nbs/GT-WW.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/pawel/anaconda3/envs/gt/lib/python3.7/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n",
      "PyTorch is available but CUDA is not. Defaulting to SciPy for SVD\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Numpy version: 1.21.6\n",
      "Rdkit version: 2022.09.5\n",
      "Torch version: 1.13.1\n",
      "TorchMetrics version: 0.11.4\n",
      "WeightWatcher version: 0.7.1.5\n"
     ]
    }
   ],
   "source": [
    "# Standard\n",
    "import logging\n",
    "\n",
    "# Third party\n",
    "import numpy as np\n",
    "import rdkit\n",
    "from rdkit import RDLogger\n",
    "import torch\n",
    "from torch import nn\n",
    "import matplotlib.pyplot as plt\n",
    "from torch_geometric.loader import DataLoader\n",
    "import torchmetrics\n",
    "from torchmetrics import MeanAbsoluteError\n",
    "import weightwatcher as ww\n",
    "\n",
    "# GT-PyG\n",
    "from gt_pyg.data.utils import (\n",
    "    get_tensor_data, \n",
    "    get_node_dim, \n",
    "    get_edge_dim, \n",
    "    get_train_valid_test_data\n",
    ")\n",
    "from gt_pyg.nn.model import GraphTransformerNet\n",
    "from gt_pyg.util import get_ww_statistic\n",
    "from gt_pyg.util import get_correction_factor, prepare_params\n",
    "\n",
    "# Turn off majority of RDKit warnings\n",
    "lg = RDLogger.logger()\n",
    "lg.setLevel(RDLogger.CRITICAL)\n",
    "\n",
    "\n",
    "# Set a random seed for a reproducibility purposes\n",
    "torch.manual_seed(192837465)\n",
    "\n",
    "# Setup the logger\n",
    "logger = logging.getLogger()\n",
    "logger.setLevel(logging.INFO)\n",
    "\n",
    "# Log the used versions of RDkit and torch\n",
    "print(f'Numpy version: {np.__version__}')\n",
    "print(f'Rdkit version: {rdkit.__version__}')\n",
    "print(f'Torch version: {torch.__version__}')\n",
    "print(f'TorchMetrics version: {torchmetrics.__version__}')\n",
    "print(f'WeightWatcher version: {ww.__version__}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get the ADME@TDC data\n",
    "\n",
    "**Note**: To use the code below, make sure that the chosen endpoint is a regression task."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Available endpoints:\n",
      "\n",
      "1. caco2_wang\n",
      "2. hia_hou\n",
      "3. pgp_broccatelli\n",
      "4. bioavailability_ma\n",
      "5. lipophilicity_astrazeneca\n",
      "6. solubility_aqsoldb\n",
      "7. bbb_martins\n",
      "8. ppbr_az\n",
      "9. vdss_lombardo\n",
      "10. cyp2d6_veith\n",
      "11. cyp3a4_veith\n",
      "12. cyp2c9_veith\n",
      "13. cyp2d6_substrate_carbonmangels\n",
      "14. cyp3a4_substrate_carbonmangels\n",
      "15. cyp2c9_substrate_carbonmangels\n",
      "16. half_life_obach\n",
      "17. clearance_microsome_az\n",
      "18. clearance_hepatocyte_az\n",
      "19. herg\n",
      "20. ames\n",
      "21. dili\n",
      "22. ld50_zhu\n"
     ]
    }
   ],
   "source": [
    "from tdc import utils\n",
    "names = utils.retrieve_benchmark_names('ADMET_Group')\n",
    "output = \"\\n\".join([f\"{index}. {name}\" for index, name in enumerate(names, start=1)])\n",
    "print(\"Available endpoints:\\n\\n\" + output)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Regression endpoints with MAE metric:\n",
    "1. caco2_wang (Best: 0.285 ± 0.005)\n",
    "2. lipophilicity_astrazeneca (Best: 0.535 ± 0.012)\n",
    "3. solubility_aqsoldb (Best: 0.776 ± 0.008)\n",
    "4. ppbr_az (Best: 9.185 ± 0.000)\n",
    "5. ld50_zhu (Best: 0.588 ± 0.005)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found local copy...\n",
      "Loading...\n",
      "Done!\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of training examples: 1130\n",
      "Number of validation examples: 161\n",
      "Number of test examples: 323\n"
     ]
    }
   ],
   "source": [
    "PE_DIM = 6\n",
    "(tr, va, te) = get_train_valid_test_data('ppbr_az', min_num_atoms=0)\n",
    "tr_dataset = get_tensor_data(tr.Drug.to_list(), tr.Y.to_list(), pe_dim=PE_DIM)\n",
    "va_dataset = get_tensor_data(va.Drug.to_list(), va.Y.to_list(), pe_dim=PE_DIM)\n",
    "te_dataset = get_tensor_data(te.Drug.to_list(), te.Y.to_list(), pe_dim=PE_DIM)\n",
    "\n",
    "NODE_DIM = get_node_dim()\n",
    "EDGE_DIM = get_edge_dim()\n",
    "\n",
    "print(f'Number of training examples: {len(tr_dataset)}')\n",
    "print(f'Number of validation examples: {len(va_dataset)}')\n",
    "print(f'Number of test examples: {len(te_dataset)}')\n",
    "\n",
    "train_loader = DataLoader(tr_dataset, batch_size=64)\n",
    "val_loader = DataLoader(va_dataset, batch_size=512)\n",
    "test_loader = DataLoader(te_dataset, batch_size=512)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train and eval the GT model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Auxiliary functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(epoch, loss_func, warmup_epochs=5, loader=None):\n",
    "    model.train()\n",
    "    train_mae = MeanAbsoluteError()\n",
    "    \n",
    "    if epoch > warmup_epochs:\n",
    "        corrections = get_correction_factor(model, optimizer)\n",
    "        for idx, val in enumerate(corrections):\n",
    "            optimizer.param_groups[idx]['lr'] = val * optimizer.param_groups[idx]['base_lr']\n",
    "\n",
    "\n",
    "    if loader is None:\n",
    "        loader = train_loader\n",
    "        \n",
    "    for data in loader:\n",
    "        data = data.to(device)\n",
    "        optimizer.zero_grad()\n",
    "        # randomly flip sign of eigenvectors\n",
    "        batch_pe = data.pe * (2 * torch.randint(low=0, high=2, size=(1, PE_DIM)).float() - 1.0)\n",
    "        (out,_) = model(data.x, data.edge_index, data.edge_attr, batch_pe, data.batch, zero_var=False)\n",
    "        loss = loss_func(out.squeeze(), data.y)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        train_mae.update(out.squeeze(), data.y)\n",
    "\n",
    "    return train_mae.compute()\n",
    "\n",
    "\n",
    "@torch.no_grad()\n",
    "def test(loader):\n",
    "    model.eval()\n",
    "    test_mae = MeanAbsoluteError()\n",
    "\n",
    "    total_error = 0\n",
    "    for data in loader:\n",
    "        data = data.to(device)\n",
    "        # randomly flip sign of eigenvectors\n",
    "        batch_pe = data.pe * (2 * torch.randint(low=0, high=2, size=(1, PE_DIM)).float() - 1.0)\n",
    "        (out,_) = model(data.x, data.edge_index, data.edge_attr, batch_pe, data.batch)\n",
    "        \n",
    "        test_mae.update(out.squeeze(), data.y)\n",
    "        \n",
    "    return test_mae.compute()\n",
    "\n",
    "train_loss = nn.L1Loss(reduction='mean')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Slighlty optimized Graph Transformer architecture\n",
    "\n",
    "1. `gelu` activation is used instead of `relu`\n",
    "2. Multiaggregator used for global pooling\n",
    "3. Multiaggregator used for message passing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GraphTransformerNet(\n",
      "  (node_emb): Linear(in_features=76, out_features=128, bias=False)\n",
      "  (edge_emb): Linear(in_features=10, out_features=128, bias=False)\n",
      "  (pe_emb): Linear(in_features=6, out_features=128, bias=False)\n",
      "  (gt_layers): ModuleList(\n",
      "    (0): GTConv(128, 128, heads=8, aggrs: sum,mean)\n",
      "    (1): GTConv(128, 128, heads=8, aggrs: sum,mean)\n",
      "    (2): GTConv(128, 128, heads=8, aggrs: sum,mean)\n",
      "    (3): GTConv(128, 128, heads=8, aggrs: sum,mean)\n",
      "  )\n",
      "  (global_pool): MultiAggregation([\n",
      "    SumAggregation(),\n",
      "    MeanAggregation(),\n",
      "    MaxAggregation(),\n",
      "    StdAggregation(),\n",
      "  ], mode=cat)\n",
      "  (mu_mlp): MLP(\n",
      "    (mlp): Sequential(\n",
      "      (0): Linear(in_features=512, out_features=128, bias=True)\n",
      "      (1): GELU(approximate='none')\n",
      "      (2): Linear(in_features=128, out_features=1, bias=True)\n",
      "    )\n",
      "  )\n",
      "  (log_var_mlp): MLP(\n",
      "    (mlp): Sequential(\n",
      "      (0): Linear(in_features=512, out_features=128, bias=True)\n",
      "      (1): GELU(approximate='none')\n",
      "      (2): Linear(in_features=128, out_features=1, bias=True)\n",
      "    )\n",
      "  )\n",
      ")\n",
      "Number of params: 873 k\n",
      "Epoch: 01, Loss: 73.0038, Val: 43.4897, Test: 42.6628\n",
      "Epoch: 02, Loss: 34.4925, Val: 27.9337, Test: 29.6309\n",
      "Epoch: 03, Loss: 20.9169, Val: 18.4203, Test: 18.6810\n",
      "Epoch: 04, Loss: 14.3110, Val: 14.9553, Test: 15.5055\n",
      "Epoch: 05, Loss: 12.2134, Val: 13.2725, Test: 13.3194\n",
      "Epoch: 06, Loss: 10.8291, Val: 11.6561, Test: 13.0165\n",
      "Epoch: 07, Loss: 10.4653, Val: 12.6163, Test: 13.8170\n",
      "Epoch: 08, Loss: 9.9319, Val: 12.7853, Test: 13.6035\n",
      "Epoch: 09, Loss: 9.6708, Val: 10.2309, Test: 11.6717\n",
      "Epoch: 10, Loss: 9.4152, Val: 11.2856, Test: 12.2924\n",
      "Epoch: 11, Loss: 9.3355, Val: 11.1456, Test: 11.8804\n",
      "Epoch: 12, Loss: 9.0907, Val: 9.6557, Test: 10.7695\n",
      "Epoch: 13, Loss: 9.0444, Val: 8.8203, Test: 9.0495\n",
      "Epoch: 14, Loss: 8.7749, Val: 8.5575, Test: 9.6170\n",
      "Epoch: 15, Loss: 8.5673, Val: 9.1264, Test: 9.9014\n",
      "Epoch: 16, Loss: 8.4035, Val: 8.2152, Test: 9.6629\n",
      "Epoch: 17, Loss: 8.5148, Val: 8.2582, Test: 9.2558\n",
      "Epoch: 18, Loss: 8.3382, Val: 8.4825, Test: 8.9306\n",
      "Epoch: 19, Loss: 8.3214, Val: 9.2002, Test: 9.9109\n",
      "Epoch: 20, Loss: 8.1751, Val: 8.9631, Test: 9.4534\n",
      "Epoch: 21, Loss: 8.0793, Val: 7.8940, Test: 9.0627\n",
      "Epoch: 22, Loss: 8.1311, Val: 7.9150, Test: 8.9080\n",
      "Epoch: 23, Loss: 7.9963, Val: 9.7513, Test: 10.3538\n",
      "Epoch: 24, Loss: 7.8681, Val: 9.3987, Test: 10.0551\n",
      "Epoch: 25, Loss: 7.6924, Val: 8.8650, Test: 9.3738\n",
      "Epoch: 26, Loss: 7.7726, Val: 9.5887, Test: 10.3783\n",
      "Epoch: 27, Loss: 7.9235, Val: 8.5639, Test: 9.6307\n",
      "Epoch: 28, Loss: 7.6205, Val: 9.4727, Test: 10.0131\n",
      "Epoch: 29, Loss: 7.4557, Val: 9.9107, Test: 10.7079\n",
      "Epoch: 30, Loss: 7.4749, Val: 8.2288, Test: 9.1693\n",
      "Epoch: 31, Loss: 7.4120, Val: 10.1037, Test: 10.8165\n",
      "Epoch: 32, Loss: 7.4974, Val: 9.0049, Test: 10.0360\n",
      "Epoch: 33, Loss: 7.3337, Val: 9.2138, Test: 9.9442\n",
      "Epoch: 34, Loss: 7.4727, Val: 8.7333, Test: 9.5281\n",
      "Epoch: 35, Loss: 7.1943, Val: 9.3923, Test: 9.9146\n",
      "Epoch: 36, Loss: 7.0553, Val: 8.2890, Test: 9.6029\n",
      "Epoch: 37, Loss: 7.0105, Val: 9.0578, Test: 9.1749\n",
      "Epoch: 38, Loss: 6.8983, Val: 9.0623, Test: 10.1774\n",
      "Epoch: 39, Loss: 6.8776, Val: 8.6391, Test: 8.7246\n",
      "Epoch: 40, Loss: 6.7561, Val: 7.8969, Test: 8.8341\n",
      "Epoch: 41, Loss: 7.0976, Val: 7.6596, Test: 8.7276\n",
      "Epoch: 42, Loss: 6.8170, Val: 9.7972, Test: 10.4631\n",
      "Epoch: 43, Loss: 6.9077, Val: 9.8627, Test: 10.7778\n",
      "Epoch: 44, Loss: 8.0548, Val: 10.0288, Test: 10.3725\n",
      "Epoch: 45, Loss: 8.6449, Val: 10.9263, Test: 11.4043\n",
      "Epoch: 46, Loss: 8.6481, Val: 7.3768, Test: 9.0311\n",
      "Epoch: 47, Loss: 8.1030, Val: 7.9729, Test: 9.4648\n",
      "Epoch: 48, Loss: 7.3890, Val: 9.0494, Test: 10.2644\n",
      "Epoch: 49, Loss: 8.1218, Val: 10.6721, Test: 11.0485\n",
      "Epoch: 50, Loss: 8.2360, Val: 8.4801, Test: 9.5449\n",
      "Epoch: 51, Loss: 7.6507, Val: 8.2154, Test: 9.7640\n",
      "Epoch: 52, Loss: 7.4247, Val: 8.6879, Test: 9.7683\n",
      "Epoch: 53, Loss: 7.2474, Val: 8.2885, Test: 9.5068\n",
      "Epoch: 54, Loss: 7.2048, Val: 7.8499, Test: 9.0714\n",
      "Epoch: 55, Loss: 6.9900, Val: 8.3778, Test: 9.5312\n",
      "Epoch: 56, Loss: 6.9134, Val: 8.2541, Test: 9.5073\n",
      "Epoch: 57, Loss: 6.7340, Val: 7.8074, Test: 9.2098\n",
      "Epoch: 58, Loss: 6.7107, Val: 7.5478, Test: 8.9694\n",
      "Epoch: 59, Loss: 6.4008, Val: 7.4883, Test: 9.1563\n",
      "Epoch: 60, Loss: 6.3573, Val: 7.9640, Test: 9.2417\n",
      "Epoch: 61, Loss: 6.2407, Val: 8.1437, Test: 9.2167\n",
      "Epoch: 62, Loss: 6.2745, Val: 7.6815, Test: 8.9004\n",
      "Epoch: 63, Loss: 6.1593, Val: 9.2787, Test: 9.9975\n",
      "Epoch: 64, Loss: 6.0572, Val: 8.7959, Test: 10.2082\n",
      "Epoch: 65, Loss: 6.1397, Val: 8.0711, Test: 9.0588\n",
      "Epoch: 66, Loss: 6.1090, Val: 7.2397, Test: 8.5835\n",
      "Epoch: 67, Loss: 6.1390, Val: 7.9588, Test: 8.7244\n",
      "Epoch: 68, Loss: 5.9959, Val: 7.8243, Test: 9.5478\n",
      "Epoch: 69, Loss: 5.9648, Val: 7.6747, Test: 8.7924\n",
      "Epoch: 70, Loss: 5.7798, Val: 8.4013, Test: 9.4248\n",
      "Epoch: 71, Loss: 5.6151, Val: 8.2777, Test: 9.2158\n",
      "Epoch: 72, Loss: 5.5840, Val: 7.7219, Test: 8.7750\n",
      "Epoch: 73, Loss: 5.4647, Val: 8.0911, Test: 8.6893\n",
      "Epoch: 74, Loss: 5.3871, Val: 8.8925, Test: 9.8021\n",
      "Epoch: 75, Loss: 5.3358, Val: 8.4654, Test: 9.5260\n",
      "Epoch: 76, Loss: 5.3821, Val: 7.5922, Test: 8.4422\n",
      "Epoch: 77, Loss: 5.3879, Val: 7.4869, Test: 8.1202\n",
      "Epoch: 78, Loss: 5.3040, Val: 8.9503, Test: 9.6016\n",
      "Epoch: 79, Loss: 5.1637, Val: 7.9185, Test: 8.5534\n",
      "Epoch: 80, Loss: 5.3539, Val: 7.5355, Test: 8.6849\n",
      "Epoch: 81, Loss: 5.2096, Val: 8.3674, Test: 9.2123\n",
      "Epoch: 82, Loss: 5.0186, Val: 9.1351, Test: 9.5483\n",
      "Epoch: 83, Loss: 5.0709, Val: 8.2844, Test: 9.2384\n",
      "Epoch: 84, Loss: 4.8916, Val: 7.2765, Test: 8.1723\n",
      "Epoch: 85, Loss: 4.9839, Val: 7.8760, Test: 8.3959\n",
      "Epoch: 86, Loss: 4.9310, Val: 8.0774, Test: 8.8119\n",
      "Epoch: 87, Loss: 4.8487, Val: 7.6321, Test: 8.0318\n",
      "Epoch: 88, Loss: 4.7514, Val: 7.8996, Test: 8.4974\n",
      "Epoch: 89, Loss: 4.7261, Val: 8.1784, Test: 8.8217\n",
      "Epoch: 90, Loss: 4.7961, Val: 8.9264, Test: 9.0509\n",
      "Epoch: 91, Loss: 4.9144, Val: 8.5532, Test: 8.9183\n",
      "Epoch: 92, Loss: 4.6913, Val: 8.1263, Test: 8.5326\n",
      "Epoch: 93, Loss: 4.5176, Val: 7.6975, Test: 8.3713\n",
      "Epoch: 94, Loss: 4.3224, Val: 7.9602, Test: 8.1766\n",
      "Epoch: 95, Loss: 4.6630, Val: 8.0078, Test: 8.1294\n",
      "Epoch: 96, Loss: 4.5855, Val: 7.5279, Test: 8.5025\n",
      "Epoch: 97, Loss: 4.6785, Val: 7.9649, Test: 8.5751\n",
      "Epoch: 98, Loss: 4.4595, Val: 7.4608, Test: 7.7662\n",
      "Epoch: 99, Loss: 4.4506, Val: 7.7798, Test: 8.5978\n",
      "Epoch: 100, Loss: 4.2632, Val: 7.9070, Test: 8.5385\n",
      "\n",
      "Model's performance on the test set\n",
      "===================================\n",
      "MAE=8.58346939086914\n",
      "Epoch=66\n"
     ]
    }
   ],
   "source": [
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "model = GraphTransformerNet(node_dim_in=NODE_DIM,\n",
    "                            edge_dim_in=EDGE_DIM,\n",
    "                            pe_in_dim=PE_DIM,\n",
    "                            num_gt_layers=4, \n",
    "                            hidden_dim=128,\n",
    "                            num_heads=8,\n",
    "                            norm='bn',\n",
    "                            gt_aggregators=['sum', 'mean'],\n",
    "                            aggregators=['sum','mean','max', 'std'],\n",
    "                            dropout=0.1,\n",
    "                            act='gelu').to(device)\n",
    "\n",
    "if int(torch.__version__.split('.')[0]) >= 2:\n",
    "    model = torch_geometric.compile(model) \n",
    "\n",
    "lr = 0.001\n",
    "params = prepare_params(model, lr)\n",
    "optimizer = torch.optim.AdamW(params)\n",
    "\n",
    "print(model)\n",
    "print(f\"Number of params: {model.num_parameters()//1000} k\")\n",
    "\n",
    "\n",
    "best_epoch = 0\n",
    "best_validation_loss = np.inf\n",
    "test_set_mae = np.inf\n",
    "\n",
    "epochs_list = []\n",
    "train_loss_list = []\n",
    "valid_loss_list = []\n",
    "test_loss_list = []\n",
    "ww_list = []\n",
    "\n",
    "EPOCHS=100\n",
    "for epoch in range(1, EPOCHS + 1):\n",
    "    tr_loss = train(epoch, loss_func=train_loss, warmup_epochs=2)\n",
    "    va_loss = test(val_loader)\n",
    "    te_loss = test(test_loader)\n",
    "    print(f'Epoch: {epoch:02d}, Loss: {tr_loss:.4f}, Val: {va_loss:.4f}, '\n",
    "          f'Test: {te_loss:.4f}')\n",
    "    if va_loss < best_validation_loss:\n",
    "        best_epoch = epoch\n",
    "        best_validation_loss = va_loss\n",
    "        test_set_mae = te_loss\n",
    "    \n",
    "    epochs_list.append(epoch)\n",
    "    train_loss_list.append(tr_loss)\n",
    "    valid_loss_list.append(va_loss)\n",
    "    test_loss_list.append(te_loss)\n",
    "    ww_list.append(get_ww_statistic(model))\n",
    "    \n",
    "print(\"\\nModel's performance on the test set\\n\"\n",
    "        \"===================================\\n\"\n",
    "        f'MAE={test_set_mae}\\n'\n",
    "        f'Epoch={best_epoch}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fc9a9807990>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(test_loss_list, ww_list, label='WW statistic')\n",
    "plt.legend(loc=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fc997dab3d0>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(test_loss_list, valid_loss_list, label='Valid loss')\n",
    "plt.scatter(test_loss_list, train_loss_list, label='Train loss')\n",
    "plt.legend(loc=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor(8.5385)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_loss_list[np.argmin(ww_list)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor(8.5835)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_loss_list[np.argmin(valid_loss_list)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor(8.5385)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_loss_list[np.argmin(train_loss_list)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7fc998498b90>]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plt.plot(epochs_list, train_loss_list)\n",
    "#plt.plot(epochs_list, valid_loss_list)\n",
    "#plt.plot(epochs_list, test_loss_list)\n",
    "plt.plot(epochs_list, ww_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>layer_id</th>\n",
       "      <th>name</th>\n",
       "      <th>D</th>\n",
       "      <th>M</th>\n",
       "      <th>N</th>\n",
       "      <th>Q</th>\n",
       "      <th>alpha</th>\n",
       "      <th>alpha_weighted</th>\n",
       "      <th>entropy</th>\n",
       "      <th>has_esd</th>\n",
       "      <th>...</th>\n",
       "      <th>rf</th>\n",
       "      <th>sigma</th>\n",
       "      <th>spectral_norm</th>\n",
       "      <th>stable_rank</th>\n",
       "      <th>status</th>\n",
       "      <th>sv_max</th>\n",
       "      <th>warning</th>\n",
       "      <th>weak_rank_loss</th>\n",
       "      <th>xmax</th>\n",
       "      <th>xmin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.122949</td>\n",
       "      <td>76</td>\n",
       "      <td>128</td>\n",
       "      <td>1.684211</td>\n",
       "      <td>4.973395</td>\n",
       "      <td>3.236582</td>\n",
       "      <td>0.926830</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.025926</td>\n",
       "      <td>4.474856</td>\n",
       "      <td>22.481977</td>\n",
       "      <td>success</td>\n",
       "      <td>2.115386</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>4.474856</td>\n",
       "      <td>2.363489</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.117601</td>\n",
       "      <td>10</td>\n",
       "      <td>128</td>\n",
       "      <td>12.800000</td>\n",
       "      <td>3.754971</td>\n",
       "      <td>2.743191</td>\n",
       "      <td>0.951755</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.041281</td>\n",
       "      <td>5.377115</td>\n",
       "      <td>4.295238</td>\n",
       "      <td>success</td>\n",
       "      <td>2.318861</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>5.377115</td>\n",
       "      <td>1.773913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.054225</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.358480</td>\n",
       "      <td>3.520788</td>\n",
       "      <td>0.872303</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.453890</td>\n",
       "      <td>11.177059</td>\n",
       "      <td>13.487674</td>\n",
       "      <td>success</td>\n",
       "      <td>3.343211</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>11.177059</td>\n",
       "      <td>2.016812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>12</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.159696</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.798255</td>\n",
       "      <td>1.651993</td>\n",
       "      <td>0.895927</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.271097</td>\n",
       "      <td>3.893725</td>\n",
       "      <td>33.281034</td>\n",
       "      <td>success</td>\n",
       "      <td>1.973252</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>3.893725</td>\n",
       "      <td>1.205171</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>13</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.081044</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.996231</td>\n",
       "      <td>3.282225</td>\n",
       "      <td>0.881519</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.726693</td>\n",
       "      <td>6.627202</td>\n",
       "      <td>22.023015</td>\n",
       "      <td>success</td>\n",
       "      <td>2.574335</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>6.627202</td>\n",
       "      <td>2.613040</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>14</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.049863</td>\n",
       "      <td>128</td>\n",
       "      <td>256</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>3.332465</td>\n",
       "      <td>3.437236</td>\n",
       "      <td>0.924163</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.412326</td>\n",
       "      <td>10.750766</td>\n",
       "      <td>19.852441</td>\n",
       "      <td>success</td>\n",
       "      <td>3.278836</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>10.750766</td>\n",
       "      <td>2.261996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>15</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.079410</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.484427</td>\n",
       "      <td>4.679551</td>\n",
       "      <td>0.847146</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.496885</td>\n",
       "      <td>22.028764</td>\n",
       "      <td>7.017210</td>\n",
       "      <td>success</td>\n",
       "      <td>4.693481</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>22.028764</td>\n",
       "      <td>2.034805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>16</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.112982</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.934383</td>\n",
       "      <td>3.203324</td>\n",
       "      <td>0.882510</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.673194</td>\n",
       "      <td>6.519092</td>\n",
       "      <td>24.837336</td>\n",
       "      <td>success</td>\n",
       "      <td>2.553251</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>6.519092</td>\n",
       "      <td>2.778342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>19</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.086711</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.497365</td>\n",
       "      <td>1.507070</td>\n",
       "      <td>0.886656</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.848236</td>\n",
       "      <td>2.163220</td>\n",
       "      <td>22.412828</td>\n",
       "      <td>success</td>\n",
       "      <td>1.470789</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>2.163220</td>\n",
       "      <td>0.883871</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>22</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.095188</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.880667</td>\n",
       "      <td>1.048817</td>\n",
       "      <td>0.889625</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.588014</td>\n",
       "      <td>1.863233</td>\n",
       "      <td>24.716977</td>\n",
       "      <td>success</td>\n",
       "      <td>1.365003</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>1.863233</td>\n",
       "      <td>0.696106</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>30</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.141328</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.962835</td>\n",
       "      <td>0.499265</td>\n",
       "      <td>0.894711</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.310351</td>\n",
       "      <td>1.474040</td>\n",
       "      <td>30.572927</td>\n",
       "      <td>success</td>\n",
       "      <td>1.214100</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>1.474040</td>\n",
       "      <td>0.459847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>33</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.089664</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.355681</td>\n",
       "      <td>1.475290</td>\n",
       "      <td>0.876933</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.491194</td>\n",
       "      <td>2.751944</td>\n",
       "      <td>20.253079</td>\n",
       "      <td>success</td>\n",
       "      <td>1.658898</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>2.751944</td>\n",
       "      <td>0.823048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>40</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.157742</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.844603</td>\n",
       "      <td>1.782890</td>\n",
       "      <td>0.895532</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.284628</td>\n",
       "      <td>4.234110</td>\n",
       "      <td>30.766720</td>\n",
       "      <td>success</td>\n",
       "      <td>2.057695</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>4.234110</td>\n",
       "      <td>1.259803</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>41</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.155480</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.047716</td>\n",
       "      <td>1.825896</td>\n",
       "      <td>0.897028</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.323772</td>\n",
       "      <td>3.972857</td>\n",
       "      <td>33.225338</td>\n",
       "      <td>success</td>\n",
       "      <td>1.993203</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>3.972857</td>\n",
       "      <td>1.382264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>42</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.150011</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.721778</td>\n",
       "      <td>2.152830</td>\n",
       "      <td>0.896832</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.488846</td>\n",
       "      <td>3.788272</td>\n",
       "      <td>33.838592</td>\n",
       "      <td>success</td>\n",
       "      <td>1.946348</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>3.788272</td>\n",
       "      <td>1.694731</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>43</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.080517</td>\n",
       "      <td>128</td>\n",
       "      <td>256</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>4.899960</td>\n",
       "      <td>3.921460</td>\n",
       "      <td>0.939359</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.894712</td>\n",
       "      <td>6.313999</td>\n",
       "      <td>30.382338</td>\n",
       "      <td>success</td>\n",
       "      <td>2.512767</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>6.313999</td>\n",
       "      <td>2.800793</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>44</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.066255</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.119843</td>\n",
       "      <td>3.618142</td>\n",
       "      <td>0.882698</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.735354</td>\n",
       "      <td>7.554804</td>\n",
       "      <td>18.734070</td>\n",
       "      <td>success</td>\n",
       "      <td>2.748600</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>7.554804</td>\n",
       "      <td>2.441215</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>45</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.106815</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.487403</td>\n",
       "      <td>3.887275</td>\n",
       "      <td>0.883171</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.871851</td>\n",
       "      <td>7.349601</td>\n",
       "      <td>21.100859</td>\n",
       "      <td>success</td>\n",
       "      <td>2.711015</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>7.349601</td>\n",
       "      <td>2.938988</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>48</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.073576</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.589228</td>\n",
       "      <td>1.921723</td>\n",
       "      <td>0.873748</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.539891</td>\n",
       "      <td>3.430947</td>\n",
       "      <td>15.391314</td>\n",
       "      <td>success</td>\n",
       "      <td>1.852281</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>3.430947</td>\n",
       "      <td>0.783011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>51</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.069610</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.935763</td>\n",
       "      <td>2.067334</td>\n",
       "      <td>0.874532</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.712027</td>\n",
       "      <td>3.351729</td>\n",
       "      <td>15.166483</td>\n",
       "      <td>success</td>\n",
       "      <td>1.830773</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>3.351729</td>\n",
       "      <td>0.900641</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>59</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.083307</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.793153</td>\n",
       "      <td>1.261016</td>\n",
       "      <td>0.884770</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.570150</td>\n",
       "      <td>2.150034</td>\n",
       "      <td>23.583085</td>\n",
       "      <td>success</td>\n",
       "      <td>1.466300</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>2.150034</td>\n",
       "      <td>0.766459</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>62</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.156413</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.789845</td>\n",
       "      <td>0.304383</td>\n",
       "      <td>0.897338</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.263898</td>\n",
       "      <td>1.285594</td>\n",
       "      <td>33.401656</td>\n",
       "      <td>success</td>\n",
       "      <td>1.133840</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>1.285594</td>\n",
       "      <td>0.386773</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>69</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.077046</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.744056</td>\n",
       "      <td>3.930385</td>\n",
       "      <td>0.886778</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.882482</td>\n",
       "      <td>6.737307</td>\n",
       "      <td>21.361144</td>\n",
       "      <td>success</td>\n",
       "      <td>2.595632</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>6.737307</td>\n",
       "      <td>2.620648</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>70</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.091993</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.849720</td>\n",
       "      <td>3.440158</td>\n",
       "      <td>0.883727</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.594208</td>\n",
       "      <td>7.827312</td>\n",
       "      <td>18.595292</td>\n",
       "      <td>success</td>\n",
       "      <td>2.797733</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>7.827312</td>\n",
       "      <td>2.238107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>71</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.127556</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.772543</td>\n",
       "      <td>2.197397</td>\n",
       "      <td>0.897773</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.482638</td>\n",
       "      <td>3.823587</td>\n",
       "      <td>33.520042</td>\n",
       "      <td>success</td>\n",
       "      <td>1.955400</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>3.823587</td>\n",
       "      <td>1.653919</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>72</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.058737</td>\n",
       "      <td>128</td>\n",
       "      <td>256</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>3.113399</td>\n",
       "      <td>3.704914</td>\n",
       "      <td>0.907649</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.379578</td>\n",
       "      <td>15.487808</td>\n",
       "      <td>14.855984</td>\n",
       "      <td>success</td>\n",
       "      <td>3.935455</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>15.487808</td>\n",
       "      <td>2.319567</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>73</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.058444</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.613540</td>\n",
       "      <td>4.292165</td>\n",
       "      <td>0.882170</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.933012</td>\n",
       "      <td>8.518061</td>\n",
       "      <td>16.943483</td>\n",
       "      <td>success</td>\n",
       "      <td>2.918572</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>8.518061</td>\n",
       "      <td>2.786161</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>74</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.094927</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.915185</td>\n",
       "      <td>3.288199</td>\n",
       "      <td>0.882718</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.651855</td>\n",
       "      <td>6.916047</td>\n",
       "      <td>22.477528</td>\n",
       "      <td>success</td>\n",
       "      <td>2.629838</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>6.916047</td>\n",
       "      <td>2.589915</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>77</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.101014</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.515042</td>\n",
       "      <td>2.035053</td>\n",
       "      <td>0.878363</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.014705</td>\n",
       "      <td>2.823107</td>\n",
       "      <td>18.749766</td>\n",
       "      <td>success</td>\n",
       "      <td>1.680210</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>2.823107</td>\n",
       "      <td>1.145636</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>80</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.056223</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.765778</td>\n",
       "      <td>2.495034</td>\n",
       "      <td>0.829254</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.368190</td>\n",
       "      <td>7.981959</td>\n",
       "      <td>9.038400</td>\n",
       "      <td>success</td>\n",
       "      <td>2.825236</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>7.981959</td>\n",
       "      <td>0.914474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>88</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.095086</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.752781</td>\n",
       "      <td>1.316081</td>\n",
       "      <td>0.883957</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.539865</td>\n",
       "      <td>2.242305</td>\n",
       "      <td>22.489459</td>\n",
       "      <td>success</td>\n",
       "      <td>1.497433</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>2.242305</td>\n",
       "      <td>0.737241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>91</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.084499</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.942387</td>\n",
       "      <td>1.854150</td>\n",
       "      <td>0.880894</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.713634</td>\n",
       "      <td>2.953327</td>\n",
       "      <td>17.645972</td>\n",
       "      <td>success</td>\n",
       "      <td>1.718525</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>2.953327</td>\n",
       "      <td>0.917592</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>98</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.065154</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.684967</td>\n",
       "      <td>3.866445</td>\n",
       "      <td>0.872412</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.615974</td>\n",
       "      <td>11.200774</td>\n",
       "      <td>14.015882</td>\n",
       "      <td>success</td>\n",
       "      <td>3.346756</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>11.200774</td>\n",
       "      <td>2.566822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>99</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.163128</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.837379</td>\n",
       "      <td>1.595752</td>\n",
       "      <td>0.897945</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.270907</td>\n",
       "      <td>3.650930</td>\n",
       "      <td>35.186669</td>\n",
       "      <td>success</td>\n",
       "      <td>1.910741</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>3.650930</td>\n",
       "      <td>1.178466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>100</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.138538</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.837522</td>\n",
       "      <td>2.297433</td>\n",
       "      <td>0.895183</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.526915</td>\n",
       "      <td>3.968956</td>\n",
       "      <td>32.540765</td>\n",
       "      <td>success</td>\n",
       "      <td>1.992224</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>3.968956</td>\n",
       "      <td>1.793853</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>101</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.087438</td>\n",
       "      <td>128</td>\n",
       "      <td>256</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>5.011326</td>\n",
       "      <td>3.664678</td>\n",
       "      <td>0.942412</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.855218</td>\n",
       "      <td>5.386158</td>\n",
       "      <td>33.801480</td>\n",
       "      <td>success</td>\n",
       "      <td>2.320810</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>5.386158</td>\n",
       "      <td>2.542778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>102</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.087584</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.779065</td>\n",
       "      <td>3.441933</td>\n",
       "      <td>0.881724</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.567274</td>\n",
       "      <td>8.143096</td>\n",
       "      <td>18.344527</td>\n",
       "      <td>success</td>\n",
       "      <td>2.853611</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>8.143096</td>\n",
       "      <td>2.260162</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>103</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.138732</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>7.376163</td>\n",
       "      <td>4.417350</td>\n",
       "      <td>0.897584</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.922486</td>\n",
       "      <td>3.970711</td>\n",
       "      <td>31.992515</td>\n",
       "      <td>success</td>\n",
       "      <td>1.992664</td>\n",
       "      <td>under-trained</td>\n",
       "      <td>0</td>\n",
       "      <td>3.970711</td>\n",
       "      <td>2.734524</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>106</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.144626</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>5.137781</td>\n",
       "      <td>0.502606</td>\n",
       "      <td>0.897100</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.925236</td>\n",
       "      <td>1.252638</td>\n",
       "      <td>33.764284</td>\n",
       "      <td>success</td>\n",
       "      <td>1.119213</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>1.252638</td>\n",
       "      <td>0.733808</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>109</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.144449</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.837622</td>\n",
       "      <td>0.388630</td>\n",
       "      <td>0.897236</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.509652</td>\n",
       "      <td>1.262608</td>\n",
       "      <td>33.551035</td>\n",
       "      <td>success</td>\n",
       "      <td>1.123658</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1.262608</td>\n",
       "      <td>0.567809</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>117</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.057210</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.653854</td>\n",
       "      <td>1.967298</td>\n",
       "      <td>0.873177</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.579118</td>\n",
       "      <td>3.454754</td>\n",
       "      <td>14.970084</td>\n",
       "      <td>success</td>\n",
       "      <td>1.858697</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>3.454754</td>\n",
       "      <td>0.807772</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>120</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.090682</td>\n",
       "      <td>128</td>\n",
       "      <td>128</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.599908</td>\n",
       "      <td>1.621778</td>\n",
       "      <td>0.876931</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.581357</td>\n",
       "      <td>2.821665</td>\n",
       "      <td>18.285566</td>\n",
       "      <td>success</td>\n",
       "      <td>1.679781</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>2.821665</td>\n",
       "      <td>0.829965</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>131</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.120503</td>\n",
       "      <td>128</td>\n",
       "      <td>512</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>2.649647</td>\n",
       "      <td>3.502778</td>\n",
       "      <td>0.767289</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.183294</td>\n",
       "      <td>20.988392</td>\n",
       "      <td>3.433379</td>\n",
       "      <td>success</td>\n",
       "      <td>4.581309</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>20.988392</td>\n",
       "      <td>0.249661</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>136</td>\n",
       "      <td>Linear</td>\n",
       "      <td>0.083846</td>\n",
       "      <td>128</td>\n",
       "      <td>512</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>4.592115</td>\n",
       "      <td>1.704532</td>\n",
       "      <td>0.961856</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.590540</td>\n",
       "      <td>2.350643</td>\n",
       "      <td>19.085313</td>\n",
       "      <td>success</td>\n",
       "      <td>1.533181</td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>2.350643</td>\n",
       "      <td>0.435499</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>44 rows × 31 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    layer_id    name         D    M    N          Q     alpha  alpha_weighted  \\\n",
       "0          1  Linear  0.122949   76  128   1.684211  4.973395        3.236582   \n",
       "1          2  Linear  0.117601   10  128  12.800000  3.754971        2.743191   \n",
       "2         11  Linear  0.054225  128  128   1.000000  3.358480        3.520788   \n",
       "3         12  Linear  0.159696  128  128   1.000000  2.798255        1.651993   \n",
       "4         13  Linear  0.081044  128  128   1.000000  3.996231        3.282225   \n",
       "5         14  Linear  0.049863  128  256   2.000000  3.332465        3.437236   \n",
       "6         15  Linear  0.079410  128  128   1.000000  3.484427        4.679551   \n",
       "7         16  Linear  0.112982  128  128   1.000000  3.934383        3.203324   \n",
       "8         19  Linear  0.086711  128  128   1.000000  4.497365        1.507070   \n",
       "9         22  Linear  0.095188  128  128   1.000000  3.880667        1.048817   \n",
       "10        30  Linear  0.141328  128  128   1.000000  2.962835        0.499265   \n",
       "11        33  Linear  0.089664  128  128   1.000000  3.355681        1.475290   \n",
       "12        40  Linear  0.157742  128  128   1.000000  2.844603        1.782890   \n",
       "13        41  Linear  0.155480  128  128   1.000000  3.047716        1.825896   \n",
       "14        42  Linear  0.150011  128  128   1.000000  3.721778        2.152830   \n",
       "15        43  Linear  0.080517  128  256   2.000000  4.899960        3.921460   \n",
       "16        44  Linear  0.066255  128  128   1.000000  4.119843        3.618142   \n",
       "17        45  Linear  0.106815  128  128   1.000000  4.487403        3.887275   \n",
       "18        48  Linear  0.073576  128  128   1.000000  3.589228        1.921723   \n",
       "19        51  Linear  0.069610  128  128   1.000000  3.935763        2.067334   \n",
       "20        59  Linear  0.083307  128  128   1.000000  3.793153        1.261016   \n",
       "21        62  Linear  0.156413  128  128   1.000000  2.789845        0.304383   \n",
       "22        69  Linear  0.077046  128  128   1.000000  4.744056        3.930385   \n",
       "23        70  Linear  0.091993  128  128   1.000000  3.849720        3.440158   \n",
       "24        71  Linear  0.127556  128  128   1.000000  3.772543        2.197397   \n",
       "25        72  Linear  0.058737  128  256   2.000000  3.113399        3.704914   \n",
       "26        73  Linear  0.058444  128  128   1.000000  4.613540        4.292165   \n",
       "27        74  Linear  0.094927  128  128   1.000000  3.915185        3.288199   \n",
       "28        77  Linear  0.101014  128  128   1.000000  4.515042        2.035053   \n",
       "29        80  Linear  0.056223  128  128   1.000000  2.765778        2.495034   \n",
       "30        88  Linear  0.095086  128  128   1.000000  3.752781        1.316081   \n",
       "31        91  Linear  0.084499  128  128   1.000000  3.942387        1.854150   \n",
       "32        98  Linear  0.065154  128  128   1.000000  3.684967        3.866445   \n",
       "33        99  Linear  0.163128  128  128   1.000000  2.837379        1.595752   \n",
       "34       100  Linear  0.138538  128  128   1.000000  3.837522        2.297433   \n",
       "35       101  Linear  0.087438  128  256   2.000000  5.011326        3.664678   \n",
       "36       102  Linear  0.087584  128  128   1.000000  3.779065        3.441933   \n",
       "37       103  Linear  0.138732  128  128   1.000000  7.376163        4.417350   \n",
       "38       106  Linear  0.144626  128  128   1.000000  5.137781        0.502606   \n",
       "39       109  Linear  0.144449  128  128   1.000000  3.837622        0.388630   \n",
       "40       117  Linear  0.057210  128  128   1.000000  3.653854        1.967298   \n",
       "41       120  Linear  0.090682  128  128   1.000000  3.599908        1.621778   \n",
       "42       131  Linear  0.120503  128  512   4.000000  2.649647        3.502778   \n",
       "43       136  Linear  0.083846  128  512   4.000000  4.592115        1.704532   \n",
       "\n",
       "     entropy  has_esd  ...  rf     sigma  spectral_norm  stable_rank   status  \\\n",
       "0   0.926830     True  ...   1  1.025926       4.474856    22.481977  success   \n",
       "1   0.951755     True  ...   1  1.041281       5.377115     4.295238  success   \n",
       "2   0.872303     True  ...   1  0.453890      11.177059    13.487674  success   \n",
       "3   0.895927     True  ...   1  0.271097       3.893725    33.281034  success   \n",
       "4   0.881519     True  ...   1  0.726693       6.627202    22.023015  success   \n",
       "5   0.924163     True  ...   1  0.412326      10.750766    19.852441  success   \n",
       "6   0.847146     True  ...   1  0.496885      22.028764     7.017210  success   \n",
       "7   0.882510     True  ...   1  0.673194       6.519092    24.837336  success   \n",
       "8   0.886656     True  ...   1  0.848236       2.163220    22.412828  success   \n",
       "9   0.889625     True  ...   1  0.588014       1.863233    24.716977  success   \n",
       "10  0.894711     True  ...   1  0.310351       1.474040    30.572927  success   \n",
       "11  0.876933     True  ...   1  0.491194       2.751944    20.253079  success   \n",
       "12  0.895532     True  ...   1  0.284628       4.234110    30.766720  success   \n",
       "13  0.897028     True  ...   1  0.323772       3.972857    33.225338  success   \n",
       "14  0.896832     True  ...   1  0.488846       3.788272    33.838592  success   \n",
       "15  0.939359     True  ...   1  0.894712       6.313999    30.382338  success   \n",
       "16  0.882698     True  ...   1  0.735354       7.554804    18.734070  success   \n",
       "17  0.883171     True  ...   1  0.871851       7.349601    21.100859  success   \n",
       "18  0.873748     True  ...   1  0.539891       3.430947    15.391314  success   \n",
       "19  0.874532     True  ...   1  0.712027       3.351729    15.166483  success   \n",
       "20  0.884770     True  ...   1  0.570150       2.150034    23.583085  success   \n",
       "21  0.897338     True  ...   1  0.263898       1.285594    33.401656  success   \n",
       "22  0.886778     True  ...   1  0.882482       6.737307    21.361144  success   \n",
       "23  0.883727     True  ...   1  0.594208       7.827312    18.595292  success   \n",
       "24  0.897773     True  ...   1  0.482638       3.823587    33.520042  success   \n",
       "25  0.907649     True  ...   1  0.379578      15.487808    14.855984  success   \n",
       "26  0.882170     True  ...   1  0.933012       8.518061    16.943483  success   \n",
       "27  0.882718     True  ...   1  0.651855       6.916047    22.477528  success   \n",
       "28  0.878363     True  ...   1  1.014705       2.823107    18.749766  success   \n",
       "29  0.829254     True  ...   1  0.368190       7.981959     9.038400  success   \n",
       "30  0.883957     True  ...   1  0.539865       2.242305    22.489459  success   \n",
       "31  0.880894     True  ...   1  0.713634       2.953327    17.645972  success   \n",
       "32  0.872412     True  ...   1  0.615974      11.200774    14.015882  success   \n",
       "33  0.897945     True  ...   1  0.270907       3.650930    35.186669  success   \n",
       "34  0.895183     True  ...   1  0.526915       3.968956    32.540765  success   \n",
       "35  0.942412     True  ...   1  0.855218       5.386158    33.801480  success   \n",
       "36  0.881724     True  ...   1  0.567274       8.143096    18.344527  success   \n",
       "37  0.897584     True  ...   1  1.922486       3.970711    31.992515  success   \n",
       "38  0.897100     True  ...   1  0.925236       1.252638    33.764284  success   \n",
       "39  0.897236     True  ...   1  0.509652       1.262608    33.551035  success   \n",
       "40  0.873177     True  ...   1  0.579118       3.454754    14.970084  success   \n",
       "41  0.876931     True  ...   1  0.581357       2.821665    18.285566  success   \n",
       "42  0.767289     True  ...   1  0.183294      20.988392     3.433379  success   \n",
       "43  0.961856     True  ...   1  0.590540       2.350643    19.085313  success   \n",
       "\n",
       "      sv_max        warning  weak_rank_loss       xmax      xmin  \n",
       "0   2.115386                              0   4.474856  2.363489  \n",
       "1   2.318861                              0   5.377115  1.773913  \n",
       "2   3.343211                              0  11.177059  2.016812  \n",
       "3   1.973252                              0   3.893725  1.205171  \n",
       "4   2.574335                              1   6.627202  2.613040  \n",
       "5   3.278836                              0  10.750766  2.261996  \n",
       "6   4.693481                              1  22.028764  2.034805  \n",
       "7   2.553251                              0   6.519092  2.778342  \n",
       "8   1.470789                              0   2.163220  0.883871  \n",
       "9   1.365003                              0   1.863233  0.696106  \n",
       "10  1.214100                              0   1.474040  0.459847  \n",
       "11  1.658898                              0   2.751944  0.823048  \n",
       "12  2.057695                              1   4.234110  1.259803  \n",
       "13  1.993203                              0   3.972857  1.382264  \n",
       "14  1.946348                              1   3.788272  1.694731  \n",
       "15  2.512767                              0   6.313999  2.800793  \n",
       "16  2.748600                              0   7.554804  2.441215  \n",
       "17  2.711015                              0   7.349601  2.938988  \n",
       "18  1.852281                              0   3.430947  0.783011  \n",
       "19  1.830773                              0   3.351729  0.900641  \n",
       "20  1.466300                              0   2.150034  0.766459  \n",
       "21  1.133840                              0   1.285594  0.386773  \n",
       "22  2.595632                              0   6.737307  2.620648  \n",
       "23  2.797733                              0   7.827312  2.238107  \n",
       "24  1.955400                              0   3.823587  1.653919  \n",
       "25  3.935455                              0  15.487808  2.319567  \n",
       "26  2.918572                              0   8.518061  2.786161  \n",
       "27  2.629838                              0   6.916047  2.589915  \n",
       "28  1.680210                              0   2.823107  1.145636  \n",
       "29  2.825236                              0   7.981959  0.914474  \n",
       "30  1.497433                              0   2.242305  0.737241  \n",
       "31  1.718525                              0   2.953327  0.917592  \n",
       "32  3.346756                              1  11.200774  2.566822  \n",
       "33  1.910741                              0   3.650930  1.178466  \n",
       "34  1.992224                              0   3.968956  1.793853  \n",
       "35  2.320810                              0   5.386158  2.542778  \n",
       "36  2.853611                              0   8.143096  2.260162  \n",
       "37  1.992664  under-trained               0   3.970711  2.734524  \n",
       "38  1.119213                              0   1.252638  0.733808  \n",
       "39  1.123658                              1   1.262608  0.567809  \n",
       "40  1.858697                              1   3.454754  0.807772  \n",
       "41  1.679781                              1   2.821665  0.829965  \n",
       "42  4.581309                              0  20.988392  0.249661  \n",
       "43  1.533181                              0   2.350643  0.435499  \n",
       "\n",
       "[44 rows x 31 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "watcher = ww.WeightWatcher(model=model)\n",
    "details = watcher.analyze(plot=False)\n",
    "details"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'log_norm': 1.9617880817975497,\n",
       " 'alpha': 3.8759143081438427,\n",
       " 'alpha_weighted': 2.5125688359017633,\n",
       " 'log_alpha_norm': 3.029487412316085,\n",
       " 'log_spectral_norm': 0.6560423502811896,\n",
       " 'stable_rank': 22.397062682171264}"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "watcher.get_summary(details)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'alpha')"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(list(range(len(details))), details.alpha.to_list())\n",
    "plt.ylabel('layer_id')\n",
    "plt.ylabel('alpha')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}