{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "RZiRah3QiR_G"
   },
   "source": [
    "# Allegro ClearML logging example\n",
    "\n",
    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/allegroai/clearml/blob/master/examples/reporting/Allegro_Trains_logging_example.ipynb)\n",
    "\n",
    "This example introduces ClearML [Logger](https://allegro.ai/docs/logger.html) functionality. Logger is the ClearML console log and metric interface.\n",
    "\n",
    "You can find more reporting examples [here](https://github.com/allegroai/clearml/tree/master/examples/reporting)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "RbESVEV0jl3c"
   },
   "outputs": [],
   "source": [
    "!pip install clearml\n",
    "!pip install numpy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "8p9tkR5wue2x"
   },
   "source": [
    "### Create a new Task\n",
    "Create a new Task and get a Logger object for the Task.\n",
    "\n",
    "To create a new Task object, call the `Task.init` method providing it with `project_name` (the project name for the experiment) and `task_name` (the name of the experiment). When `Task.init` executes, a link to the Web UI Results page for the newly generated Task will be printed, and the Task will be updated in real time in the ClearML demo server.\n",
    "\n",
    "You can read about the `Task` class in the docs [here](https://allegro.ai/docs/task.html).\n",
    "\n",
    "After the Task is created, get a Logger for it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "885DHN5SjsHy"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "from clearml import Task\n",
    "\n",
    "# Start a new task\n",
    "task = Task.init(project_name=\"Colab notebooks\", task_name=\"Explicit Logging\")\n",
    "\n",
    "# Get the task logger,\n",
    "# You can also call Task.current_task().get_logger() from anywhere in your code.\n",
    "logger = task.get_logger()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "SADmtLH8uwhw"
   },
   "source": [
    "### Explicit scalar logging\n",
    "Use the [Logger.report_scalar](https://allegro.ai/docs/logger.html#clearml.logger.Logger.report_scalar) method to explicitly log scalars. Scalar plots appear in the Web UI, Results > Scalars tab."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "E6YH4lNLjLs8"
   },
   "outputs": [],
   "source": [
    "# report two scalar series on the same graph\n",
    "for i in range(10):\n",
    "    logger.report_scalar(\"unified graph\", \"series A\", iteration=i, value=1./(i+1))\n",
    "    logger.report_scalar(\"unified graph\", \"series B\", iteration=i, value=10./(i+1))\n",
    "\n",
    "# report two scalar series on two different graphs\n",
    "for i in range(10):\n",
    "    logger.report_scalar(\"graph A\", \"series A\", iteration=i, value=1./(i+1))\n",
    "    logger.report_scalar(\"graph B\", \"series B\", iteration=i, value=10./(i+1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "8iqvizEzwRtk"
   },
   "source": [
    "### Explicit logging of other data\n",
    "\n",
    "You can log other data and report the data in a variety of plot types, including histograms, confusion matrices, 2D and 3D scatter diagrams, and surface diagrams. They appear in the Results > Plots tab.\n",
    "\n",
    "For information about the methods to report each type of plot, see the [Logger](https://allegro.ai/docs/logger.html) module.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_rWW7HTYjLtK"
   },
   "outputs": [],
   "source": [
    "iteration = 100\n",
    "\n",
    "# report a single histogram\n",
    "histogram = np.random.randint(10, size=10)\n",
    "logger.report_histogram(\n",
    "    \"single_histogram\",\n",
    "    \"random histogram\",\n",
    "    iteration=iteration,\n",
    "    values=histogram,\n",
    "    xaxis=\"title x\",\n",
    "    yaxis=\"title y\",\n",
    ")\n",
    "\n",
    "# report a two histograms on the same plot\n",
    "histogram1 = np.random.randint(13, size=10)\n",
    "histogram2 = histogram * 0.75\n",
    "logger.report_histogram(\n",
    "    \"two_histogram\",\n",
    "    \"series 1\",\n",
    "    iteration=iteration,\n",
    "    values=histogram1,\n",
    "    xaxis=\"title x\",\n",
    "    yaxis=\"title y\",\n",
    ")\n",
    "logger.report_histogram(\n",
    "    \"two_histogram\",\n",
    "    \"series 2\",\n",
    "    iteration=iteration,\n",
    "    values=histogram2,\n",
    "    xaxis=\"title x\",\n",
    "    yaxis=\"title y\",\n",
    ")\n",
    "\n",
    "# report confusion matrix\n",
    "confusion = np.random.randint(10, size=(10, 10))\n",
    "logger.report_matrix(\n",
    "    \"example_confusion\",\n",
    "    \"ignored\",\n",
    "    iteration=iteration,\n",
    "    matrix=confusion,\n",
    "    xaxis=\"title X\",\n",
    "    yaxis=\"title Y\",\n",
    ")\n",
    "\n",
    "scatter2d = np.hstack(\n",
    "    (np.atleast_2d(np.arange(0, 10)).T, np.random.randint(10, size=(10, 1)))\n",
    ")\n",
    "# report 2d scatter plot with markers\n",
    "logger.report_scatter2d(\n",
    "    \"example_scatter\",\n",
    "    \"series_lines+markers\",\n",
    "    iteration=iteration,\n",
    "    scatter=scatter2d,\n",
    "    xaxis=\"title x\",\n",
    "    yaxis=\"title y\",\n",
    "    mode='lines+markers'\n",
    ")\n",
    "\n",
    "# report 3d surface\n",
    "surface = np.random.randint(10, size=(10, 10))\n",
    "logger.report_surface(\n",
    "    \"example_surface\",\n",
    "    \"series1\",\n",
    "    iteration=iteration,\n",
    "    matrix=surface,\n",
    "    xaxis=\"title X\",\n",
    "    yaxis=\"title Y\",\n",
    "    zaxis=\"title Z\",\n",
    ")\n",
    "\n",
    "# report 3d scatter plot\n",
    "scatter3d = np.random.randint(10, size=(10, 3))\n",
    "logger.report_scatter3d(\n",
    "    \"example_scatter_3d\",\n",
    "    \"series_xyz\",\n",
    "    iteration=iteration,\n",
    "    scatter=scatter3d,\n",
    "    xaxis=\"title x\",\n",
    "    yaxis=\"title y\",\n",
    "    zaxis=\"title z\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "OZqPEaFRwcVY"
   },
   "source": [
    "### Explicit debug samples reporting\n",
    "\n",
    "Explicitly report debug samples, including images, audio, and video."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Downloading the files\n",
    "\n",
    "We use StorageManager to download a local copy of the files. You can use it immediately. Just provide the URL. Cache is enabled by default for all downloaded remote URLs/files.\n",
    "\n",
    "For more information, you can read about the storage manager [here](https://allegro.ai/docs/storage_manager_storagemanager.html)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "s4rf0ap0jLtb"
   },
   "outputs": [],
   "source": [
    "from clearml.storage import StorageManager\n",
    "image_local_copy = StorageManager.get_local_copy(\n",
    "    remote_url=\"https://pytorch.org/tutorials/_static/img/neural-style/picasso.jpg\",\n",
    "    name=\"picasso.jpg\"\n",
    ")\n",
    "\n",
    "print(\"Image location: {}\".format(image_local_copy))\n",
    "\n",
    "video_local_copy = StorageManager.get_local_copy(\n",
    "    remote_url=\"https://test-videos.co.uk/vids/bigbuckbunny/mp4/h264/720/Big_Buck_Bunny_720_10s_1MB.mp4\",\n",
    "    name=\"Big_Buck_Bunny_720_10s_1MB.mp4\"\n",
    ")\n",
    "\n",
    "print(\"Video location: {}\".format(video_local_copy))\n",
    "\n",
    "audio_local_copy = StorageManager.get_local_copy(\n",
    "    remote_url=\"https://www2.cs.uic.edu/~i101/SoundFiles/PinkPanther30.wav\",\n",
    "    name=\"PinkPanther30.wav\"\n",
    ")\n",
    "\n",
    "print(\"Audio location: {}\".format(audio_local_copy))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vydmi7HWw0gS"
   },
   "source": [
    "#### Report images and media\n",
    "\n",
    "Use [Logger.report_image](https://allegro.ai/docs/logger.html?highlight=report_image#clearml.logger.Logger.report_image) and [Logger.report_media](https://allegro.ai/docs/logger.html?highlight=report_media#clearml.logger.Logger.report_media) to report the downloaded samples. The debug samples appear in the Results > Debug Samples tab."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "MtEhbE4S_P66"
   },
   "outputs": [],
   "source": [
    "logger.report_image(\"image\", \"image from url\", iteration=100, local_path=image_local_copy)\n",
    "\n",
    "# Image can be uploaded via 'report_media' too\n",
    "# report video, an already uploaded video media (url)\n",
    "logger.report_media(\n",
    "    'video', 'big bunny', iteration=1, local_path=video_local_copy)\n",
    "\n",
    "\n",
    "# This will actually use the cache and will not download the file again.\n",
    "audio_local_copy_cache = StorageManager.get_local_copy(\n",
    "    remote_url=\"https://www2.cs.uic.edu/~i101/SoundFiles/PinkPanther30.wav\",\n",
    "    name=\"PinkPanther30.wav\"\n",
    ")\n",
    "\n",
    "# report audio, report an already uploaded audio media (url)\n",
    "logger.report_media(\n",
    "    'audio', 'pink panther', iteration=1, local_path=audio_local_copy)\n",
    "\n",
    "# reporting html from url to debug samples section\n",
    "logger.report_media(\"html\", \"url_html\", iteration=1, url=\"https://allegro.ai/docs/index.html\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "uDrcRCJxxCQP"
   },
   "source": [
    "### Explicit text logging\n",
    "Use [Logger.report_text](https://allegro.ai/docs/logger.html?highlight=report_text#clearml.logger.Logger.report_text) to log text message. They appear in Results > Log."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "a2UlIETAjLtk"
   },
   "outputs": [],
   "source": [
    "# report text\n",
    "logger.report_text(\"hello, this is plain text\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "aNFDbjZ7xNco"
   },
   "source": [
    "### Flushing the reports\n",
    "\n",
    "Reports are flushed in the background every couple of seconds, and at the end of the process execution.\n",
    "\n",
    "Or, flush the Logger by calling [Logger.flush](https://allegro.ai/docs/logger.html?highlight=report_text#clearml.logger.Logger.flush)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "SSyGUskJjLtr"
   },
   "outputs": [],
   "source": [
    "logger.flush()"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Allegro ClearML logging example.ipynb",
   "provenance": []
  },
  "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}