Update_GAN_and_HW5

范例/HW01/HW01.ipynb

@@ -7,7 +7,7 @@
       "provenance": [],
       "collapsed_sections": [],
       "toc_visible": true,
-      "authorship_tag": "ABX9TyMsVwxzy9cP7RXmbb8AmT4Y",
+      "authorship_tag": "ABX9TyPuibV8rME8Y2Er3TVnCm93",
       "include_colab_link": true
     },
     "kernelspec": {
@@ -45,7 +45,10 @@
         "Author: Heng-Jui Chang\n",
         "\n",
         "Slides: https://github.com/ga642381/ML2021-Spring/blob/main/HW01/HW01.pdf  \n",
-        "Video: TBA\n",
+        "Videos (Mandarin): https://cool.ntu.edu.tw/courses/4793/modules/items/172854  \n",
+        "https://cool.ntu.edu.tw/courses/4793/modules/items/172853  \n",
+        "Video (English): https://cool.ntu.edu.tw/courses/4793/modules/items/176529\n",
+        "\n",
         "\n",
         "Objectives:\n",
         "* Solve a regression problem with deep neural networks (DNN).\n",
@@ -378,7 +381,7 @@
         "\n",
         "    def cal_loss(self, pred, target):\n",
         "        ''' Calculate loss '''\n",
-        "        # TODO: you may implement L2 regularization here\n",
+        "        # TODO: you may implement L1/L2 regularization here\n",
         "        return self.criterion(pred, target)"
       ],
       "execution_count": null,

范例/HW02/HW02-1.ipynb

@@ -31,7 +31,11 @@
         "id": "OYlaRwNu7ojq"
       },
       "source": [
-        "# **Homework 2-1 Phoneme Classification**"
+        "# **Homework 2-1 Phoneme Classification**\n",
+        "\n",
+        "* Slides: https://speech.ee.ntu.edu.tw/~hylee/ml/ml2021-course-data/hw/HW02/HW02.pdf\n",
+        "* Video (Chinese): https://youtu.be/PdjXnQbu2zo\n",
+        "* Video (English): https://youtu.be/ESRr-VCykBs\n"
       ]
     },
     {

范例/HW02/HW02-2.ipynb

@@ -30,8 +30,11 @@
         "id": "eNSV4QGHS1I1"
       },
       "source": [
-        "# **Homework 2-2 Hessian Matrix**\r\n",
-        "\r\n"
+        "# **Homework 2-2 Hessian Matrix**\n",
+        "\n",
+        "* Slides: https://speech.ee.ntu.edu.tw/~hylee/ml/ml2021-course-data/hw/HW02/HW02.pdf\n",
+        "* Video (Chinese): https://youtu.be/PdjXnQbu2zo\n",
+        "* Video (English): https://youtu.be/ESRr-VCykBs\n"
       ]
     },
     {
@@ -171,7 +174,7 @@
         "id": "ZFGBCIFmVLS_"
       },
       "source": [
-        "### Import Libraries\r\n"
+        "### Import Libraries\n"
       ]
     },
     {
@@ -180,16 +183,16 @@
         "id": "_-vjBvH0uqA-"
       },
       "source": [
-        "import numpy as np\r\n",
-        "from math import pi\r\n",
-        "from collections import defaultdict\r\n",
-        "from autograd_lib import autograd_lib\r\n",
-        "\r\n",
-        "import torch\r\n",
-        "import torch.nn as nn\r\n",
-        "from torch.utils.data import DataLoader, Dataset\r\n",
-        "\r\n",
-        "import warnings\r\n",
+        "import numpy as np\n",
+        "from math import pi\n",
+        "from collections import defaultdict\n",
+        "from autograd_lib import autograd_lib\n",
+        "\n",
+        "import torch\n",
+        "import torch.nn as nn\n",
+        "from torch.utils.data import DataLoader, Dataset\n",
+        "\n",
+        "import warnings\n",
         "warnings.filterwarnings(\"ignore\")"
       ],
       "execution_count": null,
@@ -212,17 +215,17 @@
         "id": "uvdOpR9lVaJQ"
       },
       "source": [
-        "class MathRegressor(nn.Module):\r\n",
-        "    def __init__(self, num_hidden=128):\r\n",
-        "        super().__init__()\r\n",
-        "        self.regressor = nn.Sequential(\r\n",
-        "            nn.Linear(1, num_hidden),\r\n",
-        "            nn.ReLU(),\r\n",
-        "            nn.Linear(num_hidden, 1)\r\n",
-        "        )\r\n",
-        "\r\n",
-        "    def forward(self, x):\r\n",
-        "        x = self.regressor(x)\r\n",
+        "class MathRegressor(nn.Module):\n",
+        "    def __init__(self, num_hidden=128):\n",
+        "        super().__init__()\n",
+        "        self.regressor = nn.Sequential(\n",
+        "            nn.Linear(1, num_hidden),\n",
+        "            nn.ReLU(),\n",
+        "            nn.Linear(num_hidden, 1)\n",
+        "        )\n",
+        "\n",
+        "    def forward(self, x):\n",
+        "        x = self.regressor(x)\n",
         "        return x"
       ],
       "execution_count": null,
@@ -297,12 +300,12 @@
         "id": "OSU8vnXEbY6q"
       },
       "source": [
-        "# load checkpoint and data corresponding to the key\r\n",
-        "model = MathRegressor()\r\n",
-        "autograd_lib.register(model)\r\n",
-        "\r\n",
-        "data = torch.load('data.pth')[key]\r\n",
-        "model.load_state_dict(data['model'])\r\n",
+        "# load checkpoint and data corresponding to the key\n",
+        "model = MathRegressor()\n",
+        "autograd_lib.register(model)\n",
+        "\n",
+        "data = torch.load('data.pth')[key]\n",
+        "model.load_state_dict(data['model'])\n",
         "train, target = data['data']"
       ],
       "execution_count": null,
@@ -490,13 +493,13 @@
         "id": "1X-2uxwTcB9u"
       },
       "source": [
-        "# the main function to compute gradient norm and minimum ratio\r\n",
-        "def main(model, train, target):\r\n",
-        "    criterion = nn.MSELoss()\r\n",
-        "\r\n",
-        "    gradient_norm = compute_gradient_norm(model, criterion, train, target)\r\n",
-        "    minimum_ratio = compute_minimum_ratio(model, criterion, train, target)\r\n",
-        "\r\n",
+        "# the main function to compute gradient norm and minimum ratio\n",
+        "def main(model, train, target):\n",
+        "    criterion = nn.MSELoss()\n",
+        "\n",
+        "    gradient_norm = compute_gradient_norm(model, criterion, train, target)\n",
+        "    minimum_ratio = compute_minimum_ratio(model, criterion, train, target)\n",
+        "\n",
         "    print('gradient norm: {}, minimum ratio: {}'.format(gradient_norm, minimum_ratio))"
       ],
       "execution_count": null,

范例/HW04/HW04.ipynb

@@ -27,7 +27,15 @@
         "- Baselines:\n",
         "  - Easy: Run sample code and know how to use transformer.\n",
         "  - Medium: Know how to adjust parameters of transformer.\n",
-        "  - Hard: Construct [conformer](https://arxiv.org/abs/2005.08100) which is a variety of transformer. "
+        "  - Hard: Construct [conformer](https://arxiv.org/abs/2005.08100) which is a variety of transformer. \n",
+        "\n",
+        "- Other links\n",
+        "  - Kaggle: [link](https://www.kaggle.com/t/859c9ca9ede14fdea841be627c412322)\n",
+        "  - Slide: [link](https://speech.ee.ntu.edu.tw/~hylee/ml/ml2021-course-data/hw/HW04/HW04.pdf)\n",
+        "  - Data: [link](https://drive.google.com/file/d/1T0RPnu-Sg5eIPwQPfYysipfcz81MnsYe/view?usp=sharing)\n",
+        "  - Video (Chinese): [link](https://www.youtube.com/watch?v=EPerg2UnGaI)\n",
+        "  - Video (English): [link](https://www.youtube.com/watch?v=Gpz6AUvCak0)\n",
+        "  - Solution for downloading dataset fail.: [link](https://drive.google.com/drive/folders/13T0Pa_WGgQxNkqZk781qhc5T9-zfh19e?usp=sharing)"
       ]
     },
     {
@@ -36,7 +44,10 @@
         "id": "TPDoreyypeJE"
       },
       "source": [
-        "# Download dataset"
+        "# Download dataset\n",
+        "- **If all download links fail**\n",
+        "- **Please follow [here](https://drive.google.com/drive/folders/13T0Pa_WGgQxNkqZk781qhc5T9-zfh19e?usp=sharing)**\n",
+        "- **Data is [here](https://drive.google.com/file/d/1T0RPnu-Sg5eIPwQPfYysipfcz81MnsYe/view?usp=sharing)**"
       ]
     },
     {
@@ -63,8 +74,10 @@
         "# !gdown --id '1MUGBvG_JjqO0C2JYHuyV3B0lvaf1kWIm' --output Dataset.zip\n",
         "\"\"\" Download link 7 of Google drive \"\"\"\n",
         "# !gdown --id '18M91P5DHwILNyOlssZ57AiPOR0OwutOM' --output Dataset.zip\n",
+        "\"\"\" For all download links fail, Please paste link into 'Paste link here' \"\"\"\n",
+        "# !gdown --id 'Paste link here' --output Dataset.zip\n",
         "\"\"\" For Google drive, you can unzip the data by the command below. \"\"\"\n",
-        "# !unzip Dataset.zip\n",
+        "!unzip Dataset.zip\n",
         "\n",
         "\"\"\"\n",
         "    For Dropbox, we split dataset into five files. \n",
@@ -82,12 +95,12 @@
         "    For Onedrive, we split dataset into five files. \n",
         "    Please download all of them.\n",
         "\"\"\"\n",
-        "!wget --no-check-certificate \"https://onedrive.live.com/download?cid=10C95EE5FD151BFB&resid=10C95EE5FD151BFB%21106&authkey=ACB6opQR3CG9kmc\" -O Dataset.tar.gz.aa\n",
-        "!wget --no-check-certificate \"https://onedrive.live.com/download?cid=93DDDDD552E145DB&resid=93DDDDD552E145DB%21106&authkey=AP6EepjxSdvyV6Y\" -O Dataset.tar.gz.ab\n",
-        "!wget --no-check-certificate \"https://onedrive.live.com/download?cid=644545816461BCCC&resid=644545816461BCCC%21106&authkey=ALiefB0kI7Epb0Q\" -O Dataset.tar.gz.ac\n",
-        "!wget --no-check-certificate \"https://onedrive.live.com/download?cid=77CEBB3C3C512821&resid=77CEBB3C3C512821%21106&authkey=AAXCx4TTDYC0yjM\" -O Dataset.tar.gz.ad\n",
-        "!wget --no-check-certificate \"https://onedrive.live.com/download?cid=383D0E0146A11B02&resid=383D0E0146A11B02%21106&authkey=ALwVc4StVbig6QI\" -O Dataset.tar.gz.ae\n",
-        "!cat Dataset.tar.gz.* | tar zxvf -"
+        "# !wget --no-check-certificate \"https://onedrive.live.com/download?cid=10C95EE5FD151BFB&resid=10C95EE5FD151BFB%21106&authkey=ACB6opQR3CG9kmc\" -O Dataset.tar.gz.aa\n",
+        "# !wget --no-check-certificate \"https://onedrive.live.com/download?cid=93DDDDD552E145DB&resid=93DDDDD552E145DB%21106&authkey=AP6EepjxSdvyV6Y\" -O Dataset.tar.gz.ab\n",
+        "# !wget --no-check-certificate \"https://onedrive.live.com/download?cid=644545816461BCCC&resid=644545816461BCCC%21106&authkey=ALiefB0kI7Epb0Q\" -O Dataset.tar.gz.ac\n",
+        "# !wget --no-check-certificate \"https://onedrive.live.com/download?cid=77CEBB3C3C512821&resid=77CEBB3C3C512821%21106&authkey=AAXCx4TTDYC0yjM\" -O Dataset.tar.gz.ad\n",
+        "# !wget --no-check-certificate \"https://onedrive.live.com/download?cid=383D0E0146A11B02&resid=383D0E0146A11B02%21106&authkey=ALwVc4StVbig6QI\" -O Dataset.tar.gz.ae\n",
+        "# !cat Dataset.tar.gz.* | tar zxvf -"
       ],
       "execution_count": null,
       "outputs": []