IMP_DT copy.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "view-in-github"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Rohith-Rongali/IMP/blob/main/IMP_DT.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sTt4FBrCswwi"
      },
      "source": [
        "# Imports"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "TkKwoTDlskqo"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torch.optim as optim\n",
        "from torch.utils.data import Dataset\n",
        "from torch.utils.data import DataLoader\n",
        "\n",
        "import torch.nn.utils.prune as prune\n",
        "\n",
        "from copy import deepcopy\n",
        "\n",
        "from tqdm import tqdm\n",
        "import os\n",
        "\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from types import SimpleNamespace"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Yci_s3iDs1Mm"
      },
      "source": [
        "# Data functions"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "wgzl99bdtK7G"
      },
      "outputs": [],
      "source": [
        "def generate_points(num=40000,dim=18,type='normal',normal_var=1,radius=1):\n",
        "    '''\n",
        "    If type=='normal', then generate points from N(0,normal_var)\n",
        "    If type=='spherical', then simply divide the points by their norm.'''\n",
        "    X = torch.randn([num,dim]) #coordinates sampled from N(0,1)\n",
        "\n",
        "    if type=='spherical':\n",
        "      norm = torch.norm(X, p=2, dim=1, keepdim=True)\n",
        "      X_spherical = X / norm\n",
        "      return X_spherical\n",
        "\n",
        "    elif type=='normal':\n",
        "      return X\n",
        "\n",
        "    else:\n",
        "      raise ValueError('type should be either normal or spherical')\n",
        "\n",
        "\n",
        "class TreeNode:\n",
        "    '''\n",
        "    This class represents a node in the decision tree.\n",
        "    Each node has a depth, a maximum depth(of the tree), a feature index, and left and right child nodes.\n",
        "    Leaf nodes have a value, which is the predicted class.\n",
        "    '''\n",
        "    def __init__(self, depth, max_depth, feature_index):\n",
        "        self.depth = depth\n",
        "        self.max_depth = max_depth\n",
        "        self.feature = feature_index\n",
        "        self.left = None\n",
        "        self.right = None\n",
        "        self.value = None  # This will store the predicted class for leaf nodes\n",
        "\n",
        "    def build_tree(self):\n",
        "        if self.depth == self.max_depth:\n",
        "            self.value = float(self.feature % 2)\n",
        "            return\n",
        "\n",
        "        # Create left and right child nodes\n",
        "        self.left = TreeNode(self.depth + 1, self.max_depth, 2*self.feature+1)\n",
        "        self.right = TreeNode(self.depth + 1, self.max_depth, 2*self.feature+2)\n",
        "\n",
        "        # Recursively build left and right subtrees\n",
        "        self.left.build_tree()\n",
        "        self.right.build_tree()\n",
        "\n",
        "    def predict(self, x):\n",
        "        if self.value is not None:\n",
        "            return self.value\n",
        "\n",
        "        if x[self.feature] > 0:\n",
        "            return self.left.predict(x)\n",
        "        else:\n",
        "            return self.right.predict(x)\n",
        "\n",
        "\n",
        "def gen_std_basis_DT(depth, dim_in, type_data, num_points, feat_index_start=0,radius=1):\n",
        "    '''\n",
        "    Generate points uniformly random from a hypersphere. And the label is the prediction of the tree with depth = max_depth.\n",
        "    The node hyperplanes are simply characterised by standard basis vectors(for example, the root node hyperplane is x[0] = 0)\n",
        "\n",
        "    '''\n",
        "    Tree = TreeNode(depth = 0,max_depth=depth,feature_index = feat_index_start)\n",
        "    Tree.build_tree()\n",
        "    X = generate_points(num=num_points,dim=dim_in,type=type_data,radius=radius)\n",
        "\n",
        "    Y=[]\n",
        "    for item in X:\n",
        "        Y.append(Tree.predict(item))\n",
        "\n",
        "    Y = torch.tensor(Y)\n",
        "    return X,Y\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "def set_npseed(seed):\n",
        "\tnp.random.seed(seed)\n",
        "\n",
        "\n",
        "def set_torchseed(seed):\n",
        "\ttorch.manual_seed(seed)\n",
        "\ttorch.cuda.manual_seed(seed)\n",
        "\ttorch.cuda.manual_seed_all(seed)\n",
        "\ttorch.backends.cudnn.deterministic = True\n",
        "\ttorch.backends.cudnn.benchmark = False\n",
        "\n",
        "\n",
        "\n",
        "# depth = 4\n",
        "# dim_in = 18\n",
        "# type_data = 'normal'\n",
        "# feat_index_start = 0 #the index of the first feature in the tree\n",
        "# num_points = 40000\n",
        "\n",
        "\n",
        "#     Tree = TreeNode(depth = 0,max_depth=depth,feature_index = feat_index_start)\n",
        "#     Tree.build_tree()\n",
        "#     X = generate_points(num=num_points,dim=dim_in,type=type_data)\n",
        "\n",
        "#     Y=[]\n",
        "#     for item in X:\n",
        "#     Y.append(Tree.predict(item))\n",
        "\n",
        "#     Y = torch.tensor(Y)\n",
        "\n",
        "class CustomDataset(Dataset):\n",
        "    def __init__(self, x, y, transform=None):\n",
        "        self.x = x\n",
        "        self.y = y\n",
        "        self.transform = transform\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self.x)\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        sample_x = self.x[idx]\n",
        "        sample_y = self.y[idx]\n",
        "\n",
        "        if self.transform:\n",
        "            sample_x, sample_y = self.transform(sample_x, sample_y)\n",
        "\n",
        "        return sample_x, sample_y\n",
        "\n",
        "def return_data_elements(DataConfig):\n",
        "    # returns [data] and data_loader\n",
        "\n",
        "    x_train,y_train = gen_std_basis_DT(depth = DataConfig.depth, dim_in = DataConfig.dim_in, num_points = DataConfig.num_points,type_data= DataConfig.type_data, radius = DataConfig.radius)\n",
        "    x_test,y_test = gen_std_basis_DT(depth = DataConfig.depth, dim_in = DataConfig.dim_in, num_points = DataConfig.num_points,type_data= DataConfig.type_data, radius = DataConfig.radius)\n",
        "    # x_train, x_test, y_train, y_test = train_test_split(\n",
        "    # X, Y, test_size=0.1, random_state=42, stratify=Y)\n",
        "\n",
        "    data = [x_train,y_train.long(),x_test,y_test.long()]\n",
        "\n",
        "    return data\n",
        "\n",
        "def loaders(data,batch_size=32):\n",
        "    x_train,y_train,x_test,y_test = data\n",
        "    train_dataset = CustomDataset(x_train, y_train)\n",
        "    test_dataset = CustomDataset(x_test, y_test)\n",
        "    train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
        "    test_dataloader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)\n",
        "    return train_dataloader, test_dataloader"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3ikUnEJVwRei"
      },
      "source": [
        "# Models\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "LzvqA-FRwVVY"
      },
      "outputs": [],
      "source": [
        "class DNN(nn.Module):\n",
        "    def __init__(self, dim_in, dim_out, width, depth):\n",
        "        super(DNN, self).__init__()\n",
        "        self.depth = depth\n",
        "        self.layers = nn.ModuleList([nn.Linear(dim_in if i == 0 else width, width) for i in range(self.depth)])\n",
        "        self.output_layer = nn.Linear(width, dim_out)\n",
        "        self.relu = nn.ReLU()\n",
        "\n",
        "    def forward(self, x):\n",
        "        for layer in self.layers:\n",
        "            x = self.relu(layer(x))\n",
        "        x = self.output_layer(x)\n",
        "        return x"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dMpRYgpoxEJf"
      },
      "source": [
        "# Configs"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "Nz3C2i92xG4c"
      },
      "outputs": [],
      "source": [
        "\n",
        "model_config = SimpleNamespace(\n",
        "    width=256,\n",
        "    depth=2\n",
        ")\n",
        "\n",
        "data_config = SimpleNamespace(\n",
        "    dim_in=100,\n",
        "    depth=4,\n",
        "    type_data='spherical',\n",
        "    num_points=60000,\n",
        "    radius=1,\n",
        "    dim_out=2\n",
        ")\n",
        "\n",
        "train_config = SimpleNamespace(\n",
        "    batch_size=32,\n",
        "    lr=0.01,\n",
        "    epochs=100\n",
        ")\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7BEZbB2Ps18N"
      },
      "source": [
        "# Training"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "2VUhFr590ybb"
      },
      "outputs": [],
      "source": [
        "device = torch.device(\"cuda:1\" if torch.cuda.is_available() else \"cpu\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "id": "cMMdrJJitsUq"
      },
      "outputs": [],
      "source": [
        "data= return_data_elements(data_config)\n",
        "x_train,y_train,x_test,y_test = data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "XrkZpOuWvJwj"
      },
      "outputs": [],
      "source": [
        "# Create DataLoaders for train and test datasets\n",
        "batch_size = 32\n",
        "\n",
        "# Assuming you have a custom Dataset class, you can create instances like this:\n",
        "train_dataset = CustomDataset(x_train, y_train)\n",
        "test_dataset = CustomDataset(x_test, y_test)\n",
        "\n",
        "# Create DataLoaders\n",
        "train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
        "test_dataloader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "diQWbC5aw32C"
      },
      "outputs": [],
      "source": [
        "model = DNN(dim_in=data_config.dim_in, dim_out=data_config.dim_out, width=model_config.width, depth=model_config.depth)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=train_config.lr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "HkxSrBb4ziwR"
      },
      "outputs": [],
      "source": [
        "def train(model,loss_fn,optimizer,train_dataloader,\n",
        "          data,num_epochs=train_config.epochs,log_epochs=10,log_acc=10,thresh=1e-5,print_std=True):\n",
        "  [x_train,y_train,x_test,y_test]=data\n",
        "  model.to(device)\n",
        "  Train_losses=[]\n",
        "  acc_dict = {'train':[],'test':[]}\n",
        "  for epoch in range(num_epochs):\n",
        "      model.train()\n",
        "      for x_batch, y_batch in train_dataloader:\n",
        "          x_batch = x_batch.to(device)\n",
        "          y_batch = y_batch.to(device)\n",
        "          pred = model(x_batch)\n",
        "          loss = loss_fn(pred, y_batch)\n",
        "          loss.backward()\n",
        "          optimizer.step()\n",
        "          optimizer.zero_grad()\n",
        "\n",
        "\n",
        "      if epoch % log_epochs == 0:\n",
        "          loss_full = loss_fn(model(x_train.to(device)),y_train.to(device))\n",
        "          Train_losses.append(loss_full.item())\n",
        "          print(f'Epoch {epoch} Loss {loss_full.item():.4f}')\n",
        "      if log_acc!=0:\n",
        "        if epoch % log_acc == 0:\n",
        "          train_pred = model(x_train.to(device))\n",
        "          pred = torch.argmax(model(x_train.to(device)),dim=1)\n",
        "          zero_mask = (pred-y_train.to(device) == 0.0)\n",
        "          train_acc = zero_mask.sum().item()/len(y_train)\n",
        "          acc_dict['train'].append(train_acc)\n",
        "\n",
        "          test_pred = model(x_test.to(device))[:,0]\n",
        "          pred = torch.argmax(model(x_test.to(device)),dim=1)\n",
        "          zero_mask = (pred-y_test.to(device) == 0.0)\n",
        "          test_acc = zero_mask.sum().item()/len(y_test)\n",
        "          acc_dict['test'].append(test_acc)\n",
        "\n",
        "          if print_std:\n",
        "            print(f'Epoch {epoch} train_acc {train_acc} test_acc {test_acc}')\n",
        "      if loss_full.item() < thresh:\n",
        "          print(f'Early stopping at epoch {epoch} because loss is below 0.01')\n",
        "          break\n",
        "\n",
        "  return Train_losses,acc_dict"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YVP7CC6l1Il0",
        "outputId": "cbc5c26e-520e-47fa-be90-a112c01d7803"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0 Loss 0.5745\n",
            "Epoch 0 train_acc 0.70375 test_acc 0.6797333333333333\n",
            "Epoch 10 Loss 0.1822\n",
            "Epoch 10 train_acc 0.9298333333333333 test_acc 0.7782333333333333\n",
            "Epoch 20 Loss 0.1100\n",
            "Epoch 20 train_acc 0.9579833333333333 test_acc 0.7799833333333334\n",
            "Epoch 30 Loss 0.0767\n",
            "Epoch 30 train_acc 0.97135 test_acc 0.7811666666666667\n",
            "Epoch 40 Loss 0.0676\n",
            "Epoch 40 train_acc 0.97575 test_acc 0.7840666666666667\n",
            "Epoch 50 Loss 0.0716\n",
            "Epoch 50 train_acc 0.9755 test_acc 0.7831333333333333\n",
            "Epoch 60 Loss 0.0671\n",
            "Epoch 60 train_acc 0.9793833333333334 test_acc 0.7839833333333334\n",
            "Epoch 70 Loss 0.0754\n",
            "Epoch 70 train_acc 0.9793166666666666 test_acc 0.7823833333333333\n",
            "Epoch 80 Loss 0.0544\n",
            "Epoch 80 train_acc 0.9845666666666667 test_acc 0.7877333333333333\n",
            "Epoch 90 Loss 0.0389\n",
            "Epoch 90 train_acc 0.9883833333333333 test_acc 0.7886\n"
          ]
        }
      ],
      "source": [
        "losses, acc_dict = train(model,loss_fn,optimizer,train_dataloader,data)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## Large learning rate schedule"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {},
      "outputs": [],
      "source": [
        "model = DNN(dim_in=data_config.dim_in, dim_out=data_config.dim_out, width=model_config.width, depth=model_config.depth)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=0.1)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6qunqIvXVyp9"
      },
      "source": [
        "## Pruning"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "So far everything is the same. I set the pruning rate as $p ^ {1/n}$, this still works to some extent... Not sure how? I think we should still figure out some other schemes, maybe we can threshold the weights instead of removing a certain amount of smaller magnitude weights.... could try constant pruning fraction as well...."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            " \n",
            "pruning rate=0.1\n",
            "Pruning 1 0.9 0.9\n",
            "Pruning 2 0.683772233983162 0.81\n",
            "Pruning 3 0.535841116638722 0.7290000000000001\n",
            "Pruning 4 0.4376586748096509 0.6561\n",
            "Pruning 5 0.36904265551980675 0.5904900000000001\n",
            "Pruning 6 0.31870793094203864 0.531441\n",
            "Pruning 7 0.28031432699884795 0.4782969000000001\n",
            "Pruning 8 0.2501057906675441 0.4304672100000001\n",
            "Pruning 9 0.2257363173188729 0.3874204890000001\n",
            "Pruning 10 0.2056717652757185 0.3486784401000001\n",
            "Pruning 11 0.18886916921031294 0.31381059609000006\n",
            "Pruning 12 0.1745958147319816 0.2824295364810001\n",
            "Pruning 13 0.16232235993170807 0.2541865828329001\n",
            "Pruning 14 0.15165710175592795 0.2287679245496101\n",
            "Pruning 15 0.14230410140910588 0.20589113209464907\n",
            "Pruning 16 0.13403567663993465 0.18530201888518416\n",
            "Pruning 17 0.12667383761715667 0.16677181699666577\n",
            "Pruning 18 0.12007745643089296 0.15009463529699918\n",
            "Pruning 19 0.11413320958991735 0.13508517176729928\n",
            "Pruning 20 0.10874906186625444 0.12157665459056935\n",
            " \n",
            "pruning rate=0.2\n",
            "Pruning 1 0.8 0.8\n",
            "Pruning 2 0.5527864045000421 0.6400000000000001\n",
            "Pruning 3 0.41519645235742675 0.5120000000000001\n",
            "Pruning 4 0.331259695023578 0.4096000000000001\n",
            "Pruning 5 0.27522033632230447 0.3276800000000001\n",
            "Pruning 6 0.23527550866826996 0.2621440000000001\n",
            "Pruning 7 0.20540259529814775 0.20971520000000007\n",
            "Pruning 8 0.18223456604205746 0.1677721600000001\n",
            "Pruning 9 0.16374896904962655 0.13421772800000006\n",
            "Pruning 10 0.1486600774792154 0.10737418240000006\n",
            "Pruning 11 0.1361123362974096 0.08589934592000005\n",
            "Pruning 12 0.1255147277788321 0.06871947673600004\n",
            "Pruning 13 0.11644604222876276 0.054975581388800036\n",
            "Pruning 14 0.10859806781572867 0.043980465111040035\n",
            "Pruning 15 0.10174012623840034 0.03518437208883203\n",
            "Pruning 16 0.09569616059758845 0.028147497671065624\n",
            "Pruning 17 0.09032948674199881 0.022517998136852502\n",
            "Pruning 18 0.0855323784023988 0.018014398509482003\n",
            "Pruning 19 0.08121878961882534 0.014411518807585602\n",
            "Pruning 20 0.07731916540941164 0.011529215046068483\n",
            " \n",
            "pruning rate=0.3\n",
            "Pruning 1 0.7 0.7\n",
            "Pruning 2 0.4522774424948339 0.48999999999999994\n",
            "Pruning 3 0.3305670499178305 0.3429999999999999\n",
            "Pruning 4 0.25991719550771475 0.24009999999999995\n",
            "Pruning 5 0.2139969144033772 0.16806999999999994\n",
            "Pruning 6 0.18181117699997296 0.11764899999999996\n",
            "Pruning 7 0.15801755565497966 0.08235429999999996\n",
            "Pruning 8 0.13971934550852227 0.05764800999999997\n",
            "Pruning 9 0.12521291707606708 0.04035360699999998\n",
            "Pruning 10 0.11343184943478668 0.028247524899999984\n",
            "Pruning 11 0.10367487860124702 0.019773267429999988\n",
            "Pruning 12 0.09546209421604279 0.01384128720099999\n",
            "Pruning 13 0.08845406633142938 0.009688901040699992\n",
            "Pruning 14 0.08240398630714385 0.006782230728489994\n",
            "Pruning 15 0.07712811194671554 0.004747561509942996\n",
            "Pruning 16 0.07248684403320849 0.0033232930569600965\n",
            "Pruning 17 0.06837222737362458 0.0023263051398720674\n",
            "Pruning 18 0.06469946919509717 0.001628413597910447\n",
            "Pruning 19 0.06140104567447868 0.001139889518537313\n",
            "Pruning 20 0.058422520147591195 0.0007979226629761189\n",
            " \n",
            "pruning rate=0.4\n",
            "Pruning 1 0.6 0.6\n",
            "Pruning 2 0.3675444679663241 0.36\n",
            "Pruning 3 0.2631937002719227 0.21599999999999997\n",
            "Pruning 4 0.20472927123294937 0.1296\n",
            "Pruning 5 0.1674467925981269 0.07775999999999998\n",
            "Pruning 6 0.14162578106744295 0.04665599999999999\n",
            "Pruning 7 0.12269333787625847 0.027993599999999993\n",
            "Pruning 8 0.10822047076250363 0.016796159999999994\n",
            "Pruning 9 0.09679892986331773 0.010077695999999997\n",
            "Pruning 10 0.08755646344451917 0.006046617599999997\n",
            "Pruning 11 0.07992414149644911 0.0036279705599999985\n",
            "Pruning 12 0.07351512752093081 0.002176782335999999\n",
            "Pruning 13 0.06805725860180156 0.0013060694015999993\n",
            "Pruning 14 0.06335350204906998 0.0007836416409599996\n",
            "Pruning 15 0.059257713443878535 0.00047018498457599973\n",
            "Pruning 16 0.05565920916361111 0.00028211099074559984\n",
            "Pruning 17 0.05247262897373173 0.00016926659444735988\n",
            "Pruning 18 0.049631087347296376 0.00010155995666841593\n",
            "Pruning 19 0.0470814329107867 6.093597400104956e-05\n",
            "Pruning 20 0.04478089604767599 3.6561584400629733e-05\n"
          ]
        }
      ],
      "source": [
        "# To compare pruning rate heuristics\n",
        "for i in [0.1,0.2,0.3,0.4]:\n",
        "    x=1\n",
        "    print(' ')\n",
        "    print(f'pruning rate={i}')\n",
        "    for j in range(1,21):\n",
        "        print(f'Pruning {j} {x*(1-i**(1/j))} {(1-i)**j}')\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {},
      "outputs": [],
      "source": [
        "model = DNN(dim_in=data_config.dim_in, dim_out=data_config.dim_out, width=model_config.width, depth=model_config.depth)\n",
        "loss_fn = nn.CrossEntropyLoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=train_config.lr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Oaqx3cYxV1Ps",
        "outputId": "1753670f-5a54-44ea-90b1-961af83ee189"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "  0%|          | 0/6 [00:00<?, ?it/s]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0 Loss 0.5714\n",
            "Epoch 0 train_acc 0.7024166666666667 test_acc 0.6766666666666666\n",
            "Epoch 10 Loss 0.1536\n",
            "Epoch 10 train_acc 0.93955 test_acc 0.7992833333333333\n",
            "Epoch 20 Loss 0.0934\n",
            "Epoch 20 train_acc 0.96455 test_acc 0.7982666666666667\n",
            "Epoch 30 Loss 0.0545\n",
            "Epoch 30 train_acc 0.9791833333333333 test_acc 0.8057833333333333\n",
            "Epoch 40 Loss 0.0715\n",
            "Epoch 40 train_acc 0.9755 test_acc 0.8076833333333333\n",
            "Epoch 50 Loss 0.0608\n",
            "Epoch 50 train_acc 0.9808833333333333 test_acc 0.80795\n",
            "Epoch 60 Loss 0.0777\n",
            "Epoch 60 train_acc 0.9764166666666667 test_acc 0.80625\n",
            "Epoch 70 Loss 0.0533\n",
            "Epoch 70 train_acc 0.9826333333333334 test_acc 0.8088833333333333\n",
            "Epoch 80 Loss 0.0429\n",
            "Epoch 80 train_acc 0.9864166666666667 test_acc 0.8076333333333333\n",
            "Epoch 90 Loss 0.0411\n",
            "Epoch 90 train_acc 0.9877666666666667 test_acc 0.8122333333333334\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            " 17%|█▋        | 1/6 [05:07<25:37, 307.54s/it]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0 Loss 0.5680\n",
            "Epoch 0 train_acc 0.707 test_acc 0.6841666666666667\n",
            "Epoch 10 Loss 0.1177\n",
            "Epoch 10 train_acc 0.9570666666666666 test_acc 0.85495\n",
            "Epoch 20 Loss 0.0855\n",
            "Epoch 20 train_acc 0.96935 test_acc 0.8505166666666667\n",
            "Epoch 30 Loss 0.0453\n",
            "Epoch 30 train_acc 0.9842 test_acc 0.8645\n",
            "Epoch 40 Loss 0.0390\n",
            "Epoch 40 train_acc 0.98705 test_acc 0.8705666666666667\n",
            "Epoch 50 Loss 0.0382\n",
            "Epoch 50 train_acc 0.9890666666666666 test_acc 0.87175\n",
            "Epoch 60 Loss 0.0390\n",
            "Epoch 60 train_acc 0.9891166666666666 test_acc 0.8736833333333334\n",
            "Epoch 70 Loss 0.0399\n",
            "Epoch 70 train_acc 0.9896 test_acc 0.8741333333333333\n",
            "Epoch 80 Loss 0.0236\n",
            "Epoch 80 train_acc 0.99345 test_acc 0.8802833333333333\n",
            "Epoch 90 Loss 0.0237\n",
            "Epoch 90 train_acc 0.9934666666666667 test_acc 0.8797\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            " 33%|███▎      | 2/6 [10:01<19:57, 299.29s/it]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0 Loss 0.5071\n",
            "Epoch 0 train_acc 0.7523333333333333 test_acc 0.7408166666666667\n",
            "Epoch 10 Loss 0.1313\n",
            "Epoch 10 train_acc 0.94885 test_acc 0.9108\n",
            "Epoch 20 Loss 0.0666\n",
            "Epoch 20 train_acc 0.9748166666666667 test_acc 0.9221833333333334\n",
            "Epoch 30 Loss 0.0474\n",
            "Epoch 30 train_acc 0.9829666666666667 test_acc 0.9255333333333333\n",
            "Epoch 40 Loss 0.0383\n",
            "Epoch 40 train_acc 0.9861666666666666 test_acc 0.92595\n",
            "Epoch 50 Loss 0.0316\n",
            "Epoch 50 train_acc 0.9899166666666667 test_acc 0.9283166666666667\n",
            "Epoch 60 Loss 0.0242\n",
            "Epoch 60 train_acc 0.9924166666666666 test_acc 0.9267666666666666\n",
            "Epoch 70 Loss 0.0264\n",
            "Epoch 70 train_acc 0.9914666666666667 test_acc 0.9260333333333334\n",
            "Epoch 80 Loss 0.0245\n",
            "Epoch 80 train_acc 0.9927 test_acc 0.9280333333333334\n",
            "Epoch 90 Loss 0.0362\n",
            "Epoch 90 train_acc 0.9894333333333334 test_acc 0.9254666666666667\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            " 50%|█████     | 3/6 [14:30<14:17, 285.75s/it]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0 Loss 0.5239\n",
            "Epoch 0 train_acc 0.74215 test_acc 0.7321666666666666\n",
            "Epoch 10 Loss 0.1184\n",
            "Epoch 10 train_acc 0.9572333333333334 test_acc 0.9440833333333334\n",
            "Epoch 20 Loss 0.0635\n",
            "Epoch 20 train_acc 0.9770833333333333 test_acc 0.96\n",
            "Epoch 30 Loss 0.0431\n",
            "Epoch 30 train_acc 0.9853166666666666 test_acc 0.9641166666666666\n",
            "Epoch 40 Loss 0.0411\n",
            "Epoch 40 train_acc 0.98545 test_acc 0.9630333333333333\n",
            "Epoch 50 Loss 0.0453\n",
            "Epoch 50 train_acc 0.9869666666666667 test_acc 0.96435\n",
            "Epoch 60 Loss 0.0359\n",
            "Epoch 60 train_acc 0.9872333333333333 test_acc 0.9655833333333333\n",
            "Epoch 70 Loss 0.0327\n",
            "Epoch 70 train_acc 0.9900666666666667 test_acc 0.9650833333333333\n",
            "Epoch 80 Loss 0.0277\n",
            "Epoch 80 train_acc 0.9911666666666666 test_acc 0.9660333333333333\n",
            "Epoch 90 Loss 0.0353\n",
            "Epoch 90 train_acc 0.9899833333333333 test_acc 0.9658\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            " 67%|██████▋   | 4/6 [18:53<09:13, 276.62s/it]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0 Loss 0.5699\n",
            "Epoch 0 train_acc 0.7052333333333334 test_acc 0.6991\n",
            "Epoch 10 Loss 0.1965\n",
            "Epoch 10 train_acc 0.9232333333333334 test_acc 0.9156333333333333\n",
            "Epoch 20 Loss 0.1897\n",
            "Epoch 20 train_acc 0.9276666666666666 test_acc 0.9220666666666667\n",
            "Epoch 30 Loss 0.1504\n",
            "Epoch 30 train_acc 0.94545 test_acc 0.9394\n",
            "Epoch 40 Loss 0.1336\n",
            "Epoch 40 train_acc 0.9506666666666667 test_acc 0.94175\n",
            "Epoch 50 Loss 0.1326\n",
            "Epoch 50 train_acc 0.94925 test_acc 0.9429666666666666\n",
            "Epoch 60 Loss 0.1277\n",
            "Epoch 60 train_acc 0.9519666666666666 test_acc 0.9454833333333333\n",
            "Epoch 70 Loss 0.1286\n",
            "Epoch 70 train_acc 0.9511166666666667 test_acc 0.94195\n",
            "Epoch 80 Loss 0.1215\n",
            "Epoch 80 train_acc 0.9537333333333333 test_acc 0.9473\n",
            "Epoch 90 Loss 0.1277\n",
            "Epoch 90 train_acc 0.9564166666666667 test_acc 0.9470666666666666\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            " 83%|████████▎ | 5/6 [23:47<04:42, 282.85s/it]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 0 Loss 0.0864\n",
            "Epoch 0 train_acc 0.9682833333333334 test_acc 0.9624833333333334\n",
            "Epoch 10 Loss 0.0805\n",
            "Epoch 10 train_acc 0.9719666666666666 test_acc 0.967\n",
            "Epoch 20 Loss 0.0715\n",
            "Epoch 20 train_acc 0.975 test_acc 0.96985\n",
            "Epoch 30 Loss 0.0814\n",
            "Epoch 30 train_acc 0.9726833333333333 test_acc 0.9670166666666666\n",
            "Epoch 40 Loss 0.0884\n",
            "Epoch 40 train_acc 0.9709333333333333 test_acc 0.96555\n",
            "Epoch 50 Loss 0.0593\n",
            "Epoch 50 train_acc 0.97935 test_acc 0.9739166666666667\n",
            "Epoch 60 Loss 0.0842\n",
            "Epoch 60 train_acc 0.9720833333333333 test_acc 0.9661833333333333\n",
            "Epoch 70 Loss 0.0780\n",
            "Epoch 70 train_acc 0.9763166666666667 test_acc 0.9701666666666666\n",
            "Epoch 80 Loss 0.0797\n",
            "Epoch 80 train_acc 0.97245 test_acc 0.9670666666666666\n",
            "Epoch 90 Loss 0.0723\n",
            "Epoch 90 train_acc 0.9754 test_acc 0.97\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "                                              \r"
          ]
        }
      ],
      "source": [
        "INITIAL_WEIGHTS = deepcopy(model.state_dict())\n",
        "prune_epochs=6\n",
        "p = 0.2\n",
        "losses = {}\n",
        "acc_dict = {}\n",
        "for n in tqdm(range(1, prune_epochs + 1), total=prune_epochs, leave=False):\n",
        "\n",
        "  losses[n], acc_dict[n] = train(model,loss_fn,optimizer,train_dataloader,data)\n",
        "\n",
        "  pruning_rate = p ** (1 / n)\n",
        "\n",
        "  # Prune the model\n",
        "  prune.l1_unstructured(model.layers[0], name=\"weight\", amount=pruning_rate)\n",
        "  prune.l1_unstructured(model.layers[1], name=\"weight\", amount=pruning_rate)\n",
        "  prune.l1_unstructured(model.output_layer, name=\"weight\", amount=pruning_rate/2)\n",
        "\n",
        "  # Reset weights\n",
        "  reseted_weights = deepcopy(model.state_dict())\n",
        "  for param in reseted_weights.keys():\n",
        "      if param.split(\".\")[-1].replace(\"_orig\", \"\") in (\"bias\", \"weight\"):\n",
        "          reseted_weights[param] = deepcopy(INITIAL_WEIGHTS[param.replace(\"_orig\", \"\")])\n",
        "  if n< prune_epochs-1:\n",
        "    model.load_state_dict(reseted_weights)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yyyfoYEXVvK3"
      },
      "source": [
        "## Probing"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f1c900ce390>"
            ]
          },
          "execution_count": 12,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "#plot acc_dict(test)\n",
        "plt.figure()\n",
        "for key in acc_dict.keys():\n",
        "#   plt.plot(acc_dict[key]['train'],label=f'train_{key}')\n",
        "  plt.plot(acc_dict[key]['test'],label=f'test_{key}')\n",
        "plt.legend()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "Text(0, 0.5, 'Final accuracy')"
            ]
          },
          "execution_count": 13,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "final_acc_list = [acc_dict[key]['test'][-1] for key in acc_dict.keys()]\n",
        "plt.figure()\n",
        "plt.plot(final_acc_list)\n",
        "plt.title('Final accuracy vs pruning epoch')\n",
        "plt.xlabel('Pruning epoch')\n",
        "plt.ylabel('Final accuracy')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 435
        },
        "id": "lpxcyy_6GEQC",
        "outputId": "f3422a4d-fcd4-42eb-ce29-43a1288698c9"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "#normalise weights1 row-wise\n",
        "wt_mat = model.layers[0].weight.data.detach().cpu()\n",
        "w1_normed=F.normalize(wt_mat,p=2,dim=1)\n",
        "#visualise w1_normed as heat-map just in absolute values\n",
        "plt.imshow(w1_normed.abs().numpy())\n",
        "plt.colorbar()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 435
        },
        "id": "uQUG66BWHeOd",
        "outputId": "a2dd0e78-e1ca-4898-f8ad-d6bd84ed44a1"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "#normalise weights1 row-wise\n",
        "wt_mat = model.layers[1].weight.data.detach().cpu()\n",
        "w1_normed=F.normalize(wt_mat,p=2,dim=1)\n",
        "#visualise w1_normed as heat-map just in absolute values\n",
        "plt.imshow(w1_normed.abs().numpy())\n",
        "plt.colorbar()\n",
        "plt.show()"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "authorship_tag": "ABX9TyN1gu3P2Bvdjy5Le6WrC1E7",
      "include_colab_link": true,
      "provenance": []
    },
    "kernelspec": {
      "display_name": "env1",
      "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": "3.11.3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}