ_5.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# END FILE"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Run Learner again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.fit(lrs, 5, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Run Learner again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.fit(lrs, 5, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Log Loss | Accuracy (final model here)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_preds,y = learn.TTA()\n",
    "preds = np.mean(np.exp(log_preds),axis=0)\n",
    "(metrics.log_loss(y,preds), accuracy(preds,y))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# test_df = pd.read_json(PATH+\"/test.json\")\n",
    "# helper.make_test_file(log_preds,test_df,model_name='resnext50_sz75_zm15_bs32_comp_band',\n",
    "#                       classes=data.classes,\n",
    "#                       submission_folder =f'{PATH}sub'\n",
    "#                      )"
   ]
  }
 ],
 "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}