samples/MNIST_TPU_demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "view-in-github"
   },
   "source": [
    "<a href=\"https://colab.research.google.com/github/butchland/fastai_xla_extensions/blob/master/samples/MNIST_TPU_demo.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7lDQKIEBHMbN"
   },
   "source": [
    "# FastAI XLA Extensions MNIST Demo (Single Core TPU)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "yJAjBc_8HMbT"
   },
   "source": [
    "## How to use"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xv3xARMbHMbW"
   },
   "source": [
    "### Configure the Pytorch XLA package \n",
    "\n",
    "The Pytorch xla package requires an environment supporting TPUs (Kaggle kernels, GCP or Colab environments required)\n",
    "\n",
    "If running on Colab, make sure the Runtime Type is set to TPU.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 132857,
     "status": "ok",
     "timestamp": 1615472739280,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "p2g1lEnkOXg5",
    "outputId": "b13280b3-d8fd-4798-885c-dc9928fe40e4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[K     |████████████████████████████████| 735.4MB 1.2MB/s \n",
      "\u001b[K     |████████████████████████████████| 12.8MB 28.8MB/s \n",
      "\u001b[K     |████████████████████████████████| 7.0MB 5.3MB/s \n",
      "\u001b[?25h"
     ]
    }
   ],
   "source": [
    "!pip install -qqq --no-cache-dir torch==1.7.1+cu101 torchvision==0.8.2+cu101 torchtext==0.8.0 -f https://download.pytorch.org/whl/torch_stable.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "d_0If38OHMbr"
   },
   "outputs": [],
   "source": [
    "#hide_input\n",
    "#colab\n",
    "import os\n",
    "assert os.environ['COLAB_TPU_ADDR'], 'Make sure to select TPU from Edit > Notebook settings > Hardware accelerator'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "29dsMCEVHMb_"
   },
   "source": [
    "Install fastai2 and the fastai_xla_extensions packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 137289,
     "status": "ok",
     "timestamp": 1615472743721,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "ZbPDbj0oaBoO",
    "outputId": "047be875-57fa-40c6-9674-bb723da0f3af"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[K     |████████████████████████████████| 61kB 3.4MB/s eta 0:00:011\n",
      "\u001b[?25h"
     ]
    }
   ],
   "source": [
    "!pip install -Uqq fastcore --upgrade"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 164016,
     "status": "ok",
     "timestamp": 1615472770470,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "6UVX0iopHMbx",
    "outputId": "e9ee8f74-eb33-403b-e7a2-6cf2a43656fd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Updating fastai...\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "#hide\n",
    "#colab\n",
    "!curl -s https://course19.fast.ai/setup/colab | bash "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 140966,
     "status": "ok",
     "timestamp": 1615472747401,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "SDUSw4PFHMcA",
    "outputId": "ba0fc77d-1cd2-4599-f945-82770933886b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[K     |████████████████████████████████| 194kB 4.9MB/s eta 0:00:01\n",
      "\u001b[?25h"
     ]
    }
   ],
   "source": [
    "!pip install -Uqq fastai==2.3.0 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 160532,
     "status": "ok",
     "timestamp": 1615472766977,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "KELFH2r4HMb4",
    "outputId": "ea8bfbc5-7567-4bfc-fd2d-feb3d92da5fe"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[K     |████████████████████████████████| 133.6MB 52kB/s \n",
      "\u001b[K     |████████████████████████████████| 61kB 2.3MB/s \n",
      "\u001b[31mERROR: earthengine-api 0.1.254 has requirement google-api-python-client>=1.12.1, but you'll have google-api-python-client 1.8.0 which is incompatible.\u001b[0m\n",
      "\u001b[?25h"
     ]
    }
   ],
   "source": [
    "!pip install -Uqq cloud-tpu-client==0.10 https://storage.googleapis.com/tpu-pytorch/wheels/torch_xla-1.7-cp37-cp37m-linux_x86_64.whl\n",
    "# for nightly versions\n",
    "# VERSION = \"20200325\"  #@param [\"1.5\" , \"20200325\", \"nightly\"]\n",
    "# !curl https://raw.githubusercontent.com/pytorch/xla/master/contrib/scripts/env-setup.py -o pytorch-xla-env-setup.py\n",
    "# !python pytorch-xla-env-setup.py --version $VERSION"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 170933,
     "status": "ok",
     "timestamp": 1615472777390,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "qsnrEAeiHMcG",
    "outputId": "04fa4d1a-e4c5-4a46-bf94-7acb4450836f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Building wheel for fastai-xla-extensions (setup.py) ... \u001b[?25l\u001b[?25hdone\n"
     ]
    }
   ],
   "source": [
    "!pip install -Uqq fastai_xla_extensions\n",
    "# !pip install -Uqq git+https://github.com/butchland/fastai_xla_extensions.git"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 172553,
     "status": "ok",
     "timestamp": 1615472779020,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "7TSmVCkWgICi",
    "outputId": "47ce8615-c81d-4910-96fb-81a24c9fd78d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch==1.7.1+cu101\n",
      "torch-xla==1.7\n",
      "torchsummary==1.5.1\n",
      "torchtext==0.8.0\n",
      "torchvision==0.8.2+cu101\n",
      "fastai==2.2.7\n",
      "fastai-xla-extensions==0.0.8\n",
      "fastcore==1.3.19\n",
      "fastdtw==0.3.4\n",
      "fastprogress==1.0.0\n",
      "fastrlock==0.5\n"
     ]
    }
   ],
   "source": [
    "!pip freeze | grep torch\n",
    "!pip freeze | grep fast"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "UcLKp-m0HMcV"
   },
   "source": [
    "### Import the libraries\n",
    "Import the fastai2 and fastai_xla_extensions libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 214645,
     "status": "ok",
     "timestamp": 1615472821121,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "XkVauzS0HMcd",
    "outputId": "ed614f9c-84e4-43de-e2c2-86f679cb43cd"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Waiting for TPU to be start up with version pytorch-1.7...\n",
      "WARNING:root:Waiting for TPU to be start up with version pytorch-1.7...\n",
      "WARNING:root:TPU has started up successfully with version pytorch-1.7\n"
     ]
    }
   ],
   "source": [
    "#colab\n",
    "from fastai.vision.all import *\n",
    "from fastai_xla_extensions.all import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "O-5CdtT2HMcm"
   },
   "source": [
    "### Example\n",
    "Build a MNIST classifier -- adapted from fastai course [Lesson 4 notebook](https://github.com/fastai/course-v4/blob/master/nbs/04_mnist_basics.ipynb)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "OI564wSCHMcq"
   },
   "source": [
    "Load MNIST dataset "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 17
    },
    "executionInfo": {
     "elapsed": 232062,
     "status": "ok",
     "timestamp": 1615472838549,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "f7V3P4bpHMcs",
    "outputId": "b1976f70-04a8-4c46-c465-cfc57ba7441e"
   },
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "path = untar_data(URLs.MNIST)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "v8DG9UGJHMcz"
   },
   "outputs": [],
   "source": [
    "#hide\n",
    "Path.BASE_PATH = path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 232053,
     "status": "ok",
     "timestamp": 1615472838553,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "9Ki0w-i2imL2",
    "outputId": "8c7d57be-fd6a-462e-bdb6-d944eef222ec"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(#2) [Path('testing'),Path('training')]"
      ]
     },
     "execution_count": 12,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path.ls()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5pXtb0u2HMc3"
   },
   "source": [
    "Create Fastai DataBlock\n",
    "\n",
    "\n",
    "_Note that batch transforms are currently\n",
    "set to none as they seem to slow the training\n",
    "on the TPU (for investigation)._"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "YspZqFB4HMc5"
   },
   "outputs": [],
   "source": [
    "datablock = DataBlock(\n",
    "    blocks=(ImageBlock,CategoryBlock),\n",
    "    get_items=get_image_files,\n",
    "    splitter=GrandparentSplitter(train_name='training',valid_name='testing'),\n",
    "    get_y=parent_label,\n",
    "    item_tfms=Resize(28),\n",
    "    batch_tfms=[*aug_transforms(do_flip=False, min_scale=0.9), Normalize.from_stats(*imagenet_stats)]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ybNFQ98YHMc9"
   },
   "outputs": [],
   "source": [
    "#colab\n",
    "# datablock.summary(path)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "GbjmFmFJHMdB"
   },
   "source": [
    "Create the dataloader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "siIb-eqcHMdC"
   },
   "outputs": [],
   "source": [
    "dls = datablock.dataloaders(path,bs=256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 8594,
     "status": "ok",
     "timestamp": 1615473591486,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "47hpELWTHMdH",
    "outputId": "cf51cc2e-3c9e-43e4-f5ae-faa59e183846"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cpu')"
      ]
     },
     "execution_count": 42,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dls.device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 536
    },
    "executionInfo": {
     "elapsed": 9579,
     "status": "ok",
     "timestamp": 1615473592480,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "euAYHv4zHMdL",
    "outputId": "911ae7a0-41ba-4e00-8a3d-ff1f74c20c97"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x648 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dls.show_batch()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_09Ilvp5HMdP"
   },
   "source": [
    "Create a Fastai CNN Learner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "BG2ew3YfHMdR"
   },
   "outputs": [],
   "source": [
    "learner = cnn_learner(dls, resnet18, metrics=accuracy, concat_pool=False)\n",
    "                      "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 10473,
     "status": "ok",
     "timestamp": 1615473593394,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "pQ4lp95b_Ezf",
    "outputId": "7c1bb36c-1e2e-4fac-8951-b64a66d97b12"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<fastai.learner.Learner at 0x7fef14330ed0>"
      ]
     },
     "execution_count": 45,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learner.to_xla()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "mgZR8Nz6FV15"
   },
   "outputs": [],
   "source": [
    "learner.dls.device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 10459,
     "status": "ok",
     "timestamp": 1615473593395,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "9y65hL2-olxf",
    "outputId": "ec166b58-271d-4ff2-cb14-c944fd1c6c9a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XLAOptCallback"
      ]
     },
     "execution_count": 47,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learner.xla_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "5t0VEz_diyQF"
   },
   "outputs": [],
   "source": [
    "# learner.fit(1,1e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xDfjq0z3HMdU"
   },
   "source": [
    "Using the `lr_find` works "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 300
    },
    "executionInfo": {
     "elapsed": 58102,
     "status": "ok",
     "timestamp": 1615473641058,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "JHok2FWTHMdW",
    "outputId": "7f7bb15f-e80c-47ad-dafe-44b08e9ba3b9"
   },
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "SuggestedLRs(lr_min=0.05248074531555176, lr_steep=0.03981071710586548)"
      ]
     },
     "execution_count": 49,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#colab\n",
    "learner.lr_find()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "oAnRYPZoHMdb"
   },
   "source": [
    "Fine tune model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 267
    },
    "executionInfo": {
     "elapsed": 715605,
     "status": "ok",
     "timestamp": 1615474298570,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "GNeUgzFXHMdc",
    "outputId": "82dcab5f-8b77-45a3-e503-23d3f38a79f5"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>2.101047</td>\n",
       "      <td>1.452797</td>\n",
       "      <td>0.540300</td>\n",
       "      <td>01:47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>1.209540</td>\n",
       "      <td>1.039177</td>\n",
       "      <td>0.656700</td>\n",
       "      <td>01:46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.862565</td>\n",
       "      <td>0.755549</td>\n",
       "      <td>0.745100</td>\n",
       "      <td>01:54</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.295279</td>\n",
       "      <td>0.225736</td>\n",
       "      <td>0.927600</td>\n",
       "      <td>01:52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.142338</td>\n",
       "      <td>0.139108</td>\n",
       "      <td>0.956400</td>\n",
       "      <td>01:50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.122061</td>\n",
       "      <td>0.162159</td>\n",
       "      <td>0.952700</td>\n",
       "      <td>01:45</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#colab\n",
    "learner.fine_tune(3,freeze_epochs=3, base_lr=1e-3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "2yaPvcZEHMdg"
   },
   "source": [
    "Unfreeze the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "zPWfQMzgHMdh"
   },
   "outputs": [],
   "source": [
    "#colab\n",
    "learner.unfreeze()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HgWpfg1KHMdm"
   },
   "source": [
    "Run the LR Finder again. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 300
    },
    "executionInfo": {
     "elapsed": 755860,
     "status": "ok",
     "timestamp": 1615474338836,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "At3Z1MoNHMdn",
    "outputId": "a4780122-7c7c-41b6-e81f-faa1bbdcc555"
   },
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "SuggestedLRs(lr_min=3.311311274956097e-07, lr_steep=1.0964781722577754e-06)"
      ]
     },
     "execution_count": 52,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#colab\n",
    "learner.lr_find()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "sbIG1OArHMdq"
   },
   "source": [
    "Further fine-tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 80
    },
    "executionInfo": {
     "elapsed": 867171,
     "status": "ok",
     "timestamp": 1615474450158,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "lGz4fgUdHMds",
    "outputId": "460af5b8-372e-4162-8a4c-cafccf65aeff"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.115500</td>\n",
       "      <td>0.138664</td>\n",
       "      <td>0.958200</td>\n",
       "      <td>01:51</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#colab\n",
    "learner.fit_one_cycle(1,slice(1e-5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 870793,
     "status": "ok",
     "timestamp": 1615474453789,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "AdogJJXIHMdw",
    "outputId": "f7f96937-c82a-4202-cf5a-c6c72618938c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Path('models/stage-1.pth')"
      ]
     },
     "execution_count": 54,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#hide\n",
    "#colab\n",
    "learner.save('stage-1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 873686,
     "status": "ok",
     "timestamp": 1615474456690,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "3mnx2DOkHMdz",
    "outputId": "ab52fa7c-7648-4837-f48e-68c731e4c5d6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<fastai.learner.Learner at 0x7fef14330ed0>"
      ]
     },
     "execution_count": 55,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#hide\n",
    "#colab\n",
    "learner.load('stage-1')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "RMXK_lu5HMd4"
   },
   "source": [
    "Model params are using TPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 873679,
     "status": "ok",
     "timestamp": 1615474456691,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "r8noSy1iHMd5",
    "outputId": "56f4cb9c-3cd7-4590-8aa6-c1b497eb6e1b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='xla', index=1)"
      ]
     },
     "execution_count": 56,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#colab\n",
    "one_param(learner.model).device\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Cq5kT-Qb-r1j"
   },
   "source": [
    "Dataloader device is None _(DeviceMoverTransform)_ is actually the one moving the batch input to the TPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 873671,
     "status": "ok",
     "timestamp": 1615474456691,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "U9ApCcZm-l2p",
    "outputId": "f4dde63c-066b-49c9-862c-37ae05103a3f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 57,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#colab\n",
    "learner.dls.device is None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 873665,
     "status": "ok",
     "timestamp": 1615474456692,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "BqGFwCrz-6mk",
    "outputId": "6d370156-e7b5-4146-87bb-407a725ddd3f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(#6) [IntToFloatTensor -- {'div': 255.0, 'div_mask': 1}:\n",
       "encodes: (TensorImage,object) -> encodes\n",
       "(TensorMask,object) -> encodes\n",
       "decodes: (TensorImage,object) -> decodes\n",
       ",Warp -- {'magnitude': 0.2, 'p': 1.0, 'draw_x': None, 'draw_y': None, 'size': None, 'mode': 'bilinear', 'pad_mode': 'reflection', 'batch': False, 'align_corners': True, 'mode_mask': 'nearest'}:\n",
       "encodes: (TensorImage,object) -> encodes\n",
       "(TensorMask,object) -> encodes\n",
       "(TensorBBox,object) -> encodes\n",
       "(TensorPoint,object) -> encodes\n",
       "decodes: ,RandomResizedCropGPU -- {'size': None, 'min_scale': 0.9, 'ratio': (1, 1), 'mode': 'bilinear', 'valid_scale': 1.0, 'p': 1.0}:\n",
       "encodes: (TensorImage,object) -> encodes\n",
       "decodes: ,DeviceMoverTransform -- {'device_to': device(type='xla', index=1), 'device_from': device(type='cpu')}:\n",
       "encodes: (Tensor,object) -> encodes\n",
       "decodes: (Tensor,object) -> decodes\n",
       ",Brightness -- {'max_lighting': 0.2, 'p': 1.0, 'draw': None, 'batch': False}:\n",
       "encodes: (TensorImage,object) -> encodes\n",
       "decodes: ,Normalize -- {'mean': tensor([[[[0.4850]],\n",
       "\n",
       "         [[0.4560]],\n",
       "\n",
       "         [[0.4060]]]]), 'std': tensor([[[[0.2290]],\n",
       "\n",
       "         [[0.2240]],\n",
       "\n",
       "         [[0.2250]]]]), 'axes': (0, 2, 3)}:\n",
       "encodes: (TensorImage,object) -> encodes\n",
       "decodes: (TensorImage,object) -> decodes\n",
       "]"
      ]
     },
     "execution_count": 58,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#colab\n",
    "learner.dls.train.after_batch.fs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 265
    },
    "executionInfo": {
     "elapsed": 873923,
     "status": "ok",
     "timestamp": 1615474456958,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "Xdb3qz7zMtBU",
    "outputId": "5583a15a-7321-43be-b35d-c71ab2a5088e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.recorder.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 276
    },
    "executionInfo": {
     "elapsed": 874377,
     "status": "ok",
     "timestamp": 1615474457420,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "uClF6e_DHMd9",
    "outputId": "17d475f9-2d74-4d3f-b747-22ff23ba653e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#colab\n",
    "learner.recorder.plot_sched()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "YWDGI0REHMeK"
   },
   "source": [
    "Plot loss seems to be working fine."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 17
    },
    "executionInfo": {
     "elapsed": 892767,
     "status": "ok",
     "timestamp": 1615474475819,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "pdpFbBy1QX_z",
    "outputId": "860d84e0-c6eb-44f6-f07b-373729116c07"
   },
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp = ClassificationInterpretation.from_learner(learner)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 311
    },
    "executionInfo": {
     "elapsed": 893522,
     "status": "ok",
     "timestamp": 1615474476586,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "H5gQkQb7QeRI",
    "outputId": "f7b20726-ace7-4ab7-fff4-89518f08677f"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp.plot_confusion_matrix()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 893514,
     "status": "ok",
     "timestamp": 1615474476587,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "uqNCkdOFQqrW",
    "outputId": "66b3a52d-fb17-4ef1-da7f-65b8f592037a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('9', '8', 74),\n",
       " ('5', '8', 55),\n",
       " ('3', '8', 41),\n",
       " ('2', '8', 19),\n",
       " ('7', '9', 16),\n",
       " ('4', '8', 13),\n",
       " ('6', '8', 13),\n",
       " ('5', '3', 12),\n",
       " ('0', '8', 11),\n",
       " ('2', '7', 11),\n",
       " ('4', '7', 9),\n",
       " ('3', '2', 8),\n",
       " ('3', '7', 8),\n",
       " ('0', '6', 7),\n",
       " ('7', '8', 7),\n",
       " ('4', '6', 6),\n",
       " ('4', '9', 6),\n",
       " ('9', '4', 6),\n",
       " ('1', '6', 5),\n",
       " ('2', '9', 5),\n",
       " ('6', '1', 5),\n",
       " ('7', '2', 5),\n",
       " ('9', '7', 5),\n",
       " ('0', '2', 4),\n",
       " ('2', '0', 4),\n",
       " ('2', '3', 4),\n",
       " ('3', '5', 4),\n",
       " ('5', '6', 4),\n",
       " ('6', '0', 4),\n",
       " ('7', '1', 4),\n",
       " ('7', '3', 4)]"
      ]
     },
     "execution_count": 63,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interp.most_confused(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 585
    },
    "executionInfo": {
     "elapsed": 894513,
     "status": "ok",
     "timestamp": 1615474477596,
     "user": {
      "displayName": "Butch Landingin",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh6hxH8XsmO-D_xEWQfgLPP3712rG2YZXMG9j7Z5A=s64",
      "userId": "11910892068509052005"
     },
     "user_tz": -480
    },
    "id": "9AokGWUBQ1Cm",
    "outputId": "73671851-e8a6-4b6a-de1f-64025aea7c7c"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x648 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp.plot_top_losses(9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ZvhT4AJaQ74e"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "TPU",
  "colab": {
   "collapsed_sections": [],
   "name": "MNIST_TPU_demo.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}