model copy.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "1. Setup and Load Data"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "1.1 Install Dependencies and Setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda:0\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import cv2\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "import torchvision.models.segmentation\n",
    "import torch\n",
    "import torchvision.transforms as T\n",
    "from torch.utils.data import TensorDataset, DataLoader, Dataset\n",
    "from PIL import Image\n",
    "import scipy.io\n",
    "from sklearn.model_selection import train_test_split\n",
    "import random\n",
    "from torchviz import make_dot\n",
    "import torch.optim as optim\n",
    "from torchsummary import summary\n",
    "\n",
    "# Set the device to use\n",
    "# CUDA refers to the GPU\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "\n",
    "## Hyperparameters\n",
    "num_epochs = 400\n",
    "batch_size = 256\n",
    "\n",
    "## Fixing Random Seed for Reproducibility\n",
    "torch.manual_seed(0)\n",
    "np.random.seed(0)\n",
    "random.seed(0)\n",
    "\n",
    "\n",
    "# If you are on CoLab and successfully using the GPU, this print should\n",
    "#   contain \"cuda\" in it\n",
    "print(str(device))\n",
    "assert('cuda' in str(device))  # comment out this assert if you are not using a GPU\n",
    "\n",
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Initial pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/12500 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 12500/12500 [00:05<00:00, 2473.84it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[5.e+05 5.e-01 5.e+03]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "import tqdm\n",
    "\n",
    "os.chdir('../HMEI2023-bxliu-data/')\n",
    "\n",
    "theta_list = []\n",
    "\n",
    "value_directory = 'value_cats/'\n",
    "\n",
    "# iterate over files in\n",
    "# that directory\n",
    "for filename in tqdm.tqdm(os.listdir(value_directory)):\n",
    "    # print(filename)\n",
    "    f = os.path.join(value_directory, filename)\n",
    "    # checking if it is a file\n",
    "    if os.path.isfile(f):\n",
    "        mat = scipy.io.loadmat(f)\n",
    "\n",
    "        theta_list.append(mat['th0'][0])\n",
    "\n",
    "print(theta_list[0])\n",
    "\n",
    "theta_list = np.asarray(theta_list)\n",
    "\n",
    "avg = np.average(theta_list, axis = 0)\n",
    "std = np.std(theta_list, axis = 0)\n",
    "\n",
    "\n",
    "os.chdir('../HMEI2023-bxliu/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MyDataset(Dataset):\n",
    "    def __init__(self):\n",
    "        self.data_files = os.listdir('../HMEI2023-bxliu-data/image_cats')\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        filename = self.data_files[idx]\n",
    "        f = os.path.join('../HMEI2023-bxliu-data/image_cats', filename)\n",
    "        data = cv2.cvtColor(cv2.imread(f), cv2.COLOR_BGR2GRAY)\n",
    "        data = cv2.resize(data, (128,128), interpolation=cv2.INTER_CUBIC)\n",
    "        data = np.expand_dims(data, axis=2)\n",
    "\n",
    "        data = np.moveaxis(data, -1, 0)\n",
    "        data = np.divide(data, 255)\n",
    "        data = data.astype(np.float32)\n",
    "\n",
    "        matfile = '../HMEI2023-bxliu-data/value_cats/' + filename[:len(filename) - 4] + '.mat'\n",
    "        mat = scipy.io.loadmat(matfile)\n",
    "\n",
    "        theta = mat['th0'][0].astype(np.float32)\n",
    "        theta = np.divide(np.subtract(theta, avg), std).astype(np.float32)\n",
    "        \n",
    "\n",
    "        return data, theta\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.data_files)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class '__main__.MyDataset'>\n",
      "<class 'torch.utils.data.dataset.Subset'>\n"
     ]
    }
   ],
   "source": [
    "all_data = MyDataset()\n",
    "print(type(all_data))\n",
    "train, val, test = torch.utils.data.random_split(all_data, [0.75, 0.125, 0.125])\n",
    "print(type(train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_loader = DataLoader(train, batch_size=batch_size, shuffle=False)\n",
    "val_loader = DataLoader(val, batch_size=64, shuffle=False)\n",
    "test_loader = DataLoader(test, batch_size=64, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_loader = DataLoader(all_data, batch_size=batch_size, shuffle=False)\n",
    "# val_loader = DataLoader(all_data, batch_size=64, shuffle=False)\n",
    "# test_loader = DataLoader(all_data, batch_size=64, shuffle=False)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import tqdm\n",
    "\n",
    "# os.chdir('../HMEI2023-bxliu-data/')\n",
    "\n",
    "# image_list = []\n",
    "# theta_list = []\n",
    "\n",
    "# image_directory = 'image_cats'\n",
    "# value_directory = 'value_cats/'\n",
    "\n",
    "# # iterate over files in\n",
    "# # that directory\n",
    "# for filename in tqdm.tqdm(os.listdir(image_directory)):\n",
    "#     # print(filename)\n",
    "#     f = os.path.join(image_directory, filename)\n",
    "#     # checking if it is a file\n",
    "#     if os.path.isfile(f):\n",
    "#         # color\n",
    "#         # image = Image.open(f)\n",
    "#         # data = np.asarray(image)\n",
    "\n",
    "#         # gray scale\n",
    "#         data = cv2.cvtColor(cv2.imread(f), cv2.COLOR_BGR2GRAY)\n",
    "#         data = cv2.resize(data, (128,128), interpolation=cv2.INTER_CUBIC)\n",
    "#         data = np.expand_dims(data, axis=2)\n",
    "\n",
    "#         matfile = value_directory + filename[:len(filename) - 4] + '.mat'\n",
    "#         mat = scipy.io.loadmat(matfile)\n",
    "#         image_list.append(data)\n",
    "#         theta_list.append(mat['th0'][0])\n",
    "\n",
    "# print(image_list[0].shape)\n",
    "# print(theta_list[0])\n",
    "\n",
    "# theta_list = np.asarray(theta_list)\n",
    "# image_list = np.asarray(image_list)\n",
    "\n",
    "# image_list = np.moveaxis(image_list, -1, 1)\n",
    "# theta_list = np.moveaxis(theta_list, -1, 1)\n",
    "\n",
    "# print(image_list.shape)\n",
    "\n",
    "# image_list = torch.tensor(image_list).float()\n",
    "\n",
    "# theta_list = torch.tensor(theta_list)\n",
    "\n",
    "# os.chdir('../HMEI2023-bxliu/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # normalize data:\n",
    "\n",
    "# # images\n",
    "# imean = torch.std(image_list, dim=0)\n",
    "# istd = torch.mean(image_list, dim=0)\n",
    "# # image_tensor = torch.div(torch.sub(image_list, imean), istd)\n",
    "# image_tensor = torch.div(image_list, 255).float()\n",
    "# print(image_tensor)\n",
    "\n",
    "# # theta\n",
    "# tmean = torch.std(theta_list, dim=0)\n",
    "# tstd = torch.mean(theta_list, dim=0)\n",
    "# theta_tensor = torch.div(torch.sub(theta_list, tmean), tstd).float()\n",
    "# # theta_tensor = theta_list.float()\n",
    "# print(theta_tensor)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [],
   "source": [
    "# full_dataset = TensorDataset(image_tensor, theta_tensor)\n",
    "\n",
    "# train, val, test = torch.utils.data.random_split(full_dataset, [0.75, 0.125, 0.125])\n",
    "# train_loader = DataLoader(train, batch_size=42, shuffle=False)\n",
    "# val_loader = DataLoader(val, batch_size=batch_size, shuffle=False)\n",
    "# test_loader = DataLoader(test, batch_size=2, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<torch.utils.data.dataloader.DataLoader object at 0x000001DE7770E2C0>\n",
      "<torch.utils.data.dataloader.DataLoader object at 0x000001DE7770F640>\n",
      "<torch.utils.data.dataloader.DataLoader object at 0x000001DE7770F790>\n"
     ]
    }
   ],
   "source": [
    "print(train_loader)\n",
    "print(val_loader)\n",
    "print(test_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [],
   "source": [
    "# image_train, image_test, theta_train, theta_test = train_test_split(image_tensor, theta_tensor)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(len(image_train))\n",
    "# print(len(image_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_dataset = TensorDataset(image_train, theta_train)\n",
    "# print(train_dataset)\n",
    "# train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=False)\n",
    "# print(train_loader)\n",
    "\n",
    "# test_dataset = TensorDataset(image_test, theta_test)\n",
    "# print(test_dataset)\n",
    "# test_loader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)\n",
    "# print(test_loader)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Make Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ConvNet(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(ConvNet, self).__init__()\n",
    "        self.conv1 = nn.Conv2d(1, 32, 5, padding=2)\n",
    "        self.pool1 = nn.MaxPool2d(2, 2) \n",
    "        self.conv2 = nn.Conv2d(32, 64, 5, padding=2) \n",
    "        self.pool2 = nn.MaxPool2d(2, 2)\n",
    "        self.conv3 = nn.Conv2d(64, 128, 5, padding=2) \n",
    "        self.pool3 = nn.MaxPool2d(2, 2)\n",
    "\n",
    "        self.size_linear = 128*16*16\n",
    "        self.fc1 = nn.Linear(self.size_linear, 512)\n",
    "        self.fc2 = nn.Linear(512, 128)\n",
    "        self.fc3 = nn.Linear(128, 3)\n",
    "\n",
    "\n",
    "    def forward(self, x):\n",
    "            x = self.pool1(F.relu(self.conv1(x))) \n",
    "            x = self.pool2(F.relu(self.conv2(x))) \n",
    "            x = self.pool3(F.relu(self.conv3(x)))\n",
    "            x = x.view(-1, self.size_linear) # this flattens x into a 1D vector\n",
    "            x = F.relu(self.fc1(x))\n",
    "            x = F.relu(self.fc2(x)) \n",
    "            x = self.fc3(x)\n",
    "            return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[-0.0854,  0.0819,  0.0055],\n",
      "        [-0.0860,  0.0823,  0.0033],\n",
      "        [-0.0872,  0.0808,  0.0052],\n",
      "        [-0.0878,  0.0826,  0.0053],\n",
      "        [-0.0848,  0.0798,  0.0048],\n",
      "        [-0.0853,  0.0808,  0.0054],\n",
      "        [-0.0865,  0.0816,  0.0048],\n",
      "        [-0.0856,  0.0809,  0.0051],\n",
      "        [-0.0848,  0.0824,  0.0053],\n",
      "        [-0.0855,  0.0815,  0.0051],\n",
      "        [-0.0857,  0.0813,  0.0041],\n",
      "        [-0.0867,  0.0820,  0.0047],\n",
      "        [-0.0864,  0.0817,  0.0044],\n",
      "        [-0.0854,  0.0816,  0.0052],\n",
      "        [-0.0862,  0.0811,  0.0069],\n",
      "        [-0.0842,  0.0817,  0.0046],\n",
      "        [-0.0864,  0.0809,  0.0053],\n",
      "        [-0.0846,  0.0824,  0.0065],\n",
      "        [-0.0856,  0.0804,  0.0049],\n",
      "        [-0.0851,  0.0822,  0.0053],\n",
      "        [-0.0856,  0.0810,  0.0060],\n",
      "        [-0.0858,  0.0819,  0.0039],\n",
      "        [-0.0875,  0.0825,  0.0036],\n",
      "        [-0.0859,  0.0817,  0.0040],\n",
      "        [-0.0860,  0.0812,  0.0044],\n",
      "        [-0.0857,  0.0821,  0.0076],\n",
      "        [-0.0861,  0.0816,  0.0047],\n",
      "        [-0.0857,  0.0808,  0.0055],\n",
      "        [-0.0846,  0.0818,  0.0050],\n",
      "        [-0.0841,  0.0815,  0.0046],\n",
      "        [-0.0851,  0.0819,  0.0058],\n",
      "        [-0.0853,  0.0819,  0.0042],\n",
      "        [-0.0856,  0.0830,  0.0047],\n",
      "        [-0.0856,  0.0819,  0.0044],\n",
      "        [-0.0872,  0.0824,  0.0041],\n",
      "        [-0.0857,  0.0813,  0.0070],\n",
      "        [-0.0847,  0.0805,  0.0057],\n",
      "        [-0.0858,  0.0805,  0.0054],\n",
      "        [-0.0864,  0.0813,  0.0040],\n",
      "        [-0.0862,  0.0797,  0.0056],\n",
      "        [-0.0856,  0.0823,  0.0056],\n",
      "        [-0.0858,  0.0817,  0.0065],\n",
      "        [-0.0867,  0.0822,  0.0070],\n",
      "        [-0.0856,  0.0814,  0.0049],\n",
      "        [-0.0848,  0.0807,  0.0049],\n",
      "        [-0.0855,  0.0807,  0.0051],\n",
      "        [-0.0858,  0.0820,  0.0060],\n",
      "        [-0.0859,  0.0828,  0.0059],\n",
      "        [-0.0869,  0.0824,  0.0054],\n",
      "        [-0.0875,  0.0823,  0.0039],\n",
      "        [-0.0867,  0.0825,  0.0045],\n",
      "        [-0.0848,  0.0812,  0.0074],\n",
      "        [-0.0857,  0.0822,  0.0040],\n",
      "        [-0.0866,  0.0820,  0.0054],\n",
      "        [-0.0858,  0.0805,  0.0050],\n",
      "        [-0.0857,  0.0822,  0.0053],\n",
      "        [-0.0862,  0.0823,  0.0050],\n",
      "        [-0.0852,  0.0811,  0.0040],\n",
      "        [-0.0860,  0.0815,  0.0068],\n",
      "        [-0.0857,  0.0820,  0.0052],\n",
      "        [-0.0852,  0.0807,  0.0064],\n",
      "        [-0.0852,  0.0815,  0.0046],\n",
      "        [-0.0848,  0.0802,  0.0060],\n",
      "        [-0.0857,  0.0822,  0.0048],\n",
      "        [-0.0848,  0.0825,  0.0062],\n",
      "        [-0.0856,  0.0813,  0.0060],\n",
      "        [-0.0859,  0.0833,  0.0051],\n",
      "        [-0.0873,  0.0822,  0.0050],\n",
      "        [-0.0855,  0.0815,  0.0058],\n",
      "        [-0.0838,  0.0806,  0.0045],\n",
      "        [-0.0861,  0.0813,  0.0067],\n",
      "        [-0.0858,  0.0809,  0.0058],\n",
      "        [-0.0843,  0.0821,  0.0038],\n",
      "        [-0.0851,  0.0807,  0.0050],\n",
      "        [-0.0850,  0.0813,  0.0057],\n",
      "        [-0.0867,  0.0829,  0.0059],\n",
      "        [-0.0843,  0.0818,  0.0056],\n",
      "        [-0.0846,  0.0805,  0.0044],\n",
      "        [-0.0855,  0.0822,  0.0043],\n",
      "        [-0.0864,  0.0825,  0.0065],\n",
      "        [-0.0853,  0.0820,  0.0053],\n",
      "        [-0.0867,  0.0814,  0.0060],\n",
      "        [-0.0854,  0.0813,  0.0059],\n",
      "        [-0.0856,  0.0828,  0.0060],\n",
      "        [-0.0855,  0.0817,  0.0062],\n",
      "        [-0.0851,  0.0808,  0.0057],\n",
      "        [-0.0851,  0.0822,  0.0046],\n",
      "        [-0.0869,  0.0826,  0.0058],\n",
      "        [-0.0857,  0.0812,  0.0042],\n",
      "        [-0.0847,  0.0810,  0.0040],\n",
      "        [-0.0851,  0.0822,  0.0050],\n",
      "        [-0.0849,  0.0813,  0.0051],\n",
      "        [-0.0861,  0.0821,  0.0053],\n",
      "        [-0.0852,  0.0817,  0.0061],\n",
      "        [-0.0845,  0.0824,  0.0062],\n",
      "        [-0.0848,  0.0812,  0.0079],\n",
      "        [-0.0863,  0.0817,  0.0050],\n",
      "        [-0.0848,  0.0817,  0.0060],\n",
      "        [-0.0852,  0.0815,  0.0059],\n",
      "        [-0.0848,  0.0809,  0.0064]])\n",
      "torch.Size([100, 3])\n",
      "tensor(-0.0854)\n"
     ]
    }
   ],
   "source": [
    "##ASSERT: checks if CNN has the correct output shape\n",
    "with torch.no_grad():  # tells PyTorch not to track gradients here\n",
    "    # test_data is 100 random images, 1 channel, 128-by-128\n",
    "    test_data = torch.rand(100,1,128,128)\n",
    "    test_net = ConvNet()\n",
    "    out = test_net.forward(test_data)\n",
    "    # the output should have size (100,3)\n",
    "    print(out)\n",
    "    print(out.size())\n",
    "    assert(out.size()==(100,3))\n",
    "\n",
    "    print(out[0][0])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Make trainer for training network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Trainer():\n",
    "    def __init__(self,net=None,optim=None,loss_function=None, train_loader=None, val_loader=None):\n",
    "        self.net = net\n",
    "        self.optim = optim\n",
    "        self.loss_function = loss_function\n",
    "        self.train_loader = train_loader\n",
    "        self.val_loader = val_loader\n",
    "\n",
    "    def train(self,epochs):\n",
    "        train_losses = []\n",
    "        val_losses = []\n",
    "        for epoch in range(epochs):\n",
    "            print(epoch)\n",
    "            epoch_train_loss = 0.0\n",
    "            epoch_val_loss = 0.0\n",
    "            epoch_train_steps = 0\n",
    "            epoch_val_steps = 0\n",
    "            for data in self.train_loader:\n",
    "                \n",
    "                # Moving this batch to GPU\n",
    "                # Note that X has shape (batch_size, number of channels, height, width)\n",
    "                # which is equal to (256,3,128,128) since our default batch_size = 256 and \n",
    "                # the image has only 1 channel\n",
    "                X = data[0].to(device)\n",
    "                y = data[1].to(device)\n",
    "                \n",
    "                # print(X[0])\n",
    "                # plt.imshow(np.moveaxis(X[0].cpu().numpy(), 0, -1))\n",
    "\n",
    "                \n",
    "                # Zero the gradient in the optimizer i.e. self.optim\n",
    "                self.optim.zero_grad()\n",
    "\n",
    "                # Getting the output of the Network\n",
    "                out = self.net(X)\n",
    "\n",
    "                # Computing loss using loss function i.e. self.loss_function\n",
    "                loss = self.loss_function(out, y)\n",
    "\n",
    "                # Backpropagate to compute gradients of parameteres\n",
    "                loss.backward()\n",
    "\n",
    "                # Call the optimizer i.e. self.optim\n",
    "                self.optim.step()\n",
    "\n",
    "                epoch_train_loss += loss.item()\n",
    "                epoch_train_steps += 1\n",
    "            \n",
    "            # validation\n",
    "            self.net.eval()\n",
    "            for data in val_loader:\n",
    "                X = data[0].to(device)\n",
    "                y = data[1].to(device)\n",
    "        \n",
    "                out = self.net(X)\n",
    "                loss = self.loss_function(out, y)\n",
    "                epoch_val_loss += loss.item()\n",
    "                epoch_val_steps += 1\n",
    "\n",
    "            # average loss of epoch\n",
    "            train_losses.append(epoch_train_loss / epoch_train_steps)\n",
    "            print(\"epoch [%d]: train loss %.3f\" % (epoch+1, train_losses[-1]))\n",
    "            val_losses.append(epoch_val_loss / epoch_val_steps)\n",
    "            print(\"epoch [%d]: val loss %.3f\" % (epoch+1, val_losses[-1]))\n",
    "        # return train_losses\n",
    "        return train_losses, val_losses"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "epoch [1]: train loss 0.987\n",
      "epoch [1]: val loss 0.989\n",
      "1\n",
      "epoch [2]: train loss 0.938\n",
      "epoch [2]: val loss 0.903\n",
      "2\n",
      "epoch [3]: train loss 0.825\n",
      "epoch [3]: val loss 0.769\n",
      "3\n",
      "epoch [4]: train loss 0.728\n",
      "epoch [4]: val loss 0.718\n",
      "4\n",
      "epoch [5]: train loss 0.705\n",
      "epoch [5]: val loss 0.708\n",
      "5\n",
      "epoch [6]: train loss 0.696\n",
      "epoch [6]: val loss 0.702\n",
      "6\n",
      "epoch [7]: train loss 0.692\n",
      "epoch [7]: val loss 0.699\n",
      "7\n",
      "epoch [8]: train loss 0.688\n",
      "epoch [8]: val loss 0.696\n",
      "8\n",
      "epoch [9]: train loss 0.686\n",
      "epoch [9]: val loss 0.693\n",
      "9\n",
      "epoch [10]: train loss 0.683\n",
      "epoch [10]: val loss 0.691\n",
      "10\n",
      "epoch [11]: train loss 0.680\n",
      "epoch [11]: val loss 0.688\n",
      "11\n",
      "epoch [12]: train loss 0.677\n",
      "epoch [12]: val loss 0.685\n",
      "12\n",
      "epoch [13]: train loss 0.674\n",
      "epoch [13]: val loss 0.682\n",
      "13\n",
      "epoch [14]: train loss 0.671\n",
      "epoch [14]: val loss 0.679\n",
      "14\n",
      "epoch [15]: train loss 0.667\n",
      "epoch [15]: val loss 0.676\n",
      "15\n",
      "epoch [16]: train loss 0.663\n",
      "epoch [16]: val loss 0.673\n",
      "16\n",
      "epoch [17]: train loss 0.658\n",
      "epoch [17]: val loss 0.669\n",
      "17\n",
      "epoch [18]: train loss 0.652\n",
      "epoch [18]: val loss 0.663\n",
      "18\n",
      "epoch [19]: train loss 0.645\n",
      "epoch [19]: val loss 0.656\n",
      "19\n",
      "epoch [20]: train loss 0.638\n",
      "epoch [20]: val loss 0.648\n",
      "20\n",
      "epoch [21]: train loss 0.628\n",
      "epoch [21]: val loss 0.637\n",
      "21\n",
      "epoch [22]: train loss 0.616\n",
      "epoch [22]: val loss 0.624\n",
      "22\n",
      "epoch [23]: train loss 0.602\n",
      "epoch [23]: val loss 0.610\n",
      "23\n",
      "epoch [24]: train loss 0.585\n",
      "epoch [24]: val loss 0.592\n",
      "24\n",
      "epoch [25]: train loss 0.567\n",
      "epoch [25]: val loss 0.572\n",
      "25\n",
      "epoch [26]: train loss 0.545\n",
      "epoch [26]: val loss 0.550\n",
      "26\n",
      "epoch [27]: train loss 0.525\n",
      "epoch [27]: val loss 0.531\n",
      "27\n",
      "epoch [28]: train loss 0.507\n",
      "epoch [28]: val loss 0.514\n",
      "28\n",
      "epoch [29]: train loss 0.492\n",
      "epoch [29]: val loss 0.499\n",
      "29\n",
      "epoch [30]: train loss 0.481\n",
      "epoch [30]: val loss 0.488\n",
      "30\n",
      "epoch [31]: train loss 0.471\n",
      "epoch [31]: val loss 0.478\n",
      "31\n",
      "epoch [32]: train loss 0.463\n",
      "epoch [32]: val loss 0.470\n",
      "32\n",
      "epoch [33]: train loss 0.457\n",
      "epoch [33]: val loss 0.464\n",
      "33\n",
      "epoch [34]: train loss 0.451\n",
      "epoch [34]: val loss 0.458\n",
      "34\n",
      "epoch [35]: train loss 0.446\n",
      "epoch [35]: val loss 0.454\n",
      "35\n",
      "epoch [36]: train loss 0.442\n",
      "epoch [36]: val loss 0.449\n",
      "36\n",
      "epoch [37]: train loss 0.439\n",
      "epoch [37]: val loss 0.446\n",
      "37\n",
      "epoch [38]: train loss 0.436\n",
      "epoch [38]: val loss 0.443\n",
      "38\n",
      "epoch [39]: train loss 0.433\n",
      "epoch [39]: val loss 0.441\n",
      "39\n",
      "epoch [40]: train loss 0.430\n",
      "epoch [40]: val loss 0.439\n",
      "40\n",
      "epoch [41]: train loss 0.428\n",
      "epoch [41]: val loss 0.437\n",
      "41\n",
      "epoch [42]: train loss 0.426\n",
      "epoch [42]: val loss 0.434\n",
      "42\n",
      "epoch [43]: train loss 0.424\n",
      "epoch [43]: val loss 0.434\n",
      "43\n",
      "epoch [44]: train loss 0.422\n",
      "epoch [44]: val loss 0.432\n",
      "44\n",
      "epoch [45]: train loss 0.421\n",
      "epoch [45]: val loss 0.430\n",
      "45\n",
      "epoch [46]: train loss 0.419\n",
      "epoch [46]: val loss 0.430\n",
      "46\n",
      "epoch [47]: train loss 0.418\n",
      "epoch [47]: val loss 0.428\n",
      "47\n",
      "epoch [48]: train loss 0.416\n",
      "epoch [48]: val loss 0.428\n",
      "48\n",
      "epoch [49]: train loss 0.415\n",
      "epoch [49]: val loss 0.426\n",
      "49\n",
      "epoch [50]: train loss 0.414\n",
      "epoch [50]: val loss 0.427\n",
      "50\n",
      "epoch [51]: train loss 0.413\n",
      "epoch [51]: val loss 0.426\n",
      "51\n",
      "epoch [52]: train loss 0.412\n",
      "epoch [52]: val loss 0.424\n",
      "52\n",
      "epoch [53]: train loss 0.411\n",
      "epoch [53]: val loss 0.422\n",
      "53\n",
      "epoch [54]: train loss 0.410\n",
      "epoch [54]: val loss 0.423\n",
      "54\n",
      "epoch [55]: train loss 0.409\n",
      "epoch [55]: val loss 0.421\n",
      "55\n",
      "epoch [56]: train loss 0.408\n",
      "epoch [56]: val loss 0.421\n",
      "56\n",
      "epoch [57]: train loss 0.408\n",
      "epoch [57]: val loss 0.421\n",
      "57\n",
      "epoch [58]: train loss 0.407\n",
      "epoch [58]: val loss 0.418\n",
      "58\n",
      "epoch [59]: train loss 0.406\n",
      "epoch [59]: val loss 0.418\n",
      "59\n",
      "epoch [60]: train loss 0.405\n",
      "epoch [60]: val loss 0.418\n",
      "60\n",
      "epoch [61]: train loss 0.404\n",
      "epoch [61]: val loss 0.417\n",
      "61\n",
      "epoch [62]: train loss 0.403\n",
      "epoch [62]: val loss 0.417\n",
      "62\n",
      "epoch [63]: train loss 0.403\n",
      "epoch [63]: val loss 0.416\n",
      "63\n",
      "epoch [64]: train loss 0.402\n",
      "epoch [64]: val loss 0.416\n",
      "64\n",
      "epoch [65]: train loss 0.401\n",
      "epoch [65]: val loss 0.416\n",
      "65\n",
      "epoch [66]: train loss 0.401\n",
      "epoch [66]: val loss 0.416\n",
      "66\n",
      "epoch [67]: train loss 0.400\n",
      "epoch [67]: val loss 0.416\n",
      "67\n",
      "epoch [68]: train loss 0.399\n",
      "epoch [68]: val loss 0.416\n",
      "68\n",
      "epoch [69]: train loss 0.399\n",
      "epoch [69]: val loss 0.416\n",
      "69\n",
      "epoch [70]: train loss 0.398\n",
      "epoch [70]: val loss 0.414\n",
      "70\n",
      "epoch [71]: train loss 0.397\n",
      "epoch [71]: val loss 0.413\n",
      "71\n",
      "epoch [72]: train loss 0.397\n",
      "epoch [72]: val loss 0.412\n",
      "72\n",
      "epoch [73]: train loss 0.396\n",
      "epoch [73]: val loss 0.412\n",
      "73\n",
      "epoch [74]: train loss 0.395\n",
      "epoch [74]: val loss 0.411\n",
      "74\n",
      "epoch [75]: train loss 0.395\n",
      "epoch [75]: val loss 0.410\n",
      "75\n",
      "epoch [76]: train loss 0.394\n",
      "epoch [76]: val loss 0.411\n",
      "76\n",
      "epoch [77]: train loss 0.394\n",
      "epoch [77]: val loss 0.411\n",
      "77\n",
      "epoch [78]: train loss 0.393\n",
      "epoch [78]: val loss 0.410\n",
      "78\n",
      "epoch [79]: train loss 0.393\n",
      "epoch [79]: val loss 0.409\n",
      "79\n",
      "epoch [80]: train loss 0.392\n",
      "epoch [80]: val loss 0.410\n",
      "80\n",
      "epoch [81]: train loss 0.391\n",
      "epoch [81]: val loss 0.410\n",
      "81\n",
      "epoch [82]: train loss 0.391\n",
      "epoch [82]: val loss 0.409\n",
      "82\n",
      "epoch [83]: train loss 0.390\n",
      "epoch [83]: val loss 0.410\n",
      "83\n",
      "epoch [84]: train loss 0.390\n",
      "epoch [84]: val loss 0.410\n",
      "84\n",
      "epoch [85]: train loss 0.390\n",
      "epoch [85]: val loss 0.409\n",
      "85\n",
      "epoch [86]: train loss 0.389\n",
      "epoch [86]: val loss 0.409\n",
      "86\n",
      "epoch [87]: train loss 0.389\n",
      "epoch [87]: val loss 0.408\n",
      "87\n",
      "epoch [88]: train loss 0.388\n",
      "epoch [88]: val loss 0.407\n",
      "88\n",
      "epoch [89]: train loss 0.388\n",
      "epoch [89]: val loss 0.407\n",
      "89\n",
      "epoch [90]: train loss 0.387\n",
      "epoch [90]: val loss 0.406\n",
      "90\n",
      "epoch [91]: train loss 0.387\n",
      "epoch [91]: val loss 0.406\n",
      "91\n",
      "epoch [92]: train loss 0.386\n",
      "epoch [92]: val loss 0.405\n",
      "92\n",
      "epoch [93]: train loss 0.386\n",
      "epoch [93]: val loss 0.406\n",
      "93\n",
      "epoch [94]: train loss 0.385\n",
      "epoch [94]: val loss 0.406\n",
      "94\n",
      "epoch [95]: train loss 0.385\n",
      "epoch [95]: val loss 0.407\n",
      "95\n",
      "epoch [96]: train loss 0.385\n",
      "epoch [96]: val loss 0.407\n",
      "96\n",
      "epoch [97]: train loss 0.384\n",
      "epoch [97]: val loss 0.404\n",
      "97\n",
      "epoch [98]: train loss 0.384\n",
      "epoch [98]: val loss 0.403\n",
      "98\n",
      "epoch [99]: train loss 0.383\n",
      "epoch [99]: val loss 0.403\n",
      "99\n",
      "epoch [100]: train loss 0.383\n",
      "epoch [100]: val loss 0.403\n",
      "100\n",
      "epoch [101]: train loss 0.382\n",
      "epoch [101]: val loss 0.405\n",
      "101\n",
      "epoch [102]: train loss 0.382\n",
      "epoch [102]: val loss 0.405\n",
      "102\n",
      "epoch [103]: train loss 0.382\n",
      "epoch [103]: val loss 0.405\n",
      "103\n",
      "epoch [104]: train loss 0.381\n",
      "epoch [104]: val loss 0.404\n",
      "104\n",
      "epoch [105]: train loss 0.381\n",
      "epoch [105]: val loss 0.403\n",
      "105\n",
      "epoch [106]: train loss 0.380\n",
      "epoch [106]: val loss 0.404\n",
      "106\n",
      "epoch [107]: train loss 0.380\n",
      "epoch [107]: val loss 0.403\n",
      "107\n",
      "epoch [108]: train loss 0.380\n",
      "epoch [108]: val loss 0.403\n",
      "108\n",
      "epoch [109]: train loss 0.379\n",
      "epoch [109]: val loss 0.402\n",
      "109\n",
      "epoch [110]: train loss 0.379\n",
      "epoch [110]: val loss 0.401\n",
      "110\n",
      "epoch [111]: train loss 0.378\n",
      "epoch [111]: val loss 0.401\n",
      "111\n",
      "epoch [112]: train loss 0.378\n",
      "epoch [112]: val loss 0.399\n",
      "112\n",
      "epoch [113]: train loss 0.377\n",
      "epoch [113]: val loss 0.400\n",
      "113\n",
      "epoch [114]: train loss 0.377\n",
      "epoch [114]: val loss 0.401\n",
      "114\n",
      "epoch [115]: train loss 0.377\n",
      "epoch [115]: val loss 0.400\n",
      "115\n",
      "epoch [116]: train loss 0.376\n",
      "epoch [116]: val loss 0.400\n",
      "116\n",
      "epoch [117]: train loss 0.376\n",
      "epoch [117]: val loss 0.402\n",
      "117\n",
      "epoch [118]: train loss 0.376\n",
      "epoch [118]: val loss 0.401\n",
      "118\n",
      "epoch [119]: train loss 0.375\n",
      "epoch [119]: val loss 0.401\n",
      "119\n",
      "epoch [120]: train loss 0.375\n",
      "epoch [120]: val loss 0.401\n",
      "120\n",
      "epoch [121]: train loss 0.375\n",
      "epoch [121]: val loss 0.399\n",
      "121\n",
      "epoch [122]: train loss 0.374\n",
      "epoch [122]: val loss 0.397\n",
      "122\n",
      "epoch [123]: train loss 0.374\n",
      "epoch [123]: val loss 0.398\n",
      "123\n",
      "epoch [124]: train loss 0.373\n",
      "epoch [124]: val loss 0.398\n",
      "124\n",
      "epoch [125]: train loss 0.373\n",
      "epoch [125]: val loss 0.399\n",
      "125\n"
     ]
    }
   ],
   "source": [
    "### try different learning rates for SGD to see which one works (do not try learning rates greater than 1)\n",
    "### we want the last epoch loss to be less than 0.03\n",
    "learning_rate = 0.00001 # only nu, only sigma, color, gray, constant nu?, constant rho?, constant sigma?, rand 10 by 200\n",
    "# learning_rate = 0.0001 # only rho\n",
    "# learning_rate = 0.000001 # 5 by 400\n",
    "# learning_rate = 0.000001 # 10 by 200, rand 5 by 400\n",
    "\n",
    "net = ConvNet()\n",
    "net = net.to(device)\n",
    "# opt = optim.SGD(net.parameters(), lr=learning_rate, momentum=0.9)\n",
    "opt = optim.Adam(net.parameters(), lr=learning_rate)\n",
    "loss_function = nn.MSELoss()\n",
    "\n",
    "trainer = Trainer(net=net, optim=opt, loss_function=loss_function, train_loader=train_loader, val_loader=val_loader)\n",
    "\n",
    "train_losses, val_losses = trainer.train(num_epochs)\n",
    "# train_losses = trainer.train(num_epochs)\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot losses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "x and y must have same first dimension, but have shapes (500,) and (400,)",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "Cell \u001b[1;32mIn[145], line 4\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[39mimport\u001b[39;00m \u001b[39mmatplotlib\u001b[39;00m\u001b[39m.\u001b[39;00m\u001b[39mpyplot\u001b[39;00m \u001b[39mas\u001b[39;00m \u001b[39mplt\u001b[39;00m\n\u001b[0;32m      2\u001b[0m \u001b[39m### plot the training loss (y-axis) vs epoch number (x-axis)\u001b[39;00m\n\u001b[0;32m      3\u001b[0m \u001b[39m### using the losses computed\u001b[39;00m\n\u001b[1;32m----> 4\u001b[0m plt\u001b[39m.\u001b[39;49mplot(\u001b[39mrange\u001b[39;49m(\u001b[39m1\u001b[39;49m,num_epochs\u001b[39m+\u001b[39;49m\u001b[39m1\u001b[39;49m), train_losses, color \u001b[39m=\u001b[39;49m \u001b[39m'\u001b[39;49m\u001b[39mred\u001b[39;49m\u001b[39m'\u001b[39;49m)\n\u001b[0;32m      5\u001b[0m plt\u001b[39m.\u001b[39mplot(\u001b[39mrange\u001b[39m(\u001b[39m1\u001b[39m,num_epochs\u001b[39m+\u001b[39m\u001b[39m1\u001b[39m), val_losses, color \u001b[39m=\u001b[39m \u001b[39m'\u001b[39m\u001b[39mblue\u001b[39m\u001b[39m'\u001b[39m)\n\u001b[0;32m      6\u001b[0m plt\u001b[39m.\u001b[39mxlabel(\u001b[39m\"\u001b[39m\u001b[39mepoch number\u001b[39m\u001b[39m\"\u001b[39m)\n",
      "File \u001b[1;32mc:\\Users\\benli\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\matplotlib\\pyplot.py:2812\u001b[0m, in \u001b[0;36mplot\u001b[1;34m(scalex, scaley, data, *args, **kwargs)\u001b[0m\n\u001b[0;32m   2810\u001b[0m \u001b[39m@_copy_docstring_and_deprecators\u001b[39m(Axes\u001b[39m.\u001b[39mplot)\n\u001b[0;32m   2811\u001b[0m \u001b[39mdef\u001b[39;00m \u001b[39mplot\u001b[39m(\u001b[39m*\u001b[39margs, scalex\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m, scaley\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m, data\u001b[39m=\u001b[39m\u001b[39mNone\u001b[39;00m, \u001b[39m*\u001b[39m\u001b[39m*\u001b[39mkwargs):\n\u001b[1;32m-> 2812\u001b[0m     \u001b[39mreturn\u001b[39;00m gca()\u001b[39m.\u001b[39mplot(\n\u001b[0;32m   2813\u001b[0m         \u001b[39m*\u001b[39margs, scalex\u001b[39m=\u001b[39mscalex, scaley\u001b[39m=\u001b[39mscaley,\n\u001b[0;32m   2814\u001b[0m         \u001b[39m*\u001b[39m\u001b[39m*\u001b[39m({\u001b[39m\"\u001b[39m\u001b[39mdata\u001b[39m\u001b[39m\"\u001b[39m: data} \u001b[39mif\u001b[39;00m data \u001b[39mis\u001b[39;00m \u001b[39mnot\u001b[39;00m \u001b[39mNone\u001b[39;00m \u001b[39melse\u001b[39;00m {}), \u001b[39m*\u001b[39m\u001b[39m*\u001b[39mkwargs)\n",
      "File \u001b[1;32mc:\\Users\\benli\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\matplotlib\\axes\\_axes.py:1688\u001b[0m, in \u001b[0;36mAxes.plot\u001b[1;34m(self, scalex, scaley, data, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1445\u001b[0m \u001b[39m\u001b[39m\u001b[39m\"\"\"\u001b[39;00m\n\u001b[0;32m   1446\u001b[0m \u001b[39mPlot y versus x as lines and/or markers.\u001b[39;00m\n\u001b[0;32m   1447\u001b[0m \n\u001b[1;32m   (...)\u001b[0m\n\u001b[0;32m   1685\u001b[0m \u001b[39m(``'green'``) or hex strings (``'#008000'``).\u001b[39;00m\n\u001b[0;32m   1686\u001b[0m \u001b[39m\"\"\"\u001b[39;00m\n\u001b[0;32m   1687\u001b[0m kwargs \u001b[39m=\u001b[39m cbook\u001b[39m.\u001b[39mnormalize_kwargs(kwargs, mlines\u001b[39m.\u001b[39mLine2D)\n\u001b[1;32m-> 1688\u001b[0m lines \u001b[39m=\u001b[39m [\u001b[39m*\u001b[39m\u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_get_lines(\u001b[39m*\u001b[39margs, data\u001b[39m=\u001b[39mdata, \u001b[39m*\u001b[39m\u001b[39m*\u001b[39mkwargs)]\n\u001b[0;32m   1689\u001b[0m \u001b[39mfor\u001b[39;00m line \u001b[39min\u001b[39;00m lines:\n\u001b[0;32m   1690\u001b[0m     \u001b[39mself\u001b[39m\u001b[39m.\u001b[39madd_line(line)\n",
      "File \u001b[1;32mc:\\Users\\benli\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\matplotlib\\axes\\_base.py:311\u001b[0m, in \u001b[0;36m_process_plot_var_args.__call__\u001b[1;34m(self, data, *args, **kwargs)\u001b[0m\n\u001b[0;32m    309\u001b[0m     this \u001b[39m+\u001b[39m\u001b[39m=\u001b[39m args[\u001b[39m0\u001b[39m],\n\u001b[0;32m    310\u001b[0m     args \u001b[39m=\u001b[39m args[\u001b[39m1\u001b[39m:]\n\u001b[1;32m--> 311\u001b[0m \u001b[39myield from\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_plot_args(\n\u001b[0;32m    312\u001b[0m     this, kwargs, ambiguous_fmt_datakey\u001b[39m=\u001b[39;49mambiguous_fmt_datakey)\n",
      "File \u001b[1;32mc:\\Users\\benli\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\matplotlib\\axes\\_base.py:504\u001b[0m, in \u001b[0;36m_process_plot_var_args._plot_args\u001b[1;34m(self, tup, kwargs, return_kwargs, ambiguous_fmt_datakey)\u001b[0m\n\u001b[0;32m    501\u001b[0m     \u001b[39mself\u001b[39m\u001b[39m.\u001b[39maxes\u001b[39m.\u001b[39myaxis\u001b[39m.\u001b[39mupdate_units(y)\n\u001b[0;32m    503\u001b[0m \u001b[39mif\u001b[39;00m x\u001b[39m.\u001b[39mshape[\u001b[39m0\u001b[39m] \u001b[39m!=\u001b[39m y\u001b[39m.\u001b[39mshape[\u001b[39m0\u001b[39m]:\n\u001b[1;32m--> 504\u001b[0m     \u001b[39mraise\u001b[39;00m \u001b[39mValueError\u001b[39;00m(\u001b[39mf\u001b[39m\u001b[39m\"\u001b[39m\u001b[39mx and y must have same first dimension, but \u001b[39m\u001b[39m\"\u001b[39m\n\u001b[0;32m    505\u001b[0m                      \u001b[39mf\u001b[39m\u001b[39m\"\u001b[39m\u001b[39mhave shapes \u001b[39m\u001b[39m{\u001b[39;00mx\u001b[39m.\u001b[39mshape\u001b[39m}\u001b[39;00m\u001b[39m and \u001b[39m\u001b[39m{\u001b[39;00my\u001b[39m.\u001b[39mshape\u001b[39m}\u001b[39;00m\u001b[39m\"\u001b[39m)\n\u001b[0;32m    506\u001b[0m \u001b[39mif\u001b[39;00m x\u001b[39m.\u001b[39mndim \u001b[39m>\u001b[39m \u001b[39m2\u001b[39m \u001b[39mor\u001b[39;00m y\u001b[39m.\u001b[39mndim \u001b[39m>\u001b[39m \u001b[39m2\u001b[39m:\n\u001b[0;32m    507\u001b[0m     \u001b[39mraise\u001b[39;00m \u001b[39mValueError\u001b[39;00m(\u001b[39mf\u001b[39m\u001b[39m\"\u001b[39m\u001b[39mx and y can be no greater than 2D, but have \u001b[39m\u001b[39m\"\u001b[39m\n\u001b[0;32m    508\u001b[0m                      \u001b[39mf\u001b[39m\u001b[39m\"\u001b[39m\u001b[39mshapes \u001b[39m\u001b[39m{\u001b[39;00mx\u001b[39m.\u001b[39mshape\u001b[39m}\u001b[39;00m\u001b[39m and \u001b[39m\u001b[39m{\u001b[39;00my\u001b[39m.\u001b[39mshape\u001b[39m}\u001b[39;00m\u001b[39m\"\u001b[39m)\n",
      "\u001b[1;31mValueError\u001b[0m: x and y must have same first dimension, but have shapes (500,) and (400,)"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "### plot the training loss (y-axis) vs epoch number (x-axis)\n",
    "### using the losses computed\n",
    "plt.plot(range(1,num_epochs+1), train_losses, color = 'red')\n",
    "plt.plot(range(1,num_epochs+1), val_losses, color = 'blue')\n",
    "plt.xlabel(\"epoch number\")\n",
    "plt.ylabel(\"training loss\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Test network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###################################################\n",
      "tensor([9.2662e+05, 9.4158e-01, 9.4191e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6500e+06, 1.6467e+00, 1.7421e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9737e+05, 4.8538e-01, 1.8938e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.1151e+05, 9.2789e-01, 2.3247e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5847e+06, 1.5607e+00, 1.7138e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1391e+05, 5.2806e-01, 2.0145e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4673e+06, 2.4614e+00, 4.9709e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0861e+05, 4.9731e-01, 1.7520e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0089e+06, 2.0225e+00, 9.7113e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6863e+05, 9.7303e-01, 1.0366e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1661e+06, 2.1775e+00, 1.5893e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3074e+06, 2.3105e+00, 2.1276e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0304e+06, 2.0307e+00, 1.0493e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.4380e+05, 9.6870e-01, 2.3286e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4668e+06, 2.4629e+00, 1.4227e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2444e+05, 5.3225e-01, 7.7158e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0403e+06, 1.0486e+00, 4.6752e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3174e+06, 2.3344e+00, 2.1677e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9945e+05, 9.9140e-01, 9.6924e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.5089e+06, 2.5241e+00, 4.8103e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5845e+06, 1.5802e+00, 2.4084e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9455e+05, 4.8628e-01, 2.3264e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2008e+06, 2.2047e+00, 2.0763e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9301e+06, 1.9404e+00, 2.0617e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4340e+06, 2.4334e+00, 2.1037e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2788e+06, 2.2828e+00, 4.6934e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2338e+06, 2.2524e+00, 2.4912e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2822e+06, 2.2689e+00, 1.7631e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.4218e+05, 9.4442e-01, 2.6297e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1830e+06, 2.1826e+00, 2.2796e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8304e+06, 1.8218e+00, 2.0942e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0601e+06, 1.0690e+00, 4.9066e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.3677e+05, 4.2182e-01, 1.2728e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9352e+05, 5.0185e-01, 2.3688e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3610e+06, 1.3626e+00, 1.1122e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4967e+06, 1.4900e+00, 2.5260e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1321e+06, 2.1580e+00, 2.4040e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3499e+06, 2.3443e+00, 5.2755e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9243e+05, 4.9908e-01, 9.7339e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3749e+06, 2.3776e+00, 4.9172e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.8476e+05, 8.8108e-01, 4.9812e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9901e+05, 4.8983e-01, 2.2933e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4749e+06, 1.4740e+00, 1.9013e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0812e+06, 1.0910e+00, 2.3476e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2817e+06, 2.2952e+00, 9.7657e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7902e+05, 4.7668e-01, 4.9521e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8346e+05, 4.7165e-01, 1.7940e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2598e+05, 9.3246e-01, 2.1516e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2834e+05, 9.2333e-01, 4.9753e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0465e+06, 1.0409e+00, 2.5263e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3300e+06, 1.3492e+00, 2.1691e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0958e+06, 1.0729e+00, 1.9876e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0882e+06, 2.0989e+00, 1.4599e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3808e+06, 2.3729e+00, 9.6816e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2535e+06, 2.2673e+00, 4.9998e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6937e+06, 1.6978e+00, 1.2659e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1211e+06, 2.1119e+00, 2.1141e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8493e+05, 9.9951e-01, 5.0613e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8524e+06, 1.8524e+00, 1.5338e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0550e+06, 1.0499e+00, 2.1484e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0216e+05, 4.8088e-01, 1.5281e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4539e+06, 1.4494e+00, 4.9961e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.5357e+06, 2.5482e+00, 5.3913e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1124e+05, 4.9872e-01, 2.0168e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1183e+05, 5.0573e-01, 1.9461e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0169e+06, 1.0219e+00, 5.2243e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1717e+06, 2.1868e+00, 4.7546e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0052e+06, 1.0008e+00, 9.8902e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1166e+06, 2.1179e+00, 2.0760e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1285e+06, 2.1224e+00, 2.4552e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0015e+06, 9.9748e-01, 2.6572e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1925e+05, 5.2079e-01, 1.9380e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8475e+06, 1.8219e+00, 2.0666e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8718e+05, 4.8768e-01, 4.9963e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6863e+06, 1.6796e+00, 2.2095e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8615e+06, 1.8803e+00, 1.2805e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1022e+05, 5.0773e-01, 4.7847e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1964e+06, 2.1920e+00, 9.8525e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4206e+06, 1.4171e+00, 2.3115e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9541e+05, 5.0290e-01, 2.1326e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.1988e+05, 9.2588e-01, 2.6872e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9609e+06, 1.9800e+00, 4.6263e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8845e+06, 1.8854e+00, 1.0206e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4275e+06, 2.4426e+00, 4.4908e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7874e+05, 9.7653e-01, 1.4898e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8028e+05, 4.7160e-01, 2.2935e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6297e+06, 1.6287e+00, 1.3317e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5008e+06, 1.4649e+00, 9.3734e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2952e+06, 2.3182e+00, 2.2081e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1087e+05, 4.9561e-01, 2.2996e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.4146e+05, 5.3720e-01, 1.9646e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2409e+06, 2.2470e+00, 4.6300e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3534e+06, 2.3347e+00, 9.9682e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9608e+05, 9.7685e-01, 1.6463e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9216e+06, 1.9263e+00, 4.6899e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4640e+06, 2.4553e+00, 5.0407e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3200e+06, 2.3013e+00, 9.3949e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4613e+06, 2.4606e+00, 2.3183e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4974e+06, 1.4865e+00, 1.5363e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.4663e+05, 9.7093e-01, 1.7983e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9603e+05, 4.8044e-01, 1.7801e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9024e+06, 1.9192e+00, 9.4730e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1964e+05, 5.3156e-01, 1.0088e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1795e+05, 5.0262e-01, 1.9374e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9526e+06, 1.9379e+00, 1.8684e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0906e+06, 2.0946e+00, 2.3795e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0183e+06, 2.0056e+00, 9.1620e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.6691e+05, 4.5272e-01, 2.6093e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1731e+06, 2.1496e+00, 1.7207e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8162e+06, 1.8302e+00, 1.4809e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7273e+06, 1.7327e+00, 2.2833e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5297e+06, 1.5094e+00, 1.8575e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0985e+06, 2.1093e+00, 5.1284e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0261e+06, 1.0254e+00, 4.5493e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3821e+06, 2.3855e+00, 2.1733e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9646e+05, 9.8627e-01, 1.9620e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8789e+06, 1.8924e+00, 9.5580e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9526e+06, 1.9612e+00, 9.6925e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8172e+05, 9.6286e-01, 1.0864e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7058e+06, 1.6909e+00, 1.2949e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2950e+06, 2.2848e+00, 4.6328e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4749e+06, 2.4820e+00, 5.3237e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8416e+05, 4.8377e-01, 4.4625e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0023e+06, 1.0222e+00, 4.6008e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0833e+06, 2.0861e+00, 1.5497e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8198e+05, 4.8198e-01, 4.7575e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0498e+05, 5.0799e-01, 9.6814e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4513e+06, 1.4551e+00, 9.5113e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0082e+06, 1.0078e+00, 5.2303e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9812e+05, 4.8012e-01, 1.1433e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8505e+05, 4.7084e-01, 9.8876e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4146e+06, 1.3935e+00, 1.5557e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5508e+06, 1.5582e+00, 1.8327e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8313e+06, 1.8365e+00, 1.0461e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7468e+06, 1.7459e+00, 1.7289e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2984e+05, 5.4529e-01, 2.2412e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2247e+05, 5.1440e-01, 4.8687e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9229e+05, 4.8798e-01, 4.9931e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.0701e+05, 9.2947e-01, 2.3755e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.7951e+05, 5.5899e-01, 2.3159e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4512e+06, 1.4678e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2329e+05, 9.3510e-01, 1.1532e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8746e+05, 9.9512e-01, 1.5187e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1227e+05, 5.1981e-01, 5.8886e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1141e+05, 4.9795e-01, 2.3487e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([6.2481e+05, 6.3152e-01, 6.7928e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9888e+05, 5.0990e-01, 2.1832e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9818e+05, 4.7738e-01, 2.5027e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.4060e+05, 9.4921e-01, 2.5202e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6305e+05, 9.7857e-01, 9.8987e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5248e+06, 1.5277e+00, 4.5556e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2837e+06, 2.2865e+00, 4.6350e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8035e+05, 4.7622e-01, 2.4018e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9650e+06, 1.9621e+00, 5.0728e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2440e+05, 5.1273e-01, 5.3852e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8856e+06, 1.8889e+00, 1.5962e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4831e+06, 1.4947e+00, 9.8774e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3423e+06, 1.3493e+00, 2.3755e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9327e+05, 4.8863e-01, 1.4509e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2211e+06, 2.2160e+00, 2.2843e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4311e+06, 1.4450e+00, 4.9137e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2946e+06, 2.2924e+00, 2.1131e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4515e+06, 1.4750e+00, 2.3131e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7339e+06, 1.7354e+00, 1.3943e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0416e+06, 2.0466e+00, 1.0139e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8258e+06, 1.7996e+00, 5.2122e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4090e+06, 1.4405e+00, 1.0585e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2899e+06, 2.2965e+00, 2.4837e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9438e+06, 1.9497e+00, 9.7495e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9299e+05, 4.7518e-01, 2.3971e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2778e+05, 9.3908e-01, 1.6658e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4768e+06, 1.4703e+00, 1.9922e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2948e+06, 2.2915e+00, 2.2121e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2926e+06, 2.2902e+00, 9.8305e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4543e+06, 2.4630e+00, 1.0148e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9487e+05, 4.8618e-01, 5.0727e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3662e+06, 2.3613e+00, 9.6695e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1998e+06, 2.1787e+00, 1.9984e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4318e+06, 1.4214e+00, 9.3522e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9825e+05, 4.9080e-01, 2.2083e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1580e+06, 2.1419e+00, 2.4335e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3447e+06, 2.3502e+00, 9.8873e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1369e+06, 2.1372e+00, 2.2382e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8023e+06, 1.8057e+00, 2.2203e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4475e+06, 1.4334e+00, 1.7896e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.9068e+05, 9.0028e-01, 1.9983e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8272e+06, 1.8390e+00, 2.0938e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5198e+06, 1.5366e+00, 2.4169e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.5715e+06, 2.5443e+00, 2.0577e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2266e+06, 2.1978e+00, 9.5894e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9407e+05, 4.8146e-01, 9.7812e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0009e+06, 1.0102e+00, 1.0243e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.3434e+05, 9.5091e-01, 2.1759e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0673e+06, 1.0580e+00, 1.4423e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0888e+06, 1.0835e+00, 9.5398e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5368e+06, 1.5284e+00, 1.9121e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3985e+06, 1.4069e+00, 2.3767e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4584e+06, 1.4821e+00, 2.4056e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3298e+06, 2.3650e+00, 9.8845e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4771e+06, 2.4567e+00, 1.8788e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.5552e+05, 9.5254e-01, 2.1344e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4140e+06, 1.4005e+00, 1.5258e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5600e+06, 1.5706e+00, 2.2648e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0499e+06, 1.0452e+00, 9.0898e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1600e+05, 5.1564e-01, 2.3453e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7659e+06, 1.7553e+00, 1.6407e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9488e+05, 1.0069e+00, 2.1743e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9543e+06, 1.9601e+00, 1.5102e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9865e+05, 4.9607e-01, 1.7145e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1641e+06, 2.1656e+00, 2.2587e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0310e+06, 2.0460e+00, 2.3048e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0930e+06, 2.1023e+00, 2.3416e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6205e+06, 1.6228e+00, 5.0982e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9550e+06, 1.9510e+00, 9.5879e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.2992e+06, 1.3186e+00, 1.0561e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1738e+05, 5.1695e-01, 1.9252e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0772e+06, 1.0793e+00, 5.3019e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0244e+06, 1.0267e+00, 2.0810e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4306e+06, 2.4197e+00, 1.6274e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1432e+06, 2.1504e+00, 2.1830e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1091e+05, 5.1245e-01, 2.3433e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0687e+06, 2.0692e+00, 5.0462e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0728e+05, 4.9799e-01, 1.8274e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5206e+06, 1.5222e+00, 1.9461e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1563e+06, 2.1444e+00, 5.0678e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3822e+06, 1.3799e+00, 2.4767e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5169e+06, 1.5176e+00, 4.9860e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0597e+05, 5.2419e-01, 1.3536e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0169e+06, 2.0185e+00, 2.4702e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.2408e+06, 1.2391e+00, 5.1640e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3875e+06, 2.3863e+00, 9.4508e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9530e+06, 1.9646e+00, 1.0193e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0166e+06, 1.0225e+00, 1.0776e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7915e+05, 4.7743e-01, 1.3757e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6790e+05, 9.5287e-01, 1.9596e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9623e+05, 1.0032e+00, 1.8228e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2236e+05, 5.1663e-01, 4.8356e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4143e+06, 1.3675e+00, 1.6008e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1518e+06, 2.1745e+00, 2.4268e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8378e+06, 1.8385e+00, 1.5659e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1497e+06, 2.1567e+00, 2.4616e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1438e+06, 2.1331e+00, 1.4164e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9105e+05, 4.9375e-01, 4.6941e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1397e+06, 2.1301e+00, 2.3419e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4747e+06, 1.4539e+00, 1.5885e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6173e+06, 1.6021e+00, 4.7841e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.8894e+05, 9.3552e-01, 1.6661e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5484e+06, 1.5518e+00, 2.2843e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7400e+05, 4.6866e-01, 2.2232e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.5599e+06, 2.5743e+00, 5.2891e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9573e+05, 9.9297e-01, 2.4827e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7029e+05, 4.7031e-01, 1.9174e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.2936e+06, 1.3000e+00, 1.0556e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0782e+05, 5.0496e-01, 4.7960e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0447e+06, 1.0453e+00, 4.9351e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1791e+06, 2.1734e+00, 2.5337e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0496e+06, 1.0429e+00, 4.9082e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1386e+05, 5.0959e-01, 2.1110e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.5684e+05, 9.6462e-01, 1.7564e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.4935e+05, 9.5270e-01, 5.0994e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9449e+05, 4.9544e-01, 4.9606e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2883e+06, 2.3230e+00, 9.9582e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7448e+05, 4.8046e-01, 4.7343e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4154e+06, 1.4271e+00, 9.8838e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.6943e+05, 4.7305e-01, 1.6468e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([7.6724e+05, 7.8960e-01, 6.6275e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5580e+06, 1.5694e+00, 2.0565e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.9030e+05, 5.6205e-01, 2.5353e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3393e+06, 1.3404e+00, 5.0755e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0037e+06, 9.9295e-01, 2.4525e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9715e+05, 4.9478e-01, 4.6438e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2807e+05, 5.4592e-01, 1.2408e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3669e+06, 2.3813e+00, 4.5708e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9972e+06, 1.9773e+00, 1.9022e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1385e+06, 2.1414e+00, 2.3847e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0300e+06, 2.0221e+00, 4.7790e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.1689e+06, 1.1697e+00, 2.4643e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4949e+06, 1.4826e+00, 2.5632e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8384e+06, 1.8181e+00, 4.7697e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0369e+06, 1.0435e+00, 1.3544e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6661e+06, 1.6544e+00, 1.7202e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2185e+06, 2.2425e+00, 1.0334e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3720e+06, 2.3679e+00, 2.4518e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2669e+06, 2.2707e+00, 1.9513e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5662e+06, 1.5385e+00, 9.1407e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5175e+06, 1.5067e+00, 5.7448e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9924e+05, 5.0895e-01, 1.3505e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.6602e+05, 5.6640e-01, 4.9723e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1148e+05, 5.0215e-01, 2.1787e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4716e+06, 1.4838e+00, 1.0019e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8904e+05, 4.8680e-01, 4.9399e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7036e+05, 4.6907e-01, 2.5048e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4427e+06, 1.4449e+00, 2.5920e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6898e+05, 9.6756e-01, 2.2717e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0948e+05, 5.0125e-01, 4.7804e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7970e+05, 9.8329e-01, 1.6466e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0224e+05, 4.9341e-01, 2.5515e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6376e+05, 9.7543e-01, 5.0138e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1353e+05, 5.2233e-01, 9.3635e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3191e+06, 1.3233e+00, 1.0187e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3619e+06, 1.3834e+00, 9.7540e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8321e+05, 9.9026e-01, 5.1246e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0853e+06, 2.0809e+00, 1.7566e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0927e+05, 5.0101e-01, 5.0156e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3582e+06, 1.3758e+00, 1.0555e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0580e+05, 5.1045e-01, 2.1486e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7948e+05, 4.9576e-01, 4.6408e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.0140e+05, 9.2949e-01, 2.3968e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.0259e+05, 9.3424e-01, 1.6247e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6494e+06, 1.6219e+00, 4.7426e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3464e+06, 2.3385e+00, 2.1200e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5564e+06, 1.5695e+00, 2.2988e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0327e+05, 4.9625e-01, 2.0778e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5492e+06, 1.5517e+00, 2.1153e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0291e+06, 1.0279e+00, 1.0066e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.4403e+05, 8.6827e-01, 1.2487e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4621e+06, 1.4609e+00, 5.0451e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2271e+06, 2.2299e+00, 2.2417e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8076e+05, 4.7535e-01, 8.7217e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3927e+06, 1.3864e+00, 4.7687e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7119e+05, 1.0255e+00, 2.4699e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2875e+06, 2.3030e+00, 1.3379e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4587e+06, 2.4227e+00, 2.0411e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0242e+06, 2.0053e+00, 1.0044e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0587e+06, 2.0561e+00, 4.8913e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6751e+05, 9.9491e-01, 2.1299e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3157e+06, 2.3181e+00, 2.1583e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5543e+06, 1.5442e+00, 4.9677e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0551e+06, 1.0419e+00, 1.6026e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6286e+05, 9.8012e-01, 1.0052e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4743e+06, 2.4619e+00, 4.9004e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7961e+05, 9.8675e-01, 5.0843e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.9644e+05, 9.1959e-01, 1.0235e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8108e+05, 4.6818e-01, 1.4253e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2791e+05, 9.5007e-01, 2.2285e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3141e+06, 2.3206e+00, 2.0367e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1843e+06, 2.1790e+00, 1.7953e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2147e+06, 2.2259e+00, 1.5155e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.6110e+05, 4.5473e-01, 9.1987e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9570e+05, 4.8434e-01, 4.9637e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.5065e+05, 9.5537e-01, 1.8639e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4537e+06, 1.4530e+00, 2.4983e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7319e+05, 1.0005e+00, 1.4640e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0259e+06, 9.9639e-01, 9.2987e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0121e+05, 5.0600e-01, 2.0752e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8525e+06, 1.8350e+00, 4.7913e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6341e+06, 1.6204e+00, 1.9567e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.9641e+05, 8.9864e-01, 5.1741e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7725e+05, 4.7877e-01, 1.8854e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5621e+06, 1.5619e+00, 2.4573e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1462e+05, 5.1391e-01, 1.5719e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9355e+06, 1.9039e+00, 1.8890e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9316e+05, 5.0771e-01, 5.2352e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5539e+06, 1.5667e+00, 2.1751e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4211e+06, 2.4121e+00, 1.4158e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1491e+06, 2.1255e+00, 9.4517e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0117e+06, 2.0043e+00, 9.9129e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7856e+05, 4.9292e-01, 1.7380e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4799e+06, 2.4756e+00, 4.9404e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3590e+06, 1.3426e+00, 1.8833e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2258e+05, 9.4018e-01, 2.1817e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3139e+06, 2.3107e+00, 1.5268e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8939e+06, 1.8760e+00, 2.0463e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0288e+06, 1.0135e+00, 1.5543e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7675e+05, 4.8145e-01, 2.3639e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0248e+06, 2.0309e+00, 1.0132e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3525e+06, 1.3377e+00, 1.6829e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.4805e+05, 9.3999e-01, 1.0203e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4329e+06, 2.4611e+00, 1.0647e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8827e+05, 4.9389e-01, 1.8912e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4580e+06, 2.4595e+00, 1.4868e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.5157e+05, 9.6430e-01, 1.9519e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([6.3860e+05, 6.4611e-01, 8.1081e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6176e+06, 1.6190e+00, 1.8726e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7423e+05, 9.7384e-01, 1.8740e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0885e+06, 2.1045e+00, 4.3549e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1102e+06, 2.1149e+00, 1.4426e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0715e+06, 1.0691e+00, 1.3486e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5525e+06, 1.5519e+00, 9.7996e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.6439e+05, 4.8363e-01, 1.9723e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4083e+06, 1.3960e+00, 4.8174e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4358e+06, 2.4573e+00, 1.3961e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8036e+05, 4.7752e-01, 4.8503e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0712e+06, 2.0923e+00, 1.3479e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1399e+06, 2.1186e+00, 2.5817e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3995e+06, 2.4135e+00, 2.4379e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3412e+06, 2.3429e+00, 1.9545e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8639e+05, 4.7327e-01, 2.0094e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9526e+05, 4.8087e-01, 4.8243e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6128e+06, 1.6191e+00, 1.3058e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9563e+05, 4.9720e-01, 1.1225e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1577e+06, 2.1795e+00, 4.9000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3211e+06, 2.3173e+00, 9.5044e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9891e+06, 1.9748e+00, 2.0786e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3158e+06, 2.3418e+00, 4.3284e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7623e+06, 1.7525e+00, 1.4658e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9801e+05, 1.0084e+00, 1.5810e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8053e+05, 9.8224e-01, 5.0173e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1039e+06, 2.0858e+00, 1.8399e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8574e+05, 9.8668e-01, 1.4860e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4755e+06, 2.4693e+00, 4.8510e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8884e+06, 1.8865e+00, 1.8031e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6840e+05, 9.6021e-01, 9.7671e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2254e+06, 2.2190e+00, 2.1810e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0248e+06, 1.0174e+00, 2.0162e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6263e+06, 1.6300e+00, 1.8569e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9037e+05, 4.8705e-01, 4.6664e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0596e+06, 2.0728e+00, 1.5014e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0328e+06, 2.0301e+00, 1.5487e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9237e+05, 4.8295e-01, 2.4591e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5202e+06, 1.5231e+00, 2.1621e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4546e+06, 2.4856e+00, 2.5035e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4271e+06, 1.4398e+00, 4.4966e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9813e+05, 4.9794e-01, 1.8745e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0258e+06, 2.0242e+00, 1.4391e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5215e+06, 1.5304e+00, 2.2825e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4500e+06, 2.4480e+00, 4.8348e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4300e+06, 1.4398e+00, 2.0325e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.0670e+05, 9.1855e-01, 2.2131e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3947e+06, 2.3955e+00, 4.3241e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0283e+06, 2.0254e+00, 1.8778e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.7899e+05, 8.7681e-01, 1.0457e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4336e+06, 2.4442e+00, 2.3575e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0414e+06, 1.0558e+00, 2.3498e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7652e+05, 9.5859e-01, 1.0898e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3965e+06, 2.4342e+00, 2.3445e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0694e+06, 1.0402e+00, 1.6771e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3187e+06, 2.3121e+00, 1.4645e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8163e+05, 4.8333e-01, 1.1697e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4460e+06, 1.4428e+00, 2.4181e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9147e+06, 1.9164e+00, 4.6737e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9820e+05, 4.9873e-01, 1.5667e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8737e+05, 4.8611e-01, 5.0180e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3755e+06, 2.3663e+00, 5.0476e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4666e+06, 2.4650e+00, 2.4138e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4168e+06, 2.3989e+00, 1.9977e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3380e+06, 2.3471e+00, 4.4694e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6791e+05, 9.8121e-01, 2.3940e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4940e+06, 1.4780e+00, 5.0433e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.1373e+06, 1.1272e+00, 1.2595e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0027e+06, 1.0035e+00, 5.7424e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3257e+06, 2.3260e+00, 1.9137e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7492e+05, 9.7157e-01, 2.5864e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9202e+06, 1.9114e+00, 2.0909e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9663e+06, 1.9524e+00, 1.4844e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1149e+05, 5.0645e-01, 4.7716e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0641e+06, 2.0561e+00, 9.3453e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0952e+05, 5.2167e-01, 8.5318e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.1163e+06, 1.1186e+00, 2.3017e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.6230e+05, 4.6144e-01, 9.9546e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4800e+06, 1.4537e+00, 1.4024e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4307e+06, 2.4369e+00, 2.4785e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5233e+06, 1.5324e+00, 1.8979e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3692e+06, 1.3547e+00, 9.8112e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9532e+06, 1.9658e+00, 4.8653e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5448e+06, 1.5460e+00, 2.3897e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8408e+06, 1.8383e+00, 1.6179e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3170e+06, 2.3317e+00, 9.7553e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0302e+06, 1.0348e+00, 2.2918e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0099e+06, 2.0110e+00, 5.0061e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0256e+05, 4.8194e-01, 2.6352e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0634e+06, 2.0742e+00, 4.3707e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2251e+06, 2.2282e+00, 5.1682e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.1828e+05, 9.0616e-01, 2.3234e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0702e+05, 4.9907e-01, 2.2976e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0286e+06, 2.0443e+00, 2.4296e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2952e+06, 2.3135e+00, 1.0322e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0934e+06, 2.0675e+00, 1.9768e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6655e+05, 9.8694e-01, 2.5948e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2081e+06, 2.2149e+00, 1.7025e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5882e+06, 1.5730e+00, 2.1778e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9239e+05, 4.8566e-01, 9.8279e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8886e+05, 1.0115e+00, 2.2264e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3154e+06, 1.3005e+00, 1.8320e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.3659e+05, 9.4720e-01, 1.0105e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1654e+06, 2.1634e+00, 4.8503e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.0387e+05, 9.4034e-01, 2.1415e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3474e+06, 1.3592e+00, 9.8946e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3195e+06, 2.3359e+00, 2.2826e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9712e+06, 1.9843e+00, 1.4411e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.9663e+05, 9.1145e-01, 2.5020e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.9535e+05, 9.1905e-01, 2.5029e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.5313e+05, 9.6447e-01, 2.1139e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3156e+06, 2.3089e+00, 9.7304e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4180e+06, 2.4240e+00, 1.0158e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3565e+06, 1.3244e+00, 1.7079e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2475e+05, 5.0042e-01, 2.4253e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.6133e+05, 9.6212e-01, 2.1729e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3799e+06, 1.3776e+00, 1.7467e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3446e+06, 2.3609e+00, 1.6057e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7909e+05, 9.8519e-01, 4.4502e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9617e+05, 5.0223e-01, 1.6382e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2251e+05, 9.3088e-01, 1.2010e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3660e+06, 2.3551e+00, 2.0591e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0397e+05, 4.9925e-01, 1.6381e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9733e+05, 5.0876e-01, 1.7584e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0430e+06, 1.0427e+00, 4.7957e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.8865e+06, 1.8710e+00, 2.5864e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1230e+06, 2.1260e+00, 2.2617e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3508e+06, 2.3603e+00, 2.5722e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0022e+06, 1.0078e+00, 2.0221e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5759e+06, 1.5889e+00, 1.4166e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3331e+06, 2.3303e+00, 9.5830e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5231e+06, 1.5053e+00, 9.6330e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.1460e+06, 1.1410e+00, 4.8948e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4588e+06, 2.4709e+00, 2.4812e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2059e+06, 2.1947e+00, 2.1499e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9343e+05, 9.9369e-01, 1.0845e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3589e+06, 2.3686e+00, 9.6446e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.1065e+06, 1.1349e+00, 5.9115e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3394e+06, 2.3725e+00, 1.0173e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3039e+06, 2.3099e+00, 2.2024e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0975e+05, 4.8522e-01, 2.2216e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.5433e+06, 2.5473e+00, 2.3192e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1815e+06, 2.1610e+00, 2.1193e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.6761e+05, 4.9244e-01, 2.0681e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4054e+06, 2.4255e+00, 1.9910e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1263e+05, 5.2499e-01, 2.0644e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1173e+06, 2.1131e+00, 4.8759e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0123e+06, 1.0521e+00, 2.1722e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2593e+05, 5.3475e-01, 2.0866e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1236e+06, 2.1278e+00, 2.1294e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.1512e+05, 8.2365e-01, 1.4014e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4415e+06, 2.4444e+00, 2.4558e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9345e+06, 1.9348e+00, 9.9242e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.9329e+05, 9.9729e-01, 1.9677e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1481e+06, 2.1521e+00, 4.7707e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1989e+06, 2.1995e+00, 5.0407e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0976e+06, 2.1008e+00, 1.3729e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3300e+06, 2.3217e+00, 1.9621e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.1737e+06, 1.1646e+00, 1.7802e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8874e+05, 5.1035e-01, 2.1527e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9385e+06, 1.9372e+00, 2.2435e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4989e+06, 1.4775e+00, 4.7816e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2453e+05, 5.0758e-01, 2.0014e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2193e+05, 5.0934e-01, 1.8032e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0028e+06, 9.8614e-01, 9.7439e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0557e+06, 2.0529e+00, 9.3942e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9606e+05, 4.8986e-01, 4.5464e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1311e+06, 2.1129e+00, 1.9798e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0201e+06, 1.0098e+00, 9.2476e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9669e+05, 4.8623e-01, 1.7662e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0184e+05, 4.9942e-01, 4.9114e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1822e+06, 2.2198e+00, 2.3346e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9958e+06, 1.9836e+00, 4.4716e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5470e+06, 1.5641e+00, 2.1518e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0801e+05, 5.0711e-01, 4.8812e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5652e+06, 1.5666e+00, 1.8556e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4381e+06, 1.4142e+00, 1.4022e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.5738e+05, 4.7915e-01, 1.4954e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1272e+05, 5.1289e-01, 2.1397e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0046e+05, 5.0008e-01, 1.6625e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0563e+06, 1.0603e+00, 5.0745e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1596e+06, 2.1613e+00, 9.4094e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.3100e+05, 9.5727e-01, 2.1996e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9561e+06, 1.9705e+00, 1.4640e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2637e+06, 2.2576e+00, 2.5613e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9133e+06, 1.8953e+00, 2.0485e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3853e+06, 2.3950e+00, 9.9312e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5033e+06, 1.5016e+00, 2.3594e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7759e+06, 1.8010e+00, 1.0160e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9341e+06, 1.9218e+00, 2.1766e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4703e+06, 2.4837e+00, 2.3855e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0918e+05, 5.0083e-01, 1.4801e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0082e+06, 9.8735e-01, 2.3668e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8565e+05, 4.8259e-01, 1.0642e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9799e+06, 1.9973e+00, 1.4442e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1007e+06, 2.0757e+00, 9.5991e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2169e+06, 2.1993e+00, 2.2494e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7109e+05, 4.7992e-01, 1.7921e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3313e+06, 2.3384e+00, 1.5284e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3480e+06, 1.3735e+00, 1.0229e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4179e+06, 1.4139e+00, 4.9244e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4294e+06, 1.4467e+00, 2.1137e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0787e+06, 1.0620e+00, 9.2211e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.7686e+05, 4.7433e-01, 4.6631e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9969e+06, 2.0127e+00, 1.0355e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([7.8434e+05, 7.9786e-01, 5.5375e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.5145e+05, 9.4503e-01, 2.3407e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1425e+06, 2.1442e+00, 9.6395e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.4505e+05, 9.3708e-01, 4.4316e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2779e+06, 2.2882e+00, 5.0319e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2326e+06, 2.2576e+00, 1.0499e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8233e+05, 4.9814e-01, 1.8746e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0900e+06, 2.0949e+00, 1.5555e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9961e+06, 2.0007e+00, 1.6029e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0204e+06, 1.0266e+00, 1.5494e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9304e+06, 1.9201e+00, 9.9229e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.6290e+05, 4.5524e-01, 1.2614e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6000e+06, 1.6048e+00, 2.2549e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0309e+06, 2.0424e+00, 9.6317e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4214e+06, 2.4170e+00, 4.7693e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.9879e+06, 2.0012e+00, 9.7550e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5038e+06, 1.4805e+00, 1.4940e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8611e+05, 4.6311e-01, 2.6085e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4059e+06, 1.4116e+00, 2.0716e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.1730e+05, 9.2260e-01, 1.0262e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.6445e+06, 1.6388e+00, 2.3045e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8997e+05, 4.8739e-01, 2.1382e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9983e+05, 4.8166e-01, 9.9789e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.3501e+06, 1.3433e+00, 2.3614e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2274e+06, 2.2417e+00, 1.5440e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.2280e+06, 2.2175e+00, 1.5627e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.4866e+06, 2.4749e+00, 1.0080e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4815e+06, 1.4881e+00, 4.8626e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.4891e+06, 1.4908e+00, 2.1019e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.1084e+05, 5.1232e-01, 2.4460e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1576e+06, 2.1743e+00, 4.8305e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.0203e+06, 1.0193e+00, 2.3178e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.0215e+05, 4.7623e-01, 2.2571e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8065e+05, 4.9282e-01, 2.2800e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 2.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([5.2182e+05, 5.1996e-01, 5.0145e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([8.9813e+05, 8.8887e-01, 4.7739e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.5774e+06, 1.5610e+00, 1.7937e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.3931e+06, 2.4190e+00, 2.1624e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7049e+06, 1.7033e+00, 5.0559e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.5000e+06, 1.5000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.2523e+05, 9.2785e-01, 4.8401e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.0023e+06, 2.0215e+00, 1.4153e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.8149e+05, 9.8017e-01, 9.7434e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([9.7390e+05, 9.5797e-01, 2.0057e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([1.0000e+06, 1.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8001e+05, 4.9313e-01, 1.5302e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 1.5000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([2.1472e+06, 2.1485e+00, 1.0751e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.5000e+06, 2.5000e+00, 1.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.8943e+05, 4.9296e-01, 4.7793e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([1.7732e+06, 1.7615e+00, 2.2299e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([2.0000e+06, 2.0000e+00, 2.0000e+04], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "###################################################\n",
      "tensor([4.9460e+05, 4.9991e-01, 5.0583e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "tensor([5.0000e+05, 5.0000e-01, 5.0000e+03], device='cuda:0',\n",
      "       dtype=torch.float64)\n",
      "###################################################\n",
      "1250\n",
      "tensor(0.0073, device='cuda:0')\n",
      "tensor(0.0073, device='cuda:0')\n",
      "tensor(0.0102, device='cuda:0')\n",
      "Accuracy of prediction on test (1): 99.63%\n",
      "Accuracy of prediction on test (2): 99.63%\n",
      "Accuracy of prediction on test (3): 99.49%\n"
     ]
    }
   ],
   "source": [
    "hist_array = []\n",
    "out_array = []\n",
    "err1 = 0\n",
    "err2 = 0\n",
    "err3 = 0\n",
    "tot = 0\n",
    "count = 0\n",
    "with torch.no_grad():\n",
    "    for data in test_loader:\n",
    "        # retrieve X and y for this batch, from data, and \n",
    "        # move it to the device you are using (probably the GPU)\n",
    "        X = data[0].to(device)\n",
    "        y = data[1].to(device)\n",
    "\n",
    "        # raw output of network for X\n",
    "        output = net(X)\n",
    "\n",
    "        a = torch.add(torch.mul(output[0], tstd.to(device)), tmean.to(device))\n",
    "        b = torch.add(torch.mul(y[0], tstd.to(device)), tmean.to(device))\n",
    "\n",
    "        print(\"###################################################\")\n",
    "        print(a)\n",
    "        print(b)\n",
    "        # print(output[0])\n",
    "        # print(y[0])\n",
    "        print(\"###################################################\")\n",
    "\n",
    "        hist_array.append(a.cpu().numpy() - b.cpu().numpy())\n",
    "        out_array.append(a.cpu().numpy())\n",
    "        \n",
    "        # let the maximum index be our predicted class\n",
    "        # _, yh = torch.max(output, 1) \n",
    "\n",
    "        # tot will 10,000 at the end, total number of test data\n",
    "        tot += y.size(0)\n",
    "\n",
    "        ## add to err number of missclassification, i.e. number of indices that \n",
    "        ## yh and y are not equal\n",
    "        ## note that y and yh are vectors of size = batch_size = (256 in our case)\n",
    "        # err += (y != yh).sum()\n",
    "\n",
    "        errors = output.sub(y).multiply(output.sub(y)).div(2).sum(axis=0)\n",
    "        err1 += errors[0]\n",
    "        err2 += errors[1]\n",
    "        err3 += errors[2]\n",
    "        count += 1\n",
    "\n",
    "print(tot)\n",
    "print(err1/count)\n",
    "print(err2/count)\n",
    "print(err3/count)\n",
    "\n",
    "print('Accuracy of prediction on test (1): %5.2f%%' % (100-100 * err1 / tot))\n",
    "\n",
    "print('Accuracy of prediction on test (2): %5.2f%%' % (100-100 * err2 / tot))\n",
    "\n",
    "print('Accuracy of prediction on test (3): %5.2f%%' % (100-100 * err3 / tot))\n",
    "\n",
    "###ASSERTS\n",
    "# assert((100-100 * err / tot)>=98)\n",
    "# assert(tot==10*1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "hist_array = np.array(hist_array)\n",
    "a = hist_array[:, 0]\n",
    "\n",
    "fig, ax = plt.subplots(figsize =(10, 7))\n",
    "ax.hist(a)\n",
    " \n",
    "# Show plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "out_array = np.array(out_array)\n",
    "a = out_array[:, 0]\n",
    "\n",
    "fig, ax = plt.subplots(figsize =(10, 7))\n",
    "ax.hist(a)\n",
    " \n",
    "# Show plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = hist_array[:, 1]\n",
    "\n",
    "fig, ax = plt.subplots(figsize =(10, 7))\n",
    "ax.hist(a)\n",
    " \n",
    "# Show plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = out_array[:, 1]\n",
    "\n",
    "fig, ax = plt.subplots(figsize =(10, 7))\n",
    "ax.hist(a)\n",
    " \n",
    "# Show plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = hist_array[:, 2]\n",
    "\n",
    "fig, ax = plt.subplots(figsize =(10, 7))\n",
    "ax.hist(a)\n",
    " \n",
    "# Show plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "a = out_array[:, 2]\n",
    "\n",
    "fig, ax = plt.subplots(figsize =(10, 7))\n",
    "ax.hist(a)\n",
    " \n",
    "# Show plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# os.chdir('../HMEI2023-bxliu-models/')\n",
    "# save_file = 'only_sigma.pt'\n",
    "# torch.save(net.state_dict(), save_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/svg+xml": [
       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n",
       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
       "<!-- Generated by graphviz version 8.0.5 (20230430.1635)\n",
       " -->\n",
       "<!-- Pages: 1 -->\n",
       "<svg width=\"613pt\" height=\"918pt\"\n",
       " viewBox=\"0.00 0.00 612.56 918.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
       "<g id=\"graph0\" class=\"graph\" transform=\"scale(0.834545 0.834545) rotate(0) translate(4 1096)\">\n",
       "<polygon fill=\"white\" stroke=\"none\" points=\"-4,4 -4,-1096 730,-1096 730,4 -4,4\"/>\n",
       "<!-- 2144611127024 -->\n",
       "<g id=\"node1\" class=\"node\">\n",
       "<title>2144611127024</title>\n",
       "<polygon fill=\"#caff70\" stroke=\"black\" points=\"557,-32 499,-32 499,0 557,0 557,-32\"/>\n",
       "<text text-anchor=\"middle\" x=\"528\" y=\"-6.5\" font-family=\"monospace\" font-size=\"10.00\"> (1, 3)</text>\n",
       "</g>\n",
       "<!-- 2144610726480 -->\n",
       "<g id=\"node2\" class=\"node\">\n",
       "<title>2144610726480</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"578,-88 478,-88 478,-68 578,-68 578,-88\"/>\n",
       "<text text-anchor=\"middle\" x=\"528\" y=\"-74.5\" font-family=\"monospace\" font-size=\"10.00\">AddmmBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610726480&#45;&gt;2144611127024 -->\n",
       "<g id=\"edge42\" class=\"edge\">\n",
       "<title>2144610726480&#45;&gt;2144611127024</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M528,-67.62C528,-61.02 528,-51.84 528,-43.11\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"531.5,-43.14 528,-33.14 524.5,-43.14 531.5,-43.14\"/>\n",
       "</g>\n",
       "<!-- 2144610715776 -->\n",
       "<g id=\"node3\" class=\"node\">\n",
       "<title>2144610715776</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"463,-144 363,-144 363,-124 463,-124 463,-144\"/>\n",
       "<text text-anchor=\"middle\" x=\"413\" y=\"-130.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610715776&#45;&gt;2144610726480 -->\n",
       "<g id=\"edge1\" class=\"edge\">\n",
       "<title>2144610715776&#45;&gt;2144610726480</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M433.03,-123.59C450.98,-115.17 477.55,-102.69 498.03,-93.07\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"499.25,-95.9 506.81,-88.48 496.27,-89.56 499.25,-95.9\"/>\n",
       "</g>\n",
       "<!-- 2144611126624 -->\n",
       "<g id=\"node4\" class=\"node\">\n",
       "<title>2144611126624</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"440,-212 386,-212 386,-180 440,-180 440,-212\"/>\n",
       "<text text-anchor=\"middle\" x=\"413\" y=\"-186.5\" font-family=\"monospace\" font-size=\"10.00\"> (3)</text>\n",
       "</g>\n",
       "<!-- 2144611126624&#45;&gt;2144610715776 -->\n",
       "<g id=\"edge2\" class=\"edge\">\n",
       "<title>2144611126624&#45;&gt;2144610715776</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M413,-179.55C413,-172.17 413,-163.24 413,-155.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"416.5,-155.41 413,-145.41 409.5,-155.41 416.5,-155.41\"/>\n",
       "</g>\n",
       "<!-- 2144610726624 -->\n",
       "<g id=\"node5\" class=\"node\">\n",
       "<title>2144610726624</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"575,-144 481,-144 481,-124 575,-124 575,-144\"/>\n",
       "<text text-anchor=\"middle\" x=\"528\" y=\"-130.5\" font-family=\"monospace\" font-size=\"10.00\">ReluBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610726624&#45;&gt;2144610726480 -->\n",
       "<g id=\"edge3\" class=\"edge\">\n",
       "<title>2144610726624&#45;&gt;2144610726480</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M528,-123.59C528,-116.86 528,-107.53 528,-99.15\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"531.5,-99.3 528,-89.3 524.5,-99.3 531.5,-99.3\"/>\n",
       "</g>\n",
       "<!-- 2144610728256 -->\n",
       "<g id=\"node6\" class=\"node\">\n",
       "<title>2144610728256</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"560,-206 460,-206 460,-186 560,-186 560,-206\"/>\n",
       "<text text-anchor=\"middle\" x=\"510\" y=\"-192.5\" font-family=\"monospace\" font-size=\"10.00\">AddmmBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610728256&#45;&gt;2144610726624 -->\n",
       "<g id=\"edge4\" class=\"edge\">\n",
       "<title>2144610728256&#45;&gt;2144610726624</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M512.81,-185.62C515.31,-177.3 519.04,-164.86 522.18,-154.4\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"525.75,-155.7 525.27,-145.11 519.04,-153.69 525.75,-155.7\"/>\n",
       "</g>\n",
       "<!-- 2144610719664 -->\n",
       "<g id=\"node7\" class=\"node\">\n",
       "<title>2144610719664</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"427,-274 327,-274 327,-254 427,-254 427,-274\"/>\n",
       "<text text-anchor=\"middle\" x=\"377\" y=\"-260.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610719664&#45;&gt;2144610728256 -->\n",
       "<g id=\"edge5\" class=\"edge\">\n",
       "<title>2144610719664&#45;&gt;2144610728256</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M396.06,-253.54C418.38,-242.47 455.58,-224.01 481.37,-211.21\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"482.58,-214.02 489.98,-206.44 479.47,-207.75 482.58,-214.02\"/>\n",
       "</g>\n",
       "<!-- 2144611126464 -->\n",
       "<g id=\"node8\" class=\"node\">\n",
       "<title>2144611126464</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"404,-348 350,-348 350,-316 404,-316 404,-348\"/>\n",
       "<text text-anchor=\"middle\" x=\"377\" y=\"-322.5\" font-family=\"monospace\" font-size=\"10.00\"> (128)</text>\n",
       "</g>\n",
       "<!-- 2144611126464&#45;&gt;2144610719664 -->\n",
       "<g id=\"edge6\" class=\"edge\">\n",
       "<title>2144611126464&#45;&gt;2144610719664</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M377,-315.69C377,-306.6 377,-294.95 377,-285.12\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"380.5,-285.32 377,-275.32 373.5,-285.32 380.5,-285.32\"/>\n",
       "</g>\n",
       "<!-- 2144610715056 -->\n",
       "<g id=\"node9\" class=\"node\">\n",
       "<title>2144610715056</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"539,-274 445,-274 445,-254 539,-254 539,-274\"/>\n",
       "<text text-anchor=\"middle\" x=\"492\" y=\"-260.5\" font-family=\"monospace\" font-size=\"10.00\">ReluBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610715056&#45;&gt;2144610728256 -->\n",
       "<g id=\"edge7\" class=\"edge\">\n",
       "<title>2144610715056&#45;&gt;2144610728256</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M494.58,-253.54C497.17,-244.04 501.25,-229.09 504.56,-216.95\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"508.17,-218.01 507.43,-207.44 501.42,-216.17 508.17,-218.01\"/>\n",
       "</g>\n",
       "<!-- 2144610717120 -->\n",
       "<g id=\"node10\" class=\"node\">\n",
       "<title>2144610717120</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"524,-342 424,-342 424,-322 524,-322 524,-342\"/>\n",
       "<text text-anchor=\"middle\" x=\"474\" y=\"-328.5\" font-family=\"monospace\" font-size=\"10.00\">AddmmBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610717120&#45;&gt;2144610715056 -->\n",
       "<g id=\"edge8\" class=\"edge\">\n",
       "<title>2144610717120&#45;&gt;2144610715056</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M476.58,-321.54C479.17,-312.04 483.25,-297.09 486.56,-284.95\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"490.17,-286.01 489.43,-275.44 483.42,-284.17 490.17,-286.01\"/>\n",
       "</g>\n",
       "<!-- 2144610726768 -->\n",
       "<g id=\"node11\" class=\"node\">\n",
       "<title>2144610726768</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"357,-410 257,-410 257,-390 357,-390 357,-410\"/>\n",
       "<text text-anchor=\"middle\" x=\"307\" y=\"-396.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610726768&#45;&gt;2144610717120 -->\n",
       "<g id=\"edge9\" class=\"edge\">\n",
       "<title>2144610726768&#45;&gt;2144610717120</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M330.93,-389.54C359.57,-378.22 407.74,-359.19 440.18,-346.37\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"441.1,-349.37 449.12,-342.44 438.53,-342.86 441.1,-349.37\"/>\n",
       "</g>\n",
       "<!-- 2144611126304 -->\n",
       "<g id=\"node12\" class=\"node\">\n",
       "<title>2144611126304</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"322,-484 268,-484 268,-452 322,-452 322,-484\"/>\n",
       "<text text-anchor=\"middle\" x=\"295\" y=\"-458.5\" font-family=\"monospace\" font-size=\"10.00\"> (512)</text>\n",
       "</g>\n",
       "<!-- 2144611126304&#45;&gt;2144610726768 -->\n",
       "<g id=\"edge10\" class=\"edge\">\n",
       "<title>2144611126304&#45;&gt;2144610726768</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M297.78,-451.69C299.44,-442.6 301.55,-430.95 303.34,-421.12\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"306.96,-421.78 305.31,-411.32 300.07,-420.53 306.96,-421.78\"/>\n",
       "</g>\n",
       "<!-- 2144610718992 -->\n",
       "<g id=\"node13\" class=\"node\">\n",
       "<title>2144610718992</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"482,-410 388,-410 388,-390 482,-390 482,-410\"/>\n",
       "<text text-anchor=\"middle\" x=\"435\" y=\"-396.5\" font-family=\"monospace\" font-size=\"10.00\">ViewBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610718992&#45;&gt;2144610717120 -->\n",
       "<g id=\"edge11\" class=\"edge\">\n",
       "<title>2144610718992&#45;&gt;2144610717120</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M440.59,-389.54C446.38,-379.74 455.59,-364.16 462.87,-351.83\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"466.35,-353.83 468.42,-343.44 460.32,-350.27 466.35,-353.83\"/>\n",
       "</g>\n",
       "<!-- 2144610722160 -->\n",
       "<g id=\"node14\" class=\"node\">\n",
       "<title>2144610722160</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"530,-478 340,-478 340,-458 530,-458 530,-478\"/>\n",
       "<text text-anchor=\"middle\" x=\"435\" y=\"-464.5\" font-family=\"monospace\" font-size=\"10.00\">MaxPool2DWithIndicesBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610722160&#45;&gt;2144610718992 -->\n",
       "<g id=\"edge12\" class=\"edge\">\n",
       "<title>2144610722160&#45;&gt;2144610718992</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M435,-457.54C435,-448.13 435,-433.39 435,-421.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"438.5,-421.44 435,-411.44 431.5,-421.44 438.5,-421.44\"/>\n",
       "</g>\n",
       "<!-- 2144610726816 -->\n",
       "<g id=\"node15\" class=\"node\">\n",
       "<title>2144610726816</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"482,-546 388,-546 388,-526 482,-526 482,-546\"/>\n",
       "<text text-anchor=\"middle\" x=\"435\" y=\"-532.5\" font-family=\"monospace\" font-size=\"10.00\">ReluBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610726816&#45;&gt;2144610722160 -->\n",
       "<g id=\"edge13\" class=\"edge\">\n",
       "<title>2144610726816&#45;&gt;2144610722160</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M435,-525.54C435,-516.13 435,-501.39 435,-489.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"438.5,-489.44 435,-479.44 431.5,-489.44 438.5,-489.44\"/>\n",
       "</g>\n",
       "<!-- 2144610717600 -->\n",
       "<g id=\"node16\" class=\"node\">\n",
       "<title>2144610717600</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"503,-608 367,-608 367,-588 503,-588 503,-608\"/>\n",
       "<text text-anchor=\"middle\" x=\"435\" y=\"-594.5\" font-family=\"monospace\" font-size=\"10.00\">ConvolutionBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610717600&#45;&gt;2144610726816 -->\n",
       "<g id=\"edge14\" class=\"edge\">\n",
       "<title>2144610717600&#45;&gt;2144610726816</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M435,-587.62C435,-579.48 435,-567.39 435,-557.07\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"438.5,-557.11 435,-547.11 431.5,-557.11 438.5,-557.11\"/>\n",
       "</g>\n",
       "<!-- 2144610716496 -->\n",
       "<g id=\"node17\" class=\"node\">\n",
       "<title>2144610716496</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"367,-664 177,-664 177,-644 367,-644 367,-664\"/>\n",
       "<text text-anchor=\"middle\" x=\"272\" y=\"-650.5\" font-family=\"monospace\" font-size=\"10.00\">MaxPool2DWithIndicesBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610716496&#45;&gt;2144610717600 -->\n",
       "<g id=\"edge15\" class=\"edge\">\n",
       "<title>2144610716496&#45;&gt;2144610717600</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M300.39,-643.59C326.88,-634.82 366.61,-621.66 396.05,-611.9\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"396.99,-614.95 405.39,-608.48 394.79,-608.3 396.99,-614.95\"/>\n",
       "</g>\n",
       "<!-- 2144610719088 -->\n",
       "<g id=\"node18\" class=\"node\">\n",
       "<title>2144610719088</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"319,-726 225,-726 225,-706 319,-706 319,-726\"/>\n",
       "<text text-anchor=\"middle\" x=\"272\" y=\"-712.5\" font-family=\"monospace\" font-size=\"10.00\">ReluBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610719088&#45;&gt;2144610716496 -->\n",
       "<g id=\"edge16\" class=\"edge\">\n",
       "<title>2144610719088&#45;&gt;2144610716496</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M272,-705.62C272,-697.48 272,-685.39 272,-675.07\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"275.5,-675.11 272,-665.11 268.5,-675.11 275.5,-675.11\"/>\n",
       "</g>\n",
       "<!-- 2144610716352 -->\n",
       "<g id=\"node19\" class=\"node\">\n",
       "<title>2144610716352</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"340,-788 204,-788 204,-768 340,-768 340,-788\"/>\n",
       "<text text-anchor=\"middle\" x=\"272\" y=\"-774.5\" font-family=\"monospace\" font-size=\"10.00\">ConvolutionBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610716352&#45;&gt;2144610719088 -->\n",
       "<g id=\"edge17\" class=\"edge\">\n",
       "<title>2144610716352&#45;&gt;2144610719088</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M272,-767.62C272,-759.48 272,-747.39 272,-737.07\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"275.5,-737.11 272,-727.11 268.5,-737.11 275.5,-737.11\"/>\n",
       "</g>\n",
       "<!-- 2144610719616 -->\n",
       "<g id=\"node20\" class=\"node\">\n",
       "<title>2144610719616</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"204,-844 14,-844 14,-824 204,-824 204,-844\"/>\n",
       "<text text-anchor=\"middle\" x=\"109\" y=\"-830.5\" font-family=\"monospace\" font-size=\"10.00\">MaxPool2DWithIndicesBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610719616&#45;&gt;2144610716352 -->\n",
       "<g id=\"edge18\" class=\"edge\">\n",
       "<title>2144610719616&#45;&gt;2144610716352</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M137.39,-823.59C163.88,-814.82 203.61,-801.66 233.05,-791.9\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"233.99,-794.95 242.39,-788.48 231.79,-788.3 233.99,-794.95\"/>\n",
       "</g>\n",
       "<!-- 2144610716976 -->\n",
       "<g id=\"node21\" class=\"node\">\n",
       "<title>2144610716976</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"156,-906 62,-906 62,-886 156,-886 156,-906\"/>\n",
       "<text text-anchor=\"middle\" x=\"109\" y=\"-892.5\" font-family=\"monospace\" font-size=\"10.00\">ReluBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610716976&#45;&gt;2144610719616 -->\n",
       "<g id=\"edge19\" class=\"edge\">\n",
       "<title>2144610716976&#45;&gt;2144610719616</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M109,-885.62C109,-877.48 109,-865.39 109,-855.07\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"112.5,-855.11 109,-845.11 105.5,-855.11 112.5,-855.11\"/>\n",
       "</g>\n",
       "<!-- 2144610714480 -->\n",
       "<g id=\"node22\" class=\"node\">\n",
       "<title>2144610714480</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"177,-968 41,-968 41,-948 177,-948 177,-968\"/>\n",
       "<text text-anchor=\"middle\" x=\"109\" y=\"-954.5\" font-family=\"monospace\" font-size=\"10.00\">ConvolutionBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610714480&#45;&gt;2144610716976 -->\n",
       "<g id=\"edge20\" class=\"edge\">\n",
       "<title>2144610714480&#45;&gt;2144610716976</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M109,-947.62C109,-939.48 109,-927.39 109,-917.07\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"112.5,-917.11 109,-907.11 105.5,-917.11 112.5,-917.11\"/>\n",
       "</g>\n",
       "<!-- 2144610720192 -->\n",
       "<g id=\"node23\" class=\"node\">\n",
       "<title>2144610720192</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"100,-1024 0,-1024 0,-1004 100,-1004 100,-1024\"/>\n",
       "<text text-anchor=\"middle\" x=\"50\" y=\"-1010.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610720192&#45;&gt;2144610714480 -->\n",
       "<g id=\"edge21\" class=\"edge\">\n",
       "<title>2144610720192&#45;&gt;2144610714480</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M60.28,-1003.59C68.74,-995.85 80.93,-984.69 91.01,-975.47\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"92.83,-978.63 97.84,-969.3 88.1,-973.47 92.83,-978.63\"/>\n",
       "</g>\n",
       "<!-- 2144611124944 -->\n",
       "<g id=\"node24\" class=\"node\">\n",
       "<title>2144611124944</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"100,-1092 0,-1092 0,-1060 100,-1060 100,-1092\"/>\n",
       "<text text-anchor=\"middle\" x=\"50\" y=\"-1066.5\" font-family=\"monospace\" font-size=\"10.00\"> (32, 1, 5, 5)</text>\n",
       "</g>\n",
       "<!-- 2144611124944&#45;&gt;2144610720192 -->\n",
       "<g id=\"edge22\" class=\"edge\">\n",
       "<title>2144611124944&#45;&gt;2144610720192</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M50,-1059.55C50,-1052.17 50,-1043.24 50,-1035.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"53.5,-1035.41 50,-1025.41 46.5,-1035.41 53.5,-1035.41\"/>\n",
       "</g>\n",
       "<!-- 2144610717312 -->\n",
       "<g id=\"node25\" class=\"node\">\n",
       "<title>2144610717312</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"218,-1024 118,-1024 118,-1004 218,-1004 218,-1024\"/>\n",
       "<text text-anchor=\"middle\" x=\"168\" y=\"-1010.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610717312&#45;&gt;2144610714480 -->\n",
       "<g id=\"edge23\" class=\"edge\">\n",
       "<title>2144610717312&#45;&gt;2144610714480</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M157.72,-1003.59C149.26,-995.85 137.07,-984.69 126.99,-975.47\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"129.9,-973.47 120.16,-969.3 125.17,-978.63 129.9,-973.47\"/>\n",
       "</g>\n",
       "<!-- 2144611125344 -->\n",
       "<g id=\"node26\" class=\"node\">\n",
       "<title>2144611125344</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"195,-1092 141,-1092 141,-1060 195,-1060 195,-1092\"/>\n",
       "<text text-anchor=\"middle\" x=\"168\" y=\"-1066.5\" font-family=\"monospace\" font-size=\"10.00\"> (32)</text>\n",
       "</g>\n",
       "<!-- 2144611125344&#45;&gt;2144610717312 -->\n",
       "<g id=\"edge24\" class=\"edge\">\n",
       "<title>2144611125344&#45;&gt;2144610717312</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M168,-1059.55C168,-1052.17 168,-1043.24 168,-1035.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"171.5,-1035.41 168,-1025.41 164.5,-1035.41 171.5,-1035.41\"/>\n",
       "</g>\n",
       "<!-- 2144610720096 -->\n",
       "<g id=\"node27\" class=\"node\">\n",
       "<title>2144610720096</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"322,-844 222,-844 222,-824 322,-824 322,-844\"/>\n",
       "<text text-anchor=\"middle\" x=\"272\" y=\"-830.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610720096&#45;&gt;2144610716352 -->\n",
       "<g id=\"edge25\" class=\"edge\">\n",
       "<title>2144610720096&#45;&gt;2144610716352</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M272,-823.59C272,-816.86 272,-807.53 272,-799.15\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"275.5,-799.3 272,-789.3 268.5,-799.3 275.5,-799.3\"/>\n",
       "</g>\n",
       "<!-- 2144611125664 -->\n",
       "<g id=\"node28\" class=\"node\">\n",
       "<title>2144611125664</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"325,-912 219,-912 219,-880 325,-880 325,-912\"/>\n",
       "<text text-anchor=\"middle\" x=\"272\" y=\"-886.5\" font-family=\"monospace\" font-size=\"10.00\"> (64, 32, 5, 5)</text>\n",
       "</g>\n",
       "<!-- 2144611125664&#45;&gt;2144610720096 -->\n",
       "<g id=\"edge26\" class=\"edge\">\n",
       "<title>2144611125664&#45;&gt;2144610720096</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M272,-879.55C272,-872.17 272,-863.24 272,-855.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"275.5,-855.41 272,-845.41 268.5,-855.41 275.5,-855.41\"/>\n",
       "</g>\n",
       "<!-- 2144610720528 -->\n",
       "<g id=\"node29\" class=\"node\">\n",
       "<title>2144610720528</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"440,-844 340,-844 340,-824 440,-824 440,-844\"/>\n",
       "<text text-anchor=\"middle\" x=\"390\" y=\"-830.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610720528&#45;&gt;2144610716352 -->\n",
       "<g id=\"edge27\" class=\"edge\">\n",
       "<title>2144610720528&#45;&gt;2144610716352</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M369.45,-823.59C350.95,-815.13 323.52,-802.58 302.47,-792.94\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"304.26,-789.46 293.71,-788.48 301.35,-795.82 304.26,-789.46\"/>\n",
       "</g>\n",
       "<!-- 2144611125904 -->\n",
       "<g id=\"node30\" class=\"node\">\n",
       "<title>2144611125904</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"417,-912 363,-912 363,-880 417,-880 417,-912\"/>\n",
       "<text text-anchor=\"middle\" x=\"390\" y=\"-886.5\" font-family=\"monospace\" font-size=\"10.00\"> (64)</text>\n",
       "</g>\n",
       "<!-- 2144611125904&#45;&gt;2144610720528 -->\n",
       "<g id=\"edge28\" class=\"edge\">\n",
       "<title>2144611125904&#45;&gt;2144610720528</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M390,-879.55C390,-872.17 390,-863.24 390,-855.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"393.5,-855.41 390,-845.41 386.5,-855.41 393.5,-855.41\"/>\n",
       "</g>\n",
       "<!-- 2144610721392 -->\n",
       "<g id=\"node31\" class=\"node\">\n",
       "<title>2144610721392</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"485,-664 385,-664 385,-644 485,-644 485,-664\"/>\n",
       "<text text-anchor=\"middle\" x=\"435\" y=\"-650.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610721392&#45;&gt;2144610717600 -->\n",
       "<g id=\"edge29\" class=\"edge\">\n",
       "<title>2144610721392&#45;&gt;2144610717600</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M435,-643.59C435,-636.86 435,-627.53 435,-619.15\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"438.5,-619.3 435,-609.3 431.5,-619.3 438.5,-619.3\"/>\n",
       "</g>\n",
       "<!-- 2144611126064 -->\n",
       "<g id=\"node32\" class=\"node\">\n",
       "<title>2144611126064</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"491,-732 379,-732 379,-700 491,-700 491,-732\"/>\n",
       "<text text-anchor=\"middle\" x=\"435\" y=\"-706.5\" font-family=\"monospace\" font-size=\"10.00\"> (128, 64, 5, 5)</text>\n",
       "</g>\n",
       "<!-- 2144611126064&#45;&gt;2144610721392 -->\n",
       "<g id=\"edge30\" class=\"edge\">\n",
       "<title>2144611126064&#45;&gt;2144610721392</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M435,-699.55C435,-692.17 435,-683.24 435,-675.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"438.5,-675.41 435,-665.41 431.5,-675.41 438.5,-675.41\"/>\n",
       "</g>\n",
       "<!-- 2144610724416 -->\n",
       "<g id=\"node33\" class=\"node\">\n",
       "<title>2144610724416</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"603,-664 503,-664 503,-644 603,-644 603,-664\"/>\n",
       "<text text-anchor=\"middle\" x=\"553\" y=\"-650.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610724416&#45;&gt;2144610717600 -->\n",
       "<g id=\"edge31\" class=\"edge\">\n",
       "<title>2144610724416&#45;&gt;2144610717600</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M532.45,-643.59C513.95,-635.13 486.52,-622.58 465.47,-612.94\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"467.26,-609.46 456.71,-608.48 464.35,-615.82 467.26,-609.46\"/>\n",
       "</g>\n",
       "<!-- 2144611126144 -->\n",
       "<g id=\"node34\" class=\"node\">\n",
       "<title>2144611126144</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"580,-732 526,-732 526,-700 580,-700 580,-732\"/>\n",
       "<text text-anchor=\"middle\" x=\"553\" y=\"-706.5\" font-family=\"monospace\" font-size=\"10.00\"> (128)</text>\n",
       "</g>\n",
       "<!-- 2144611126144&#45;&gt;2144610724416 -->\n",
       "<g id=\"edge32\" class=\"edge\">\n",
       "<title>2144611126144&#45;&gt;2144610724416</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M553,-699.55C553,-692.17 553,-683.24 553,-675.32\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"556.5,-675.41 553,-665.41 549.5,-675.41 556.5,-675.41\"/>\n",
       "</g>\n",
       "<!-- 2144610718848 -->\n",
       "<g id=\"node35\" class=\"node\">\n",
       "<title>2144610718848</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"576,-410 500,-410 500,-390 576,-390 576,-410\"/>\n",
       "<text text-anchor=\"middle\" x=\"538\" y=\"-396.5\" font-family=\"monospace\" font-size=\"10.00\">TBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610718848&#45;&gt;2144610717120 -->\n",
       "<g id=\"edge33\" class=\"edge\">\n",
       "<title>2144610718848&#45;&gt;2144610717120</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M528.83,-389.54C518.85,-379.25 502.69,-362.58 490.49,-350.01\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"493.63,-348.18 484.15,-343.44 488.6,-353.05 493.63,-348.18\"/>\n",
       "</g>\n",
       "<!-- 2144610722640 -->\n",
       "<g id=\"node36\" class=\"node\">\n",
       "<title>2144610722640</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"648,-478 548,-478 548,-458 648,-458 648,-478\"/>\n",
       "<text text-anchor=\"middle\" x=\"598\" y=\"-464.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610722640&#45;&gt;2144610718848 -->\n",
       "<g id=\"edge34\" class=\"edge\">\n",
       "<title>2144610722640&#45;&gt;2144610718848</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M589.4,-457.54C580.14,-447.35 565.18,-430.9 553.79,-418.37\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"556.9,-416.48 547.58,-411.44 551.72,-421.19 556.9,-416.48\"/>\n",
       "</g>\n",
       "<!-- 2144611126224 -->\n",
       "<g id=\"node37\" class=\"node\">\n",
       "<title>2144611126224</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"645,-552 551,-552 551,-520 645,-520 645,-552\"/>\n",
       "<text text-anchor=\"middle\" x=\"598\" y=\"-526.5\" font-family=\"monospace\" font-size=\"10.00\"> (512, 32768)</text>\n",
       "</g>\n",
       "<!-- 2144611126224&#45;&gt;2144610722640 -->\n",
       "<g id=\"edge35\" class=\"edge\">\n",
       "<title>2144611126224&#45;&gt;2144610722640</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M598,-519.69C598,-510.6 598,-498.95 598,-489.12\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"601.5,-489.32 598,-479.32 594.5,-489.32 601.5,-489.32\"/>\n",
       "</g>\n",
       "<!-- 2144610715008 -->\n",
       "<g id=\"node38\" class=\"node\">\n",
       "<title>2144610715008</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"633,-274 557,-274 557,-254 633,-254 633,-274\"/>\n",
       "<text text-anchor=\"middle\" x=\"595\" y=\"-260.5\" font-family=\"monospace\" font-size=\"10.00\">TBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610715008&#45;&gt;2144610728256 -->\n",
       "<g id=\"edge36\" class=\"edge\">\n",
       "<title>2144610715008&#45;&gt;2144610728256</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M582.82,-253.54C569.19,-242.96 546.87,-225.63 530.53,-212.94\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"533.2,-209.81 523.16,-206.44 528.91,-215.34 533.2,-209.81\"/>\n",
       "</g>\n",
       "<!-- 2144610722736 -->\n",
       "<g id=\"node39\" class=\"node\">\n",
       "<title>2144610722736</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"665,-342 565,-342 565,-322 665,-322 665,-342\"/>\n",
       "<text text-anchor=\"middle\" x=\"615\" y=\"-328.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610722736&#45;&gt;2144610715008 -->\n",
       "<g id=\"edge37\" class=\"edge\">\n",
       "<title>2144610722736&#45;&gt;2144610715008</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M612.13,-321.54C609.25,-312.04 604.72,-297.09 601.04,-284.95\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"604.11,-283.99 597.86,-275.44 597.41,-286.02 604.11,-283.99\"/>\n",
       "</g>\n",
       "<!-- 2144611125984 -->\n",
       "<g id=\"node40\" class=\"node\">\n",
       "<title>2144611125984</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"676,-416 594,-416 594,-384 676,-384 676,-416\"/>\n",
       "<text text-anchor=\"middle\" x=\"635\" y=\"-390.5\" font-family=\"monospace\" font-size=\"10.00\"> (128, 512)</text>\n",
       "</g>\n",
       "<!-- 2144611125984&#45;&gt;2144610722736 -->\n",
       "<g id=\"edge38\" class=\"edge\">\n",
       "<title>2144611125984&#45;&gt;2144610722736</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M630.36,-383.69C627.57,-374.5 624,-362.7 621,-352.8\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"624.07,-351.87 617.82,-343.32 617.37,-353.9 624.07,-351.87\"/>\n",
       "</g>\n",
       "<!-- 2144610718944 -->\n",
       "<g id=\"node41\" class=\"node\">\n",
       "<title>2144610718944</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"705,-144 629,-144 629,-124 705,-124 705,-144\"/>\n",
       "<text text-anchor=\"middle\" x=\"667\" y=\"-130.5\" font-family=\"monospace\" font-size=\"10.00\">TBackward0</text>\n",
       "</g>\n",
       "<!-- 2144610718944&#45;&gt;2144610726480 -->\n",
       "<g id=\"edge39\" class=\"edge\">\n",
       "<title>2144610718944&#45;&gt;2144610726480</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M642.79,-123.59C620.6,-114.97 587.51,-102.12 562.55,-92.42\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"563.99,-88.84 553.4,-88.48 561.45,-95.36 563.99,-88.84\"/>\n",
       "</g>\n",
       "<!-- 2144610722832 -->\n",
       "<g id=\"node42\" class=\"node\">\n",
       "<title>2144610722832</title>\n",
       "<polygon fill=\"lightgrey\" stroke=\"black\" points=\"726,-206 626,-206 626,-186 726,-186 726,-206\"/>\n",
       "<text text-anchor=\"middle\" x=\"676\" y=\"-192.5\" font-family=\"monospace\" font-size=\"10.00\">AccumulateGrad</text>\n",
       "</g>\n",
       "<!-- 2144610722832&#45;&gt;2144610718944 -->\n",
       "<g id=\"edge40\" class=\"edge\">\n",
       "<title>2144610722832&#45;&gt;2144610718944</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M674.59,-185.62C673.36,-177.39 671.52,-165.13 669.96,-154.74\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"673.31,-154.48 668.37,-145.11 666.39,-155.52 673.31,-154.48\"/>\n",
       "</g>\n",
       "<!-- 2144611126544 -->\n",
       "<g id=\"node43\" class=\"node\">\n",
       "<title>2144611126544</title>\n",
       "<polygon fill=\"lightblue\" stroke=\"black\" points=\"721,-280 651,-280 651,-248 721,-248 721,-280\"/>\n",
       "<text text-anchor=\"middle\" x=\"686\" y=\"-254.5\" font-family=\"monospace\" font-size=\"10.00\"> (3, 128)</text>\n",
       "</g>\n",
       "<!-- 2144611126544&#45;&gt;2144610722832 -->\n",
       "<g id=\"edge41\" class=\"edge\">\n",
       "<title>2144611126544&#45;&gt;2144610722832</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M683.68,-247.69C682.3,-238.6 680.54,-226.95 679.05,-217.12\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"682.37,-216.68 677.41,-207.32 675.45,-217.73 682.37,-216.68\"/>\n",
       "</g>\n",
       "</g>\n",
       "</svg>\n"
      ],
      "text/plain": [
       "<graphviz.graphs.Digraph at 0x1f354c29de0>"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data = torch.rand(1,1,128,128)\n",
    "test_net = ConvNet()\n",
    "out = test_net.forward(test_data)\n",
    "make_dot(out)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next Steps:\n",
    "custom loss function (later)\n",
    "documentation (done)\n",
    "validation (done-ish, needs going thru again with all the datasets probably)\n",
    "more data (done)\n",
    "graph outputs not just loss (done)\n",
    "outside training set for 5by400 and 10by200\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(128, 128, 1)\n"
     ]
    }
   ],
   "source": [
    "f = '../HMEI2023-bxliu-data/image_rand_10by200/gentrain_1001.png'\n",
    "data = cv2.cvtColor(cv2.imread(f), cv2.COLOR_BGR2GRAY)\n",
    "data = cv2.resize(data, (64,64), interpolation=cv2.INTER_CUBIC)\n",
    "data = np.expand_dims(data, axis=2)\n",
    "print(data.shape)\n",
    "# out = net.forward()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ConvNet(\n",
      "  (conv1): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
      "  (pool1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (conv2): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
      "  (pool2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (conv3): Conv2d(64, 128, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
      "  (pool3): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  (fc1): Linear(in_features=32768, out_features=512, bias=True)\n",
      "  (fc2): Linear(in_features=512, out_features=128, bias=True)\n",
      "  (fc3): Linear(in_features=128, out_features=3, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "print(net)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def use_net(im_data):\n",
    "    l = []\n",
    "    l.append(im_data)\n",
    "    l = np.asarray(l)\n",
    "    l = np.moveaxis(l, -1, 1)\n",
    "    print(l.shape)\n",
    "    l = torch.tensor(l).float()\n",
    "    l = torch.div(l, 255).float()\n",
    "    l = l.to(device)\n",
    "    print(l.shape)\n",
    "    print(l)\n",
    "    print(l.stride())\n",
    "    \n",
    "\n",
    "    with torch.no_grad():  \n",
    "        test_data = torch.rand(1,1,128,128).to(device)\n",
    "        test_data[0] = l[0]\n",
    "        print(test_data.shape)\n",
    "        print(test_data)\n",
    "        print(test_data.stride())\n",
    "        out = net.forward(test_data)\n",
    "        # out = net.forward(l)\n",
    "        out = torch.add(torch.mul(out[0], tstd.to(device)), tmean.to(device))\n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 1, 128, 128)\n",
      "torch.Size([1, 1, 128, 128])\n",
      "tensor([[[[0.7294, 0.7451, 0.6000,  ..., 0.5216, 0.5569, 0.5529],\n",
      "          [0.7412, 0.7569, 0.6039,  ..., 0.5216, 0.5529, 0.5529],\n",
      "          [0.6314, 0.6392, 0.5686,  ..., 0.5373, 0.5725, 0.5686],\n",
      "          ...,\n",
      "          [0.3137, 0.2902, 0.5216,  ..., 0.3804, 0.8549, 0.8078],\n",
      "          [0.4902, 0.4980, 0.4275,  ..., 0.5373, 0.7176, 0.6980],\n",
      "          [0.4863, 0.4902, 0.4196,  ..., 0.5333, 0.7137, 0.6941]]]],\n",
      "       device='cuda:0')\n",
      "(16384, 1, 128, 1)\n",
      "torch.Size([1, 1, 128, 128])\n",
      "tensor([[[[0.7294, 0.7451, 0.6000,  ..., 0.5216, 0.5569, 0.5529],\n",
      "          [0.7412, 0.7569, 0.6039,  ..., 0.5216, 0.5529, 0.5529],\n",
      "          [0.6314, 0.6392, 0.5686,  ..., 0.5373, 0.5725, 0.5686],\n",
      "          ...,\n",
      "          [0.3137, 0.2902, 0.5216,  ..., 0.3804, 0.8549, 0.8078],\n",
      "          [0.4902, 0.4980, 0.4275,  ..., 0.5373, 0.7176, 0.6980],\n",
      "          [0.4863, 0.4902, 0.4196,  ..., 0.5333, 0.7137, 0.6941]]]],\n",
      "       device='cuda:0')\n",
      "(16384, 16384, 128, 1)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "tensor([2.4816e+06, 2.4836e+00, 4.7060e+03], device='cuda:0',\n",
       "       dtype=torch.float64)"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "use_net(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[3.34650214e+06 2.59287601e+00 4.98000000e+03]\n"
     ]
    }
   ],
   "source": [
    "matfile = '../HMEI2023-bxliu-data/value_rand_10by200/gentrain_1001.mat'\n",
    "mat = scipy.io.loadmat(matfile)\n",
    "print(mat['th0'][0])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.5"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}