diff --git a/integrations/torchvision/README.md b/integrations/torchvision/README.md
index bc58efbf18d..bc895aec666 100644
--- a/integrations/torchvision/README.md
+++ b/integrations/torchvision/README.md
@@ -39,7 +39,9 @@ pip install sparseml[torchvision]
 - [Sparse Transfer Learning with the CLI](tutorials/sparse-transfer-learning.md)
 - [Sparse Transfer Learning with the Python API](tutorials/docs-torchvision-python-transfer-imagenette.ipynb)
 - Sparsification from Scratch with the CLI (coming soon!)
-- Sparsification from Scratch with the Python API (coming soon!)
+- [Sparsification from Scratch with the Python API - ResNet-50](tutorials/docs-torchvision-sparsify-from-scatch-resnet50-beans.ipynb)
+- [Sparsification from Scratch with the Python API - MobileNetv2](tutorials/docs-torchvision-sparsify-from-scratch-mobilenetv2-beans.ipynb)
+- [Sparsification from Scratch with the Python API - DenseNet-121](tutorials/docs-torchvision-sparsify-from-scratch-densenet-flowers.ipynb )
 
 ## Quick Tour
 
diff --git a/integrations/torchvision/tutorials/docs-torchvision-python-transfer-imagenette.ipynb b/integrations/torchvision/tutorials/docs-torchvision-python-transfer-imagenette.ipynb
index bdf0a4d5685..69cc1f0138d 100644
--- a/integrations/torchvision/tutorials/docs-torchvision-python-transfer-imagenette.ipynb
+++ b/integrations/torchvision/tutorials/docs-torchvision-python-transfer-imagenette.ipynb
@@ -276,7 +276,6 @@
    ]
   },
   {
-   "attachments": {},
    "cell_type": "markdown",
    "metadata": {
     "id": "KCMRFLFN8024"
@@ -358,7 +357,6 @@
    ]
   },
   {
-   "attachments": {},
    "cell_type": "markdown",
    "metadata": {
     "id": "nrgF_ur_B66y"
@@ -764,15 +762,7 @@
     "id": "eJv4S9fkGjdO",
     "outputId": "df458b06-3ff4-4a0d-c1c2-3689799abaf5"
    },
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "6\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "img_path = \"/root/.cache/nm_datasets/imagenette/imagenette-320/val/n03417042/ILSVRC2012_val_00001144.JPEG\" \n",
     "images = [img_path]\n",
@@ -800,30 +790,7 @@
     "id": "79Xj9eqvJXaD",
     "outputId": "2bdfbe4d-ab3c-4cc4-c54f-a6be6cbb2b2b"
    },
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "<matplotlib.image.AxesImage at 0x7fd544213940>"
-      ]
-     },
-     "execution_count": 19,
-     "metadata": {},
-     "output_type": "execute_result"
-    },
-    {
-     "data": {
-      "image/png": "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",
-      "text/plain": [
-       "<Figure size 431.1x288 with 1 Axes>"
-      ]
-     },
-     "metadata": {
-      "needs_background": "light"
-     },
-     "output_type": "display_data"
-    }
-   ],
+   "outputs": [],
    "source": [
     "from PIL import Image\n",
     "import numpy as np\n",
@@ -857,7 +824,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.6"
+   "version": "3.8.10"
   },
   "widgets": {
    "application/vnd.jupyter.widget-state+json": {
diff --git a/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-densenet-flowers-pruned-quant.ipynb b/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-densenet-flowers-pruned-quant.ipynb
new file mode 100644
index 00000000000..725441a825c
--- /dev/null
+++ b/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-densenet-flowers-pruned-quant.ipynb
@@ -0,0 +1,2457 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "00e081b7",
+   "metadata": {},
+   "source": [
+    "# Sparsifying DenseNet121 from Scratch (Flower102)\n",
+    "\n",
+    "In this example, we will demonstrate how to sparsify an image classification model from scratch using SparseML's PyTorch integration. We train and prune [DenseNet121](https://pytorch.org/vision/main/models/generated/torchvision.models.densenet121.html) on the downstream [Oxford Flower 102 dataset](https://pytorch.org/vision/main/generated/torchvision.datasets.Flowers102.html#:~:text=Oxford%20102%20Flower%20is%20an,scale%2C%20pose%20and%20light%20variations) using the Global Magnitude Pruning algorithm. \n",
+    "\n",
+    "## Agenda\n",
+    "\n",
+    "There are a few steps:\n",
+    "\n",
+    " 1. Setup the dataset\n",
+    " 2. Setup the PyTorch training loop\n",
+    " 3. Train a dense version of DenseNet121\n",
+    " 4. Run the GMP pruning algorithm on the dense model\n",
+    " \n",
+    "## Installation\n",
+    "\n",
+    "Install SparseML with `pip`:\n",
+    "\n",
+    "```\n",
+    "pip install sparseml[torchvision]\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "1ad80edf",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "import sparseml\n",
+    "import torchvision\n",
+    "from sparseml.pytorch.optim import ScheduledModifierManager\n",
+    "from sparseml.pytorch.utils import TensorBoardLogger, ModuleExporter, get_prunable_layers, tensor_sparsity\n",
+    "\n",
+    "from torch.utils.data import DataLoader\n",
+    "from torch.nn import CrossEntropyLoss\n",
+    "from torch.optim import Adam\n",
+    "\n",
+    "from torchvision import transforms\n",
+    "from tqdm.auto import tqdm\n",
+    "import math"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "3a128dcc",
+   "metadata": {},
+   "source": [
+    "## **Step 1: Setup Dataset**\n",
+    "\n",
+    "Oxford 102 Flower is an image classification dataset consisting of 102 flower categories. The flowers were chosen to be flowers commonly occurring in the United Kingdom. Each class consists of between 40 and 258 images. The images have large scale, pose and light variations. In addition, there are categories that have large variations within the category, and several very similar categories.\n",
+    "\n",
+    "We use the standard PyTorch `datasets` and `dataloaders` to manage the dataset."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "7d180bbc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "NUM_LABELS = 102\n",
+    "BATCH_SIZE = 16\n",
+    "\n",
+    "# imagenet transforms\n",
+    "imagenet_transform = transforms.Compose([\n",
+    "   transforms.Resize(size=256, interpolation=transforms.InterpolationMode.BILINEAR, max_size=None, antialias=None),\n",
+    "   transforms.CenterCrop(size=(224, 224)),\n",
+    "   transforms.ToTensor(),\n",
+    "   transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])\n",
+    "])\n",
+    "\n",
+    "# datasets\n",
+    "train_dataset = torchvision.datasets.Flowers102(\n",
+    "    root=\"./data\",\n",
+    "    split=\"train\",\n",
+    "    transform=imagenet_transform,\n",
+    "    download=True\n",
+    ")\n",
+    "\n",
+    "val_dataset = torchvision.datasets.Flowers102(\n",
+    "    root=\"./data\",\n",
+    "    split=\"val\",\n",
+    "    transform=imagenet_transform,\n",
+    "    download=True\n",
+    ")\n",
+    "\n",
+    "# dataloaders\n",
+    "train_loader = DataLoader(train_dataset, BATCH_SIZE, shuffle=True, pin_memory=True, num_workers=16)\n",
+    "val_loader = DataLoader(val_dataset, BATCH_SIZE, shuffle=False, pin_memory=True, num_workers=16)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "890cca4a",
+   "metadata": {},
+   "source": [
+    "## Step 2: Setup PyTorch Training Loop\n",
+    "\n",
+    "We will use this training loop below. This is standard PyTorch functionality."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "5f1b878a",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "cuda\n"
+     ]
+    }
+   ],
+   "source": [
+    "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
+    "print(device)\n",
+    "\n",
+    "def run_model_one_epoch(model, data_loader, criterion, device, train=False, optimizer=None):\n",
+    "    if train:\n",
+    "        model.train()\n",
+    "    else:\n",
+    "        model.eval()\n",
+    "\n",
+    "    running_loss = 0.0\n",
+    "    total_correct = 0\n",
+    "    total_predictions = 0\n",
+    "\n",
+    "    # loop through batches\n",
+    "    for step, (inputs, labels) in tqdm(enumerate(data_loader), total=len(data_loader)):\n",
+    "        inputs = inputs.to(device)\n",
+    "        labels = labels.to(device)\n",
+    "\n",
+    "        if train:\n",
+    "            optimizer.zero_grad()\n",
+    "\n",
+    "        # compute loss, run backpropogation\n",
+    "        outputs = model(inputs)  # model returns logits\n",
+    "        loss = criterion(outputs, labels)\n",
+    "        if train:\n",
+    "            loss.backward()\n",
+    "            optimizer.step()\n",
+    "\n",
+    "        running_loss += loss.item()\n",
+    "\n",
+    "        # run evaluation\n",
+    "        predictions = outputs.argmax(dim=1)\n",
+    "        total_correct += torch.sum(predictions == labels).item()\n",
+    "        total_predictions += inputs.size(0)\n",
+    "\n",
+    "    # return loss and evaluation metric\n",
+    "    loss = running_loss / (step + 1.0)\n",
+    "    accuracy = total_correct / total_predictions\n",
+    "    return loss, accuracy"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "b385497a",
+   "metadata": {},
+   "source": [
+    "## **Step 3: Train DenseNet121 on Flowers102**\n",
+    "\n",
+    "First, we will train a dense version of DenseNet121 on the Flowers dataset."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "4d554578",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [],
+   "source": [
+    "# download pre-trained model, setup classification head\n",
+    "model = torchvision.models.densenet121(weights=torchvision.models.DenseNet121_Weights.DEFAULT)\n",
+    "model.classifier = torch.nn.Linear(model.classifier.in_features, NUM_LABELS)\n",
+    "model.to(device)\n",
+    "\n",
+    "# setup loss function and optimizer\n",
+    "criterion = CrossEntropyLoss()\n",
+    "optimizer = Adam(model.parameters(), lr=8e-3) # lr will be override by sparseml"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "eb400dc3",
+   "metadata": {
+    "scrolled": true
+   },
+   "source": [
+    "Next, we will use SparseML's recipes to set the hyperparameters of training loop. In this case, we will use the following recipe:\n",
+    "\n",
+    "```yaml\n",
+    "# Epoch and Learning-Rate variables\n",
+    "num_epochs: 10.0\n",
+    "init_lr: 0.0005\n",
+    "\n",
+    "training_modifiers:\n",
+    "  - !EpochRangeModifier\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: 0.0\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "```\n",
+    "\n",
+    "As you can see, the recipe includes an `!EpochRangeModifier` and a `!LearningRateFunctionModifier`. These modifiers simply set the number of epochs to train for and the learning rate schedule. As a result, the final model will be dense."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "5f6a53f1",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dense_recipe_path = \"./recipes/densenet-flowers-dense-recipe.yaml\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "8653c9bc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!cat ./recipes/densenet-flowers-dense-recipe.yaml"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c31aed1b",
+   "metadata": {},
+   "source": [
+    "Next, we use SparseML's `ScheduledModifierManager` to parse and apply the recipe. The `manager.modify` function modifies and wraps the `model` and `optimizer` with the instructions from the recipe. You can use the `model` and `optimizer` just like standard PyTorch objects."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "1749e00d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# create ScheduledModifierManager and Optimizer wrapper\n",
+    "manager = ScheduledModifierManager.from_yaml(dense_recipe_path)\n",
+    "optimizer = manager.modify(model, optimizer, steps_per_epoch=len(train_loader))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "ee67c917",
+   "metadata": {},
+   "source": [
+    "Kick off the transfer learning loop. Our run reached ~91.8% validation accuracy after 10 epochs."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "d00d175b",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running Training Epoch 1/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e3bff056f89743c3a030bf38186113dd",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 1/10\n",
+      "Training Loss: 3.705597158521414\n",
+      "Top 1 Acc: 0.25392156862745097\n",
+      "\n",
+      "Running Validation Epoch 1/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b6979449794948349a2ccc8599f9958d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 1/10\n",
+      "Val Loss: 2.1234971778467298\n",
+      "Top 1 Acc: 0.5637254901960784\n",
+      "\n",
+      "Running Training Epoch 2/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "6a116d25fa6a45d193d8203737d4ecfb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 2/10\n",
+      "Training Loss: 1.32985198777169\n",
+      "Top 1 Acc: 0.8382352941176471\n",
+      "\n",
+      "Running Validation Epoch 2/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "1181f510948745b69dbd75ff476135c8",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 2/10\n",
+      "Val Loss: 0.956481144297868\n",
+      "Top 1 Acc: 0.8303921568627451\n",
+      "\n",
+      "Running Training Epoch 3/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "c130a002f8944a3ea6a3a4c1544c3c7a",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 3/10\n",
+      "Training Loss: 0.4278648835606873\n",
+      "Top 1 Acc: 0.9803921568627451\n",
+      "\n",
+      "Running Validation Epoch 3/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "a5cc3447dac84178b3800c1a30325e23",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 3/10\n",
+      "Val Loss: 0.6528619016171433\n",
+      "Top 1 Acc: 0.8578431372549019\n",
+      "\n",
+      "Running Training Epoch 4/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d68ce2162a92413d9a74ae4b8c200a95",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 4/10\n",
+      "Training Loss: 0.15628680447116494\n",
+      "Top 1 Acc: 0.9970588235294118\n",
+      "\n",
+      "Running Validation Epoch 4/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f527bc8cfa6c4222b5703eeeaa2c07e9",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 4/10\n",
+      "Val Loss: 0.5102186386066023\n",
+      "Top 1 Acc: 0.9049019607843137\n",
+      "\n",
+      "Running Training Epoch 5/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d60459fa3c46401baa940ff07489ee04",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 5/10\n",
+      "Training Loss: 0.060788551229052246\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 5/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f2b5ab2844704170b08ffe5857d3bf0d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 5/10\n",
+      "Val Loss: 0.42239774897461757\n",
+      "Top 1 Acc: 0.9107843137254902\n",
+      "\n",
+      "Running Training Epoch 6/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "65c3b9cd9e844ee096d6152cc07146ae",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 6/10\n",
+      "Training Loss: 0.0473741386376787\n",
+      "Top 1 Acc: 0.9990196078431373\n",
+      "\n",
+      "Running Validation Epoch 6/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9ee91c900ebb419c8da0edb371d73dda",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 6/10\n",
+      "Val Loss: 0.40489213640103117\n",
+      "Top 1 Acc: 0.9098039215686274\n",
+      "\n",
+      "Running Training Epoch 7/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "c29b7b84bbaa42629bd39eb7b233f111",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 7/10\n",
+      "Training Loss: 0.03090776341559831\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 7/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "a8f37fdd27b647f0a8b00a987906e355",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 7/10\n",
+      "Val Loss: 0.39284458926704247\n",
+      "Top 1 Acc: 0.9166666666666666\n",
+      "\n",
+      "Running Training Epoch 8/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f09644d377b843d6969b52284d3b80f4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 8/10\n",
+      "Training Loss: 0.027710435475455597\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 8/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "1f0d5025929b4cc2837a4e4b27e969ad",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 8/10\n",
+      "Val Loss: 0.3832304854586255\n",
+      "Top 1 Acc: 0.9176470588235294\n",
+      "\n",
+      "Running Training Epoch 9/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "81aee6700e444b8cb6cb99d31c69980a",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 9/10\n",
+      "Training Loss: 0.026527045309194364\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 9/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "c0e66e476ac049f487204671bb3c41ce",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 9/10\n",
+      "Val Loss: 0.36842548519780394\n",
+      "Top 1 Acc: 0.9186274509803921\n",
+      "\n",
+      "Running Training Epoch 10/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "756a007a3b02480b8872b65fb88277ab",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 10/10\n",
+      "Training Loss: 0.025542080067680217\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 10/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "affa32f0824e4f77a830bf1392c8b3cb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 10/10\n",
+      "Val Loss: 0.3675568748876685\n",
+      "Top 1 Acc: 0.9186274509803921\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "epoch = 0\n",
+    "for epoch in range(manager.max_epochs):\n",
+    "    # run training loop\n",
+    "    epoch_name = f\"{epoch + 1}/{manager.max_epochs}\"\n",
+    "    \n",
+    "    print(f\"Running Training Epoch {epoch_name}\")\n",
+    "    train_loss, train_acc = run_model_one_epoch(model, train_loader, criterion, device, train=True, optimizer=optimizer)\n",
+    "    print(f\"Training Epoch: {epoch_name}\\nTraining Loss: {train_loss}\\nTop 1 Acc: {train_acc}\\n\")\n",
+    "\n",
+    "    # run validation loop\n",
+    "    print(f\"Running Validation Epoch {epoch_name}\")\n",
+    "    val_loss, val_acc = run_model_one_epoch(model, val_loader, criterion, device)\n",
+    "    print(f\"Validation Epoch: {epoch_name}\\nVal Loss: {val_loss}\\nTop 1 Acc: {val_acc}\\n\")\n",
+    "\n",
+    "# clean up\n",
+    "manager.finalize(model)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "1eb3d14e",
+   "metadata": {},
+   "source": [
+    "Export the model in case we want to reload in the future, so we do not have to rerun."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "11fa4cf2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "save_dir = \"densenet-models\"\n",
+    "exporter = ModuleExporter(model, output_dir=save_dir)\n",
+    "exporter.export_pytorch(name=\"dense-model.pth\")\n",
+    "exporter.export_onnx(torch.randn(1, 3, 224, 224), name=\"dense-model.onnx\", convert_qat=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "4532442e",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "torch.cuda.empty_cache()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "8a7d6309",
+   "metadata": {},
+   "source": [
+    "## Step 4: Prune The Model\n",
+    "\n",
+    "With a model trained on Flowers, we are now ready to apply the GMP algorithm to prune the model. The GMP algorithm is an interative pruning algorithm. At the end of each epoch, we identify the lowest magnitude weights (those closest to 0) and remove them from the network starting from an initial level of sparsity until a final level of sparsity. The remaining nonzero weights are then fine-tuned onto training dataset."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "8447d074",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# first, load the trained model from Part 3\n",
+    "checkpoint = torch.load(\"./densenet-models/training/dense-model.pth\")\n",
+    "model = torchvision.models.densenet121()\n",
+    "model.classifier = torch.nn.Linear(model.classifier.in_features, NUM_LABELS)\n",
+    "model.load_state_dict(checkpoint['state_dict'])\n",
+    "model.to(device)\n",
+    "\n",
+    "# setup loss function and optimizer, LR will be overriden by sparseml\n",
+    "criterion = CrossEntropyLoss()\n",
+    "optimizer = Adam(model.parameters(), lr=8e-3)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7a05d884",
+   "metadata": {},
+   "source": [
+    "Next, we need to create a SparseML recipe which includes the GMP algorithm. The `!GlobalMagnitudePruningModifier` modifier instructs SparseML to apply the GMP algorithm at a global level (pruning the lowest magnitude weights across all layers).\n",
+    "\n",
+    "Firstly, we need to decide identify which parameters of the model to apply the GMP algorithm to. We can use the `get_prunable_layers` function to inspect:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "id": "d367905a",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "features.conv0\n",
+      "features.denseblock1.denselayer1.conv1\n",
+      "features.denseblock1.denselayer1.conv2\n",
+      "features.denseblock1.denselayer2.conv1\n",
+      "features.denseblock1.denselayer2.conv2\n",
+      "features.denseblock1.denselayer3.conv1\n",
+      "features.denseblock1.denselayer3.conv2\n",
+      "features.denseblock1.denselayer4.conv1\n",
+      "features.denseblock1.denselayer4.conv2\n",
+      "features.denseblock1.denselayer5.conv1\n",
+      "features.denseblock1.denselayer5.conv2\n",
+      "features.denseblock1.denselayer6.conv1\n",
+      "features.denseblock1.denselayer6.conv2\n",
+      "features.transition1.conv\n",
+      "features.denseblock2.denselayer1.conv1\n",
+      "features.denseblock2.denselayer1.conv2\n",
+      "features.denseblock2.denselayer2.conv1\n",
+      "features.denseblock2.denselayer2.conv2\n",
+      "features.denseblock2.denselayer3.conv1\n",
+      "features.denseblock2.denselayer3.conv2\n",
+      "features.denseblock2.denselayer4.conv1\n",
+      "features.denseblock2.denselayer4.conv2\n",
+      "features.denseblock2.denselayer5.conv1\n",
+      "features.denseblock2.denselayer5.conv2\n",
+      "features.denseblock2.denselayer6.conv1\n",
+      "features.denseblock2.denselayer6.conv2\n",
+      "features.denseblock2.denselayer7.conv1\n",
+      "features.denseblock2.denselayer7.conv2\n",
+      "features.denseblock2.denselayer8.conv1\n",
+      "features.denseblock2.denselayer8.conv2\n",
+      "features.denseblock2.denselayer9.conv1\n",
+      "features.denseblock2.denselayer9.conv2\n",
+      "features.denseblock2.denselayer10.conv1\n",
+      "features.denseblock2.denselayer10.conv2\n",
+      "features.denseblock2.denselayer11.conv1\n",
+      "features.denseblock2.denselayer11.conv2\n",
+      "features.denseblock2.denselayer12.conv1\n",
+      "features.denseblock2.denselayer12.conv2\n",
+      "features.transition2.conv\n",
+      "features.denseblock3.denselayer1.conv1\n",
+      "features.denseblock3.denselayer1.conv2\n",
+      "features.denseblock3.denselayer2.conv1\n",
+      "features.denseblock3.denselayer2.conv2\n",
+      "features.denseblock3.denselayer3.conv1\n",
+      "features.denseblock3.denselayer3.conv2\n",
+      "features.denseblock3.denselayer4.conv1\n",
+      "features.denseblock3.denselayer4.conv2\n",
+      "features.denseblock3.denselayer5.conv1\n",
+      "features.denseblock3.denselayer5.conv2\n",
+      "features.denseblock3.denselayer6.conv1\n",
+      "features.denseblock3.denselayer6.conv2\n",
+      "features.denseblock3.denselayer7.conv1\n",
+      "features.denseblock3.denselayer7.conv2\n",
+      "features.denseblock3.denselayer8.conv1\n",
+      "features.denseblock3.denselayer8.conv2\n",
+      "features.denseblock3.denselayer9.conv1\n",
+      "features.denseblock3.denselayer9.conv2\n",
+      "features.denseblock3.denselayer10.conv1\n",
+      "features.denseblock3.denselayer10.conv2\n",
+      "features.denseblock3.denselayer11.conv1\n",
+      "features.denseblock3.denselayer11.conv2\n",
+      "features.denseblock3.denselayer12.conv1\n",
+      "features.denseblock3.denselayer12.conv2\n",
+      "features.denseblock3.denselayer13.conv1\n",
+      "features.denseblock3.denselayer13.conv2\n",
+      "features.denseblock3.denselayer14.conv1\n",
+      "features.denseblock3.denselayer14.conv2\n",
+      "features.denseblock3.denselayer15.conv1\n",
+      "features.denseblock3.denselayer15.conv2\n",
+      "features.denseblock3.denselayer16.conv1\n",
+      "features.denseblock3.denselayer16.conv2\n",
+      "features.denseblock3.denselayer17.conv1\n",
+      "features.denseblock3.denselayer17.conv2\n",
+      "features.denseblock3.denselayer18.conv1\n",
+      "features.denseblock3.denselayer18.conv2\n",
+      "features.denseblock3.denselayer19.conv1\n",
+      "features.denseblock3.denselayer19.conv2\n",
+      "features.denseblock3.denselayer20.conv1\n",
+      "features.denseblock3.denselayer20.conv2\n",
+      "features.denseblock3.denselayer21.conv1\n",
+      "features.denseblock3.denselayer21.conv2\n",
+      "features.denseblock3.denselayer22.conv1\n",
+      "features.denseblock3.denselayer22.conv2\n",
+      "features.denseblock3.denselayer23.conv1\n",
+      "features.denseblock3.denselayer23.conv2\n",
+      "features.denseblock3.denselayer24.conv1\n",
+      "features.denseblock3.denselayer24.conv2\n",
+      "features.transition3.conv\n",
+      "features.denseblock4.denselayer1.conv1\n",
+      "features.denseblock4.denselayer1.conv2\n",
+      "features.denseblock4.denselayer2.conv1\n",
+      "features.denseblock4.denselayer2.conv2\n",
+      "features.denseblock4.denselayer3.conv1\n",
+      "features.denseblock4.denselayer3.conv2\n",
+      "features.denseblock4.denselayer4.conv1\n",
+      "features.denseblock4.denselayer4.conv2\n",
+      "features.denseblock4.denselayer5.conv1\n",
+      "features.denseblock4.denselayer5.conv2\n",
+      "features.denseblock4.denselayer6.conv1\n",
+      "features.denseblock4.denselayer6.conv2\n",
+      "features.denseblock4.denselayer7.conv1\n",
+      "features.denseblock4.denselayer7.conv2\n",
+      "features.denseblock4.denselayer8.conv1\n",
+      "features.denseblock4.denselayer8.conv2\n",
+      "features.denseblock4.denselayer9.conv1\n",
+      "features.denseblock4.denselayer9.conv2\n",
+      "features.denseblock4.denselayer10.conv1\n",
+      "features.denseblock4.denselayer10.conv2\n",
+      "features.denseblock4.denselayer11.conv1\n",
+      "features.denseblock4.denselayer11.conv2\n",
+      "features.denseblock4.denselayer12.conv1\n",
+      "features.denseblock4.denselayer12.conv2\n",
+      "features.denseblock4.denselayer13.conv1\n",
+      "features.denseblock4.denselayer13.conv2\n",
+      "features.denseblock4.denselayer14.conv1\n",
+      "features.denseblock4.denselayer14.conv2\n",
+      "features.denseblock4.denselayer15.conv1\n",
+      "features.denseblock4.denselayer15.conv2\n",
+      "features.denseblock4.denselayer16.conv1\n",
+      "features.denseblock4.denselayer16.conv2\n",
+      "classifier\n"
+     ]
+    }
+   ],
+   "source": [
+    "# print parameters\n",
+    "for (name, layer) in get_prunable_layers(model):\n",
+    "    print(f\"{name}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "322fa136",
+   "metadata": {},
+   "source": [
+    "We will apply pruning to each of the `convs` and exclude the `classifier` (which is the final projection head). Fortunately, SparseML allows us to pass regexes to identify layers in the network.\n",
+    "\n",
+    "Here is what the recipe looks like:\n",
+    "\n",
+    "```yaml\n",
+    "# Epoch and Learning-Rate variables\n",
+    "num_epochs: 13.0\n",
+    "pruning_epochs: 10.0\n",
+    "init_lr: 0.0005\n",
+    "final_lr: 0.0001\n",
+    "inter_func: cubic\n",
+    "mask_type: unstructured\n",
+    "\n",
+    "training_modifiers:\n",
+    "  - !EpochRangeModifier\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: eval(final_lr)\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(pruning_epochs)\n",
+    "    \n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: eval(final_lr)\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: eval(pruning_epochs)\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "# Pruning\n",
+    "pruning_modifiers:\n",
+    "  - !GlobalMagnitudePruningModifier\n",
+    "    init_sparsity: 0.05\n",
+    "    final_sparsity: 0.9\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(pruning_epochs)\n",
+    "    update_frequency: 0.5\n",
+    "    params: \n",
+    "        - 'features.conv0.weight'\n",
+    "        - 're:features.denseblock1.*.conv1.weight'\n",
+    "        - 're:features.denseblock1.*.conv2.weight'\n",
+    "        - 're:features.transition1.conv.weight'\n",
+    "        - 're:features.denseblock2.*.conv1.weight'\n",
+    "        - 're:features.denseblock2.*.conv2.weight'\n",
+    "        - 're:features.transition2.conv.weight'\n",
+    "        - 're:features.denseblock3.*.conv1.weight'\n",
+    "        - 're:features.denseblock3.*.conv2.weight'\n",
+    "        - 're:features.transition3.conv.weight'\n",
+    "        - 're:features.denseblock4.*.conv1.weight'\n",
+    "        - 're:features.denseblock4.*.conv2.weight'\n",
+    "    leave_enabled: True\n",
+    "    inter_func: eval(inter_func)\n",
+    "    mask_type: eval(mask_type)\n",
+    "```\n",
+    "\n",
+    "This recipe specifies that we will run the GMP algorithm for the first 10 epochs. We start at an `init_sparsity` level of 5% and gradually increase sparsity to a `final_sparsity` level of 90% following a `cubic` curve across each of the `convs` in the network. The pruning is applied in an `unstructured` manner, meaning that any weight can be pruned in any pattern.\n",
+    "\n",
+    "Over the final 3 epochs, we fine-tune the 90% pruned model further. Since we set `leave_enabled=True` the sparsity level will be maintained as the fine-tuning occurs."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "b8518b39",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "pruning_recipe_path = \"./recipes/densenet-flowers-pruning-recipe.yaml\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9b17fc31",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [],
+   "source": [
+    "!cat ./recipes/densenet-flowers-pruning-recipe.yaml"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "id": "68646bc7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# create ScheduledModifierManager and Optimizer wrapper\n",
+    "manager = ScheduledModifierManager.from_yaml(pruning_recipe_path)\n",
+    "logger = TensorBoardLogger(log_path=\"./tensorboard_outputs/densenet/pruning-run\")\n",
+    "optimizer = manager.modify(model, optimizer, loggers=[logger], steps_per_epoch=len(train_loader))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "da0a5909",
+   "metadata": {},
+   "source": [
+    "Next, kick off the GMP training loop. \n",
+    "\n",
+    "As you can see, we use the wrapped `optimizer` and `model` in the same way as above. SparseML parsed the recipe and updated the `optimizer` with the logic of GMP algorithm from the recipe. This allows you to use the `optimizer` and `model` as usual, with all of the pruning-related logic handled by SparseML.\n",
+    "\n",
+    "Our 90% pruned model reaches ~91.8% validation accuracy (vs ~91.9% for the dense model)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "id": "601c8c21",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running Training Epoch 1/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "5790f1d7b4ae4329af773cc7a7537fe2",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 1/13\n",
+      "Training Loss: 0.22910855180816725\n",
+      "Top 1 Acc: 0.9735294117647059\n",
+      "\n",
+      "Running Validation Epoch 1/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "bb1be77737da4057a7553cc3546c5092",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 1/13\n",
+      "Val Loss: 0.9752587381517515\n",
+      "Top 1 Acc: 0.7333333333333333\n",
+      "\n",
+      "Running Training Epoch 2/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b7c7a74efde14715a4a7148cbc748a67",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 2/13\n",
+      "Training Loss: 0.1926842465181835\n",
+      "Top 1 Acc: 0.9676470588235294\n",
+      "\n",
+      "Running Validation Epoch 2/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "6b841d8f660e415781a589f4c2537c6c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 2/13\n",
+      "Val Loss: 0.5537524847895838\n",
+      "Top 1 Acc: 0.85\n",
+      "\n",
+      "Running Training Epoch 3/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0b04f316fb1b480bb9e53ba6745aca99",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 3/13\n",
+      "Training Loss: 0.0884120583650656\n",
+      "Top 1 Acc: 0.9882352941176471\n",
+      "\n",
+      "Running Validation Epoch 3/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "4db527bfb864464a818217a04dc6b9c3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 3/13\n",
+      "Val Loss: 0.458682878066611\n",
+      "Top 1 Acc: 0.8764705882352941\n",
+      "\n",
+      "Running Training Epoch 4/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9715e9c734da4f598febbe6f6115f3f4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 4/13\n",
+      "Training Loss: 0.07008300906454679\n",
+      "Top 1 Acc: 0.9941176470588236\n",
+      "\n",
+      "Running Validation Epoch 4/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "10fca0f38b4d486eb37fbe85151eae00",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 4/13\n",
+      "Val Loss: 0.4294259559828788\n",
+      "Top 1 Acc: 0.8872549019607843\n",
+      "\n",
+      "Running Training Epoch 5/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "50143e51efae42bdb8e27f2a103f187c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 5/13\n",
+      "Training Loss: 0.058840728845098056\n",
+      "Top 1 Acc: 0.9941176470588236\n",
+      "\n",
+      "Running Validation Epoch 5/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "612376311bfd4f47805fd0a943d5da8e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 5/13\n",
+      "Val Loss: 0.4169776111084502\n",
+      "Top 1 Acc: 0.8970588235294118\n",
+      "\n",
+      "Running Training Epoch 6/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "c2b54036b759447a9f11b1d38b85be1d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 6/13\n",
+      "Training Loss: 0.06827072701707948\n",
+      "Top 1 Acc: 0.9931372549019608\n",
+      "\n",
+      "Running Validation Epoch 6/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "1b944f8a61c241df9c9b81cd97fda4e3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 6/13\n",
+      "Val Loss: 0.4382545033295173\n",
+      "Top 1 Acc: 0.8921568627450981\n",
+      "\n",
+      "Running Training Epoch 7/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "637f8d0ce14943cabdab66efe3755b21",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 7/13\n",
+      "Training Loss: 0.07268102490343153\n",
+      "Top 1 Acc: 0.9921568627450981\n",
+      "\n",
+      "Running Validation Epoch 7/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e80a8bd36ef14a20983353949775e5a4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 7/13\n",
+      "Val Loss: 0.4186342091416009\n",
+      "Top 1 Acc: 0.8941176470588236\n",
+      "\n",
+      "Running Training Epoch 8/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ef52b8ed50b746409cfe942bf1043cdd",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 8/13\n",
+      "Training Loss: 0.062260200167656876\n",
+      "Top 1 Acc: 0.9950980392156863\n",
+      "\n",
+      "Running Validation Epoch 8/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3998fd77d6ae4769b9e829f6b7303a75",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 8/13\n",
+      "Val Loss: 0.4408616948639974\n",
+      "Top 1 Acc: 0.8872549019607843\n",
+      "\n",
+      "Running Training Epoch 9/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ec575690c5074b558e9ca9cc9b00b91c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 9/13\n",
+      "Training Loss: 0.0411255351791624\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 9/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "91c6e3618a1c4d3fa6c19768dd6d2bdb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 9/13\n",
+      "Val Loss: 0.4180755140114343\n",
+      "Top 1 Acc: 0.9019607843137255\n",
+      "\n",
+      "Running Training Epoch 10/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "781584fe5b2d4d5ea951292b374846e5",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 10/13\n",
+      "Training Loss: 0.027908188028959557\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 10/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "5017b756091946a69fc2bc93809ef239",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 10/13\n",
+      "Val Loss: 0.40239793025830295\n",
+      "Top 1 Acc: 0.8980392156862745\n",
+      "\n",
+      "Running Training Epoch 11/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "caf1167eb6d04a2e87614e712cee2815",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 11/13\n",
+      "Training Loss: 0.02035851358959917\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 11/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0b54ca171d254220b64003390124cde4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 11/13\n",
+      "Val Loss: 0.3589820931883878\n",
+      "Top 1 Acc: 0.9088235294117647\n",
+      "\n",
+      "Running Training Epoch 12/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0f60ba49298345a48864fa001c26b8ce",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 12/13\n",
+      "Training Loss: 0.012050284800352529\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 12/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "4b98b4757b144440b63af638152e34c2",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 12/13\n",
+      "Val Loss: 0.35020815586904064\n",
+      "Top 1 Acc: 0.9127450980392157\n",
+      "\n",
+      "Running Training Epoch 13/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e242702241c045308936b241588e3fd4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 13/13\n",
+      "Training Loss: 0.007597094325319631\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 13/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "80fa6f95af8a43b8bdbeffaa502061e1",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 13/13\n",
+      "Val Loss: 0.3375227072065172\n",
+      "Top 1 Acc: 0.9186274509803921\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "# run GMP algorithm\n",
+    "epoch = 0\n",
+    "for epoch in range(manager.max_epochs):\n",
+    "    # run training loop\n",
+    "    epoch_name = f\"{epoch + 1}/{manager.max_epochs}\"\n",
+    "    \n",
+    "    print(f\"Running Training Epoch {epoch_name}\")\n",
+    "    train_loss, train_acc = run_model_one_epoch(model, train_loader, criterion, device, train=True, optimizer=optimizer)\n",
+    "    print(f\"Training Epoch: {epoch_name}\\nTraining Loss: {train_loss}\\nTop 1 Acc: {train_acc}\\n\")\n",
+    "\n",
+    "    # run validation loop\n",
+    "    print(f\"Running Validation Epoch {epoch_name}\")\n",
+    "    val_loss, val_acc = run_model_one_epoch(model, val_loader, criterion, device)\n",
+    "    print(f\"Validation Epoch: {epoch_name}\\nVal Loss: {val_loss}\\nTop 1 Acc: {val_acc}\\n\")\n",
+    "    \n",
+    "    logger.log_scalar(\"Metrics/Loss (Train)\", train_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Train)\", train_acc, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Loss (Validation)\", val_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Validation)\", val_acc, epoch)\n",
+    "\n",
+    "manager.finalize(model)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "22c96232",
+   "metadata": {},
+   "source": [
+    "Here is a sample of the TensorBoard output, showing the validation accuracy, a particular layer's sparsity level, and the learning rate over time.\n",
+    "\n",
+    "![tensorboard output](./images/densenet-flowers-tensorboard-output.png)\n",
+    "\n",
+    "The resulting model is is 90% sparse and achieves validation accuracy of ~91.8% (vs the unoptimized dense model at ~92.5%) without much hyperparameter search. Key hyperparameter experiments you may want to run include:\n",
+    "- Learning rate\n",
+    "- Learning rate schedule\n",
+    "- Sparsity level\n",
+    "- Number of pruning epochs"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "id": "df5afa94",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Sparsity By Layer:\n",
+      "features.conv0.weight: 0.4737\n",
+      "features.denseblock1.denselayer1.conv1.weight: 0.7423\n",
+      "features.denseblock1.denselayer1.conv2.weight: 0.8321\n",
+      "features.denseblock1.denselayer2.conv1.weight: 0.7501\n",
+      "features.denseblock1.denselayer2.conv2.weight: 0.8680\n",
+      "features.denseblock1.denselayer3.conv1.weight: 0.7543\n",
+      "features.denseblock1.denselayer3.conv2.weight: 0.8259\n",
+      "features.denseblock1.denselayer4.conv1.weight: 0.8161\n",
+      "features.denseblock1.denselayer4.conv2.weight: 0.8302\n",
+      "features.denseblock1.denselayer5.conv1.weight: 0.8723\n",
+      "features.denseblock1.denselayer5.conv2.weight: 0.8858\n",
+      "features.denseblock1.denselayer6.conv1.weight: 0.8428\n",
+      "features.denseblock1.denselayer6.conv2.weight: 0.8443\n",
+      "features.transition1.conv.weight: 0.7129\n",
+      "features.denseblock2.denselayer1.conv1.weight: 0.9314\n",
+      "features.denseblock2.denselayer1.conv2.weight: 0.8886\n",
+      "features.denseblock2.denselayer2.conv1.weight: 0.8830\n",
+      "features.denseblock2.denselayer2.conv2.weight: 0.8611\n",
+      "features.denseblock2.denselayer3.conv1.weight: 0.8865\n",
+      "features.denseblock2.denselayer3.conv2.weight: 0.8476\n",
+      "features.denseblock2.denselayer4.conv1.weight: 0.8632\n",
+      "features.denseblock2.denselayer4.conv2.weight: 0.8669\n",
+      "features.denseblock2.denselayer5.conv1.weight: 0.8920\n",
+      "features.denseblock2.denselayer5.conv2.weight: 0.8509\n",
+      "features.denseblock2.denselayer6.conv1.weight: 0.9111\n",
+      "features.denseblock2.denselayer6.conv2.weight: 0.8686\n",
+      "features.denseblock2.denselayer7.conv1.weight: 0.8802\n",
+      "features.denseblock2.denselayer7.conv2.weight: 0.8758\n",
+      "features.denseblock2.denselayer8.conv1.weight: 0.8886\n",
+      "features.denseblock2.denselayer8.conv2.weight: 0.8620\n",
+      "features.denseblock2.denselayer9.conv1.weight: 0.9019\n",
+      "features.denseblock2.denselayer9.conv2.weight: 0.8794\n",
+      "features.denseblock2.denselayer10.conv1.weight: 0.8742\n",
+      "features.denseblock2.denselayer10.conv2.weight: 0.8851\n",
+      "features.denseblock2.denselayer11.conv1.weight: 0.8860\n",
+      "features.denseblock2.denselayer11.conv2.weight: 0.8894\n",
+      "features.denseblock2.denselayer12.conv1.weight: 0.8729\n",
+      "features.denseblock2.denselayer12.conv2.weight: 0.9024\n",
+      "features.transition2.conv.weight: 0.7744\n",
+      "features.denseblock3.denselayer1.conv1.weight: 0.9163\n",
+      "features.denseblock3.denselayer1.conv2.weight: 0.8984\n",
+      "features.denseblock3.denselayer2.conv1.weight: 0.9366\n",
+      "features.denseblock3.denselayer2.conv2.weight: 0.8999\n",
+      "features.denseblock3.denselayer3.conv1.weight: 0.9175\n",
+      "features.denseblock3.denselayer3.conv2.weight: 0.9029\n",
+      "features.denseblock3.denselayer4.conv1.weight: 0.9195\n",
+      "features.denseblock3.denselayer4.conv2.weight: 0.8846\n",
+      "features.denseblock3.denselayer5.conv1.weight: 0.9337\n",
+      "features.denseblock3.denselayer5.conv2.weight: 0.8808\n",
+      "features.denseblock3.denselayer6.conv1.weight: 0.9185\n",
+      "features.denseblock3.denselayer6.conv2.weight: 0.8808\n",
+      "features.denseblock3.denselayer7.conv1.weight: 0.9392\n",
+      "features.denseblock3.denselayer7.conv2.weight: 0.8726\n",
+      "features.denseblock3.denselayer8.conv1.weight: 0.9046\n",
+      "features.denseblock3.denselayer8.conv2.weight: 0.8859\n",
+      "features.denseblock3.denselayer9.conv1.weight: 0.9096\n",
+      "features.denseblock3.denselayer9.conv2.weight: 0.8791\n",
+      "features.denseblock3.denselayer10.conv1.weight: 0.9393\n",
+      "features.denseblock3.denselayer10.conv2.weight: 0.8824\n",
+      "features.denseblock3.denselayer11.conv1.weight: 0.9334\n",
+      "features.denseblock3.denselayer11.conv2.weight: 0.8691\n",
+      "features.denseblock3.denselayer12.conv1.weight: 0.9233\n",
+      "features.denseblock3.denselayer12.conv2.weight: 0.8895\n",
+      "features.denseblock3.denselayer13.conv1.weight: 0.9379\n",
+      "features.denseblock3.denselayer13.conv2.weight: 0.8606\n",
+      "features.denseblock3.denselayer14.conv1.weight: 0.9445\n",
+      "features.denseblock3.denselayer14.conv2.weight: 0.8798\n",
+      "features.denseblock3.denselayer15.conv1.weight: 0.9243\n",
+      "features.denseblock3.denselayer15.conv2.weight: 0.8870\n",
+      "features.denseblock3.denselayer16.conv1.weight: 0.9368\n",
+      "features.denseblock3.denselayer16.conv2.weight: 0.9131\n",
+      "features.denseblock3.denselayer17.conv1.weight: 0.9187\n",
+      "features.denseblock3.denselayer17.conv2.weight: 0.9029\n",
+      "features.denseblock3.denselayer18.conv1.weight: 0.9293\n",
+      "features.denseblock3.denselayer18.conv2.weight: 0.8975\n",
+      "features.denseblock3.denselayer19.conv1.weight: 0.9232\n",
+      "features.denseblock3.denselayer19.conv2.weight: 0.8956\n",
+      "features.denseblock3.denselayer20.conv1.weight: 0.9250\n",
+      "features.denseblock3.denselayer20.conv2.weight: 0.9041\n",
+      "features.denseblock3.denselayer21.conv1.weight: 0.9216\n",
+      "features.denseblock3.denselayer21.conv2.weight: 0.8955\n",
+      "features.denseblock3.denselayer22.conv1.weight: 0.9245\n",
+      "features.denseblock3.denselayer22.conv2.weight: 0.8802\n",
+      "features.denseblock3.denselayer23.conv1.weight: 0.9230\n",
+      "features.denseblock3.denselayer23.conv2.weight: 0.9111\n",
+      "features.denseblock3.denselayer24.conv1.weight: 0.9264\n",
+      "features.denseblock3.denselayer24.conv2.weight: 0.9092\n",
+      "features.transition3.conv.weight: 0.8496\n",
+      "features.denseblock4.denselayer1.conv1.weight: 0.8750\n",
+      "features.denseblock4.denselayer1.conv2.weight: 0.9133\n",
+      "features.denseblock4.denselayer2.conv1.weight: 0.8727\n",
+      "features.denseblock4.denselayer2.conv2.weight: 0.9468\n",
+      "features.denseblock4.denselayer3.conv1.weight: 0.8816\n",
+      "features.denseblock4.denselayer3.conv2.weight: 0.9228\n",
+      "features.denseblock4.denselayer4.conv1.weight: 0.9007\n",
+      "features.denseblock4.denselayer4.conv2.weight: 0.9406\n",
+      "features.denseblock4.denselayer5.conv1.weight: 0.9005\n",
+      "features.denseblock4.denselayer5.conv2.weight: 0.9582\n",
+      "features.denseblock4.denselayer6.conv1.weight: 0.8966\n",
+      "features.denseblock4.denselayer6.conv2.weight: 0.9577\n",
+      "features.denseblock4.denselayer7.conv1.weight: 0.9111\n",
+      "features.denseblock4.denselayer7.conv2.weight: 0.9601\n",
+      "features.denseblock4.denselayer8.conv1.weight: 0.9141\n",
+      "features.denseblock4.denselayer8.conv2.weight: 0.9624\n",
+      "features.denseblock4.denselayer9.conv1.weight: 0.9165\n",
+      "features.denseblock4.denselayer9.conv2.weight: 0.9656\n",
+      "features.denseblock4.denselayer10.conv1.weight: 0.9182\n",
+      "features.denseblock4.denselayer10.conv2.weight: 0.9647\n",
+      "features.denseblock4.denselayer11.conv1.weight: 0.9191\n",
+      "features.denseblock4.denselayer11.conv2.weight: 0.9645\n",
+      "features.denseblock4.denselayer12.conv1.weight: 0.9226\n",
+      "features.denseblock4.denselayer12.conv2.weight: 0.9655\n",
+      "features.denseblock4.denselayer13.conv1.weight: 0.9245\n",
+      "features.denseblock4.denselayer13.conv2.weight: 0.9638\n",
+      "features.denseblock4.denselayer14.conv1.weight: 0.9280\n",
+      "features.denseblock4.denselayer14.conv2.weight: 0.9642\n",
+      "features.denseblock4.denselayer15.conv1.weight: 0.9290\n",
+      "features.denseblock4.denselayer15.conv2.weight: 0.9638\n",
+      "features.denseblock4.denselayer16.conv1.weight: 0.9251\n",
+      "features.denseblock4.denselayer16.conv2.weight: 0.9610\n",
+      "classifier.weight: 0.0000\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(\"Sparsity By Layer:\")\n",
+    "for (name, layer) in get_prunable_layers(model):\n",
+    "    print(f\"{name}.weight: {tensor_sparsity(layer.weight).item():.4f}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "18fed85b",
+   "metadata": {},
+   "source": [
+    "Finally, export your model to ONNX."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "417ac105",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "save_dir = \"densenet-models\"\n",
+    "exporter = ModuleExporter(model, output_dir=save_dir)\n",
+    "exporter.export_pytorch(name=\"densenet-pruned-fp32.pth\")\n",
+    "exporter.export_onnx(torch.randn(1, 3, 224, 224), name=\"pruned-fp32-model.onnx\", convert_qat=True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c379d623",
+   "metadata": {},
+   "source": [
+    "## Step 5: Quantize the Model\n",
+    "\n",
+    "With a sparse model trained on Flowers, we are now ready to apply the QAT algorithm to quantize the model. Generally, we can run QAT for 5-6 epochs. Quantizing the model together with pruning allows us to gain a speedup when running with DeepSparse."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 38,
+   "id": "11f97312",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "torch.cuda.empty_cache()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 40,
+   "id": "fda0599c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# first, load the trained model from Step 4\n",
+    "checkpoint = torch.load(\"./densenet-models/training/densenet-pruned-fp32.pth\")\n",
+    "model = torchvision.models.densenet121()\n",
+    "model.classifier = torch.nn.Linear(model.classifier.in_features, NUM_LABELS)\n",
+    "model.load_state_dict(checkpoint['state_dict'])\n",
+    "model.to(device)\n",
+    "\n",
+    "# setup loss function and optimizer, LR will be overriden by sparseml\n",
+    "criterion = CrossEntropyLoss()\n",
+    "optimizer = Adam(model.parameters(), lr=8e-3)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "0473fcbd",
+   "metadata": {},
+   "source": [
+    "Next, we will create a Quantization recipe. We will us the `QuantizationModifier` (which instructs SparseML to run QAT) along with the `ConstantPruningModifier` (which instructs SparseML to maintain sparsity).\n",
+    "\n",
+    "Here is what the recipe looks like:\n",
+    "\n",
+    "```yaml\n",
+    "# Epoch and Learning-Rate variables\n",
+    "num_epochs: 6.0\n",
+    "init_lr: 0.0004\n",
+    "\n",
+    "# quantization variables\n",
+    "quantization_epochs: 6.0\n",
+    "\n",
+    "training_modifiers:\n",
+    "  - !EpochRangeModifier\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: 0.0\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "# Phase 1 Sparse Transfer Learning / Recovery\n",
+    "sparse_transfer_learning_modifiers:\n",
+    "  - !ConstantPruningModifier\n",
+    "    start_epoch: 0.0\n",
+    "    params: __ALL_PRUNABLE__\n",
+    "\n",
+    "# Phase 2 Apply quantization\n",
+    "sparse_quantized_transfer_learning_modifiers:\n",
+    "  - !QuantizationModifier\n",
+    "    start_epoch: eval(num_epochs - quantization_epochs)\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 41,
+   "id": "6d2c1be3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "quantization_recipe_path = \"./recipes/quantization-recipe.yaml\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 42,
+   "id": "6b8a4324",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# create ScheduledModifierManager and Optimizer wrapper\n",
+    "manager = ScheduledModifierManager.from_yaml(quantization_recipe_path)\n",
+    "logger = TensorBoardLogger(log_path=\"./tensorboard_outputs/densenet/quantization-run\")\n",
+    "optimizer = manager.modify(model, optimizer, loggers=[logger], steps_per_epoch=len(train_loader))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 43,
+   "id": "208fc85c",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running Training Epoch 1/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "4ed2d6e3d1ac42369f8a9ea020909d50",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 1/5\n",
+      "Training Loss: 0.020167092072369996\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 1/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "8426880f7f1c4c2aa5aa022b0909288d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 1/5\n",
+      "Val Loss: 0.3898960210317455\n",
+      "Top 1 Acc: 0.8980392156862745\n",
+      "\n",
+      "Running Training Epoch 2/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "2c368eb0488549a0b50aca44c594bbf5",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 2/5\n",
+      "Training Loss: 0.011033405537091312\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 2/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "03148f10352a4213a5b49f7db26e2b4f",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 2/5\n",
+      "Val Loss: 0.381910876960319\n",
+      "Top 1 Acc: 0.8970588235294118\n",
+      "\n",
+      "Running Training Epoch 3/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "73988d5735404cc08795116a2d27e19e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 3/5\n",
+      "Training Loss: 0.006457825023971964\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 3/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "61e8db2842564befb05c6c18582093ff",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 3/5\n",
+      "Val Loss: 0.33566539320781885\n",
+      "Top 1 Acc: 0.9137254901960784\n",
+      "\n",
+      "Running Training Epoch 4/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d548a06770834517ad9db8fdb99a1ae3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 4/5\n",
+      "Training Loss: 0.003771477568989212\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 4/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "13afa51fa11b44aaa237067823ae9137",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 4/5\n",
+      "Val Loss: 0.34420355190604823\n",
+      "Top 1 Acc: 0.907843137254902\n",
+      "\n",
+      "Running Training Epoch 5/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "857e1593d5ca457ba73918723f04c008",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 5/5\n",
+      "Training Loss: 0.002937028719316004\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 5/5\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3be1fd3d62ed487fb2d6bacecebc4160",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/64 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 5/5\n",
+      "Val Loss: 0.3459806305090751\n",
+      "Top 1 Acc: 0.9088235294117647\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "# run QAT algorithm\n",
+    "epoch = 0\n",
+    "for epoch in range(manager.max_epochs):\n",
+    "    # run training loop\n",
+    "    epoch_name = f\"{epoch + 1}/{manager.max_epochs}\"\n",
+    "    \n",
+    "    print(f\"Running Training Epoch {epoch_name}\")\n",
+    "    train_loss, train_acc = run_model_one_epoch(model, train_loader, criterion, device, train=True, optimizer=optimizer)\n",
+    "    print(f\"Training Epoch: {epoch_name}\\nTraining Loss: {train_loss}\\nTop 1 Acc: {train_acc}\\n\")\n",
+    "\n",
+    "    # run validation loop\n",
+    "    print(f\"Running Validation Epoch {epoch_name}\")\n",
+    "    val_loss, val_acc = run_model_one_epoch(model, val_loader, criterion, device)\n",
+    "    print(f\"Validation Epoch: {epoch_name}\\nVal Loss: {val_loss}\\nTop 1 Acc: {val_acc}\\n\")\n",
+    "    \n",
+    "    logger.log_scalar(\"Metrics/Loss (Train)\", train_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Train)\", train_acc, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Loss (Validation)\", val_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Validation)\", val_acc, epoch)\n",
+    "\n",
+    "manager.finalize(model)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "52ac51e0",
+   "metadata": {},
+   "source": [
+    "The resulting model is is 90% sparse and INT8 quantized and achieves validation accuracy of ~90.9% (vs the unoptimized dense model at ~91.9%) without much hyperparameter search. Key hyperparameter experiments you may want to run for QAT are:\n",
+    "- Learning rate"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "eec331b7",
+   "metadata": {},
+   "source": [
+    "Finally, export to ONNX:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 44,
+   "id": "87f21bc7",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "2023-03-14 23:27:05 sparseml.pytorch.sparsification.quantization.quantize_qat_export INFO     Converted 59 quantizable Conv ops with weight and bias to ConvInteger and Add\n",
+      "2023-03-14 23:27:05 sparseml.pytorch.sparsification.quantization.quantize_qat_export INFO     Converted 1 quantizable Gemm ops with weight and bias to MatMulInteger and Add\n"
+     ]
+    }
+   ],
+   "source": [
+    "save_dir = \"densenet-models\"\n",
+    "exporter = ModuleExporter(model, output_dir=save_dir)\n",
+    "exporter.export_pytorch(name=\"densenet-pruned-int8.pth\")\n",
+    "exporter.export_onnx(torch.randn(1, 3, 224, 224), name=\"pruned-int8-model.onnx\", convert_qat=True)"
+   ]
+  }
+ ],
+ "metadata": {
+  "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": "3.8.10"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-mobilenetv2-beans.ipynb b/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-mobilenetv2-beans.ipynb
new file mode 100644
index 00000000000..8e1fec22d4b
--- /dev/null
+++ b/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-mobilenetv2-beans.ipynb
@@ -0,0 +1,1905 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "00e081b7",
+   "metadata": {},
+   "source": [
+    "# Sparsifying MobileNetv2 from Scratch (Beans)\n",
+    "\n",
+    "In this example, we will demonstrate how to sparsify an image classification model from scratch using SparseML's PyTorch integration. We train and prune [MobileNetv2](https://pytorch.org/vision/main/models/generated/torchvision.models.mobilenet_v2.html) on the downstream [Beans dataset](https://huggingface.co/datasets/beans) using the Global Magnitude Pruning algorithm. \n",
+    "\n",
+    "## Agenda\n",
+    "\n",
+    "There are a few steps:\n",
+    "\n",
+    " 1. Setup the dataset\n",
+    " 2. Setup the PyTorch training loop\n",
+    " 3. Train a dense version of MobileNetv2\n",
+    " 4. Run the GMP pruning algorithm on the dense model\n",
+    " \n",
+    "## Installation\n",
+    "\n",
+    "Install SparseML with `pip`:\n",
+    "\n",
+    "```\n",
+    "pip install sparseml[torchvision]\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "1ad80edf",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "import sparseml\n",
+    "import torchvision\n",
+    "from sparseml.pytorch.optim import ScheduledModifierManager\n",
+    "from sparseml.pytorch.utils import TensorBoardLogger, ModuleExporter, get_prunable_layers, tensor_sparsity\n",
+    "from torch.utils.data import DataLoader\n",
+    "from torch.nn import CrossEntropyLoss\n",
+    "from torch.optim import Adam\n",
+    "from torchvision import transforms\n",
+    "from tqdm.auto import tqdm\n",
+    "import math\n",
+    "import datasets"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "0d30f71c",
+   "metadata": {},
+   "source": [
+    "## Step 1: Setup Dataset\n",
+    "\n",
+    "Beans leaf dataset is a set of images of diseased and healthy leaves. Based on a leaf image, the goal of this task is to predict the disease type (Angular Leaf Spot and Bean Rust), if any.\n",
+    "\n",
+    "We will use the Hugging Face `datasets` library to download the data and the torchvision `ImageFolder` in the training loop.\n",
+    "\n",
+    "[Checkout the dataset card](https://huggingface.co/datasets/beans)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f8eb3642",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "beans_dataset = datasets.load_dataset(\"beans\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "b06fc4fd",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/eeb026374cf5ecfd5f40131a3159be9b9055ac21a3da11690e7eb4d117c99eee/train/bean_rust/bean_rust_train.84.jpg\n",
+      "/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/f287261265d2f9a3e8f87a5526a54d1847b17f7c3ec5714e5719432f2b3e4a73/validation/bean_rust/bean_rust_val.36.jpg\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(beans_dataset[\"train\"][0][\"image_file_path\"])\n",
+    "print(beans_dataset[\"validation\"][0][\"image_file_path\"])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "4ecdebad",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_path = \"/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/eeb026374cf5ecfd5f40131a3159be9b9055ac21a3da11690e7eb4d117c99eee/train\"\n",
+    "val_path = \"/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/f287261265d2f9a3e8f87a5526a54d1847b17f7c3ec5714e5719432f2b3e4a73/validation\" "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "7d180bbc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "NUM_LABELS = 3\n",
+    "BATCH_SIZE = 32\n",
+    "\n",
+    "# imagenet transforms\n",
+    "imagenet_transform = transforms.Compose([\n",
+    "   transforms.Resize(size=256, interpolation=transforms.InterpolationMode.BILINEAR, max_size=None, antialias=None),\n",
+    "   transforms.CenterCrop(size=(224, 224)),\n",
+    "   transforms.ToTensor(),\n",
+    "   transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])\n",
+    "])\n",
+    "\n",
+    "# datasets\n",
+    "train_dataset = torchvision.datasets.ImageFolder(\n",
+    "    root=train_path,\n",
+    "    transform=imagenet_transform\n",
+    ")\n",
+    "\n",
+    "val_dataset = torchvision.datasets.ImageFolder(\n",
+    "    root=val_path,\n",
+    "    transform=imagenet_transform\n",
+    ")\n",
+    "\n",
+    "# dataloaders\n",
+    "train_loader = DataLoader(train_dataset, BATCH_SIZE, shuffle=True, pin_memory=True, num_workers=16)\n",
+    "val_loader = DataLoader(val_dataset, BATCH_SIZE, shuffle=False, pin_memory=True, num_workers=16)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "132fc514",
+   "metadata": {},
+   "source": [
+    "## Step 2: Setup PyTorch Training Loop\n",
+    "\n",
+    "We will use this training loop below. This is standard PyTorch functionality."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "c968c098",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "cuda\n"
+     ]
+    }
+   ],
+   "source": [
+    "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
+    "print(device)\n",
+    "\n",
+    "def run_model_one_epoch(model, data_loader, criterion, device, train=False, optimizer=None):\n",
+    "    if train:\n",
+    "        model.train()\n",
+    "    else:\n",
+    "        model.eval()\n",
+    "\n",
+    "    running_loss = 0.0\n",
+    "    total_correct = 0\n",
+    "    total_predictions = 0\n",
+    "\n",
+    "    # loop through batches\n",
+    "    for step, (inputs, labels) in tqdm(enumerate(data_loader), total=len(data_loader)):\n",
+    "        inputs = inputs.to(device)\n",
+    "        labels = labels.to(device)\n",
+    "\n",
+    "        if train:\n",
+    "            optimizer.zero_grad()\n",
+    "\n",
+    "        # compute loss, run backpropogation\n",
+    "        outputs = model(inputs)  # model returns logits\n",
+    "        loss = criterion(outputs, labels)\n",
+    "        if train:\n",
+    "            loss.backward()\n",
+    "            optimizer.step()\n",
+    "\n",
+    "        running_loss += loss.item()\n",
+    "\n",
+    "        # run evaluation\n",
+    "        predictions = outputs.argmax(dim=1)\n",
+    "        total_correct += torch.sum(predictions == labels).item()\n",
+    "        total_predictions += inputs.size(0)\n",
+    "\n",
+    "    # return loss and evaluation metric\n",
+    "    loss = running_loss / (step + 1.0)\n",
+    "    accuracy = total_correct / total_predictions\n",
+    "    return loss, accuracy"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a3521e78",
+   "metadata": {},
+   "source": [
+    "## **Step 3: Train MobileNet-v2 on Beans**\n",
+    "\n",
+    "First, we will train a dense version of MobileNetv2 on the Beans dataset."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "4d554578",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [],
+   "source": [
+    "# download pre-trained model, setup classification head\n",
+    "model = torchvision.models.mobilenet_v2(weights=torchvision.models.MobileNet_V2_Weights.DEFAULT)\n",
+    "model.classifier[1] = torch.nn.Linear(model.classifier[1].in_features, NUM_LABELS)\n",
+    "model.to(device)\n",
+    "\n",
+    "# setup loss function and optimizer\n",
+    "criterion = CrossEntropyLoss()\n",
+    "optimizer = Adam(model.parameters(), lr=8e-3) # lr will be override by sparseml"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "31064d5c",
+   "metadata": {},
+   "source": [
+    "Next, we will use SparseML's recipes to set the hyperparameters of training loop. In this case, we will use the following recipe:\n",
+    "\n",
+    "```yaml\n",
+    "# Epoch and Learning-Rate variables\n",
+    "num_epochs: 10.0\n",
+    "init_lr: 0.0005\n",
+    "\n",
+    "training_modifiers:\n",
+    "  - !EpochRangeModifier\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: 0.0\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "```\n",
+    "\n",
+    "As you can see, the recipe includes an `!EpochRangeModifier` and a `!LearningRateFunctionModifier`. These modifiers simply set the number of epochs to train for and the learning rate schedule. As a result, the final model will be dense."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "2723a40f",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!cat ./recipes/mobilenetv2-beans-dense-recipe.yaml"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "id": "cec41d57",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dense_recipe_path = \"./recipes/mobilenetv2-beans-dense-recipe.yaml\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "f7cba342",
+   "metadata": {},
+   "source": [
+    "Next, we use SparseML's `ScheduledModifierManager` to parse and apply the recipe. The `manager.modify` function modifies and wraps the `model` and `optimizer` with the instructions from the recipe. You can use the `model` and `optimizer` just like standard PyTorch objects."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "1749e00d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# create ScheduledModifierManager and Optimizer wrapper\n",
+    "manager = ScheduledModifierManager.from_yaml(dense_recipe_path)\n",
+    "optimizer = manager.modify(model, optimizer, steps_per_epoch=len(train_loader))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a00c4ceb",
+   "metadata": {},
+   "source": [
+    "Kick off the transfer learning loop. Our run reached ~99% validation accuracy after 10 epochs."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "d00d175b",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running Training Epoch 1/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "aaa6edfc1be94d68a8eee93e1691e3aa",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 1/10\n",
+      "Training Loss: 0.47380749520027277\n",
+      "Top 1 Acc: 0.8152804642166345\n",
+      "\n",
+      "Running Validation Epoch 1/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b703e41c6e804bbeacf25aace9f67748",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 1/10\n",
+      "Val Loss: 0.12894474416971208\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 2/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e03a4c8f3706490ab3642e81b149c890",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 2/10\n",
+      "Training Loss: 0.09328031384696563\n",
+      "Top 1 Acc: 0.9690522243713733\n",
+      "\n",
+      "Running Validation Epoch 2/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "4a00338db9b54e1ebf5be2154d026bd4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 2/10\n",
+      "Val Loss: 0.1064663665369153\n",
+      "Top 1 Acc: 0.9548872180451128\n",
+      "\n",
+      "Running Training Epoch 3/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "87566c12df134f38ad2d9f831998f01c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 3/10\n",
+      "Training Loss: 0.04011029944839803\n",
+      "Top 1 Acc: 0.9903288201160542\n",
+      "\n",
+      "Running Validation Epoch 3/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f66ea6ca87944d63ab5d8a2a3fb07969",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 3/10\n",
+      "Val Loss: 0.06654092976823449\n",
+      "Top 1 Acc: 0.9699248120300752\n",
+      "\n",
+      "Running Training Epoch 4/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9c0ab06fdf344eb4a98aa54b02f6e4b0",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 4/10\n",
+      "Training Loss: 0.013448385923931544\n",
+      "Top 1 Acc: 0.9990328820116054\n",
+      "\n",
+      "Running Validation Epoch 4/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f2d999410c2f42fe8ef5c695c51c908d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 4/10\n",
+      "Val Loss: 0.04685080051422119\n",
+      "Top 1 Acc: 0.9699248120300752\n",
+      "\n",
+      "Running Training Epoch 5/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "178d61945ca94dffa214bdb40d7d6bfd",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 5/10\n",
+      "Training Loss: 0.011129878745343762\n",
+      "Top 1 Acc: 0.9970986460348162\n",
+      "\n",
+      "Running Validation Epoch 5/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "37456b087b96435dadb09b871e32694b",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 5/10\n",
+      "Val Loss: 0.04289984308416024\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 6/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "4b8d82516d1d40659e6b2ae7aabf59fb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 6/10\n",
+      "Training Loss: 0.007474540554650241\n",
+      "Top 1 Acc: 0.9990328820116054\n",
+      "\n",
+      "Running Validation Epoch 6/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b5bc30bbadd84ddebcc25510617ae2ce",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 6/10\n",
+      "Val Loss: 0.038813505135476586\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 7/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3df44744907d4833b6e925f8ed7ef4be",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 7/10\n",
+      "Training Loss: 0.022992545157621586\n",
+      "Top 1 Acc: 0.9961315280464217\n",
+      "\n",
+      "Running Validation Epoch 7/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9b42a80e927a41559038ed25a5ba85d9",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 7/10\n",
+      "Val Loss: 0.038337701372802256\n",
+      "Top 1 Acc: 0.9699248120300752\n",
+      "\n",
+      "Running Training Epoch 8/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "cfb91384c3ee4dc0aa18605ca65295b3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 8/10\n",
+      "Training Loss: 0.01563219685693074\n",
+      "Top 1 Acc: 0.9941972920696325\n",
+      "\n",
+      "Running Validation Epoch 8/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ce3a2318d0d14f12b79e6d011afecaaa",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 8/10\n",
+      "Val Loss: 0.052491285931319\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 9/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "15077760bb514451a4fc7fbc45762a6d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 9/10\n",
+      "Training Loss: 0.004495378677154694\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 9/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "770b9912502a4f5eaf820fda80fb0b5c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 9/10\n",
+      "Val Loss: 0.03483513812534511\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 10/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ec9f06d2c1b44fcb8273d49bd106a542",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 10/10\n",
+      "Training Loss: 0.003678643000911865\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 10/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "57a78c45e414444593c0dbcc0e08761e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 10/10\n",
+      "Val Loss: 0.04160818513482809\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "# run transfer learning\n",
+    "epoch = 0\n",
+    "for epoch in range(manager.max_epochs):\n",
+    "    # run training loop\n",
+    "    epoch_name = f\"{epoch + 1}/{manager.max_epochs}\"\n",
+    "    print(f\"Running Training Epoch {epoch_name}\")\n",
+    "    train_loss, train_acc = run_model_one_epoch(model, train_loader, criterion, device, train=True, optimizer=optimizer)\n",
+    "    print(f\"Training Epoch: {epoch_name}\\nTraining Loss: {train_loss}\\nTop 1 Acc: {train_acc}\\n\")\n",
+    "\n",
+    "    # run validation loop\n",
+    "    print(f\"Running Validation Epoch {epoch_name}\")\n",
+    "    val_loss, val_acc = run_model_one_epoch(model, val_loader, criterion, device)\n",
+    "    print(f\"Validation Epoch: {epoch_name}\\nVal Loss: {val_loss}\\nTop 1 Acc: {val_acc}\\n\")\n",
+    "\n",
+    "manager.finalize(model)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "11fa4cf2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from sparseml.pytorch.utils import ModuleExporter\n",
+    "\n",
+    "save_dir = \"dense_model\"\n",
+    "exporter = ModuleExporter(model, output_dir=save_dir)\n",
+    "exporter.export_pytorch(name=\"mobilenet-v2-dense-beans.pth\")\n",
+    "exporter.export_onnx(torch.randn(1, 3, 224, 224), name=\"dense-model.onnx\", convert_qat=True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "273ce478",
+   "metadata": {},
+   "source": [
+    "## Step 4: Prune The Model\n",
+    "\n",
+    "With a model trained on Beans, we are now ready to apply the GMP algorithm to prune the model. The GMP algorithm is an interative pruning algorithm. At the end of each epoch, we identify the lowest magnitude weights (those closest to 0) and remove them from the network starting from an initial level of sparsity until a final level of sparsity. The remaining nonzero weights are then fine-tuned onto training dataset."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "id": "8447d074",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "checkpoint = torch.load(\"./dense_model/training/mobilenet-v2-dense-beans.pth\")\n",
+    "model = torchvision.models.mobilenet_v2()\n",
+    "model.classifier[1] = torch.nn.Linear(model.classifier[1].in_features, NUM_LABELS)\n",
+    "model.load_state_dict(checkpoint['state_dict'])\n",
+    "model.to(device)\n",
+    "\n",
+    "# setup loss function and optimizer, LR will be overriden by sparseml\n",
+    "criterion = CrossEntropyLoss()\n",
+    "optimizer = Adam(model.parameters(), lr=8e-3)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "1b188d60",
+   "metadata": {},
+   "source": [
+    "Next, we need to create a SparseML recipe which includes the GMP algorithm. The `!GlobalMagnitudePruningModifier` modifier instructs SparseML to apply the GMP algorithm at a global level (pruning the lowest magnitude weights across all layers).\n",
+    "\n",
+    "Firstly, we need to decide identify which parameters of the model to apply the GMP algorithm to. We can use the `get_prunable_layers` function to inspect:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "id": "44578399",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "features.0.0\n",
+      "features.1.conv.0.0\n",
+      "features.1.conv.1\n",
+      "features.2.conv.0.0\n",
+      "features.2.conv.1.0\n",
+      "features.2.conv.2\n",
+      "features.3.conv.0.0\n",
+      "features.3.conv.1.0\n",
+      "features.3.conv.2\n",
+      "features.4.conv.0.0\n",
+      "features.4.conv.1.0\n",
+      "features.4.conv.2\n",
+      "features.5.conv.0.0\n",
+      "features.5.conv.1.0\n",
+      "features.5.conv.2\n",
+      "features.6.conv.0.0\n",
+      "features.6.conv.1.0\n",
+      "features.6.conv.2\n",
+      "features.7.conv.0.0\n",
+      "features.7.conv.1.0\n",
+      "features.7.conv.2\n",
+      "features.8.conv.0.0\n",
+      "features.8.conv.1.0\n",
+      "features.8.conv.2\n",
+      "features.9.conv.0.0\n",
+      "features.9.conv.1.0\n",
+      "features.9.conv.2\n",
+      "features.10.conv.0.0\n",
+      "features.10.conv.1.0\n",
+      "features.10.conv.2\n",
+      "features.11.conv.0.0\n",
+      "features.11.conv.1.0\n",
+      "features.11.conv.2\n",
+      "features.12.conv.0.0\n",
+      "features.12.conv.1.0\n",
+      "features.12.conv.2\n",
+      "features.13.conv.0.0\n",
+      "features.13.conv.1.0\n",
+      "features.13.conv.2\n",
+      "features.14.conv.0.0\n",
+      "features.14.conv.1.0\n",
+      "features.14.conv.2\n",
+      "features.15.conv.0.0\n",
+      "features.15.conv.1.0\n",
+      "features.15.conv.2\n",
+      "features.16.conv.0.0\n",
+      "features.16.conv.1.0\n",
+      "features.16.conv.2\n",
+      "features.17.conv.0.0\n",
+      "features.17.conv.1.0\n",
+      "features.17.conv.2\n",
+      "features.18.0\n",
+      "classifier.1\n"
+     ]
+    }
+   ],
+   "source": [
+    "# print parameters\n",
+    "for (name, layer) in get_prunable_layers(model):\n",
+    "    print(f\"{name}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7249f54d",
+   "metadata": {},
+   "source": [
+    "We will apply pruning to each of the convs and exclude the classifier layer (which is the final projection head). Fortunately, SparseML allows us to pass regexes to identify layers in the network, so we can use the following list to identify the relevant layers for pruning:\n",
+    "    \n",
+    "    - 'features.0.0.weight'\n",
+    "    - 'features.18.0.weight'\n",
+    "    - 're:features.*.conv.*.weight'\n",
+    "    - 're:features.*.conv.*.*.weight'\n",
+    "\n",
+    "Here is what the recipe looks like:\n",
+    "\n",
+    "```yaml\n",
+    "# Epoch and Learning-Rate variables\n",
+    "num_epochs: 13.0\n",
+    "pruning_epochs: 10.0\n",
+    "init_lr: 0.0005\n",
+    "inter_func: cubic\n",
+    "mask_type: unstructured\n",
+    "\n",
+    "training_modifiers:\n",
+    "  - !EpochRangeModifier\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: 0.0\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "# Pruning\n",
+    "pruning_modifiers:\n",
+    "  - !GlobalMagnitudePruningModifier\n",
+    "    init_sparsity: 0.05\n",
+    "    final_sparsity: 0.90\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(pruning_epochs)\n",
+    "    update_frequency: 1.0\n",
+    "    params: \n",
+    "        - 'features.0.0.weight'\n",
+    "        - 'features.18.0.weight'\n",
+    "        - 're:features.*.conv.*.weight'\n",
+    "        - 're:features.*.conv.*.*.weight'\n",
+    "    leave_enabled: True\n",
+    "    inter_func: eval(inter_func)\n",
+    "    mask_type: eval(mask_type)\n",
+    "```\n",
+    "\n",
+    "This recipe specifies that we will run the GMP algorithm for the first 10 epochs. We start at an init_sparsity level of 5% and gradually increase sparsity to a final_sparsity level of 90% following a cubic curve. The pruning is applied in an unstructured manner, meaning that any weight can be pruned.\n",
+    "\n",
+    "Over the final 3 epochs, we will fine-tune the 90% pruned model further. Since we set leave_enabled=True the sparsity level will be maintained as the fine-tuning occurs."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ca5c28b2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!cat ./recipes/mobilenetv2-beans-pruning-recipe.yaml"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "id": "b55a7c74",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "pruning_recipe_path = \"./recipes/mobilenetv2-beans-pruning-recipe.yaml\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "id": "68646bc7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# create ScheduledModifierManager and Optimizer wrapper\n",
+    "manager = ScheduledModifierManager.from_yaml(pruning_recipe_path)\n",
+    "logger = TensorBoardLogger(log_path=\"./tensorboard_outputs\")\n",
+    "optimizer = manager.modify(model, optimizer, loggers=[logger], steps_per_epoch=len(train_loader))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "37d4b6a4",
+   "metadata": {},
+   "source": [
+    "Next, kick off the GMP training loop. \n",
+    "\n",
+    "As you can see, we use the wrapped `optimizer` and `model` in the same way as above. SparseML parsed the recipe and updated the `optimizer` with the logic of GMP algorithm from the recipe. This allows you to just the `optimizer` and `model` as usual, with all of the pruning-related logic specified by the declarative recipe interface.\n",
+    "\n",
+    "Our 90% pruned model reaches ~99% validation accuracy (vs ~99% for the dense model)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 33,
+   "id": "601c8c21",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running Training Epoch 1/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "1532a057991c4f92ae19103ed357b47d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 1/13\n",
+      "Training Loss: 0.030314430911940606\n",
+      "Top 1 Acc: 0.9893617021276596\n",
+      "\n",
+      "Running Validation Epoch 1/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f8229930b27d4f22a4b9071e2047f0cd",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 1/13\n",
+      "Val Loss: 0.10016061989590526\n",
+      "Top 1 Acc: 0.9699248120300752\n",
+      "\n",
+      "Running Training Epoch 2/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b044a0e5d8414ea68ee41547c4a27210",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 2/13\n",
+      "Training Loss: 0.03320332106721418\n",
+      "Top 1 Acc: 0.9903288201160542\n",
+      "\n",
+      "Running Validation Epoch 2/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e5dd99cf59ab499db75f0cebe474a798",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 2/13\n",
+      "Val Loss: 0.05205255227629095\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 3/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ff9d00403a5c4a4a951d7c4c7a280da0",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 3/13\n",
+      "Training Loss: 0.02863485404558367\n",
+      "Top 1 Acc: 0.9922630560928434\n",
+      "\n",
+      "Running Validation Epoch 3/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e239549148104b41a2c845856c105288",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 3/13\n",
+      "Val Loss: 0.04446154618635774\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 4/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0394c6e5c0d749daae91ac40c14d9577",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 4/13\n",
+      "Training Loss: 0.028403193393552847\n",
+      "Top 1 Acc: 0.9932301740812379\n",
+      "\n",
+      "Running Validation Epoch 4/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "7210b33d08c14e4191c0d843ff49bf2c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 4/13\n",
+      "Val Loss: 0.021256184950470925\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 5/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "afb5a222c0314969b10b4f8bf123c607",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 5/13\n",
+      "Training Loss: 0.10820412065720919\n",
+      "Top 1 Acc: 0.9671179883945842\n",
+      "\n",
+      "Running Validation Epoch 5/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ed07d03c2d4e4dfc8fb1d5b39d67a789",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 5/13\n",
+      "Val Loss: 0.07088811788707972\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 6/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e94599ac85bd46b0ac5ca2200745ce42",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 6/13\n",
+      "Training Loss: 0.21380004890714632\n",
+      "Top 1 Acc: 0.9216634429400387\n",
+      "\n",
+      "Running Validation Epoch 6/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "75a137efb0df4a78879678d686263c69",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 6/13\n",
+      "Val Loss: 0.07981351003982126\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 7/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "18bf542d5d09438d93969342de4fab9e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 7/13\n",
+      "Training Loss: 0.19059311598539352\n",
+      "Top 1 Acc: 0.9332688588007737\n",
+      "\n",
+      "Running Validation Epoch 7/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "177ed7c1ade646f092ba59bafb462b5e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 7/13\n",
+      "Val Loss: 0.07885787561535836\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 8/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "74e04f218de44f8ea10e4b261f9473d0",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 8/13\n",
+      "Training Loss: 0.14277072397596907\n",
+      "Top 1 Acc: 0.965183752417795\n",
+      "\n",
+      "Running Validation Epoch 8/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "be70a6fffa6f419f82f87212155b98fe",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 8/13\n",
+      "Val Loss: 0.08072680607438087\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 9/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "246b3bf84568417f97c4bae88f83b4fb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 9/13\n",
+      "Training Loss: 0.11926381854396878\n",
+      "Top 1 Acc: 0.9680851063829787\n",
+      "\n",
+      "Running Validation Epoch 9/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b41b1ccbde8c4857897e7b16758260dc",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 9/13\n",
+      "Val Loss: 0.08937856331467628\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 10/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "eacdaf91b1e9436890b621b5095bfe05",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 10/13\n",
+      "Training Loss: 0.08645325391130014\n",
+      "Top 1 Acc: 0.9854932301740812\n",
+      "\n",
+      "Running Validation Epoch 10/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0d39663296c0495f9a9578847a5678f3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 10/13\n",
+      "Val Loss: 0.06663795448839664\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 11/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e738fd26e239454387c7b528b6b0e33c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 11/13\n",
+      "Training Loss: 0.07962510655775215\n",
+      "Top 1 Acc: 0.9816247582205029\n",
+      "\n",
+      "Running Validation Epoch 11/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "fcfbbe2836864d618f6dea028dd7c37f",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 11/13\n",
+      "Val Loss: 0.0655035775154829\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 12/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0ba18ba8e68a46169c9faef52291598a",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 12/13\n",
+      "Training Loss: 0.08157054532432195\n",
+      "Top 1 Acc: 0.9825918762088974\n",
+      "\n",
+      "Running Validation Epoch 12/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "cc213871f6bf4851817a59c8a282b98d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 12/13\n",
+      "Val Loss: 0.061416388303041455\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 13/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ac56ff40a6e348ae88f943de31efce6d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/33 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 13/13\n",
+      "Training Loss: 0.06805486869857166\n",
+      "Top 1 Acc: 0.9874274661508704\n",
+      "\n",
+      "Running Validation Epoch 13/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9cf25c82c70c449ebbdd8b6bf4493075",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/5 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 13/13\n",
+      "Val Loss: 0.06589164286851883\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "# run GMP algorithm\n",
+    "epoch = 0\n",
+    "for epoch in range(manager.max_epochs):\n",
+    "    # run training loop\n",
+    "    epoch_name = f\"{epoch + 1}/{manager.max_epochs}\"\n",
+    "    print(f\"Running Training Epoch {epoch_name}\")\n",
+    "    train_loss, train_acc = run_model_one_epoch(model, train_loader, criterion, device, train=True, optimizer=optimizer)\n",
+    "    print(f\"Training Epoch: {epoch_name}\\nTraining Loss: {train_loss}\\nTop 1 Acc: {train_acc}\\n\")\n",
+    "\n",
+    "    # run validation loop\n",
+    "    print(f\"Running Validation Epoch {epoch_name}\")\n",
+    "    val_loss, val_acc = run_model_one_epoch(model, val_loader, criterion, device)\n",
+    "    print(f\"Validation Epoch: {epoch_name}\\nVal Loss: {val_loss}\\nTop 1 Acc: {val_acc}\\n\")\n",
+    "    \n",
+    "    logger.log_scalar(\"Metrics/Loss (Train)\", train_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Train)\", train_acc, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Loss (Validation)\", val_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Validation)\", val_acc, epoch)\n",
+    "    \n",
+    "manager.finalize(model)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "64aabe37",
+   "metadata": {},
+   "source": [
+    "Here is a sample of the TensorBoard output, showing the validation accuracy, a particular layer's sparsity level, and the learning rate over time.\n",
+    "\n",
+    "![tensorboard output](./images/mobilenetv2-beans-tensorboard-output.png)\n",
+    "\n",
+    "We can print layer-by-layer sparsity as well."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 35,
+   "id": "2bf48ed2",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "features.0.0.weight: 0.4861\n",
+      "features.1.conv.0.0.weight: 0.4688\n",
+      "features.1.conv.1.weight: 0.5156\n",
+      "features.2.conv.0.0.weight: 0.5547\n",
+      "features.2.conv.1.0.weight: 0.2905\n",
+      "features.2.conv.2.weight: 0.4865\n",
+      "features.3.conv.0.0.weight: 0.6102\n",
+      "features.3.conv.1.0.weight: 0.4961\n",
+      "features.3.conv.2.weight: 0.6111\n",
+      "features.4.conv.0.0.weight: 0.5229\n",
+      "features.4.conv.1.0.weight: 0.2731\n",
+      "features.4.conv.2.weight: 0.5187\n",
+      "features.5.conv.0.0.weight: 0.6901\n",
+      "features.5.conv.1.0.weight: 0.5880\n",
+      "features.5.conv.2.weight: 0.7126\n",
+      "features.6.conv.0.0.weight: 0.6934\n",
+      "features.6.conv.1.0.weight: 0.6111\n",
+      "features.6.conv.2.weight: 0.7402\n",
+      "features.7.conv.0.0.weight: 0.5495\n",
+      "features.7.conv.1.0.weight: 0.3634\n",
+      "features.7.conv.2.weight: 0.6127\n",
+      "features.8.conv.0.0.weight: 0.8370\n",
+      "features.8.conv.1.0.weight: 0.7405\n",
+      "features.8.conv.2.weight: 0.8673\n",
+      "features.9.conv.0.0.weight: 0.8272\n",
+      "features.9.conv.1.0.weight: 0.7384\n",
+      "features.9.conv.2.weight: 0.8634\n",
+      "features.10.conv.0.0.weight: 0.8271\n",
+      "features.10.conv.1.0.weight: 0.7870\n",
+      "features.10.conv.2.weight: 0.8725\n",
+      "features.11.conv.0.0.weight: 0.7251\n",
+      "features.11.conv.1.0.weight: 0.7980\n",
+      "features.11.conv.2.weight: 0.7725\n",
+      "features.12.conv.0.0.weight: 0.8820\n",
+      "features.12.conv.1.0.weight: 0.7932\n",
+      "features.12.conv.2.weight: 0.9047\n",
+      "features.13.conv.0.0.weight: 0.8708\n",
+      "features.13.conv.1.0.weight: 0.8169\n",
+      "features.13.conv.2.weight: 0.9118\n",
+      "features.14.conv.0.0.weight: 0.7866\n",
+      "features.14.conv.1.0.weight: 0.6883\n",
+      "features.14.conv.2.weight: 0.8266\n",
+      "features.15.conv.0.0.weight: 0.9452\n",
+      "features.15.conv.1.0.weight: 0.7866\n",
+      "features.15.conv.2.weight: 0.9565\n",
+      "features.16.conv.0.0.weight: 0.9238\n",
+      "features.16.conv.1.0.weight: 0.7926\n",
+      "features.16.conv.2.weight: 0.9577\n",
+      "features.17.conv.0.0.weight: 0.8770\n",
+      "features.17.conv.1.0.weight: 0.9112\n",
+      "features.17.conv.2.weight: 0.9516\n",
+      "features.18.0.weight: 0.9784\n",
+      "classifier.1.weight: 0.0000\n"
+     ]
+    }
+   ],
+   "source": [
+    "for (name, layer) in get_prunable_layers(model):\n",
+    "    print(f\"{name}.weight: {tensor_sparsity(layer.weight).item():.4f}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "00805b02",
+   "metadata": {},
+   "source": [
+    "Finally, export your model to ONNX."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 37,
+   "id": "48fb7c0e",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "save_dir = \"experiment-0\"\n",
+    "exporter = ModuleExporter(model, output_dir=save_dir)\n",
+    "exporter.export_pytorch(name=\"mobilenet_v2-sparse-beans.pth\")\n",
+    "exporter.export_onnx(torch.randn(1, 3, 224, 224), name=\"sparse-model.onnx\", convert_qat=True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7960612c",
+   "metadata": {},
+   "source": [
+    "## Wrap Up\n",
+    "\n",
+    "The resulting model is is 90% sparse and achieves validation accuracy of ~99% (vs the unoptimized dense model at ~99%) without much hyperparameter search.\n",
+    "\n",
+    "Key hyperparameter experiments you may want to run include:\n",
+    "- Learning rate\n",
+    "- Learning rate schedule\n",
+    "- Sparsity level\n",
+    "- Number of pruning epochs\n",
+    "\n",
+    "DeepSparse supports speedup from pruning and quantization. To reach maximum performance, check out our examples of quantizing a model!"
+   ]
+  }
+ ],
+ "metadata": {
+  "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": "3.8.10"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-resnet50-beans.ipynb b/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-resnet50-beans.ipynb
new file mode 100644
index 00000000000..f3726bb5a39
--- /dev/null
+++ b/integrations/torchvision/tutorials/docs-torchvision-sparsify-from-scratch-resnet50-beans.ipynb
@@ -0,0 +1,1944 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "00e081b7",
+   "metadata": {},
+   "source": [
+    "# Sparsifying ResNet-50 from Scratch (Beans)\n",
+    "\n",
+    "In this example, we will demonstrate how to sparsify an image classification model from scratch using SparseML's PyTorch integration. We train and prune [ResNet-50](https://pytorch.org/vision/main/models/generated/torchvision.models.resnet50.html) on the downstream [Beans dataset](https://huggingface.co/datasets/beans) using the Global Magnitude Pruning algorithm. \n",
+    "\n",
+    "## Agenda\n",
+    "\n",
+    "There are a few steps:\n",
+    "\n",
+    " 1. Setup the dataset\n",
+    " 2. Setup the PyTorch training loop\n",
+    " 3. Train a dense version of ResNet-50\n",
+    " 4. Run the GMP pruning algorithm on the dense model\n",
+    " \n",
+    "## Installation\n",
+    "\n",
+    "Install SparseML with `pip`:\n",
+    "\n",
+    "```\n",
+    "pip install sparseml[torchvision]\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "1ad80edf",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "import sparseml\n",
+    "import torchvision\n",
+    "from sparseml.pytorch.optim import ScheduledModifierManager\n",
+    "from sparseml.pytorch.utils import TensorBoardLogger, ModuleExporter, get_prunable_layers, tensor_sparsity\n",
+    "from torch.utils.data import DataLoader\n",
+    "from torch.nn import CrossEntropyLoss\n",
+    "from torch.optim import Adam\n",
+    "from torchvision import transforms\n",
+    "from tqdm.auto import tqdm\n",
+    "import math\n",
+    "import datasets"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c421a838",
+   "metadata": {},
+   "source": [
+    "## Step 1: Setup Dataset\n",
+    "\n",
+    "Beans leaf dataset is a set of images of diseased and healthy leaves. Based on a leaf image, the goal of this task is to predict the disease type (Angular Leaf Spot and Bean Rust), if any.\n",
+    "\n",
+    "We will use the Hugging Face `datasets` library to download the data and the torchvision `ImageFolder` in the training loop.\n",
+    "\n",
+    "[Checkout the dataset card](https://huggingface.co/datasets/beans)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3f49303e",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "beans_dataset = datasets.load_dataset(\"beans\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "4115c9d8",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/eeb026374cf5ecfd5f40131a3159be9b9055ac21a3da11690e7eb4d117c99eee/train/bean_rust/bean_rust_train.84.jpg\n",
+      "/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/f287261265d2f9a3e8f87a5526a54d1847b17f7c3ec5714e5719432f2b3e4a73/validation/bean_rust/bean_rust_val.36.jpg\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(beans_dataset[\"train\"][0][\"image_file_path\"])\n",
+    "print(beans_dataset[\"validation\"][0][\"image_file_path\"])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "00aaf463",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_path = \"/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/eeb026374cf5ecfd5f40131a3159be9b9055ac21a3da11690e7eb4d117c99eee/train\"\n",
+    "val_path = \"/home/ubuntu/.cache/huggingface/datasets/downloads/extracted/f287261265d2f9a3e8f87a5526a54d1847b17f7c3ec5714e5719432f2b3e4a73/validation\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "7d180bbc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "NUM_LABELS = 3\n",
+    "BATCH_SIZE = 64\n",
+    "\n",
+    "# imagenet transforms\n",
+    "imagenet_transform = transforms.Compose([\n",
+    "   transforms.Resize(size=256, interpolation=transforms.InterpolationMode.BILINEAR, max_size=None, antialias=None),\n",
+    "   transforms.CenterCrop(size=(224, 224)),\n",
+    "   transforms.ToTensor(),\n",
+    "   transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])\n",
+    "])\n",
+    "\n",
+    "# datasets\n",
+    "train_dataset = torchvision.datasets.ImageFolder(\n",
+    "    root=train_path,\n",
+    "    transform=imagenet_transform\n",
+    ")\n",
+    "\n",
+    "val_dataset = torchvision.datasets.ImageFolder(\n",
+    "    root=val_path,\n",
+    "    transform=imagenet_transform\n",
+    ")\n",
+    "\n",
+    "\n",
+    "# dataloaders\n",
+    "train_loader = DataLoader(train_dataset, BATCH_SIZE, shuffle=True, pin_memory=True, num_workers=16)\n",
+    "val_loader = DataLoader(val_dataset, BATCH_SIZE, shuffle=False, pin_memory=True, num_workers=16)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "132fc514",
+   "metadata": {},
+   "source": [
+    "## Step 2: Setup PyTorch Training Loop\n",
+    "\n",
+    "We will use this training loop below. This is standard PyTorch functionality."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "5f1b878a",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "cuda\n"
+     ]
+    }
+   ],
+   "source": [
+    "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
+    "print(device)\n",
+    "\n",
+    "def run_model_one_epoch(model, data_loader, criterion, device, train=False, optimizer=None):\n",
+    "    if train:\n",
+    "        model.train()\n",
+    "    else:\n",
+    "        model.eval()\n",
+    "\n",
+    "    running_loss = 0.0\n",
+    "    total_correct = 0\n",
+    "    total_predictions = 0\n",
+    "\n",
+    "    # loop through batches\n",
+    "    for step, (inputs, labels) in tqdm(enumerate(data_loader), total=len(data_loader)):\n",
+    "        inputs = inputs.to(device)\n",
+    "        labels = labels.to(device)\n",
+    "\n",
+    "        if train:\n",
+    "            optimizer.zero_grad()\n",
+    "\n",
+    "        # compute loss, run backpropogation\n",
+    "        outputs = model(inputs)  # model returns logits\n",
+    "        loss = criterion(outputs, labels)\n",
+    "        if train:\n",
+    "            loss.backward()\n",
+    "            optimizer.step()\n",
+    "\n",
+    "        running_loss += loss.item()\n",
+    "\n",
+    "        # run evaluation\n",
+    "        predictions = outputs.argmax(dim=1)\n",
+    "        total_correct += torch.sum(predictions == labels).item()\n",
+    "        total_predictions += inputs.size(0)\n",
+    "\n",
+    "    # return loss and evaluation metric\n",
+    "    loss = running_loss / (step + 1.0)\n",
+    "    accuracy = total_correct / total_predictions\n",
+    "    return loss, accuracy"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "21e250c6",
+   "metadata": {},
+   "source": [
+    "## **Step 3: Train ResNet-50 on Beans**\n",
+    "\n",
+    "First, we will train a dense version of ResNet-50 on the Beans dataset."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "4d554578",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# download pre-trained model, setup classification head\n",
+    "model = torchvision.models.resnet50(weights=torchvision.models.ResNet50_Weights.DEFAULT)\n",
+    "model.fc = torch.nn.Linear(model.fc.in_features, NUM_LABELS)\n",
+    "model.to(device)\n",
+    "\n",
+    "# setup loss function and optimizer\n",
+    "criterion = CrossEntropyLoss()\n",
+    "optimizer = Adam(model.parameters(), lr=8e-3) # lr will be override by sparseml"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "39a82a98",
+   "metadata": {},
+   "source": [
+    "Next, we will use SparseML's recipes to set the hyperparameters of training loop. In this case, we will use the following recipe:\n",
+    "\n",
+    "```yaml\n",
+    "# Epoch and Learning-Rate variables\n",
+    "num_epochs: 10.0\n",
+    "init_lr: 0.0005\n",
+    "\n",
+    "training_modifiers:\n",
+    "  - !EpochRangeModifier\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: 0.0\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "```\n",
+    "\n",
+    "As you can see, the recipe includes an `!EpochRangeModifier` and a `!LearningRateFunctionModifier`. These modifiers simply set the number of epochs to train for and the learning rate schedule. As a result, the final model will be dense."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d4553beb",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!cat ./recipes/resnet50-beans-dense-recipe.yaml"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "1b19a8d8",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dense_recipe_path = \"./recipes/resnet50-beans-dense-recipe.yaml\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "9e00a7a9",
+   "metadata": {},
+   "source": [
+    "Next, we use SparseML's `ScheduledModifierManager` to parse and apply the recipe. The `manager.modify` function modifies and wraps the `model` and `optimizer` with the instructions from the recipe. You can use the `model` and `optimizer` just like standard PyTorch objects."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "1749e00d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# create ScheduledModifierManager and Optimizer wrapper\n",
+    "manager = ScheduledModifierManager.from_yaml(dense_recipe_path)\n",
+    "optimizer = manager.modify(model, optimizer, steps_per_epoch=len(train_loader))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "f3b56352",
+   "metadata": {},
+   "source": [
+    "Kick off the transfer learning loop. Our run reached ~99% validation accuracy after 10 epochs."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "d00d175b",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running Training Epoch 1/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "975dfb9440f748c296156a532b0e4cb2",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 1/10\n",
+      "Training Loss: 0.4227903384496184\n",
+      "Top 1 Acc: 0.8355899419729207\n",
+      "\n",
+      "Running Validation Epoch 1/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "1ac44c8fd5ab41e4945a053976c1ce4f",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 1/10\n",
+      "Val Loss: 0.166859846872588\n",
+      "Top 1 Acc: 0.9398496240601504\n",
+      "\n",
+      "Running Training Epoch 2/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "5ac4ab50f4b74673a929620a215c5265",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 2/10\n",
+      "Training Loss: 0.06358965676661361\n",
+      "Top 1 Acc: 0.9816247582205029\n",
+      "\n",
+      "Running Validation Epoch 2/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "fac7b5df3ae64f94bb6ed1b19a795c19",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 2/10\n",
+      "Val Loss: 0.6292845177813433\n",
+      "Top 1 Acc: 0.9548872180451128\n",
+      "\n",
+      "Running Training Epoch 3/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "673576da2a604abdacb88963abddb880",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 3/10\n",
+      "Training Loss: 0.032505514428896064\n",
+      "Top 1 Acc: 0.9893617021276596\n",
+      "\n",
+      "Running Validation Epoch 3/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "8d988e7a3c164f0a9c8ef194b7100d51",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 3/10\n",
+      "Val Loss: 0.047614920418709517\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 4/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "475850d6ad78458184acaa2a21f5b9db",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 4/10\n",
+      "Training Loss: 0.03107739576850744\n",
+      "Top 1 Acc: 0.9903288201160542\n",
+      "\n",
+      "Running Validation Epoch 4/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ec810e59cf784d58956449df7a77779d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 4/10\n",
+      "Val Loss: 0.038283314109624676\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 5/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "60680c7c2e9b4ed281ed4b444ce475a8",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 5/10\n",
+      "Training Loss: 0.11940328791217111\n",
+      "Top 1 Acc: 0.9941972920696325\n",
+      "\n",
+      "Running Validation Epoch 5/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b4444fa067714efda9fc5f6eb7354599",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 5/10\n",
+      "Val Loss: 0.04354451159209324\n",
+      "Top 1 Acc: 0.9774436090225563\n",
+      "\n",
+      "Running Training Epoch 6/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "c49c726e3ce141ea8a4236cc1d1e3eb2",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 6/10\n",
+      "Training Loss: 0.05448708479183123\n",
+      "Top 1 Acc: 0.9893617021276596\n",
+      "\n",
+      "Running Validation Epoch 6/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f84aaa08b92b4486ada7d57d2e96338e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 6/10\n",
+      "Val Loss: 0.05759213243921598\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 7/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "49a0611edd8740aea4ebe3861df90e64",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 7/10\n",
+      "Training Loss: 0.01337281458408517\n",
+      "Top 1 Acc: 0.9990328820116054\n",
+      "\n",
+      "Running Validation Epoch 7/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "7b13f9fccf7e4bac95d533bff9d40c21",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 7/10\n",
+      "Val Loss: 0.026064214219028752\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 8/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "5516a804e8714ed2a0a326593304fc31",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 8/10\n",
+      "Training Loss: 0.0028814816066776126\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 8/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "36497f568ff9499d823b742bfb627d07",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 8/10\n",
+      "Val Loss: 0.026710195777316887\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 9/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "faaf8be44fc44587b5cb96fa338c8dba",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 9/10\n",
+      "Training Loss: 0.0022678900106043063\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 9/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "6e4f22651d6e41cb9d1e98e8cc536314",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 9/10\n",
+      "Val Loss: 0.024938953380721312\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 10/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f0844eedab5d436e890e55eb5401fe96",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 10/10\n",
+      "Training Loss: 0.0028374234782805774\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 10/10\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d244e2fc3059410a8f6a01cda2e77417",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 10/10\n",
+      "Val Loss: 0.021825105145884056\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "# run transfer learning\n",
+    "epoch = 0\n",
+    "for epoch in range(manager.max_epochs):\n",
+    "    # run training loop\n",
+    "    epoch_name = f\"{epoch + 1}/{manager.max_epochs}\"\n",
+    "    print(f\"Running Training Epoch {epoch_name}\")\n",
+    "    train_loss, train_acc = run_model_one_epoch(model, train_loader, criterion, device, train=True, optimizer=optimizer)\n",
+    "    print(f\"Training Epoch: {epoch_name}\\nTraining Loss: {train_loss}\\nTop 1 Acc: {train_acc}\\n\")\n",
+    "\n",
+    "    # run validation loop\n",
+    "    print(f\"Running Validation Epoch {epoch_name}\")\n",
+    "    val_loss, val_acc = run_model_one_epoch(model, val_loader, criterion, device)\n",
+    "    print(f\"Validation Epoch: {epoch_name}\\nVal Loss: {val_loss}\\nTop 1 Acc: {val_acc}\\n\")\n",
+    "\n",
+    "manager.finalize(model)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "id": "11fa4cf2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from sparseml.pytorch.utils import ModuleExporter\n",
+    "\n",
+    "save_dir = \"dense_model\"\n",
+    "exporter = ModuleExporter(model, output_dir=save_dir)\n",
+    "exporter.export_pytorch(name=\"resnet-50-dense-beans.pth\")\n",
+    "exporter.export_onnx(torch.randn(1, 3, 224, 224), name=\"dense-model.onnx\", convert_qat=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "1617f476",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "torch.cuda.empty_cache()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "848572dd",
+   "metadata": {},
+   "source": [
+    "## Step 4: Prune The Model\n",
+    "\n",
+    "With a model trained on Beans, we are now ready to apply the GMP algorithm to prune the model. The GMP algorithm is an interative pruning algorithm. At the end of each epoch, we identify the lowest magnitude weights (those closest to 0) and remove them from the network starting from an initial level of sparsity until a final level of sparsity. The remaining nonzero weights are then fine-tuned onto training dataset."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "8447d074",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "checkpoint = torch.load(\"./dense_model/training/resnet-50-dense-beans.pth\")\n",
+    "model = torchvision.models.resnet50()\n",
+    "model.fc = torch.nn.Linear(model.fc.in_features, NUM_LABELS)\n",
+    "model.load_state_dict(checkpoint['state_dict'])\n",
+    "model.to(device)\n",
+    "\n",
+    "# setup loss function and optimizer, LR will be overriden by sparseml\n",
+    "criterion = CrossEntropyLoss()\n",
+    "optimizer = Adam(model.parameters(), lr=8e-3)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "edb62dc9",
+   "metadata": {
+    "scrolled": false
+   },
+   "source": [
+    "Next, we need to create a SparseML recipe which includes the GMP algorithm. The `!GlobalMagnitudePruningModifier` modifier instructs SparseML to apply the GMP algorithm at a global level (pruning the lowest magnitude weights across all layers).\n",
+    "\n",
+    "Firstly, we need to decide identify which parameters of the model to apply the GMP algorithm to. We can use the `get_prunable_layers` function to inspect:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "8950185f",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "conv1\n",
+      "layer1.0.conv1\n",
+      "layer1.0.conv2\n",
+      "layer1.0.conv3\n",
+      "layer1.0.downsample.0\n",
+      "layer1.1.conv1\n",
+      "layer1.1.conv2\n",
+      "layer1.1.conv3\n",
+      "layer1.2.conv1\n",
+      "layer1.2.conv2\n",
+      "layer1.2.conv3\n",
+      "layer2.0.conv1\n",
+      "layer2.0.conv2\n",
+      "layer2.0.conv3\n",
+      "layer2.0.downsample.0\n",
+      "layer2.1.conv1\n",
+      "layer2.1.conv2\n",
+      "layer2.1.conv3\n",
+      "layer2.2.conv1\n",
+      "layer2.2.conv2\n",
+      "layer2.2.conv3\n",
+      "layer2.3.conv1\n",
+      "layer2.3.conv2\n",
+      "layer2.3.conv3\n",
+      "layer3.0.conv1\n",
+      "layer3.0.conv2\n",
+      "layer3.0.conv3\n",
+      "layer3.0.downsample.0\n",
+      "layer3.1.conv1\n",
+      "layer3.1.conv2\n",
+      "layer3.1.conv3\n",
+      "layer3.2.conv1\n",
+      "layer3.2.conv2\n",
+      "layer3.2.conv3\n",
+      "layer3.3.conv1\n",
+      "layer3.3.conv2\n",
+      "layer3.3.conv3\n",
+      "layer3.4.conv1\n",
+      "layer3.4.conv2\n",
+      "layer3.4.conv3\n",
+      "layer3.5.conv1\n",
+      "layer3.5.conv2\n",
+      "layer3.5.conv3\n",
+      "layer4.0.conv1\n",
+      "layer4.0.conv2\n",
+      "layer4.0.conv3\n",
+      "layer4.0.downsample.0\n",
+      "layer4.1.conv1\n",
+      "layer4.1.conv2\n",
+      "layer4.1.conv3\n",
+      "layer4.2.conv1\n",
+      "layer4.2.conv2\n",
+      "layer4.2.conv3\n",
+      "fc\n"
+     ]
+    }
+   ],
+   "source": [
+    "# print parameters\n",
+    "for (name, layer) in get_prunable_layers(model):\n",
+    "    print(f\"{name}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "f5ef3bdf",
+   "metadata": {},
+   "source": [
+    "We will apply pruning to each of the convs and exclude the fc layer (which is the final projection head). Fortunately, SparseML allows us to pass regexes to identify layers in the network, so we can use the following list to identify the relevant layers for pruning:\n",
+    "    \n",
+    "    - 'conv1.weight'\n",
+    "    - 're:layer1.*.conv1.weight'\n",
+    "    - 're:layer1.*.conv2.weight'\n",
+    "    - 're:layer1.*.conv3.weight'\n",
+    "    - 're:layer1.0.downsample.0.weight'\n",
+    "    - 're:layer2.*.conv1.weight'\n",
+    "    - 're:layer2.*.conv2.weight'\n",
+    "    - 're:layer2.*.conv3.weight'\n",
+    "    - 're:layer2.0.downsample.0.weight'\n",
+    "    - 're:layer3.*.conv1.weight'\n",
+    "    - 're:layer3.*.conv2.weight'\n",
+    "    - 're:layer3.*.conv3.weight'\n",
+    "    - 're:layer3.0.downsample.0.weight'\n",
+    "    - 're:layer4.*.conv1.weight'\n",
+    "    - 're:layer4.*.conv2.weight'\n",
+    "    - 're:layer4.*.conv3.weight'\n",
+    "    - 're:layer4.0.downsample.0.weight'\n",
+    "\n",
+    "Here is what the recipe looks like:\n",
+    "\n",
+    "```yaml\n",
+    "# Epoch and Learning-Rate variables\n",
+    "num_epochs: 13.0\n",
+    "pruning_epochs: 10.0\n",
+    "init_lr: 0.00025\n",
+    "inter_func: cubic\n",
+    "mask_type: unstructured\n",
+    "\n",
+    "training_modifiers:\n",
+    "  - !EpochRangeModifier\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "  - !LearningRateFunctionModifier\n",
+    "    final_lr: 0.0\n",
+    "    init_lr: eval(init_lr)\n",
+    "    lr_func: cosine\n",
+    "    start_epoch: 0.0\n",
+    "    end_epoch: eval(num_epochs)\n",
+    "\n",
+    "# Pruning\n",
+    "pruning_modifiers:\n",
+    "   - !GlobalMagnitudePruningModifier\n",
+    "     init_sparsity: 0.05\n",
+    "     final_sparsity: 0.90\n",
+    "     start_epoch: 0.0\n",
+    "     end_epoch: eval(pruning_epochs)\n",
+    "     update_frequency: 1.0\n",
+    "     params: \n",
+    "        - 'conv1.weight'\n",
+    "        - 're:layer1.*.conv1.weight'\n",
+    "        - 're:layer1.*.conv2.weight'\n",
+    "        - 're:layer1.*.conv3.weight'\n",
+    "        - 're:layer1.0.downsample.0.weight'\n",
+    "        - 're:layer2.*.conv1.weight'\n",
+    "        - 're:layer2.*.conv2.weight'\n",
+    "        - 're:layer2.*.conv3.weight'\n",
+    "        - 're:layer2.0.downsample.0.weight'\n",
+    "        - 're:layer3.*.conv1.weight'\n",
+    "        - 're:layer3.*.conv2.weight'\n",
+    "        - 're:layer3.*.conv3.weight'\n",
+    "        - 're:layer3.0.downsample.0.weight'\n",
+    "        - 're:layer4.*.conv1.weight'\n",
+    "        - 're:layer4.*.conv2.weight'\n",
+    "        - 're:layer4.*.conv3.weight'\n",
+    "        - 're:layer4.0.downsample.0.weight'\n",
+    "     leave_enabled: True\n",
+    "     inter_func: eval(inter_func)\n",
+    "     mask_type: eval(mask_type)\n",
+    "```\n",
+    "\n",
+    "This recipe specifies that we will run the GMP algorithm for the first 10 epochs. We start at an init_sparsity level of 5% and gradually increase sparsity to a final_sparsity level of 90% following a cubic curve. The pruning is applied in an unstructured manner, meaning that any weight can be pruned.\n",
+    "\n",
+    "Over the final 3 epochs, we will fine-tune the 90% pruned model further. Since we set leave_enabled=True the sparsity level will be maintained as the fine-tuning occurs."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "0c492e33",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!cat recipes/resnet50-beans-pruning-recipe.yaml"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "9cc0e8ab",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "pruning_recipe_path = \"./recipes/resnet50-beans-pruning-recipe.yaml\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "68646bc7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# create ScheduledModifierManager and Optimizer wrapper\n",
+    "manager = ScheduledModifierManager.from_yaml(pruning_recipe_path)\n",
+    "logger = TensorBoardLogger(log_path=\"./tensorboard_outputs\")\n",
+    "optimizer = manager.modify(model, optimizer, loggers=[logger], steps_per_epoch=len(train_loader))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "1ab5b75b",
+   "metadata": {},
+   "source": [
+    "Next, kick off the GMP training loop. \n",
+    "\n",
+    "As you can see, we use the wrapped `optimizer` and `model` in the same way as above. SparseML parsed the recipe and updated the `optimizer` with the logic of GMP algorithm from the recipe. This allows you to just the `optimizer` and `model` as usual, with all of the pruning-related logic specified by the declarative recipe interface.\n",
+    "\n",
+    "Our 90% pruned model reaches ~99% validation accuracy (vs ~99% for the dense model)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "601c8c21",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running Training Epoch 1/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "a687385d2d3244c189ffac1e58c63f4b",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 1/13\n",
+      "Training Loss: 0.016460627817329678\n",
+      "Top 1 Acc: 0.995164410058027\n",
+      "\n",
+      "Running Validation Epoch 1/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f75f7df5e4dd444c990312a33b5e55dd",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 1/13\n",
+      "Val Loss: 0.025849940720945597\n",
+      "Top 1 Acc: 0.9849624060150376\n",
+      "\n",
+      "Running Training Epoch 2/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "4b9800318d914258896ddf34e4a34c54",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 2/13\n",
+      "Training Loss: 0.009291439395918347\n",
+      "Top 1 Acc: 0.9970986460348162\n",
+      "\n",
+      "Running Validation Epoch 2/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "895ac43d40ba4fd0a4f52caaa6e97dcb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 2/13\n",
+      "Val Loss: 0.02041603586015602\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 3/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "1ac0de20101d4e94888bc5c25dc36203",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 3/13\n",
+      "Training Loss: 0.012328761325026041\n",
+      "Top 1 Acc: 0.9941972920696325\n",
+      "\n",
+      "Running Validation Epoch 3/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "28fd64b3fe6343beb2d66230be57f937",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 3/13\n",
+      "Val Loss: 0.010778552386909723\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Training Epoch 4/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "1c3382a6ee094132b0dce1a8be81d09c",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 4/13\n",
+      "Training Loss: 0.006804226791513536\n",
+      "Top 1 Acc: 0.9980657640232108\n",
+      "\n",
+      "Running Validation Epoch 4/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "6787b4d7046f437a917554555849a56e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 4/13\n",
+      "Val Loss: 0.006039845136304696\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Training Epoch 5/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "fb378f0699fa4a09b849c195d4575e65",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 5/13\n",
+      "Training Loss: 0.012682446326567409\n",
+      "Top 1 Acc: 0.9980657640232108\n",
+      "\n",
+      "Running Validation Epoch 5/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0bf4e5a11b234dcaacb8b08a570b9e17",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 5/13\n",
+      "Val Loss: 0.00994978565722704\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Training Epoch 6/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "cc4c1ed11e5a4f4f80904b210ce2fce6",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 6/13\n",
+      "Training Loss: 0.0117055864265079\n",
+      "Top 1 Acc: 0.9990328820116054\n",
+      "\n",
+      "Running Validation Epoch 6/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "a8867c5c5759450bbe91369534e73e1d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 6/13\n",
+      "Val Loss: 0.01689507276751101\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 7/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "728e29c381be4ffe95fa4e2607aaead2",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 7/13\n",
+      "Training Loss: 0.012449477751301053\n",
+      "Top 1 Acc: 0.9990328820116054\n",
+      "\n",
+      "Running Validation Epoch 7/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f089807ef4704eac91a29faa131fc1d4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 7/13\n",
+      "Val Loss: 0.02214170355970661\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 8/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d09f08b9aa7e4a459f0b5ecc68415fa0",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 8/13\n",
+      "Training Loss: 0.01196294802460162\n",
+      "Top 1 Acc: 0.9990328820116054\n",
+      "\n",
+      "Running Validation Epoch 8/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "6b784f2a60e147d9b93808a1f64398c8",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 8/13\n",
+      "Val Loss: 0.02145360441257556\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Training Epoch 9/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9f81d0020ba0413185f316c800afb0ee",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 9/13\n",
+      "Training Loss: 0.01053273505257333\n",
+      "Top 1 Acc: 0.9990328820116054\n",
+      "\n",
+      "Running Validation Epoch 9/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "96bace92d85e4cef9f7d65846f6a8756",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 9/13\n",
+      "Val Loss: 0.02385075483471155\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Training Epoch 10/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "bd969b4487d34464944cd720f734fc9d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 10/13\n",
+      "Training Loss: 0.004325077581383726\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 10/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "c8549b27bfb54e208303eddf59acf0b1",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 10/13\n",
+      "Val Loss: 0.017839766883601744\n",
+      "Top 1 Acc: 0.9924812030075187\n",
+      "\n",
+      "Running Training Epoch 11/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "33b3406e7b4f44ffab8765971a5eba20",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 11/13\n",
+      "Training Loss: 0.003297735289807486\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 11/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "93e5a79272034e3da9fe9f7b2ce79d5a",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 11/13\n",
+      "Val Loss: 0.012569367730369171\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Training Epoch 12/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "5d1a07c0564a4be8a55c882e95c58de0",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 12/13\n",
+      "Training Loss: 0.004005543901311124\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 12/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "5003c9fb24924629ae11f10bda705979",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 12/13\n",
+      "Val Loss: 0.013252861564978957\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Training Epoch 13/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ce3f08d614364ae0bc2a09f80dee38a3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/17 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Training Epoch: 13/13\n",
+      "Training Loss: 0.005194885152912534\n",
+      "Top 1 Acc: 1.0\n",
+      "\n",
+      "Running Validation Epoch 13/13\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9362c25ec45a43f1992d5a501bfd3df8",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/3 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Validation Epoch: 13/13\n",
+      "Val Loss: 0.012128440197557211\n",
+      "Top 1 Acc: 1.0\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "# run GMP algorithm\n",
+    "epoch = 0\n",
+    "for epoch in range(manager.max_epochs):\n",
+    "    # run training loop\n",
+    "    epoch_name = f\"{epoch + 1}/{manager.max_epochs}\"\n",
+    "    print(f\"Running Training Epoch {epoch_name}\")\n",
+    "    train_loss, train_acc = run_model_one_epoch(model, train_loader, criterion, device, train=True, optimizer=optimizer)\n",
+    "    print(f\"Training Epoch: {epoch_name}\\nTraining Loss: {train_loss}\\nTop 1 Acc: {train_acc}\\n\")\n",
+    "\n",
+    "    # run validation loop\n",
+    "    print(f\"Running Validation Epoch {epoch_name}\")\n",
+    "    val_loss, val_acc = run_model_one_epoch(model, val_loader, criterion, device)\n",
+    "    print(f\"Validation Epoch: {epoch_name}\\nVal Loss: {val_loss}\\nTop 1 Acc: {val_acc}\\n\")\n",
+    "    \n",
+    "    logger.log_scalar(\"Metrics/Loss (Train)\", train_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Train)\", train_acc, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Loss (Validation)\", val_loss, epoch)\n",
+    "    logger.log_scalar(\"Metrics/Accuracy (Validation)\", val_acc, epoch)\n",
+    "    \n",
+    "manager.finalize(model)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7aa91abe",
+   "metadata": {},
+   "source": [
+    "Here is a sample of the TensorBoard output, showing the validation accuracy, a particular layer's sparsity level, and the learning rate over time.\n",
+    "\n",
+    "![tensorboard output](./images/resnet-beans-tensorboard-output.png)\n",
+    "\n",
+    "We can print layer-by-layer sparsity as well."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "id": "65fb7b12",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "conv1.weight: 0.4039\n",
+      "layer1.0.conv1.weight: 0.4758\n",
+      "layer1.0.conv2.weight: 0.6846\n",
+      "layer1.0.conv3.weight: 0.5506\n",
+      "layer1.0.downsample.0.weight: 0.5157\n",
+      "layer1.1.conv1.weight: 0.5690\n",
+      "layer1.1.conv2.weight: 0.6410\n",
+      "layer1.1.conv3.weight: 0.5604\n",
+      "layer1.2.conv1.weight: 0.5430\n",
+      "layer1.2.conv2.weight: 0.6075\n",
+      "layer1.2.conv3.weight: 0.6064\n",
+      "layer2.0.conv1.weight: 0.5498\n",
+      "layer2.0.conv2.weight: 0.7556\n",
+      "layer2.0.conv3.weight: 0.6881\n",
+      "layer2.0.downsample.0.weight: 0.8303\n",
+      "layer2.1.conv1.weight: 0.7918\n",
+      "layer2.1.conv2.weight: 0.8089\n",
+      "layer2.1.conv3.weight: 0.7450\n",
+      "layer2.2.conv1.weight: 0.7564\n",
+      "layer2.2.conv2.weight: 0.8100\n",
+      "layer2.2.conv3.weight: 0.7706\n",
+      "layer2.3.conv1.weight: 0.7239\n",
+      "layer2.3.conv2.weight: 0.7862\n",
+      "layer2.3.conv3.weight: 0.7415\n",
+      "layer3.0.conv1.weight: 0.6814\n",
+      "layer3.0.conv2.weight: 0.8697\n",
+      "layer3.0.conv3.weight: 0.7558\n",
+      "layer3.0.downsample.0.weight: 0.9209\n",
+      "layer3.1.conv1.weight: 0.8875\n",
+      "layer3.1.conv2.weight: 0.9177\n",
+      "layer3.1.conv3.weight: 0.8482\n",
+      "layer3.2.conv1.weight: 0.9054\n",
+      "layer3.2.conv2.weight: 0.9143\n",
+      "layer3.2.conv3.weight: 0.8672\n",
+      "layer3.3.conv1.weight: 0.8686\n",
+      "layer3.3.conv2.weight: 0.9209\n",
+      "layer3.3.conv3.weight: 0.8573\n",
+      "layer3.4.conv1.weight: 0.8710\n",
+      "layer3.4.conv2.weight: 0.9158\n",
+      "layer3.4.conv3.weight: 0.8599\n",
+      "layer3.5.conv1.weight: 0.8293\n",
+      "layer3.5.conv2.weight: 0.9191\n",
+      "layer3.5.conv3.weight: 0.8144\n",
+      "layer4.0.conv1.weight: 0.7612\n",
+      "layer4.0.conv2.weight: 0.9458\n",
+      "layer4.0.conv3.weight: 0.8048\n",
+      "layer4.0.downsample.0.weight: 0.9347\n",
+      "layer4.1.conv1.weight: 0.9039\n",
+      "layer4.1.conv2.weight: 0.9626\n",
+      "layer4.1.conv3.weight: 0.8643\n",
+      "layer4.2.conv1.weight: 0.9441\n",
+      "layer4.2.conv2.weight: 0.9775\n",
+      "layer4.2.conv3.weight: 0.8976\n",
+      "fc.weight: 0.0000\n"
+     ]
+    }
+   ],
+   "source": [
+    "for (name, layer) in get_prunable_layers(model):\n",
+    "    print(f\"{name}.weight: {tensor_sparsity(layer.weight).item():.4f}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "00805b02",
+   "metadata": {},
+   "source": [
+    "Finally, export your model to ONNX."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "id": "417ac105",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "save_dir = \"experiment-0\"\n",
+    "exporter = ModuleExporter(model, output_dir=save_dir)\n",
+    "exporter.export_pytorch(name=\"resnet-50-sparse-beans.pth\")\n",
+    "exporter.export_onnx(torch.randn(1, 3, 224, 224), name=\"sparse-model.onnx\", convert_qat=True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7960612c",
+   "metadata": {},
+   "source": [
+    "## Wrap Up\n",
+    "\n",
+    "The resulting model is is 90% sparse and achieves validation accuracy of ~100% (vs the unoptimized dense model at ~99%) without much hyperparameter search.\n",
+    "\n",
+    "Key hyperparameter experiments you may want to run include:\n",
+    "- Learning rate\n",
+    "- Learning rate schedule\n",
+    "- Sparsity level\n",
+    "- Number of pruning epochs\n",
+    "\n",
+    "DeepSparse supports speedup from pruning and quantization. To reach maximum performance, check out our examples of quantizing a model!"
+   ]
+  }
+ ],
+ "metadata": {
+  "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": "3.8.10"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/integrations/torchvision/tutorials/images/densenet-flowers-tensorboard-output.png b/integrations/torchvision/tutorials/images/densenet-flowers-tensorboard-output.png
new file mode 100644
index 00000000000..d73aaca887d
Binary files /dev/null and b/integrations/torchvision/tutorials/images/densenet-flowers-tensorboard-output.png differ
diff --git a/integrations/torchvision/tutorials/images/mobilenetv2-beans-tensorboard-output.png b/integrations/torchvision/tutorials/images/mobilenetv2-beans-tensorboard-output.png
new file mode 100644
index 00000000000..97ba818a942
Binary files /dev/null and b/integrations/torchvision/tutorials/images/mobilenetv2-beans-tensorboard-output.png differ
diff --git a/integrations/torchvision/tutorials/images/resnet-beans-tensorboard-output.png b/integrations/torchvision/tutorials/images/resnet-beans-tensorboard-output.png
new file mode 100644
index 00000000000..420638f0879
Binary files /dev/null and b/integrations/torchvision/tutorials/images/resnet-beans-tensorboard-output.png differ
diff --git a/integrations/torchvision/tutorials/recipes/densenet-flowers-dense-recipe.yaml b/integrations/torchvision/tutorials/recipes/densenet-flowers-dense-recipe.yaml
new file mode 100644
index 00000000000..e5afb1d4499
--- /dev/null
+++ b/integrations/torchvision/tutorials/recipes/densenet-flowers-dense-recipe.yaml
@@ -0,0 +1,16 @@
+
+# Epoch and Learning-Rate variables
+num_epochs: 10.0
+init_lr: 0.0005
+
+training_modifiers:
+  - !EpochRangeModifier
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+  - !LearningRateFunctionModifier
+    final_lr: 0.0
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
diff --git a/integrations/torchvision/tutorials/recipes/densenet-flowers-pruning-recipe.yaml b/integrations/torchvision/tutorials/recipes/densenet-flowers-pruning-recipe.yaml
new file mode 100644
index 00000000000..b23986e4d24
--- /dev/null
+++ b/integrations/torchvision/tutorials/recipes/densenet-flowers-pruning-recipe.yaml
@@ -0,0 +1,56 @@
+# Epoch and Learning-Rate variables
+num_epochs: 13.0
+pruning_epochs: 10.0
+init_lr: 0.0005
+final_lr: 0.0001
+inter_func: cubic
+mask_type: unstructured
+
+training_modifiers:
+  - !EpochRangeModifier
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+  - !LearningRateFunctionModifier
+    final_lr: eval(final_lr)
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: 0.0
+    end_epoch: eval(pruning_epochs)
+    
+  - !LearningRateFunctionModifier
+    final_lr: eval(final_lr)
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: eval(pruning_epochs)
+    end_epoch: eval(num_epochs)
+
+# Pruning
+pruning_modifiers:
+  - !GlobalMagnitudePruningModifier
+    init_sparsity: 0.05
+    final_sparsity: 0.9
+    start_epoch: 0.0
+    end_epoch: eval(pruning_epochs)
+    update_frequency: 0.5
+    params: 
+        - 'features.conv0.weight'
+        - 're:features.denseblock1.*.conv1.weight'
+        - 're:features.denseblock1.*.conv2.weight'
+        - 're:features.transition1.conv.weight'
+        - 're:features.denseblock2.*.conv1.weight'
+        - 're:features.denseblock2.*.conv2.weight'
+        - 're:features.transition2.conv.weight'
+        - 're:features.denseblock3.*.conv1.weight'
+        - 're:features.denseblock3.*.conv2.weight'
+        - 're:features.transition3.conv.weight'
+        - 're:features.denseblock4.*.conv1.weight'
+        - 're:features.denseblock4.*.conv2.weight'
+    leave_enabled: True
+    inter_func: eval(inter_func)
+    mask_type: eval(mask_type)
+
+finetuning_modifiers:
+  - !ConstantPruningModifier
+    start_epoch: eval(pruning_epochs)
+    params: __ALL_PRUNABLE__
\ No newline at end of file
diff --git a/integrations/torchvision/tutorials/recipes/mobilenetv2-beans-dense-recipe.yaml b/integrations/torchvision/tutorials/recipes/mobilenetv2-beans-dense-recipe.yaml
new file mode 100644
index 00000000000..cd66d90908f
--- /dev/null
+++ b/integrations/torchvision/tutorials/recipes/mobilenetv2-beans-dense-recipe.yaml
@@ -0,0 +1,16 @@
+
+# Epoch and Learning-Rate variables
+num_epochs: 10.0
+init_lr: 0.0005
+
+training_modifiers:
+  - !EpochRangeModifier
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+  - !LearningRateFunctionModifier
+    final_lr: 0.0
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
\ No newline at end of file
diff --git a/integrations/torchvision/tutorials/recipes/mobilenetv2-beans-pruning-recipe.yaml b/integrations/torchvision/tutorials/recipes/mobilenetv2-beans-pruning-recipe.yaml
new file mode 100644
index 00000000000..f995f3194af
--- /dev/null
+++ b/integrations/torchvision/tutorials/recipes/mobilenetv2-beans-pruning-recipe.yaml
@@ -0,0 +1,35 @@
+# Epoch and Learning-Rate variables
+num_epochs: 13.0
+pruning_epochs: 10.0
+init_lr: 0.0005
+inter_func: cubic
+mask_type: unstructured
+
+training_modifiers:
+  - !EpochRangeModifier
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+  - !LearningRateFunctionModifier
+    final_lr: 0.0
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+# Pruning
+pruning_modifiers:
+  - !GlobalMagnitudePruningModifier
+    init_sparsity: 0.05
+    final_sparsity: 0.90
+    start_epoch: 0.0
+    end_epoch: eval(pruning_epochs)
+    update_frequency: 1.0
+    params: 
+        - 'features.0.0.weight'
+        - 'features.18.0.weight'
+        - 're:features.*.conv.*.weight'
+        - 're:features.*.conv.*.*.weight'
+    leave_enabled: True
+    inter_func: eval(inter_func)
+    mask_type: eval(mask_type)
\ No newline at end of file
diff --git a/integrations/torchvision/tutorials/recipes/quantization-recipe.yaml b/integrations/torchvision/tutorials/recipes/quantization-recipe.yaml
new file mode 100644
index 00000000000..6df4f5ff921
--- /dev/null
+++ b/integrations/torchvision/tutorials/recipes/quantization-recipe.yaml
@@ -0,0 +1,29 @@
+# Epoch and Learning-Rate variables
+num_epochs: 5.0
+init_lr: 0.0004
+
+# quantization variables
+quantization_epochs: 6.0
+
+training_modifiers:
+  - !EpochRangeModifier
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+  - !LearningRateFunctionModifier
+    final_lr: 0.0
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+# Phase 1 Sparse Transfer Learning / Recovery
+sparse_transfer_learning_modifiers:
+  - !ConstantPruningModifier
+    start_epoch: 0.0
+    params: __ALL_PRUNABLE__
+
+# Phase 2 Apply quantization
+sparse_quantized_transfer_learning_modifiers:
+  - !QuantizationModifier
+    start_epoch: eval(num_epochs - quantization_epochs)
\ No newline at end of file
diff --git a/integrations/torchvision/tutorials/recipes/resnet50-beans-dense-recipe.yaml b/integrations/torchvision/tutorials/recipes/resnet50-beans-dense-recipe.yaml
new file mode 100644
index 00000000000..cd66d90908f
--- /dev/null
+++ b/integrations/torchvision/tutorials/recipes/resnet50-beans-dense-recipe.yaml
@@ -0,0 +1,16 @@
+
+# Epoch and Learning-Rate variables
+num_epochs: 10.0
+init_lr: 0.0005
+
+training_modifiers:
+  - !EpochRangeModifier
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+  - !LearningRateFunctionModifier
+    final_lr: 0.0
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
\ No newline at end of file
diff --git a/integrations/torchvision/tutorials/recipes/resnet50-beans-pruning-recipe.yaml b/integrations/torchvision/tutorials/recipes/resnet50-beans-pruning-recipe.yaml
new file mode 100644
index 00000000000..1bc94036805
--- /dev/null
+++ b/integrations/torchvision/tutorials/recipes/resnet50-beans-pruning-recipe.yaml
@@ -0,0 +1,48 @@
+# Epoch and Learning-Rate variables
+num_epochs: 13.0
+pruning_epochs: 10.0
+init_lr: 0.00025
+inter_func: cubic
+mask_type: unstructured
+
+training_modifiers:
+  - !EpochRangeModifier
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+  - !LearningRateFunctionModifier
+    final_lr: 0.0
+    init_lr: eval(init_lr)
+    lr_func: cosine
+    start_epoch: 0.0
+    end_epoch: eval(num_epochs)
+
+# Pruning
+pruning_modifiers:
+   - !GlobalMagnitudePruningModifier
+     init_sparsity: 0.05
+     final_sparsity: 0.90
+     start_epoch: 0.0
+     end_epoch: eval(pruning_epochs)
+     update_frequency: 1.0
+     params: 
+        - 'conv1.weight'
+        - 're:layer1.*.conv1.weight'
+        - 're:layer1.*.conv2.weight'
+        - 're:layer1.*.conv3.weight'
+        - 're:layer1.0.downsample.0.weight'
+        - 're:layer2.*.conv1.weight'
+        - 're:layer2.*.conv2.weight'
+        - 're:layer2.*.conv3.weight'
+        - 're:layer2.0.downsample.0.weight'
+        - 're:layer3.*.conv1.weight'
+        - 're:layer3.*.conv2.weight'
+        - 're:layer3.*.conv3.weight'
+        - 're:layer3.0.downsample.0.weight'
+        - 're:layer4.*.conv1.weight'
+        - 're:layer4.*.conv2.weight'
+        - 're:layer4.*.conv3.weight'
+        - 're:layer4.0.downsample.0.weight'
+     leave_enabled: True
+     inter_func: eval(inter_func)
+     mask_type: eval(mask_type)
\ No newline at end of file