04_02_wgan_cifar_train.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# WGAN Horse Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n"
     ]
    }
   ],
   "source": [
    "import torch as t\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm import tqdm\n",
    "from torch.utils.data import DataLoader\n",
    "from models.GAN import build_generator, build_critic\n",
    "from torchvision import datasets, transforms\n",
    "\n",
    "bs = 128\n",
    "z_dim = 100\n",
    "\n",
    "transform = transforms.Compose([transforms.ToTensor()])\n",
    "ds = datasets.CIFAR10(root='./data', download=True, transform=transform)\n",
    "ds.targets = t.tensor(ds.targets)\n",
    "idx = ds.targets == 7  # 7 is the label for horses\n",
    "ds.targets = ds.targets[idx]\n",
    "ds.data = ds.data[idx.numpy().astype(np.bool)]\n",
    "dl = DataLoader(dataset=ds, batch_size=bs, shuffle=True, drop_last=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_tensor(x):\n",
    "    _, ax = plt.subplots(figsize=(3, 3))\n",
    "    def image2np(image):\n",
    "        res = image.permute(1, 2, 0)\n",
    "        return res[..., 0] if res.shape[2] == 1 else res\n",
    "    ax.imshow(image2np(t.from_numpy(x.cpu().detach().numpy())), cmap=\"viridis\", alpha=None)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Sequential(\n",
       "  (0): Conv2d(3, 32, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2))\n",
       "  (1): LeakyReLU(negative_slope=0.2)\n",
       "  (2): Dropout(p=0.4)\n",
       "  (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2))\n",
       "  (4): LeakyReLU(negative_slope=0.2)\n",
       "  (5): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (6): Dropout(p=0.4)\n",
       "  (7): Conv2d(64, 128, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2))\n",
       "  (8): LeakyReLU(negative_slope=0.2)\n",
       "  (9): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (10): Dropout(p=0.4)\n",
       "  (11): Conv2d(128, 128, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))\n",
       "  (12): LeakyReLU(negative_slope=0.2)\n",
       "  (13): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (14): Dropout(p=0.4)\n",
       "  (15): Flatten()\n",
       "  (16): Linear(in_features=2048, out_features=1, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "generator = build_generator(gen_init_size=(128, 4, 4), gen_upsample_flags=[True, True, True, False],\n",
    "                            gen_c=[128, 64, 32, 3], gen_ks=[5, 5, 5, 5], gen_strides=[1, 1, 1, 1],\n",
    "                            gen_pads=[2, 2, 2, 2], z_dim=z_dim, output_act=t.nn.Tanh, \n",
    "                            leaky_relu=True, first_gen_c=128)\n",
    "\n",
    "critic = build_critic(ds[0][0][None], critic_c=[3, 32, 64, 128, 128], critic_ks=[5, 5, 5, 5],\n",
    "                      critic_strides=[2, 2, 2, 1], critic_pads=[2, 2, 2, 2], wgan=True, leaky_relu=True)\n",
    "\n",
    "g_opt = t.optim.RMSprop(generator.parameters(), lr=5e-5)\n",
    "c_opt = t.optim.RMSprop(critic.parameters(), lr=5e-5)\n",
    "device = t.device('cuda') if t.cuda.is_available() else 'cpu'\n",
    "generator.cuda(device)\n",
    "critic.cuda(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/12000 [00:00<?, ?it/s]/home/p/.installed_software/anaconda3/lib/python3.7/site-packages/torch/nn/modules/upsampling.py:129: UserWarning: nn.UpsamplingBilinear2d is deprecated. Use nn.functional.interpolate instead.\n",
      "  warnings.warn(\"nn.{} is deprecated. Use nn.functional.interpolate instead.\".format(self.name))\n",
      "  1%|          | 100/12000 [05:08<10:04:50,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.08996586501598358 -0.047341324388980865 0.14166073501110077\n",
      "gen_imgs.mean 0.45774388313293457\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  2%|▏         | 200/12000 [10:14<10:16:21,  3.13s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.08020076155662537 -0.05226001888513565 0.12244974821805954\n",
      "gen_imgs.mean 0.46653109788894653\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  2%|▎         | 300/12000 [15:20<9:51:12,  3.03s/it] "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.08393855392932892 -0.057059481739997864 0.07831021398305893\n",
      "gen_imgs.mean 0.4719371795654297\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  3%|▎         | 400/12000 [20:23<9:48:24,  3.04s/it] "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.07525461167097092 -0.04008696973323822 0.12553457915782928\n",
      "gen_imgs.mean 0.4631131589412689\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  4%|▍         | 500/12000 [25:28<9:39:42,  3.02s/it] "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.024202944710850716 -0.07060792297124863 0.06444558501243591\n",
      "gen_imgs.mean 0.46151766180992126\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  5%|▌         | 600/12000 [30:31<9:33:56,  3.02s/it] "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.06400828808546066 -0.011739307083189487 0.08050434291362762\n",
      "gen_imgs.mean 0.4617008566856384\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  6%|▌         | 700/12000 [35:38<10:04:13,  3.21s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03313614800572395 -0.07435403019189835 0.06897317618131638\n",
      "gen_imgs.mean 0.46777233481407166\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  7%|▋         | 800/12000 [40:45<9:25:50,  3.03s/it] "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.0008977563120424747 -0.07989709079265594 0.05888156592845917\n",
      "gen_imgs.mean 0.465684175491333\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  8%|▊         | 900/12000 [45:49<9:20:59,  3.03s/it] "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.0014652255922555923 -0.08714429289102554 0.037689194083213806\n",
      "gen_imgs.mean 0.4744569659233093\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  8%|▊         | 999/12000 [50:50<9:14:00,  3.02s/it] Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03520442545413971 -0.04503607377409935 0.05924616754055023\n",
      "gen_imgs.mean 0.4648718535900116\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMUAAADDCAYAAAAyYdXtAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAbfklEQVR4nO2dWYxkZ3XH/6duLd1VvXdPz9o9G2MztrHHxnFMTCTCEjkIySAlCB4iP1jAA0hB4cUiUkKkPBApgPIQERnFwpEISwIEC5kkjgVYhNjYnuBh8Hg8+9ru6a26q6prvycPVW267v/UTLl7pqbbnJ80mu7Td/m+e+vUvWf9RFXhOM5viN3sATjORsOVwnEiuFI4TgRXCseJ4ErhOBFcKRwnwrqUQkQeFJHjInJSRB69XoNynJuJrDVOISIBgNcAfADARQAvAPi4qr7Sbp/R0VGdnJxskdVDPn9MLF0VkryZoXe8LZ/G3szYzpK9mdO8mf07OmJnoo3Hmxi3dVutex39mF04fxZzs7PmYeNXH91VuQ/ASVU9DQAi8i0ADwFoqxSTk5P4yU9+3CJbXA5pu95UL8kC5fFXK/al0tDYtm5sZ+0c8L5iXKW4JTN0OdbmbgbGtnHr3MYgrXGL8UUSM44X6+z75mpiHk+HY7SE1hdBEHS2XbtzV4wv2mKlVfa+3/9d+4BY3+vTTgAXVv1+sSlrQUQ+KSIvisiLc3Oz6zid43SH9SiFpbukoqr6mKreq6r3jo6OreN0jtMd1vP6dBHAxKrfdwG4fLUdYrEYUql0i6xWqtB2lZBfqRIxfqYm02102niVSFjjMcfY0eHMfdu8zJlS67UqZgit7Yy3A4TGe4Q1l8B6pboRhkbH714diUzbEwBqxmtxoczCYr71cxbW+TO2wnqeFC8AOCAie0UkCeBjAJ5cx/EcZ0Ow5ieFqtZE5DMA/hNAAOBxVf31dRuZ49wk1vP6BFV9CsBT12ksjrMh8Ii240RY15PizSIQJOKtBvNohocghlM6Zlhu7XzXMct5bUbbWGSEQ2w3m2UgKhtvYRsDsRrWTHmUngRfnyDOToeEEfhYXzCwS1iOBOOalaq2YVwzLmO+yNtWIgGf9ma2Pykch3ClcJwIrhSOE8GVwnEidNXQVg1RrUQii8qWUiaR6uh4NSucCUCNmGhoJQRaVrUR3rUM1tA6R80wtOu2QV2tlUlWzC/zGA0Dc3Qrp8v0DWSMs3TH0rYzVY3rY2QqhMaNyRUKJCsb1xYA4kEPycSIVg/2tMoCK9OyiT8pHCeCK4XjRHClcJwIrhSOE6GrhnYxX8DRnz/XIpPMAG03sbOfZGror1WpBgBi5EyXy2zw1iVNsqAnyec2DHKts6FWLbPxXC0XzTHG41WSxSpsaFcLSyTLK+8bYCvLkjwXK8BeMsYNANmlRZJZhWJXpq+Q7LUTZ0k2dfkMyeY1T7I6+DqMbdtujlHqO0iWMSL+9969r+X35SIb8yv4k8JxIrhSOE4EVwrHieBK4TgRXCkcJ8K6vE8ichZADkAdQE1V773a9rncIp797x+1yIYTQ7Td2fSrvHOet5u461bzPKk+9s4sl9nLpQPs0Uj18XmgnHaiNfZI5XNZkpUW7LY+YZW9H/Eke6qSg5zSEU6XSHbh6RdIdnr2dZKNj7H7SWp2ysP0DI99PstzrBj1C4VlnotU2MtVF/YKJhIsu3iBPWEAsDj9U5IVl7aQbH6q9V4vLS6YxwOuj0v2D1TVGzo5bxn89clxIqxXKRTAf4nISyLySWuD1R0CrUeq42w01vv69ICqXhaRcQBPi8irqvrs6g1U9TEAjwHAxPZtvuqks+FZb4uby83/r4jI99Fouvxsu+3DWhXFhdYmgvEqh/Qr9RMkG97+dpLl5//HPE91jnP0l+qTJEuFRopBjY00CDd8rpRZvwtzbDzPXpozx1gpXyJZbIQN+qECzyWf5X6HF7NsiPYLnzud6yNZQoyOxgDSvVyrsB28fyHORuuS8FtBzUhtKVSMJgNGnUx/r/1RTffzvbkccjrJhZnWMVZq7KxYYc2vTyKSEZH+lZ8B/CGAo2s9nuNsFNbzpNgK4PvNdjRxAP+iqv9xXUblODeR9bTNPA3grus4FsfZELhL1nEidLdDYEwQRDoCDhud8lID+0gWpNlwy2anzPOk4xypliE2jGslPmaQYGNZhSOx2dc5sntlhg3OSp7rIQCgb4RrRkaCUZLNn+Ko9NQSx0ondw2SbHSQmxnEjTYDEtqucjWaK1QTbAQvG4b6gPJHqxzjfbPG/Z8psvMlaOO31DrXjPT3Gh0Gp1vvl1pLWzXxJ4XjRHClcJwIrhSOE8GVwnEidNnQDhFPtxpRYyOcqp0DG34zU1wcL/0ccQWAwTQbfmGJDd5UnSPDYcAR0mKOjbIzJ18j2dzlaT5emy6Gk/XdJKsNGivzCY97x7DRNbCXU9l7YrxdKsZGaDJgYxUA4kYkX/v5PLUR3tfKcyssGenyM/MkW7h4lmTVZbu5QsowtIdHxkk2m20dj1zleeBPCseJ4ErhOBFcKRwngiuF40TorqGdiKF3S2vqcTnOxuCVKU55nq9wWnVGuO4aALJlPma5wBFoKeVI1p9imbVwe26ODcRqoUKyeoVlAHDpOK+unNy7k2Tj4+yIiBvOgJTRWj4dcKp2Ks6GcmCs1QcAatRuV8t8zFrFkBWMkoBFTm+vGJ36akanxErZ7gbZG3DUfmyYPxfVeqvDIwiOmccD/EnhOIQrheNEcKVwnAiuFI4T4ZqGtog8DuBDAK6o6h1N2QiAbwPYA+AsgI+qavvuUk0UQCkSZT1/luuxp+a5frYUssF5vDBjnqdupDL3VDlVe8xYL20qxcccKhtt23NWKjob80nY0eKBpLG2Xo4dDNUUG8GJQV5CIGmkZfem2AhNpYzxtIm61+p8H4rLPMelKzzu2SxH4udyXBM/bxjz+SI7JyqL9hgrNSNlfj9fn6EtrUsVBHEje6BJJ0+KrwN4MCJ7FMAzqnoAwDPN3x3nLcE1laLZsibqf3wIwBPNn58A8OHrPC7HuWms1abYqqpTAND8nzOwmrQ0Q8u3byviOBuFG25oq+pjqnqvqt6b6bOzWh1nI7HWiPa0iGxX1SkR2Q6A87oNwjBEYbk1Wlm9zPXGhRhHdi9VuZb3+VOXSQYAuSmOnH7o995Jsn33v41ki68fJ9nx46dJljbqiOMhG4iZDBv4ANCXZgMRRs14mGeDPN5rrHlX5yi3WFFgZYPVWsAeAOrGk72a5zkWlni7wjIfs15nB0ilyCnhlTp/V5eK9kc1bzg8kkXePzPZ2r1d4u0/+mt9UjwJ4OHmzw8D+MEaj+M4G45rKoWIfBPA/wK4VUQuisgjAL4I4AMicgLAB5q/O85bgmu+Pqnqx9v86X3XeSyOsyHwiLbjROhq6jigiEXSlGNqLKc1ZNRJn+eU7ksvc3QVANJJNt4ygzzV4hwvI3b+xVdIljXSvIeGuDt52opSGwvBA0AVbJz2pDnKKik2vsuGcRpTjl7DSPMup/ja1irsxACAopEKv1Aw7oNR/35xlueXNwz3ntFdJNu3Zy+PMW1fx/NZo+lanrfNL7ZG06Op5KvxJ4XjRHClcJwIrhSOE8GVwnEiuFI4ToSuep+CWIDBvtai8qUBQy8T7EmJzV8g2S277RSKd95zO8n2b91GssPP/DvJXnmOU0duNRw7g8PsvcgEnMaQNGoSACAosAdJ41wHUF5mz1D+HJeuJHrZq9ST5tqJitFIIZuzW/EXje57+RqP8dXTnPJy5hLXoKT7uaHAA+/gezW+k5dieGWKOzICwKWFIyRT4yNV0PMtv5cq7Vfq9SeF40RwpXCcCK4UjhPBlcJxInTV0I7HE9gy3Loe3VTsHG0XlrnAfTDJ1u6B2+3GBbfv5nqMwhlukDBzims5eE9gK9uWSBZ434xhKCcS9iVWcEpIKsnfUcVZLlXJXmFDu8cwYnuHdpCsUuNzaMpItQHQlxkj2cAWXk+wUGKjumeS7yEybLjnwLUvs8deItmzL71sjjGo8vIHE9sPkSxRa62oDoy6khX8SeE4EVwpHCeCK4XjRHClcJwIa+0Q+AUAnwCwYul+XlWfuubJ4gmMjLYaf/MFNoCLYCNN+1iWNlrAA0D+DC/IfuTnPyVZZYaju7+zs49kY8ZadIkKG4g9ytFrrXGUGwAkzksLJAzju7DEBuHiBZbVxvlWapKj3Ok+NshrbYr4gz4eo1X+sjgwRbKx3VtJFggb2q+c4ZqW5BRfh8m4/f092Mfdlfr72OFRqLd+fhLrXPPu6+AOgQDwFVU91Px3TYVwnM3CWjsEOs5blvXYFJ8RkSMi8riIDLfbaHWHwMVFw3ftOBuMtSrFVwHsB3AIwBSAL7XbcHWHwMFBfl93nI3GmiLaqvpGGFFEvgbgh53sFwQJDAy2Fvyne/ghM1czGg6WuTg+lje67AEoLJwnWe4KG9XW421kgI3B0Qx3oYuljCYDhjFYarPmnRpGZ2hEWYNeNozFSFsvFnmM/VUj0txrdAgcsp0B/RN8ffMh7z8xxtkGt+29k2RjI3y8t8f4Olypcfr+2Rw7NgAgO8fR/ekZlgU9rSn4YZ2v1wprelI0W2Wu8BEAR9dyHMfZiHTikv0mgPcAGBORiwD+CsB7ROQQGuuwnAXwqRs4RsfpKmvtEPhPN2AsjrMh8Ii240Toauq4AggjUdtUhoewzQibzp3jNPEwb9fZ5qpc/7x7nD1f44Zhm45xlDxhbBdL8PdJLeT282qviQ4NeN75Is9nqcBR5eSwUY+d4O227uPrmJ7g61C2ipoBFIy2+0eNbomj47eQbMcIJ+H3bx8hWSnHHQcrJR5Pz3KbqLux3EA1Yy1B0GrQq7S5MfAnheMQrhSOE8GVwnEiuFI4ToSuGtphGGK52BplLRbZsBXD2B0Sjuxm1c5TnFviVObtcY6cDqV4+gnj3Fbss25EdpdLvG89w+MGgJixFl5+nnPDpma50djACEfTB7ewEZvZyw3gFqu8fMHUGTta/NosR7rPz3G0eSTH17Z3+DmS7Rjnmu/Zi5d4jMZi9T1Jjs4DwMRtPMeeQR63RtL6E0fsKD7gTwrHIVwpHCeCK4XjRHClcJwIXTW0YxAkg54WmRoLqlfBsliV9TeWbmMszbLBGxrWsoK3qxrfExXhy1SrsqGdM0rGUwN2o7FqmhPXZ6bYqD5bYiN4nIPA6Itzk7JzC7zhmSnODJjOcYQcACppjn5rlg36U+eN5mwZ7hIuIXdQj8V5jH3bOBU9s9seoyYMh0WJHTBJaf3cBUZGwhtjavsXx/ktxZXCcSK4UjhOBFcKx4nQSeXdBIB/BrANQAjgMVX9exEZAfBtAHvQqL77qKqyxdVyrBh6Y62NqrYNs8E5tcBRzmKd07LrZqwZqNRZ1xNG7XXMWMy9muCU4lD43EWjxjdrNLIeTLQxEHvZmKxu5+hsucbdzfPKBvBsjBuAvb7A0efZEs8ZCY40A0B9gVPwAT7PrffwuO+8Y5JPExiL1ef52o7287UNMnaX8DKMaHqBr3kKrU6ZWLA+Q7sG4HOqehDA/QA+LSK3AXgUwDOqegDAM83fHWfT00kztClVPdz8OQfgGBrLODwE4InmZk8A+PCNGqTjdJM3ZVOIyB4AdwN4HsBWVZ0CGooDgJt6orUZWjbLbWocZ6PRsVKISB+A7wL4rKp2/Ole3QxtaMjOGHWcjURHSiEiCTQU4huq+r2meHql/1Pzf6ODmeNsPjrxPgkaLW2OqeqXV/3pSQAPA/hi8/8fXPNYKgjCVj3cMsLep+ks62pumR1by2p7n8oBe1jqCU4JKRuzLwSG96nO+Rt54eOFI+z1yPdyGgIALIecypLeye3rd9W4tqB3gN9U44PcuCCo8xj74ny9l9ssaXD57EWSDUzw0z4zzuc5lz1JssNHeCH45UX2kB0a56YHo5N8bQBgcIy9T2qsjxjWWueo7PR6g05ynx4A8KcAfiUiv2zKPo+GMnxHRB4BcB7An3RwLMfZ8HTSDO1ngLGaSIP3Xd/hOM7NxyPajhPBlcJxInS1ngIApBbpEJgwcvY57R7ZChumpZptaJeUp1UyjO9Fw9oq1lhWNs5TSXC6w95D95FsvmhMBsCFOU5luX3nbpJN9h4g2XKNDfpyjesu+lNbSBYzGjicOPwrc4ynZi+Q7ODut5EsAza0E72c+rGfd4UUuA4ktsjpJfPG+oQAUFri9Q0XjTqQnkiWSL1i3xfAnxSOQ7hSOE4EVwrHieBK4TgRumpoiwKxiMGTCDj6GBht5YtG/vt8we4QWCqwUTY7wIZ2xWpmYBTXLxuW/7Zdu0gW38Kyk0e5Ux4AzMzyMd+V4WsRNxZ4X7o8TbKwxAvbG8vqYXaBMwMunOM1AgEg7GGnQ63A5zl9gqPSt9/NDoI7JyZItpxlozgnfLxixS7ViRnr+lULxhqF9db7r1Yni5Vt2/7FcX5LcaVwnAiuFI4TwZXCcSJ0N6ItgnhkrbeeuJHyHHC6dSneQ7Lpil3rVDLE/YahnUuyIVkNjWUAxnlR9O233EOygvEdc3badgZs6+X06NBYe65irN+XX+LSlVqOczbr4I6Dr7zGRuzMMhvPAHCP0eZ+IM2R/HMnT5MsmeYMhMoId0tMldnhkCjzfQmMDAIASCaN6L6R/l8tRo/phrbjdIwrheNEcKVwnAiuFI4TYT0dAr8A4BMAVnq7f15Vn7rqyeIJDG9tTRUevsTrvM1M8bpssxd4Hbtym4JANYyoopHebGSJo7eHDfo7Dr2bZPtuvZtkh09yDfKu7XeaYzx4kCO+6DcWp1/iVvUDRpS7mmRHQnqAt8v0cdR8ZDdHmgFg5x5uux+W2QlSneH7lV1k430YRsv+Mt+rwZThfDHuHwAkYvwZGOzhbQv11u1ixn4rdOJ9WukQeFhE+gG8JCJPN//2FVX9uw6O4Tibhk5qtKcArDQ9y4nISodAx3lLsp4OgQDwGRE5IiKPiwj3TkFrh8A5IxnNcTYa6+kQ+FUA+wEcQuNJ8iVrv9UdAkeNDuOOs9HoKKJtdQhU1elVf/8agB9e82TJJEZ2tbZo31fl/Oblx9m4DIVThN82wTXNADBtRHKRYgNa4pxivnOC3wzvvOdekqmxMP3cJR7jtsF95hj37+ba5KE4R3fj/XyeQSO1vlThyHdPL0eB79ppRPaX7HrlVD8b4OMj/LS/fJyzAAqzfG2DAW75359m50JSjYZ07VK9Q/5eTxqN6qqR2nRp27WpgydFuw6BKy0zm3wEwNFrHctxNgPr6RD4cRE5hEYSyVkAn7ohI3ScLrOeDoFXjUk4zmbFI9qOE6G7qeOxGOKRiPGI0Wn7wNvvINnyIBuxA1sPmqe5vHiYZPOUOgzsGuDo7L5b3kGybTt4/bbpOV6kPSZcbxz02u2tK8ucUh7rYyM4HrAjQhLsNIjX2VguG93E+/pYVq7YjcYWFjgqPXkrX4uhIU5Hv3CRm7PltrGhPbaFnQFxMZrPtWleVjRS/evGGoWxZOu9kZgvLu84HeNK4TgRXCkcJ4IrheNE6K6hHYaoRSKvlSIbeVIzukbXuE66auwLAK9f4lRm1LlmOJEaJVnWME6zRg1zrmSkece4Xvhcnjt3A0DhpGHEDrGh3Zfg763yMi84X8txpDks8YLsNWPcI6HdDG36ubMkm429i2T3HWTHyFjsNZJVz58j2eXX2Wkwvo3vNeJ2RDtnLZPWz9exNxKdj7mh7Tid40rhOBFcKRwngiuF40RwpXCcCF31PmkYohrx5MzPcOOCC1NnSVae4UXWb9lheCkATG5jT8X501yjMTvHXopCntsLXpnmjny5RT7eYILrOMJlOz0hXOC0lSDkIqz0EKdB9CU51UIHuT4D7NhBMcm3fO8Er40HAKemeN6zyvPeMb6fZNt6uL7j9Kuc+nEl5PufHuZOgqO9nCICAAj4Xvfu4O/6VP+elt9jCWOdgpW/tf2L4/yW4krhOBFcKRwnQiflqD0i8gsReVlEfi0if92U7xWR50XkhIh8W8RaTMpxNh+dGNplAO9V1XyzgcHPRORHAP4cjWZo3xKRfwTwCBodPtqiAOqR7n1lo/38rkMfItnBDHfp2zps6+EpTu9HsfxLkt16YJxktxzYS7JSjWsfTpx4mWTlGU7diGPAHGPfABuTw0muLcmkjM54MU7VkF6+lcrZMkiMsmF6YKDNd+M5lpfqfC0uzfG6fgNGzcfkHqMVf4bTavr7OT1lqE0jmPkyOzeCNH8uhkdbjxm0SRsBOnhSaIMVF0Gi+U8BvBfAvzXlTwD48LWO5TibgY5sChEJmk0LrgB4GsApAFnVN5YNvYg2XQNXN0ObnWO3quNsNDpSClWtq+ohALsA3AfAqgM1n0erm6GNjXJWquNsNN6U90lVswB+AuB+AEMisvIiuwuA8SbvOJuPTlrxbwFQVdWsiPQCeD+AvwXwYwB/DOBbAB4G8INrHatWKWPmXGuefWmBu+c8cIijl0GGjWLJclQYAN6/n7vy3Z1mg3VkmCO5gwWOIC+f4cXcd5zhyHehx4hSj7DRCACDvWyIJgM+ZkX5ASxGwyE1bmXF6LRXNdYfiMFe8264j8eenzbWCTQcEUmjRerQFu44GBqTqVa5RuLyjF07U8uyoR7P8LWoXzrVut8yR9zf2L/tX37DdgBPiEiAxpPlO6r6QxF5BcC3RORvAPwfGl0EHWfT00kztCNodBqPyk+jYV84zlsKj2g7TgRXCseJIGoYcjfsZCIzAM4BGAPA1febE5/LxuRac9mtqmbOfFeV4o2TiryoqrzowybE57IxWc9c/PXJcSK4UjhOhJulFI/dpPPeCHwuG5M1z+Wm2BSOs5Hx1yfHieBK4TgRuq4UIvKgiBwXkZMi8mi3z78eRORxEbkiIkdXyUZE5OlmWe7TIrIpFgsXkQkR+bGIHGuWGf9ZU77p5nO9S6a7qhTNpMJ/APBHAG5DY4XV27o5hnXydQAPRmSPAnhGVQ8AeKb5+2agBuBzqnoQjVKATzfvxWacz0rJ9F0ADgF4UETuRyOb+yvNuSygUTJ9Tbr9pLgPwElVPa2qFTTSzh/q8hjWjKo+CyCaJ/0QGuW4wCYqy1XVKVU93Pw5B+AYGtWTm24+17tkuttKsRPA6gUb2paxbiK2quoU0PigAeDCjw2OiOxBIxP6eWzS+aynZDpKt5XCWo/bfcI3ERHpA/BdAJ9VVa5y2iSsp2Q6SreV4iKA1eVXb4Uy1mkR2Q4Azf+5AesGpdmy6LsAvqGq32uKN+18gOtTMt1tpXgBwIGmVyAJ4GMAnuzyGK43T6JRjgt0WJa7ERARQaNa8piqfnnVnzbdfERki4gMNX9eKZk+ht+UTANvZi6q2tV/AD4I4DU03vn+otvnX+fYvwlgCkAVjafeIwBG0fDSnGj+P3Kzx9nhXN6NxuvEEQC/bP774GacD4A70SiJPgLgKIC/bMr3AfgFgJMA/hVAqpPjeZqH40TwiLbjRHClcJwIrhSOE8GVwnEiuFI4TgRXCseJ4ErhOBH+H6aE/hH02kbLAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  9%|▉         | 1100/12000 [55:56<9:55:03,  3.28s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.036880526691675186 -0.05326002836227417 0.09265013039112091\n",
      "gen_imgs.mean 0.46660733222961426\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 10%|█         | 1200/12000 [1:01:00<9:07:28,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.005367569159716368 -0.08897556364536285 0.06396696716547012\n",
      "gen_imgs.mean 0.46841105818748474\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 11%|█         | 1300/12000 [1:06:06<9:03:27,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.05638543888926506 -0.029951633885502815 0.09165964275598526\n",
      "gen_imgs.mean 0.45955026149749756\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 12%|█▏        | 1400/12000 [1:11:09<8:54:49,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.0073172771371901035 -0.06609690934419632 0.023974347859621048\n",
      "gen_imgs.mean 0.4703606367111206\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 12%|█▎        | 1500/12000 [1:16:15<8:48:28,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.007617192342877388 -0.09684007614850998 0.01663091778755188\n",
      "gen_imgs.mean 0.4720819890499115\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 13%|█▎        | 1600/12000 [1:21:21<8:48:25,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.05063028633594513 -0.10843576490879059 0.01553542073816061\n",
      "gen_imgs.mean 0.46021753549575806\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 14%|█▍        | 1700/12000 [1:26:25<8:44:21,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.005094860680401325 -0.055441681295633316 0.043895214796066284\n",
      "gen_imgs.mean 0.4656561017036438\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 15%|█▌        | 1800/12000 [1:31:27<8:38:19,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.050303831696510315 -0.030268793925642967 0.09887729585170746\n",
      "gen_imgs.mean 0.4634999930858612\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 16%|█▌        | 1900/12000 [1:36:29<8:29:50,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0015906114131212234 -0.05672073736786842 0.03545389696955681\n",
      "gen_imgs.mean 0.4713971018791199\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 17%|█▋        | 1999/12000 [1:41:30<8:20:27,  3.00s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.05545973777770996 -0.00830682273954153 0.07281221449375153\n",
      "gen_imgs.mean 0.4646250605583191\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 18%|█▊        | 2100/12000 [1:46:39<8:20:31,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02995067834854126 -0.019153613597154617 0.008454700000584126\n",
      "gen_imgs.mean 0.46218857169151306\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 18%|█▊        | 2200/12000 [1:51:43<8:20:44,  3.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.00041656545363366604 -0.0697510689496994 0.03711428493261337\n",
      "gen_imgs.mean 0.4653418958187103\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 19%|█▉        | 2300/12000 [1:56:49<8:16:59,  3.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03677413985133171 -0.03296459838747978 0.016417251899838448\n",
      "gen_imgs.mean 0.46846920251846313\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 20%|██        | 2400/12000 [2:01:57<8:09:35,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.010257536545395851 -0.04953543469309807 0.04912066459655762\n",
      "gen_imgs.mean 0.46396565437316895\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 21%|██        | 2500/12000 [2:07:01<7:57:52,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.08727440237998962 0.017892993986606598 0.10061384737491608\n",
      "gen_imgs.mean 0.46373891830444336\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 22%|██▏       | 2600/12000 [2:12:04<7:55:50,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.030369244515895844 -0.03493635728955269 0.07220106571912766\n",
      "gen_imgs.mean 0.4628511369228363\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 22%|██▎       | 2700/12000 [2:17:09<7:48:41,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.015235072001814842 -0.05248366296291351 0.04428020119667053\n",
      "gen_imgs.mean 0.4696338176727295\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 23%|██▎       | 2800/12000 [2:22:16<7:48:40,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0081519465893507 -0.052034758031368256 0.016321245580911636\n",
      "gen_imgs.mean 0.46381625533103943\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 24%|██▍       | 2900/12000 [2:27:23<7:39:29,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.055283695459365845 -0.010396141558885574 0.0753137618303299\n",
      "gen_imgs.mean 0.4599538743495941\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 25%|██▍       | 2999/12000 [2:32:25<7:33:30,  3.02s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.012220792472362518 -0.0641164630651474 0.06410261243581772\n",
      "gen_imgs.mean 0.46390652656555176\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 26%|██▌       | 3100/12000 [2:37:38<7:52:51,  3.19s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.002498524496331811 -0.06764405965805054 0.04168661683797836\n",
      "gen_imgs.mean 0.47338151931762695\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 27%|██▋       | 3200/12000 [2:42:43<7:24:42,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.018907170742750168 -0.047685157507658005 0.07153502851724625\n",
      "gen_imgs.mean 0.4689357280731201\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 28%|██▊       | 3300/12000 [2:47:48<7:20:00,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.06947597861289978 0.022859662771224976 0.0789358988404274\n",
      "gen_imgs.mean 0.4686150550842285\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 28%|██▊       | 3400/12000 [2:52:51<7:14:26,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.008124801330268383 -0.06718215346336365 0.02421717718243599\n",
      "gen_imgs.mean 0.4638507068157196\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 29%|██▉       | 3500/12000 [2:57:54<7:11:11,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03140705078840256 -0.01164068840444088 0.08085578680038452\n",
      "gen_imgs.mean 0.4659085273742676\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 30%|███       | 3600/12000 [3:02:58<7:06:52,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.03348320350050926 -0.07944139093160629 0.05080718547105789\n",
      "gen_imgs.mean 0.47067850828170776\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 31%|███       | 3700/12000 [3:08:02<7:02:55,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03477197885513306 -0.028427790850400925 0.07839027792215347\n",
      "gen_imgs.mean 0.4674418866634369\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 32%|███▏      | 3800/12000 [3:13:08<6:58:07,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.005515861324965954 -0.06254743039608002 0.05056347697973251\n",
      "gen_imgs.mean 0.4646446108818054\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 32%|███▎      | 3900/12000 [3:18:12<6:54:11,  3.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.026879820972681046 -0.05015713348984718 0.04673448204994202\n",
      "gen_imgs.mean 0.4625251889228821\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 33%|███▎      | 3999/12000 [3:23:14<6:42:13,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03217826783657074 -0.013668491505086422 0.03314321115612984\n",
      "gen_imgs.mean 0.46510979533195496\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 34%|███▍      | 4100/12000 [3:28:24<6:42:50,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.01049075648188591 -0.03298107907176018 0.052899401634931564\n",
      "gen_imgs.mean 0.4695844054222107\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 35%|███▌      | 4200/12000 [3:33:29<6:38:26,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03198910132050514 -0.025025632232427597 0.06304821372032166\n",
      "gen_imgs.mean 0.46508175134658813\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 36%|███▌      | 4300/12000 [3:38:33<6:30:33,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.043452188372612 -0.019948534667491913 0.06731146574020386\n",
      "gen_imgs.mean 0.467080295085907\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 37%|███▋      | 4400/12000 [3:43:39<6:22:48,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.014548787847161293 -0.0574144646525383 0.07601208984851837\n",
      "gen_imgs.mean 0.4632507264614105\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 38%|███▊      | 4500/12000 [3:48:50<6:20:51,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.006383850239217281 -0.0656614750623703 0.03854114189743996\n",
      "gen_imgs.mean 0.4623304307460785\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 38%|███▊      | 4600/12000 [3:53:58<6:17:06,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0312783345580101 -0.052377961575984955 0.04429083317518234\n",
      "gen_imgs.mean 0.46832454204559326\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 39%|███▉      | 4700/12000 [3:59:03<6:29:57,  3.21s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02495190128684044 -0.03397838771343231 0.08441822230815887\n",
      "gen_imgs.mean 0.46647679805755615\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 40%|████      | 4800/12000 [4:04:09<6:04:21,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.016454853117465973 -0.07489921152591705 0.06737953424453735\n",
      "gen_imgs.mean 0.46772727370262146\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 41%|████      | 4900/12000 [4:09:11<6:00:37,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.002426616381853819 -0.04823368787765503 0.027504798024892807\n",
      "gen_imgs.mean 0.4635281562805176\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 42%|████▏     | 4999/12000 [4:14:12<5:56:04,  3.05s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.006387769244611263 -0.060256779193878174 0.07012110948562622\n",
      "gen_imgs.mean 0.46888023614883423\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 42%|████▎     | 5100/12000 [4:19:23<5:55:40,  3.09s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.0044378736056387424 -0.06167544424533844 0.06462767720222473\n",
      "gen_imgs.mean 0.4681699275970459\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 43%|████▎     | 5200/12000 [4:24:32<5:47:44,  3.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.05596540868282318 -0.006055266596376896 0.07421935349702835\n",
      "gen_imgs.mean 0.4658535122871399\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 44%|████▍     | 5300/12000 [4:29:38<5:39:24,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03369522839784622 -0.03833942487835884 0.02910400927066803\n",
      "gen_imgs.mean 0.46815454959869385\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 45%|████▌     | 5400/12000 [4:34:47<5:55:38,  3.23s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02061627060174942 -0.018087320029735565 0.04349461942911148\n",
      "gen_imgs.mean 0.46555909514427185\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 46%|████▌     | 5500/12000 [4:39:53<5:32:37,  3.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0033055124804377556 -0.05906042456626892 0.015971196815371513\n",
      "gen_imgs.mean 0.4637012481689453\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 47%|████▋     | 5600/12000 [4:44:58<5:25:12,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.026341786608099937 -0.04254632443189621 0.08985714614391327\n",
      "gen_imgs.mean 0.4707804322242737\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 48%|████▊     | 5700/12000 [4:50:02<5:20:50,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.04216202348470688 -0.04443178325891495 0.060427386313676834\n",
      "gen_imgs.mean 0.46612948179244995\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 48%|████▊     | 5800/12000 [4:55:06<5:14:37,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.026912018656730652 -0.048537563532590866 0.02737792581319809\n",
      "gen_imgs.mean 0.4665070176124573\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 49%|████▉     | 5900/12000 [5:00:11<5:08:12,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.003532296046614647 -0.07122485339641571 0.05046042799949646\n",
      "gen_imgs.mean 0.4634568691253662\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 50%|████▉     | 5999/12000 [5:05:10<5:00:22,  3.00s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02033229172229767 -0.025989804416894913 0.01628934033215046\n",
      "gen_imgs.mean 0.46281808614730835\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 51%|█████     | 6100/12000 [5:10:18<4:58:51,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0635150596499443 0.014438417740166187 0.0540640614926815\n",
      "gen_imgs.mean 0.4705491065979004\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 52%|█████▏    | 6200/12000 [5:15:21<4:53:45,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03158463537693024 -0.028736189007759094 0.025781258940696716\n",
      "gen_imgs.mean 0.46351051330566406\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 52%|█████▎    | 6300/12000 [5:20:24<4:47:33,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.00811070017516613 -0.0466257743537426 0.04883085936307907\n",
      "gen_imgs.mean 0.46576008200645447\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 53%|█████▎    | 6400/12000 [5:25:30<4:55:12,  3.16s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.01706499606370926 -0.0639447346329689 0.029139520600438118\n",
      "gen_imgs.mean 0.46530184149742126\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 54%|█████▍    | 6500/12000 [5:30:37<4:39:38,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.05827535688877106 -0.013938076794147491 0.05963161587715149\n",
      "gen_imgs.mean 0.46324193477630615\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 55%|█████▌    | 6600/12000 [5:35:45<4:32:24,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.023584958165884018 -0.04039771109819412 0.04742307960987091\n",
      "gen_imgs.mean 0.46901464462280273\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 56%|█████▌    | 6700/12000 [5:40:48<4:27:12,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.05757440999150276 -0.03045307844877243 0.04665646702051163\n",
      "gen_imgs.mean 0.46474742889404297\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 57%|█████▋    | 6800/12000 [5:45:53<4:25:21,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.008958575315773487 -0.05796854943037033 0.03618098050355911\n",
      "gen_imgs.mean 0.46452391147613525\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 57%|█████▊    | 6900/12000 [5:50:58<4:19:20,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.038843151181936264 -0.026032987982034683 0.05699191987514496\n",
      "gen_imgs.mean 0.4654899537563324\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 58%|█████▊    | 6999/12000 [5:55:59<4:13:25,  3.04s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0222290251404047 -0.02025645598769188 0.01323558110743761\n",
      "gen_imgs.mean 0.46498048305511475\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 59%|█████▉    | 7100/12000 [6:01:08<4:08:59,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.002944235224276781 -0.04501693695783615 0.03484257683157921\n",
      "gen_imgs.mean 0.466913640499115\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 60%|██████    | 7200/12000 [6:06:11<4:04:18,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.021935326978564262 -0.025204092264175415 0.025847110897302628\n",
      "gen_imgs.mean 0.4609273672103882\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 61%|██████    | 7300/12000 [6:11:18<3:55:57,  3.01s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.036265432834625244 -0.030588656663894653 0.032263755798339844\n",
      "gen_imgs.mean 0.4677545428276062\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 62%|██████▏   | 7400/12000 [6:16:27<4:03:18,  3.17s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.00959406141191721 -0.07301812618970871 0.026071056723594666\n",
      "gen_imgs.mean 0.4674316346645355\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 62%|██████▎   | 7500/12000 [6:21:32<3:53:33,  3.11s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.011828131973743439 -0.02558360993862152 0.027179604396224022\n",
      "gen_imgs.mean 0.4619220197200775\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 63%|██████▎   | 7600/12000 [6:26:38<3:44:50,  3.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03225094825029373 -0.02386529929935932 0.01056681852787733\n",
      "gen_imgs.mean 0.4631762206554413\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 64%|██████▍   | 7700/12000 [6:31:45<3:38:35,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.012481573969125748 -0.05556647107005119 0.04865887016057968\n",
      "gen_imgs.mean 0.469479501247406\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 65%|██████▌   | 7800/12000 [6:36:49<3:33:11,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.0018354484345763922 -0.0587601475417614 0.041957657784223557\n",
      "gen_imgs.mean 0.46179279685020447\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 66%|██████▌   | 7900/12000 [6:41:53<3:27:38,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.027303606271743774 -0.05019494518637657 0.017468981444835663\n",
      "gen_imgs.mean 0.4651910662651062\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 67%|██████▋   | 7999/12000 [6:46:59<3:24:37,  3.07s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02722887322306633 -0.03956994414329529 0.05073058232665062\n",
      "gen_imgs.mean 0.4698219299316406\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMUAAADDCAYAAAAyYdXtAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAboElEQVR4nO2dWYxkZ3XH/+feW7equrbee1bPeMx4vAAeJ2PHhCDFJo6MlcQgEYQfiEUs4AGkoPBiESkhUh6IFEB5iIiMsHAkwpJgZGM5gDNiVzA2NjY2Xsfb9Ex7esbTe+33fnmoatN1/6dmyt09NdXD+Umt7jp9l++7VafuPd/ZxDkHwzB+i3e+B2AYg4YphWEkMKUwjASmFIaRwJTCMBKYUhhGgg0phYjcJCLPisgLInLHZg3KMM4nsl4/hYj4AJ4DcCOAaQAPA7jVOfebbvsEgefCdNAh8z3RDk6iqNEkme/rOu0pYl8RxrEyd+V6pFIpHk89JlkzYpnvq0NE4CI+j3ItvB6vj4MiU3YNM6FyDv0zUK9VWOh4juLxJD2fzxNmciTzwwzvq4zb9/k9AAARfl+VtwGCzjEem3kNc/PzypmAQBP2yLUAXnDOvdganHwdwC0AuipFmA5w4MqJDlk+yxfFD3iic7MnSVYqDKnnyQ/xtPIZPk+9xorWrNVItn1qJ8nmjy2R7PV5/hCNFHWtGHNzJNs5xB+kobTyFikf7KpTlD7gfXdddhGfI1dVxzh95AmSRRWeYyY3yrIin2fnlb9PsrEdB0gWhvxZHR7dro4xI1mSzVf4WngY7nj9gQ//tXq81rbrZyeAo2teT7dlHYjIR0XkERF5pNlUVNgwBoyNKIV266H7sHPuTufcIefcoUC5AxjGoLGRx6dpALvXvN4F4PiZdvA8QS7T+WzoK7f9lOPHg1KpQLKJUZYBQLW8zOcWfoxJK7fpZsTP+rWIH6mOnX6dZGGKxz05NqWO0Z85TbJikGdZiR9NFgN+vl4u86Pgi6/xI+dioUyy0S6PeCdPK+9DqDyyNvkRJshPkKyw+yqSpSdHSLa88grJMukFdYyl3DDJxop8zYuFXR2vwzQ/Tq+yka/uhwHsF5GLRSQE8EEA923geIYxEKz7TuGca4rIJwB8D4AP4C7n3FObNjLDOE9s5PEJzrkHADywSWMxjIHALF/DSLChO8WbxROHMNXolKFB24WKsTtSKpFscrKonmf6GBuxmpMnrTiTghQbkimPl5LHxtm4HB5m4zIY4XEDgFtiA9obH+cNC2x8VxS3wqtLbIgeq/NCQuMUG9qvL6XVMaI8RqJciQ3b4+VFkl1a3EeyOLuHZLNlNuYLBfZJ5Id5LAAQKH6uxdd5YWQk4dsRzcPbxu4UhpHAlMIwEphSGEYCUwrDSNBfQzvwURjtNNSiKhvFr8/MkkyEvdee4gEGgDhm4z2K2aBzwtGhzmfZQpm911AWA+Iae5qPHnlVHWMxYkN9rs6e5foKG8HTc2xpv3yUAxTrynWYb7C3/+VZ3b100e6LSZbNsaE+t7jC2514jWWz7Kk+qkTiXjLJhvaOcd3QXq7zeJ4++hLJhoc7FzGiuHscnt0pDCOBKYVhJDClMIwEphSGkaCvhnY2HeLyA50hvFJmr/SRkEOeR8c4xHjHDj0b69Ujz5Os2WQDOorZWI6UNNGasl0DLDs1zwb58gIvJADAyDbO5ju1wkbnS6fmSfbqAhvL5Sob2tksL0ScPsUGcHlZNzr9NM9xpVzn/ets+E+//CLJUlM7SDYjfI6HHvkJyXaN6tELu7ZzmPj4GEcGlGud1yxW3udV7E5hGAlMKQwjgSmFYSQwpTCMBKYUhpFgQ6tPIvIygCUAEYCmc+7QGXfwgCDsDGXwhUMjghSvhgQppWBXl0pjtYaymhJxYn+YVgp7+ZyD4Cv1k4bGedVjcU4pcDakJ8gHeQ7fqHJUBk6vKCtaNV5pCtNKUbA6r1I1K7zvvkv3qmPM53jeKSUvZWyE81IWl3kF8ZVnHydZqIRvPPvLH5Lsp/t4tQ4A3v8Xt5JsosR5Lc1G58remYoAbsaS7PXOuVObcBzDGAjs8ckwEmxUKRyA74vIL0Xko9oGaysEllfY8WMYg8ZGH5/e6Zw7LiKTAB4UkWeccz9eu4Fz7k4AdwLA9p0l6zppDDwbLXFzvP17VkS+jVbR5R932z6KYiwsdca/hxGHNiwoRXyrM8dIlhI9n8KLlSp2StXqZkMxqhUDbGp0kmT1Ml+6ssex/bGvh1CcWmJDtFbjbReX2VjWqioGolQTz7Lhr1XPy2T1MTaVUIisUqg6oxjkgXLI+VM859mlEyQbV3Z++MH/Vcd44FIuhrB/xw0kk0ZnYYeoqaxqtFn345OI5KSd+SMiOQB/CuDJ9R7PMAaFjdwppgB8W1q9EgIA/+mc++6mjMowziMbKZv5IgCumGsYWxxbkjWMBH3Np4hjh3K508CJmmwAb9vGcffNKld9m5/Xy7PHmtGptPJydZbF3GIDY2NsaB+vc55DNs/FCMq+3iVoscHzaSgtw6p1NnZzyjqCKMZ3Ns1e84yyb9UpbbwAFNPsqR4q8jG13JLY8fxiJdIgLvPiRBG8gDI3O62O8fD995Ks/I5tJJsodL6vjcY5MLQN40LFlMIwEphSGEYCUwrDSNBXQxvgkN1FpbpcPs/WYDDEBrkX6zo9Ns6e6oV5Pk/cVHpzK2HixRKHN08f54IETtkXKf0Si9KPulBQjOUM758JOGQ+VjrPeik2gF3IslKBDWoAKI2ywespPcX9mGWiLGzEStuFZpPfq9DjBYuJot4e+pVHXiDZjj1cnGHPwc5qh56V4jeM3jGlMIwEphSGkcCUwjAS9NfQdkCU8NrWy+xNPa0kKw+NsaG1Y0Rv3J7z2JhsVri8f1Tlc6eFPbYpsKyywIZ7FLFhmlG86wBQq7HH1/fYgB4u8bm1/n31MnvOnXJqGeJ980pfPQDIj3J/O0T8kYkr/H4FytdtShSPdpWvYzrDA5eYDXIAmHuJz704wyHqY9df0zk+v/tH3+4UhpHAlMIwEphSGEYCUwrDSHBWQ1tE7gLwZwBmnXNvbctGAXwDwF4ALwP4gHNu7mzHimOHWqXTCBbHxqVTcqdzSn+7tKd7Yp1w1ZCMr4Q8e+x1DRXDeEXxhjfKbOClweP2/C6Gdpnn7edZlvJ5gcFFfO4wxftGAc/PC3iMkRIuDwCVCi9Y+LFynoby3RrxvH3HYeIZxeNfynN/w0Zdv47Ls5y7/9iPHiXZRz58c8drZa3iDXq5U3wFwE0J2R0ADjvn9gM43H5tGBcEZ1WKdsmaZKDPLQDubv99N4D3bvK4DOO8sV6bYso5NwMA7d+cmtZmbTG0WpUD8Axj0DjnhrZz7k7n3CHn3KG0EvFpGIPGej+lJ0Rku3NuRkS2A2B3sUIcOVQWO+8Wac0gC5Qw8QaHE0cV3ctZXlDKcyrGOyI2GrVc58W5RR6PUiisXuPCZTWl+TkASFPrucaLAVprtqaS9l1T8s1dyHOeGOMw+HRGX7BYUaqoo8HXNhC+jss1nvdShb34kRJCnxoZJdmrT+k52tWIr/lb9/DiRG6o8/PjefpnB1j/neI+ALe1/74NAGePG8YW5axKISJfA/B/AA6IyLSI3A7gswBuFJHnAdzYfm0YFwRnfXxyznFXjBbv3uSxGMZAYB5tw0jQ1+UgFwPNhP2VDtiAhhImvHCSw7xTXbycC6d526ZmiCqVsWNFuKTkkbsmf58sL3NxtlhpKwYA2kqclrddV8KyG0q4fSNWcrSV6+iqfM1cl49BTfHkLy3wHPMlNtSDHHulh5Xc+5VgO8nCCV4M2PsWvZXbdbt2kezPb72ZZEOJKnDeGVzadqcwjASmFIaRwJTCMBKYUhhGAlMKw0jQ58IFAjQ6VxEaSoU4P1QS82u8CrOywmEDABBHSuECpYJeQymH74d8SRorvIJUr3O4Q7nCqzW5bs3llcx+pa89mjWO6dDO45S8lHzAq0JxVVmFU3IsAKChFJAol/mabdvHpe/3ve0akkUZDt/52eM/J9lolq/tlde9Qx3jpJKDUiwoK5oUvrP5YR6GccFiSmEYCUwpDCOBKYVhJOiroS0QiOs0tppKM/dAKTKQyrD+xkq5d0A3qmtKsr9TerVpoR81pe/cSoVDLWLH5/C8LpdYyUGQmA36mmLQxxEP0inFBwKlmEEq4DCPjNIwHtAXMvYfuIRkl739D3g8Snn/uMiykyucq3L1/reT7OCBd6pjjGqcytNsKG0AunxWNOxOYRgJTCkMI4EphWEkMKUwjATrrRD4GQAfAbBa8/zTzrkHzn46gSedp9TyxzWbKJVSGsYrRQYAoKok15drStn9QGnmrjRAb0Ix3GvKOarsfQ67VN9LKxOvNNjQbigFDtKKYdxUeqWr5eaV/I6Kkp/R2pTnuH0nVzPKl7jQwKzSGiDMKUZ+bpxkQ0OcT5Efm1DHePLYDMmadeVD1d2BTay3QiAAfME5d7D904NCGMbWYL0VAg3jgmUjNsUnROQJEblLREa6bbS2QmC9odzjDWPAWK9SfBHAJQAOApgB8LluG66tEBgqPZgNY9BYl0fbOXdi9W8R+RKA+3vZTwD4CUNblHL4cZNloniAu9iwiBwbkytVpVJfXik+sMQGop/i7arg41VXFM+3Un4eAEo59tojUsLJteIDSm+8ULmO2gWqK5742dlT6hiHlFDvsWF+KEhnuCDBWJ5bJ8zW+Ppkh4skW4l4jEtVrgQIACdPv0IyV99Jsl1nqAiYZF13inapzFXeB+DJ9RzHMAaRXpZkvwbgjwGMi8g0gH8A8McichCtr6KXAXzsHI7RMPrKeisEfvkcjMUwBgLzaBtGgj43jPDgeZ0GpuexUaxparOh5V3r1fdqiic2ViroNRTbXdJsXBZLbEjmFUNyIc3V83aM62XuJ8bZYK0vspF/6jVulL4yx+epOn4rm1Wec67E+cvlFc75BoBtF3GYeOCUNgmKx780zB7ouQq/L2Pb9/HxMrxvQ2lMDwDF4YtIVi2z8S7J3oNiOdqG0TOmFIaRwJTCMBKYUhhGgv7maIsgFXSGPYvH8VC+kqu8tKTlROvF0JaUsOWaYmiLY6NzpMBGdWGYQ6MDsJVe8tl7PTHCRh8ApPPsbZ6a5O+oHTu4VP0zzx0h2fJJPnegNKZPeexJ3zVRUseYAYeonzjOveey29mDnE/ztZ0q8Xn2RntJVnVKsTdfD1/I5zmUPaV42IXy/s3QNoyeMaUwjASmFIaRwJTCMBL01dD2PB+ZbKcnuNlUetkFbHxXKmw8V7okLS0qRbw8pTp1mJ8iWXaMe7VN7bmMZPmAPbspcA5yyukGXejYu5su8LhDKLnlymJA1FSqjmd4gaAZ81t+5OgxdYyVZQ7XPjZ9gmSjl3BBsmF3Bcm2DQ+TLAjeQrIJcCi7p1SSB4BanT8DkuXzwEt6tK3nnWH0jCmFYSQwpTCMBKYUhpGgl8y73QD+A8A2ADGAO51z/yoiowC+AWAvWtl3H3DOzZ3pWJ7nI5vrNILiiI3TUNij7ZTCZeUFPXR8ZJK9nJe+9SBvqLSBSqc41HuXYmgPh2zsajnRQagXa8gp9vdIkbddrrLR6RefJ5nwOgTGJtkbPjzOstLjT6lj/M493yHZytF5koVT3N6rdDFvd+m+PSQbLbLHv6AUh4mX9I9WQ6kcn82wod1M1AY4Uw3yXu4UTQCfcs5dDuA6AB8XkSsA3AHgsHNuP4DD7deGseXppRjajHPu0fbfSwCeBrATwC0A7m5vdjeA956rQRpGP3lTNoWI7AVwNYCHAEw552aAluIA4GcWdBZDqyj1XA1j0OhZKUQkD+BbAD7pnOP2M11YWwwtq0ROGsag0ZNSiEgKLYX4qnPunrb4xGr9p/Zvdmsaxhakl9UnQaukzdPOuc+v+dd9AG4D8Nn273vPdiwvSKEw2hlaIY4fqUIoq0rLrL/5Lq766/+Qe7BN7uJwgudPsB6fPs15CfUq5xVUmpyX4AfK5Qz1MaaUggZhjle0JiJeiRlSwlOCrJYvwqtZU1O8byqjF1f42UO/INnDv+KVqjjzOMkmLuLVvgNXX02y8RyH1aTB17aqLzTCUz7CQyVeffKCzuU+OUOYRy+xT+8E8CEAvxaRX7Vln0ZLGb4pIrcDeBXAX/ZwLMMYeHophvZTdE9TevfmDscwzj/m0TaMBKYUhpGgr/kUvh+gUOp0Z4jPOQSB0uG9rqjvRUrIAgBc9XvXsjDFy8HHKhyL//wRXm0eS3EMRdXnMaYznNMwUWDDFgCWy3zpF6t87skJNr6LSlhFpPTbi5Xy8yeUhYR0qDeXP/Su60k2fXKJZM/9msvhN+/9Acmuec8NJNs+xv3thkI2vjPJfIg2rsj7R8q26cTl9qxCoGH0jimFYSQwpTCMBKYUhpGgv4UL/ADZQqeH1nlsIHpKsn9RUd/RbWoMIsZG2RCtxVwoYLyoGHQ+e4FLRfb4Tii92tJZ3rcwonuLRTmPi9jgdUqeRH2IDXqvyK0B/AZfx0bEhvbCol5pcceO3ST78F99iGQ/+snPSfb977Lne3aaO08Hl3GkQagUhfC6NBEVra9fU+kdmLhkZ+qAZ3cKw0hgSmEYCUwpDCOBKYVhJOhvKX7PR2qosxy7F7KnOdmAHgAKSuW+MaVvHACEaTZYvZgT3ItZ3m58hMczOqqEdI9zWflAaUI/rBjpAJDJsNdVktYggJrS169e40WDoTwbouUUG6Gvn2Jj9+QJPWdscojHuO/A5SQbUd6HfHqcZG6Fxx0onmXP4+vYzTDWvtVTSi9DvhLm0TaMnjGlMIwEphSGkcCUwjASbKRC4GcAfATAavfzTzvnHjjjsTwPqUQecirNxpznsdGYVcKJw4we8lyts1HdUIzT2VmuvpdJ8XiaEVtus/Psag4VQzlS9gWAYoFNv6ziEU8pxnuU4rdNmrzvxDDnOr/kHyfZPd87rI7x5mvfRrLxSQ7VFqVaYmmC86QfUyoRXns9522nFI+2dCvpp6QZxJFiRKc6ZbHrXiOwl9Wn1QqBj4pIAcAvReTB9v++4Jz7lx6OYRhbhl5ytGcArBY9WxKR1QqBhnFBspEKgQDwCRF5QkTuElGq4qKzQuDyEhfdNYxBYyMVAr8I4BIAB9G6k3xO229thcB8QWm7ZBgDRk8eba1CoHPuxJr/fwnA/T2dMaGGzmODp6EYQY0KG8+LNW5CDgDViI3qWpVLuT/2xDMkK2Y5HP21Wc5LPrl8kmSFNBu220Y4pBsApsZZXsyzkZ9RevUFoRISXlO2Czi8ffsoL1jMnjqqjrFSuZRk5TIXr5udY4/49Al+Knjh2Oske+bZGZLl3saGu+LkBgBUK0p9Yo6ORzDceW2jiA30N87V9T9tulUIXC2Z2eZ9AJ4827EMYyuwkQqBt4rIQbTCSl4G8LFzMkLD6DMbqRB4Rp+EYWxVzKNtGAn6HDoOZBMFw/yADegIbDT6wrJml85lyxUl3LrOhtXF+3aRLO2zsVuP2Hs9ojiqc4osSOnlsiMlZ1wbIxZYlM4o4dZKT8DpI1xVvZHm633ju9hzDQDHT7PH/y1VXtxIpzk83g84BD9Qcsvnj79GsqU9XOQuo3j7ASBu8PVNRVzkrrHSOW8Xb8DQNozfNUwpDCOBKYVhJDClMIwEfTW04zhGudLpbvSUquPOUxqGJ8tGAwhCffhlJa/ZKYm749u4IviyUvk7aLBRDMWrXK6ygYeKUs0MQBzxvJeEjfyUx+POKdciO8TjiZo87qoSBTAccvE4AHhy+tckmz/B++/ezQsWF++4iGSNKr+vwSLL6qeWeTslTx4A8lm+FjklN72eaL3mKxXZV7E7hWEkMKUwjASmFIaRwJTCMBKYUhhGgj6vPkVYWknELcS8ChPmlPCNiPU3hN4HDVBWOZTFBqdkwwcer1z5vrKapYQXQJQG78LhDq1/KKX4teb0Sgl6UasL8gQzyupcs8nXZgScdwEA11xyFckm8pwoNpzhHI13HbqSZJfv4RUpL+C5RPO8iicF/fvbLyjFHkI+ZpioBukrRSbeGFPX/xjG7yimFIaRwJTCMBL0ko6aEZFfiMjjIvKUiPxjW36xiDwkIs+LyDdEFHesYWxBejG0awBucM4ttwsY/FRE/gfA36JVDO3rIvLvAG5Hq8LHGYghfqcLP06zsdv0ORm9phiS1bqStA4gEjYmh7JcVCBQKtGlAjYkK46N6qjE3ydeiisWSkYPT/CyLPeVXI5I+d5aURIh68LbZZQy9/UcG9VRRl8MuLLE89k2rFVvZFmY5VYFkxdzdUFo/e2UZve5LgUgtDwLL+L330suYpyh6d1Z7xSuxeonOdX+cQBuAPDfbfndAN57tmMZxlagJ5tCRPx20YJZAA8COAJg3rk3vkKn0aVq4NpiaCvLXCrGMAaNnpTCORc55w4C2AXgWgDczkZrFoPOYmi5PK9nG8ag8aZWn5xz8wB+COA6AMMib/Th2gWAy1kbxhakl1L8EwAazrl5abln/wTAPwP4AYD3A/g6gNsA3Hu2Y3muiUytMxlelFQF8dmjmc6yMZjJ6f3kwlAxqrUy+UryulMW0SKlNYCvnGN0ROn91mWMmtc1pRisacWjHfja28aWY1OZX3OKjdC4zoUCAKCU4jkWMsq1VTznKa1dgFLmz/M1Gc9FWUdoybXFhCY/tCwsdOa1RNHGSvFvB3C3iPho3Vm+6Zy7X0R+A+DrIvJPAB5Dq4qgYWx5eimG9gRalcaT8hfRsi8M44LCPNqGkcCUwjASiDtD769NP5nISQCvABgHwOXntiY2l8HkbHPZ45yb0P7RV6V446QijzjnDvX9xOcAm8tgspG52OOTYSQwpTCMBOdLKe48T+c9F9hcBpN1z+W82BSGMcjY45NhJDClMIwEfVcKEblJRJ4VkRdE5I5+n38jiMhdIjIrIk+ukY2KyIPttNwHRYSjAgcQEdktIj8QkafbacZ/05Zvuflsdsp0X5WiHVT4bwDeA+AKtDqsXtHPMWyQrwC4KSG7A8Bh59x+AIfbr7cCTQCfcs5djlYqwMfb78VWnM9qyvRVAA4CuElErkMrmvsL7bnMoZUyfVb6fae4FsALzrkXnXN1tMLOb+nzGNaNc+7HAE4nxLeglY4LbKG0XOfcjHPu0fbfSwCeRit7csvNZ7NTpvutFDsBHF3zumsa6xZiyjk3A7Q+aAAmz/N43jQishetSOiHsEXns5GU6ST9VgqthoKtCZ9HRCQP4FsAPumc4441W4SNpEwn6bdSTAPYveb1hZDGekJEtgNA+zf36R1Q2iWLvgXgq865e9riLTsfYHNSpvutFA8D2N9eFQgBfBDAfX0ew2ZzH1rpuECPabmDgLTyOL8M4Gnn3OfX/GvLzUdEJkRkuP33asr00/htyjTwZubinOvrD4CbATyH1jPf3/X7/Bsc+9cAzABooHXXux3AGFqrNM+3f4+e73H2OJc/Qutx4gkAv2r/3LwV5wPg7WilRD8B4EkAf9+W7wPwCwAvAPgvAOlejmdhHoaRwDzahpHAlMIwEphSGEYCUwrDSGBKYRgJTCkMI4EphWEk+H8oYDQNd+G2wgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 68%|██████▊   | 8100/12000 [6:52:07<3:16:26,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03831042721867561 -0.041904691606760025 0.06348445266485214\n",
      "gen_imgs.mean 0.4693276286125183\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 68%|██████▊   | 8200/12000 [6:57:13<3:13:04,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.017091920599341393 -0.05215686559677124 0.05199699103832245\n",
      "gen_imgs.mean 0.46518629789352417\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 69%|██████▉   | 8300/12000 [7:02:16<3:08:40,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.032980576157569885 -0.05763408914208412 0.03720521181821823\n",
      "gen_imgs.mean 0.46465104818344116\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 70%|███████   | 8400/12000 [7:07:21<3:02:54,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.015162385068833828 -0.06263493746519089 0.037846051156520844\n",
      "gen_imgs.mean 0.4662405848503113\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 71%|███████   | 8500/12000 [7:12:24<2:56:29,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0332280695438385 -0.0519314706325531 0.05270325392484665\n",
      "gen_imgs.mean 0.4660720229148865\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 72%|███████▏  | 8600/12000 [7:17:26<2:52:26,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.05299679934978485 -0.007328640669584274 0.08143793046474457\n",
      "gen_imgs.mean 0.4676799774169922\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 72%|███████▎  | 8700/12000 [7:22:31<2:46:06,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.04375810921192169 -0.030583815649151802 0.03936494141817093\n",
      "gen_imgs.mean 0.4674837291240692\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 73%|███████▎  | 8800/12000 [7:27:38<2:42:10,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.03046143800020218 -0.04791726917028427 0.0641246885061264\n",
      "gen_imgs.mean 0.4695182740688324\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 74%|███████▍  | 8900/12000 [7:32:43<2:37:44,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.003479274921119213 -0.04956168681383133 0.05324667692184448\n",
      "gen_imgs.mean 0.46828562021255493\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 75%|███████▍  | 8999/12000 [7:37:43<2:30:43,  3.01s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.025975149124860764 -0.019836146384477615 0.08984681218862534\n",
      "gen_imgs.mean 0.4642788767814636\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 76%|███████▌  | 9100/12000 [7:42:49<2:27:28,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.013370276428759098 -0.07032689452171326 0.019896389916539192\n",
      "gen_imgs.mean 0.46596890687942505\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 77%|███████▋  | 9200/12000 [7:47:52<2:20:50,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.020568879321217537 -0.06147242337465286 0.039850957691669464\n",
      "gen_imgs.mean 0.4672893285751343\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 78%|███████▊  | 9300/12000 [7:52:57<2:16:37,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.029060829430818558 -0.04377158731222153 0.05583301931619644\n",
      "gen_imgs.mean 0.4700668752193451\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 78%|███████▊  | 9400/12000 [7:58:01<2:12:52,  3.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.004457790404558182 -0.07097277790307999 0.047657646238803864\n",
      "gen_imgs.mean 0.47014057636260986\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 79%|███████▉  | 9500/12000 [8:03:06<2:06:29,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.010563511401414871 -0.04057137295603752 0.05396250635385513\n",
      "gen_imgs.mean 0.4668115973472595\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 80%|████████  | 9600/12000 [8:08:08<2:00:59,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.015503671020269394 -0.05511099472641945 0.0122295543551445\n",
      "gen_imgs.mean 0.46340319514274597\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 81%|████████  | 9700/12000 [8:13:11<1:56:37,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.04305802658200264 -0.02312442846596241 0.06715657562017441\n",
      "gen_imgs.mean 0.46673649549484253\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 82%|████████▏ | 9800/12000 [8:18:16<1:50:52,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.012068257667124271 -0.035277362912893295 0.025819458067417145\n",
      "gen_imgs.mean 0.4682740569114685\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 82%|████████▎ | 9900/12000 [8:23:22<1:47:50,  3.08s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02172965370118618 -0.038560204207897186 0.04205422103404999\n",
      "gen_imgs.mean 0.4628715515136719\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 83%|████████▎ | 9999/12000 [8:28:23<1:41:52,  3.05s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.031565405428409576 -0.06329628825187683 0.02770078182220459\n",
      "gen_imgs.mean 0.4698084592819214\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 84%|████████▍ | 10100/12000 [8:33:28<1:35:49,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.01067162211984396 -0.035079143941402435 0.04182063043117523\n",
      "gen_imgs.mean 0.46504050493240356\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 85%|████████▌ | 10200/12000 [8:38:33<1:31:16,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.0062630996108055115 -0.05730415880680084 0.04802761226892471\n",
      "gen_imgs.mean 0.47346100211143494\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 86%|████████▌ | 10300/12000 [8:43:36<1:25:39,  3.02s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.005590924993157387 -0.04305866360664368 0.03726928308606148\n",
      "gen_imgs.mean 0.46804279088974\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 87%|████████▋ | 10400/12000 [8:48:38<1:20:48,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.013489264994859695 -0.021582478657364845 0.03437265753746033\n",
      "gen_imgs.mean 0.46503743529319763\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 88%|████████▊ | 10500/12000 [8:53:42<1:17:39,  3.11s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.006425923202186823 -0.030460458248853683 0.006330698262900114\n",
      "gen_imgs.mean 0.4597504734992981\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 88%|████████▊ | 10600/12000 [8:58:45<1:11:54,  3.08s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.012597919441759586 -0.04552415758371353 0.039260830730199814\n",
      "gen_imgs.mean 0.46669620275497437\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 89%|████████▉ | 10700/12000 [9:03:52<1:07:32,  3.12s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.01210830919444561 -0.05689021944999695 0.060676656663417816\n",
      "gen_imgs.mean 0.4616939425468445\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 90%|█████████ | 10800/12000 [9:08:56<1:00:40,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.013785630464553833 -0.04134773835539818 0.0532836951315403\n",
      "gen_imgs.mean 0.46502405405044556\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 91%|█████████ | 10900/12000 [9:14:03<55:57,  3.05s/it]  "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.010696999728679657 -0.05325273051857948 0.024496428668498993\n",
      "gen_imgs.mean 0.460718035697937\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 92%|█████████▏| 10999/12000 [9:19:07<50:29,  3.03s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.025265660136938095 -0.048382364213466644 0.034355372190475464\n",
      "gen_imgs.mean 0.4655207097530365\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMUAAADDCAYAAAAyYdXtAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAcQ0lEQVR4nO2deWxld3XHv+ft73n3jGfGs3rWMEkgkxDSYWkhgUCgVQNVi0BVG1UIqARSUfNPRKWWSpVKpQLqH4gqtClBogmURaRRCkRhCQGSzGRIJrNlZjL72B57vD7bz2+59/SP95z43e9xxrE9b+xwPpJl+/i+e3+/e+/xved3NlFVOI7zKrFrPQDHWW64UjhOBFcKx4ngSuE4EVwpHCeCK4XjRFiUUojIXSLykoicFJH7lmpQjnMtkYX6KUQkDuA4gDsBXACwD8DHVfXIa3xGo2qYa0pZG5IolmD9Tcbj5nHKxQp/PsbbptJNJGttbuHtUjzGWIzHUw7KJCuWS+YYC9MFkikCkiWTfGyFcX6Mf2/JBM85CPjcBKE5RAQh3xsa8hitHYjx2aBkHLvMslQqyfub4/93vsjnt2yMMayUI78HCIOQTySAhHmk+XEbgJOqegoARORhAHcDmFMpEAOQqx/H7r0bebMEj7VpdY5ka1rbzcMMnB4iWTrdRrItO24l2fvfdQfJNm7u4fHkMiTrHe4l2ene8+YYX3yZT1NJR0m2du0Wkqmp4HzONnTx+Rkdu0SysSnz3kB+mm+u4gSPEWOTJMoYnx07P0Cy4YuDJNu8ie+JlPD5BoAnT14kWV9hmGT5ofp553v5HplhMa9PGwDMvuIXarI6RORTIrJfRPbDnefOCmAxTwrr3wvd9qp6P4D7AUDi4mrhLHsWoxQXAGya9ftGAPz+MBuJAYn69/iePW+nzcKwSLL+SX70vjQ5YR5mAtMkaw54n8HFZ0hWlN8jWS5j2DMpfoU53fsSyX783JPmGKfLF0hWbm7mMcoqkqWa1pBsyxp6SGMizu/b+46OkWzIPo0oF9hGOjvK1wGGLdVhvOJlMmwrdF3Pr4c37b2dZD/65v+aY3z2qZ8bUj4OwGOci8W8Pu0DsFNEtopICsDHADyyiP05zrJgwU8KVa2IyGcB/BhAHMADqnp4yUbmONeIxbw+QVUfA/DYEo3FcZYF7tF2nAiLelK87oOl01i1a1udbO1q1svWOBucrQEbjZfG2PEDAKemWN5l+TRGeV3gqZ8+RLLmJJ+mXHsnyXov8/6On7PfKBPC80m3Gg7BbjYaC2V2lg2W2X+QSbL/YMIwnle3sA8HANq6O0i2PmAjf2CU/QIb03zOVsXZ1/THt72bZLs37STZ088fMMeIn1hCdsqiJXJPTE7Z+4M/KRyHcKVwnAiuFI4TwZXCcSI01NBOptNYv6Xe0E5mu2i7rZ2bSNYMdrtqhoPbAGB6qptkWWUDsynHxz5wnAP1ptueItm5fg5EQ5IjXzesYoMcAHovG0bwhBFlW2CD/NTIaZKdDPMkW9PMXuWL59go7tnJ3nAA2JVlb/qmNEcR37CaDfWmuLGAony7ScjRB8N5vq7vveMd5hiHjcWW7//6FyR705b6xZvTT50w9wf4k8JxCFcKx4ngSuE4EVwpHCdCQw3tVCqLjVtuqJM1t7ORt2P7jSRbU+TQ7zNDz5rHWdXMHs1yPxtvkwU20takd5FMB9jY7TvWT7J8aYTHsrbVHGPCCG9O5djjmynyJQrGOatNp9mjvb5rOx+4nVNa1mcMDzCA5gwfO2HkvXZmOCsuboTqJ5U/e+78SZI9vu9RkmXVXgy4cc1W3ueGYyTriURS9O/n0P0Z/EnhOBFcKRwngiuF40RwpXCcCK4UjhNhUatPInIGQB5AAKCiqlxIaTYKaGTBp/cc1/1Jb+fVjI40hyxki3a9Ik1kSTYZ49WQUsCrOOvWcyJ9LsWhGiWjVlIhz3kO/ZNcKAAAurp5n6k0b5er8BjbjOJlqTR/+K1bt5Hs+jX84dK0nVvQFnAhtoRx7P6zfA3DSV4NE6Pw3vAYH/vsCV4ZOn74eXOMoXHStm9dS7KWpvrwFKs43gxLsSR7u6peXoL9OM6ywF+fHCfCYpVCAfxERJ4TkU9ZG8yuEFgqcGSo4yw3Fvv69E5V7RWRNQAeF5FjqlpX/Wt2hcD2teu9QqCz7FlsiZve2vcBEfkBqkWX7ZJ4AIJKCaOD9cn9OWVDSQN+okzmOdSiNDxuHqdimDihUXR3epKL7K7ddQvJbtj1NpL99DeHSHbyDOdirF/HOQkAMAw2wEslzqcY7edwkol+oziwUZx5YoxzLFIxzodY07neHGNXKy8G9A/yeKZHeTFgqJ/nN1ngMJh8nq/16QtclDrZZlSnB7Bzxw7eNsu39anz9QUkimXOfZlhwa9PItIkIi0zPwN4PwC+UxxnhbGYJ8VaAD+Qai+JBID/VtUfLcmoHOcaspiymacA3LSEY3GcZYEvyTpOhIbmUwABoPXGVjbFhujoGBtph45xsv7pi2fMo8SUDbqBc1xCvlxi4/Rj27hows238QPxM3/15yS795/ZCJ3M23H7WSNhP4izoV0e4blkpthIHBrjubz4K84r2LWZvdx/8IF3mmMcHGBP9bFjnP/wlu27SdaqHEHwq8NnSZZq4bySd/3+bSTLZTlKAQCKo7zo8Pj+/SSbLtZfm3KRz/8M/qRwnAiuFI4TwZXCcSK4UjhOhIYa2qIVZIJ6w7E4xt7QIye47PqRk1zmfnTUqNIHIAQbeRcGjHZ8Rjh55zr2+DY1cyj7bXvZ+L73r/+SZPuO2CHP2Rgby4Uye3dHx9kTPzFseK+HeHHi9KHjJOtp5lL6+w7sM8d47BwbsZu7V5NsZIiLQvz20M9INjrMiw7BZb4FB8/zosrouNGqGEBphM/ZoQGOaNi8rd47r6/RqtefFI4TwZXCcSK4UjhOBFcKx4nQUEM7Hhe0NNfr4SXDE/vc4d+S7PIIeyAvDfaZx4nl2DOcambPaTnkhOPnjrLRuWs7p563tHNfvtvfvZdkG7atM8dYBM8nP2Z4kF94kWSp4ik+To4XAz54150k27p5M8n2H7bL0u/YyPnqOzZwm4P/ePirJHv619y+oHstG+kx4VuwnOQFkEza/v/d2s0VGN/cxZUWs+31xxk8xK0dXhnTnH9xnN9RXCkcJ4IrheNEcKVwnAhXNLRF5AEAfwRgQFVvrMk6AXwbQA+AMwA+qmrEa0eIxeLI5ur7o40UOI94pMAFsibzLAtidjG01mbO+5Ym9uSWhb3KvUNGcbYsG7Gr2jh/ua2Ve79t6ObCXABw6DR7m58fYg/yuk6jONsm9rrvuI4N4L3v4NzypiaeS/8gh9UDwLoO7gmYCHjBojTO16a1iceTTvL/YOsSJrNcqCzVat+qCeHFkjDgOaY768cdT8z9PJjPk+IbAO6KyO4D8ISq7gTwRO13x3lDcEWlqJWsiQbg3A3gwdrPDwL48BKPy3GuGQu1Kdaqah8A1L7zu0mN2cXQpgu8/uw4y42rbmir6v2qequq3prJGhWEHWeZsVCP9iUR6VbVPhHpBmBbahHCIMTUZL0ntzLFhlIiwx7pXBvr7w0bbW9xtp2tt1C5mJYqG8Y338TF0Kz84HiCjcFYnGWr5mguf2OSi3h1dbCXPJzmfR49dpBkY5c4/3lskMPOcwk2gFe1tptj7DbO7/Q4F6BLTfNt1Jllr3JXh9HTL8uGe2CUNk8Z1x8ARI3+f0UOC8+0RQzt+OIMbYtHANxT+/keAD9c4H4cZ9lxRaUQkYcA/AbAdSJyQUQ+AeCLAO4UkRMA7qz97jhvCK74+qSqH5/jT+9d4rE4zrLAPdqOE6GhoeOhBpgq14eKl+Pc4D2VY0OrOccGZ0cbe3YBINfOq1zZFOt/sciez9UdvE8BjxFGjq8Y3llLBgCZFBvvPeu4EFtgtPeKhezlPhVwjvboCHvn29u4kXx3Fx8XADpbO0gWtrFRns3xIkZqkMfd3cmh44kWvgaTgVEt3V4LgITG+anwvSIt9RciHlt6Q9tx3rC4UjhOBFcKx4ngSuE4EVwpHCdCQ1efMpkUdl+3sU52+iJXjUtlOT7fWDRBS7u1KgSsXsd5AC1pXuUYm+A+en1DZ0g2ZfTLSyZ4fxOTPO7hEbuyXb9R5n7Hxq0kq1Q4wT5hNGlft55XpCaGuVJescRjzGTsVbzxPK9odW/glaoPvO8PSfat7/4XyYZGeDzdLby/7T3bSRZm+VoBQN/gOZLFjcbxseidPteyIPxJ4TiEK4XjRHClcJwIrhSOE6GhhnYum8Fbb7y+Trapm428RJqNqua0UeEvaRtLsSRPq1LifQbgvI2T/S+R7NJlLg0fM07dk7/4BckuD9uLAW2dPSRb1c65Ck1pDmOIpzmMpW0VV8pLpdkgbzLyHLJz9JMbGuRFgnFj2w3dbOS/7c3Xk2yowIsLhRKX0peYkYMS2KXzw9DInQGHCYHCQbwUv+PMG1cKx4ngSuE4EVwpHCfCQisEfgHAJwHMWE6fV9XHrrSvWAxI5+oNo644J+tnMhyfb+WZF2NshAJAaHh8J5SnmsvydpUKG26Hj3PfurDCpfQPvfBzku3e/SFzjF1tPJ7SNHuQO5p5IWIq5EUDUTYuM2k+j6EaRSFa2UgHgEqJ9/nsM78kWSnPiwndXRtItmM1e+xP93MbgIlJNr6Hi+wNB4AgMJrEh7yYgExkjMY9MsNCKwQCwFdUdU/t64oK4TgrhYVWCHScNyyLsSk+KyIHReQBEeG8xRqzKwSOj3EwmuMsNxaqFF8DsB3AHgB9AL4014azKwS2thnveo6zzFiQR1tVX+kmLiJfB/DofD4XhAHGpurLzafibPjF4xwnXihyMnsBdjhxaPRRM0RoznKFwHSCDe2zF7jHXGuGDbVt27eRrDTFxjMAHHqxn2R73sL96DqauJR/7+BJkiWUjd2gwrV7J/N8HjXG1wAAkjH2nF/o5XNx/ASH/xvdEHBLM4eEd2S5mMF0xriuc/S8673MoeMp4W1TsfpQ/9hSh47XSmXO8BEAhxayH8dZjsxnSfYhAO8BsFpELgD4BwDvEZE9qAaQnAHw6as4RsdpKAutEPifV2EsjrMscI+240RoaOg4hCuzWY7FacNbXDY8trY/G4Cw4ZiMsXc3m+Ppa5lDo8MW3l9lipeXkwne3zPPcdl8AChOc1j2ju2c910OWVYK2OM7rWxUlw1ZkGCD/OXeI+YYV3dyL54NG1aR7NDJYyTrHeLjhM+8SLLuzbzY0fPm3XzcnH21B0Z6SWYUnUQqcqsLPEfbceaNK4XjRHClcJwIrhSOE6GhhrYAiEe6iZcqbMSWy1Y4MOtvKHb+cxiwUS5GmHkqxXnfTRl2xU7m2NidKvC4UwEb87968tfmGDvX87ZDU+zx3RKyIVoA53KXAy6aVjbCxK0rHljbARjI8yrIYMDtDaeFDf+0EZVw8iLnaD/9wgsk+5NWDpff9iY+NwAwaRSbKysvlmxsioxnkaHjjvM7hSuF40RwpXCcCK4UjhOhsR5tABKxb8QqXGU0DLc2i4khBKAhG1FhaHjEYxyirMY+0xmjENsUe4sT4GrXLx2zvcU3t2wk2aXxiyQbGOP86Uv5SySrlDkkPF9mA7hQZqM6ASPOG0CpZPQJzHMSZjFlJI8VeHEiZRRie7mX53Lk5fMk27x9pznGTIIXIvITfC6KhfpjGwESr+BPCseJ4ErhOBFcKRwngiuF40SYT+bdJgDfBLAOVXP3flX9NxHpBPBtAD2oZt99VFVHXnNnqggjHmxLK+NG/qwaLZvm8kqGgeElD9gw1gp/PpZiIw1GIbWKdWxjf4ZvHgDwpq1cGGzUyDc+dpLncmaCPdpx4fMzHfLRxwtsFM9pdAaGxz+6UgKgrcUIyzcK2hWNlmhWFfS4UfnOWsQAgNYUF5Lpm2DPebmtftyLNbQrAO5V1d0A9gL4jIhcD+A+AE+o6k4AT9R+d5wVz3yKofWp6oHaz3kARwFsAHA3gAdrmz0I4MNXa5CO00hel00hIj0AbgbwDIC1qtoHVBUHAKdpob4YWn68sLjROk4DmLdSiEgzgO8B+Jyq8kvtHMwuhtbSanfMcZzlxLyUQkSSqCrEt1T1+zXxpZn6T7XvHFPsOCuQ+aw+CaolbY6q6pdn/ekRAPcA+GLt+w+vtC9VRVCqz4HQGC8DxIzcBzFWgKxwDgAoF3mlySqxH8RZJmXOS1DlVRNVDhGJvY6X0c1GI/nRES5ycHByiGR9RoN4iXH+Qtk4j4USnxsxKuoBQFOKV4ZyOZalhFe5tm1ZT7LRFIedtKR4Veniy1zg4Hg3t2wAgP7zXLFwuszznhitr9QYBHOWvZhX7NM7AfwFgBdFZKZRw+dRVYbviMgnAJwD8Gfz2JfjLHvmUwztKWDOeiDvXdrhOM61xz3ajhPBlcJxIjQ0n0JDRaVYb6BqzAi/CDm0waqcXgnswgXlIhvBlTLrv8TYUFfDWjbscaSTHAIRj/PpzHK0AwCgZORtdK/qJln/BId+9Be4vH8xyXMuGOEuU4aBmTMqGwJA0ph4S4YrBHa28KLB5i2cL9IU47513evZgG42KjdO5G0vwGSejfeAGskDE4n662Xl18zgTwrHieBK4TgRXCkcJ4IrheNEaKihHYYhpibqDSONc5BgImkY0IZhWg5tQ3u6wPsslwxr2cjRSBlVB9NJ/mwuxwnzcSPkv7PDdvGcOcHl699z+00kqwj3twtjfOz2NvZoB0n2xIcxHk9M7VyFpmwnyXq6biFZNs4V/bo62dBOGm0Obnn7jSS7bjcXKejs7DLHuGYTL068dPZlkhWm6hcd4nHu0zeDPykcJ4IrheNEcKVwnAiuFI4TobGGdhAgP14fmm0Z2skke2JDo5F8eQ6PdqHAocyWRxtGyHTSKDVgeaWN1nFob2PDdOt1PeYY9x9hQ/u6G9jQbuvawh+e5nDppHEpc3H2uqeN0O/mJg7zBoBVbWzwXh7hUPYfPf1LknV2sKc6l2GDPshyVEGQ4iiHF88+TzIAOHWGFyLG83xPhZHbJ5hjkQbwJ4XjEK4UjhPBlcJxIrhSOE6ExVQI/AKATwKYKcf2eVV97LX2FQQhxsfqPdqBcEW+mBg51iEbaRUjRBgAikbouJbZ6LQM7ZTRhF5zhufbyBkPW9nr3trCnmYAODJ9mmSXR3jeLV3c62104iyPx/A+Z6Z5jGXl8bQ3m9WJ8MLzh1l2jL3FodGDLyy3k2xTj+GJ7+Q2AJeGuQbGxCT3tgOAppxxvYxrOJ2vz2sXw7M/w3xWn2YqBB4QkRYAz4nI47W/fUVV/3Ue+3CcFcN8crT7AMwUPcuLyEyFQMd5Q7KYCoEA8FkROSgiD4gIV7pFfYXAqSm7Na3jLCcWUyHwawC2A9iD6pPkS9bnZlcIzOWM93rHWWbMy6NtVQhU1Uuz/v51AI/OY0+Ia70RFAY8hCBgr7JZdV9tnY4bJeTVOE4I3qlVF6xseMjHAzYuK138JMymbINu1xY2eLvXsbd5bIyN/GPHeCHB8vivijZUB6BFlh3uPWOO8ekXufF7Ksee4LVr2agugPOxzw+yzIjUhyZ5AaXViBYAADUMZk0Y4fHp+usvVtL/zLZz/uXVD5sVAmdKZtb4CIBDV9qX46wEFlMh8OMisgeAotq05dNXZYSO02AWUyHwNX0SjrNScY+240RoaOi4aAwS1BvacXDeblBiw1aEvZTJhJ1bDLDBa7SjAwyPeGhUMi8b2yWMaunDw9zyr9XI5QaAA/1cxOv4medItm4Nh29nWjknuu8yH7toNIcvFbhxe//lPnOMk2VuJN8pPJ/+cd5ubRcbxpLmczahXOU9meRrXTaK5gFAYPTgq8TYS16MREmE8GJojjNvXCkcJ4IrheNEcKVwnAgNNbSDIMTEaL0RrWI0gi8abs4EG0aSsHXaKpIWlPnzgfJ25h6FjWJJs/F9+NQBkp0dOGOOkf3PwMGz7EGOZXjLjjY2Ti/nOXd6tMBN1oOyseKQZCMdABJp4zwm2CvdZnjO01mjcrxxDVvibBQnUhyRIHE7bi6b4oJvlTxfxalyfeh4+Brd5f1J4TgRXCkcJ4IrheNEcKVwnAiuFI4TobEVAlVRiDSXtxIlStMss9zyGrcLF1SMVSU1ci/Kyp8PjH57oTHGMM0rJFMFXikaMZrVA0CSK8ij3MRzvFDoJVnJ6Nc2ptxwfrrE84uDx10xSvYDgIJDKxJxPo8J4c+Pj/K8xThnaqwCibEiGUvY17qtiY9dLhq9DMNoTOsi8ikc53cNVwrHieBK4TgR5pOOmhGRZ0XkBRE5LCL/WJNvFZFnROSEiHxbrNhux1mBzMfQLgK4Q1UnagUMnhKR/wPwt6gWQ3tYRP4dwCdQrfAxJ0GoGIuEcEyV2KgqGTEQqQyHA9jVDIBKYDSSNyoMqljFDAwj3fjXMVnm0IZ8kfenqXXmGNszPO8mo+a/GoZxoFaBBM6xSBpJ/WGFPxs3chIAIIyzEVsUHk8x0cpjNIpPpIzonbjRKNC6rFqxc2fGx9gAHxniGyiRjd4/izC0tcrMUkKy9qUA7gDw3Zr8QQAfvtK+HGclMC+bQkTitaIFAwAeB/AygFHVV9Y+L2COqoGzi6EVrUA/x1lmzEspVDVQ1T0ANgK4DcBua7M5PvtKMbR0uqFuEcdZEK9r9UlVRwH8HMBeAO0ir7yUbwTAXibHWYHMpxR/F4Cyqo6KSBbA+wD8C4CfAfhTAA8DuAfAD6+8rxiqu3iVorLBk2xbTbJ4lo0+naPKW7loNJcv86ubGA3rY8KGm8ZYNml4gctthgHczIUZADu3INvMl6OpiY3vprgxb8s6rfCcK9PspQ6KcxQFMLbNGcZ3RwsXM0imjNYJFaNvoeGdt/5XVyq2R1sMwz+V4PyQeKx+O3mN58F83me6ATwoInFUR/sdVX1URI4AeFhE/gnAb1GtIug4K575FEM7iGql8aj8FKr2heO8oXCPtuNEcKVwnAiic3iFr8rBRAYBnAWwGjBqta9MfC7LkyvNZYuqdll/aKhSvHJQkf2qemvDD3wV8LksTxYzF399cpwIrhSOE+FaKcX91+i4VwOfy/JkwXO5JjaF4yxn/PXJcSK4UjhOhIYrhYjcJSIvichJEbmv0cdfDCLygIgMiMihWbJOEXm8lpb7uIh0XMsxzhcR2SQiPxORo7U047+pyVfcfJY6ZbqhSlELKvwqgA8CuB7VDqvXN3IMi+QbAO6KyO4D8ISq7gTwRO33lUAFwL2quhvVVIDP1K7FSpzPTMr0TQD2ALhLRPaiGs39ldpcRlBNmb4ijX5S3AbgpKqeUtUSqmHndzd4DAtGVZ8EEG3wdjeq6bjACkrLVdU+VT1Q+zkP4Ciq2ZMrbj5LnTLdaKXYAGB2J8I501hXEGtVtQ+o3mgA1lzj8bxuRKQH1UjoZ7BC57OYlOkojVYKKyvI14SvISLSDOB7AD6nquPXejwLZTEp01EarRQXAGya9fsbIY31koh0A0Dt+8A1Hs+8qZUs+h6Ab6nq92viFTsfYGlSphutFPsA7KytCqQAfAzAIw0ew1LzCKrpuMA803KXAyIiqGZLHlXVL8/604qbj4h0iUh77eeZlOmjeDVlGng9c1HVhn4B+BCA46i+8/1do4+/yLE/BKAP1e71F1BdzViF6irNidr3zms9znnO5V2ovk4cBPB87etDK3E+AN6Cakr0QQCHAPx9Tb4NwLMATgL4HwDp+ezPwzwcJ4J7tB0ngiuF40RwpXCcCK4UjhPBlcJxIrhSOE4EVwrHifD/0zIxtHn/bPUAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 92%|█████████▎| 11100/12000 [9:24:15<49:51,  3.32s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.016583183780312538 -0.06951490044593811 0.021266750991344452\n",
      "gen_imgs.mean 0.4695119261741638\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 93%|█████████▎| 11200/12000 [9:29:20<40:42,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.01966857723891735 -0.06366699188947678 0.0457640178501606\n",
      "gen_imgs.mean 0.47464412450790405\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 94%|█████████▍| 11300/12000 [9:34:25<35:27,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02274242229759693 -0.0781189501285553 0.04306270554661751\n",
      "gen_imgs.mean 0.46650251746177673\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 95%|█████████▌| 11400/12000 [9:39:33<30:33,  3.06s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.010916903614997864 -0.05608580261468887 0.0355733223259449\n",
      "gen_imgs.mean 0.46471700072288513\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 96%|█████████▌| 11500/12000 [9:44:38<25:25,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.007861430756747723 -0.041074488312006 0.04980313777923584\n",
      "gen_imgs.mean 0.4717915654182434\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 97%|█████████▋| 11600/12000 [9:49:42<20:17,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.003600078634917736 -0.05118304863572121 0.020463228225708008\n",
      "gen_imgs.mean 0.46514007449150085\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 98%|█████████▊| 11700/12000 [9:54:45<15:11,  3.04s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.04324229806661606 -0.04115341603755951 -0.009917104616761208\n",
      "gen_imgs.mean 0.46633291244506836\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 98%|█████████▊| 11800/12000 [9:59:50<10:06,  3.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) 0.0011308398097753525 -0.06693989038467407 0.014361164532601833\n",
      "gen_imgs.mean 0.46394217014312744\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 99%|█████████▉| 11900/12000 [10:04:55<05:04,  3.05s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.023788832128047943 -0.03997358679771423 0.039204586297273636\n",
      "gen_imgs.mean 0.46472617983818054\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████▉| 11999/12000 [10:09:56<00:03,  3.02s/it]Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred) -0.02750680409371853 -0.05895908921957016 0.050397854298353195\n",
      "gen_imgs.mean 0.4641554057598114\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 12000/12000 [10:10:00<00:00,  3.16s/it]\n"
     ]
    }
   ],
   "source": [
    "for epoch in tqdm(range(1, 12001)):\n",
    "    if epoch % 6001 == 0:\n",
    "        g_opt = t.optim.RMSprop(generator.parameters(), lr=4e-5)\n",
    "        c_opt = t.optim.RMSprop(critic.parameters(), lr=4e-5)\n",
    "\n",
    "    for i, (imgs, _) in enumerate(dl):\n",
    "        # train critic\n",
    "        real_imgs = imgs.to(device)\n",
    "        randn = t.randn(bs, z_dim)\n",
    "        gen_imgs = generator(randn.to(device))\n",
    "\n",
    "        c_opt.zero_grad()\n",
    "        ri_pred = critic(real_imgs)\n",
    "        gi_pred = critic(gen_imgs)\n",
    "        c_loss = (-t.mean(ri_pred) + t.mean(gi_pred)) / 2\n",
    "        c_loss.backward()\n",
    "        c_opt.step()\n",
    "\n",
    "        # clip discriminator weights\n",
    "        for p in critic.parameters():\n",
    "            p.data.clamp_(-.01, .01)\n",
    "\n",
    "        # train generator\n",
    "        if i % 3 == 0:\n",
    "            gen_imgs = generator(randn.to(device))\n",
    "            g_opt.zero_grad()\n",
    "            gi_critic_pred = critic(gen_imgs)\n",
    "            g_loss = -t.mean(gi_critic_pred)\n",
    "            g_loss.backward()\n",
    "            g_opt.step()\n",
    "\n",
    "    if epoch % 100 == 0:\n",
    "        print('-t.mean(ri_pred), t.mean(gi_pred), -t.mean(gi_critic_pred)', -t.mean(ri_pred).item(), t.mean(gi_pred).item(), -t.mean(gi_critic_pred).item())\n",
    "        print('gen_imgs.mean', gen_imgs.mean().item())\n",
    "\n",
    "    if epoch % 1000 == 0:\n",
    "        for img in generator(t.randn(3, z_dim).to(device)):\n",
    "            plot_tensor(img)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "t.save(generator.state_dict(), 'models/state_dicts/04_02_horse_gen.pth')\n",
    "t.save(critic.state_dict(), 'models/state_dicts/04_02_horse_critic.pth')"
   ]
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}