2.Classifier_free_guidance.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "51618a18-7bda-4d87-bce4-06ecfe9a0889",
   "metadata": {},
   "source": [
    "# ISBI 2024 DiMEDIA: MONAI Tutorial: DDIM + Classifier Free Guidance\n",
    "\n",
    "This tutorial illustrates how to use MONAI for training a denoising diffusion implicit model (DDIM) to create synthetic 2D images using the classifier-free guidance technique [1] to perform conditioning.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1565072f-c76a-43d4-b557-105b01eee9d5",
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install -q matplotlib monai monailabel monai-generative\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "226301c8-105a-43fd-8179-38692f5ca53b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MONAI version: 1.3.0\n",
      "Numpy version: 1.26.2\n",
      "Pytorch version: 2.2.2+cu118\n",
      "MONAI flags: HAS_EXT = False, USE_COMPILED = False, USE_META_DICT = False\n",
      "MONAI rev id: 865972f7a791bf7b42efbcd87c8402bd865b329e\n",
      "MONAI __file__: /remote/rds/users/<username>/miniforge3/envs/dev/lib/python3.10/site-packages/monai/__init__.py\n",
      "\n",
      "Optional dependencies:\n",
      "Pytorch Ignite version: 0.4.11\n",
      "ITK version: 5.3.0\n",
      "Nibabel version: 5.2.1\n",
      "scikit-image version: 0.22.0\n",
      "scipy version: 1.11.4\n",
      "Pillow version: 10.3.0\n",
      "Tensorboard version: 2.16.2\n",
      "gdown version: 4.7.3\n",
      "TorchVision version: 0.17.2+cu118\n",
      "tqdm version: 4.66.1\n",
      "lmdb version: 1.4.1\n",
      "psutil version: 5.9.5\n",
      "pandas version: 2.1.4\n",
      "einops version: 0.7.0\n",
      "transformers version: 4.39.3\n",
      "mlflow version: 2.11.3\n",
      "pynrrd version: 1.0.0\n",
      "clearml version: NOT INSTALLED or UNKNOWN VERSION.\n",
      "\n",
      "For details about installing the optional dependencies, please visit:\n",
      "    https://docs.monai.io/en/latest/installation.html#installing-the-recommended-dependencies\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import shutil\n",
    "import tempfile\n",
    "import time\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn.functional as F\n",
    "from monai import transforms\n",
    "from monai.apps import MedNISTDataset\n",
    "from monai.config import print_config\n",
    "from monai.data import CacheDataset, DataLoader\n",
    "from monai.utils import first, set_determinism\n",
    "from torch.cuda.amp import GradScaler, autocast\n",
    "from tqdm import tqdm\n",
    "\n",
    "from generative.inferers import DiffusionInferer\n",
    "from generative.networks.nets import DiffusionModelUNet\n",
    "from generative.networks.schedulers import DDPMScheduler, DDIMScheduler\n",
    "\n",
    "print_config()\n",
    "set_determinism(42)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "707db2fb-aa0c-4681-9d49-2768fafb697e",
   "metadata": {},
   "source": [
    "## Setup MedNIST Dataset\n",
    "\n",
    "In this tutorial, we will train our models on the MedNIST dataset available on MONAI `https://docs.monai.io/en/stable/apps.html#monai.apps.MedNISTDataset`. \n",
    "\n",
    "Here, we will use the `Hand` and `HeadCT`, where our conditioning variable `class` will specify the modality."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "15986f04-84e5-4b5e-8d4f-4a10047cca89",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2024-05-24 10:58:52,027 - INFO - Verified 'MedNIST.tar.gz', md5: 0bc7306e7427e00ad1c5526a6677552d.\n",
      "2024-05-24 10:58:52,029 - INFO - File exists: data/MedNIST.tar.gz, skipped downloading.\n",
      "2024-05-24 10:58:52,031 - INFO - Non-empty folder exists in data/MedNIST, skipped extracting.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Loading dataset: 100%|██████████████████████████████████████████████████████████████████████████████████████| 15990/15990 [00:29<00:00, 533.13it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2024-05-24 11:01:01,852 - INFO - Verified 'MedNIST.tar.gz', md5: 0bc7306e7427e00ad1c5526a6677552d.\n",
      "2024-05-24 11:01:01,854 - INFO - File exists: data/MedNIST.tar.gz, skipped downloading.\n",
      "2024-05-24 11:01:01,857 - INFO - Non-empty folder exists in data/MedNIST, skipped extracting.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Loading dataset: 100%|████████████████████████████████████████████████████████████████████████████████████████| 1977/1977 [00:03<00:00, 556.34it/s]\n"
     ]
    }
   ],
   "source": [
    "root_dir = './'\n",
    "data_dir = './data/'\n",
    "os.makedirs(data_dir, exist_ok=True)\n",
    "train_data = MedNISTDataset(root_dir=data_dir, section=\"training\", download=True, progress=False, seed=0)\n",
    "train_datalist = []\n",
    "for item in train_data.data:\n",
    "    if item[\"class_name\"] in [\"Hand\", \"HeadCT\"]:\n",
    "        train_datalist.append({\"image\": item[\"image\"], \"class\": 1 if item[\"class_name\"] == \"Hand\" else 2})\n",
    "\n",
    "train_transforms = transforms.Compose(\n",
    "    [\n",
    "        transforms.LoadImaged(keys=[\"image\"]),\n",
    "        transforms.EnsureChannelFirstd(keys=[\"image\"]),\n",
    "        transforms.ScaleIntensityRanged(keys=[\"image\"], a_min=0.0, a_max=255.0, b_min=0.0, b_max=1.0, clip=True),\n",
    "        transforms.RandAffined(\n",
    "            keys=[\"image\"],\n",
    "            rotate_range=[(-np.pi / 36, np.pi / 36), (-np.pi / 36, np.pi / 36)],\n",
    "            translate_range=[(-1, 1), (-1, 1)],\n",
    "            scale_range=[(-0.05, 0.05), (-0.05, 0.05)],\n",
    "            spatial_size=[64, 64],\n",
    "            padding_mode=\"zeros\",\n",
    "            prob=0.5,\n",
    "        ),\n",
    "        transforms.RandLambdad(keys=[\"class\"], prob=0.15, func=lambda x: -1 * torch.ones_like(x)),\n",
    "        transforms.Lambdad(\n",
    "            keys=[\"class\"], func=lambda x: torch.tensor(x, dtype=torch.float32).unsqueeze(0).unsqueeze(0)\n",
    "        ),\n",
    "    ]\n",
    ")\n",
    "train_ds = CacheDataset(data=train_datalist, transform=train_transforms)\n",
    "train_loader = DataLoader(train_ds, batch_size=128, shuffle=True, num_workers=4, persistent_workers=True)\n",
    "\n",
    "val_data = MedNISTDataset(root_dir=data_dir, section=\"validation\", download=True, progress=False, seed=0)\n",
    "val_datalist = []\n",
    "for item in val_data.data:\n",
    "    if item[\"class_name\"] in [\"Hand\", \"HeadCT\"]:\n",
    "        val_datalist.append({\"image\": item[\"image\"], \"class\": 1 if item[\"class_name\"] == \"Hand\" else 2})\n",
    "\n",
    "\n",
    "val_transforms = transforms.Compose(\n",
    "    [\n",
    "        transforms.LoadImaged(keys=[\"image\"]),\n",
    "        transforms.EnsureChannelFirstd(keys=[\"image\"]),\n",
    "        transforms.ScaleIntensityRanged(keys=[\"image\"], a_min=0.0, a_max=255.0, b_min=0.0, b_max=1.0, clip=True),\n",
    "        transforms.Lambdad(\n",
    "            keys=[\"class\"], func=lambda x: torch.tensor(x, dtype=torch.float32).unsqueeze(0).unsqueeze(0)\n",
    "        ),\n",
    "    ]\n",
    ")\n",
    "val_ds = CacheDataset(data=val_datalist, transform=val_transforms)\n",
    "val_loader = DataLoader(val_ds, batch_size=128, shuffle=False, num_workers=4, persistent_workers=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "763fc3ee-ccc6-4edd-b7db-cec65c3dde4a",
   "metadata": {},
   "source": [
    "### Visualisation of the training images\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "75f4f7b9-a66d-4626-ac71-6ec1d5a36873",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
      "To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
      "To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
      "To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "batch shape: torch.Size([128, 1, 64, 64])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "check_data = first(train_loader)\n",
    "print(f\"batch shape: {check_data['image'].shape}\")\n",
    "image_visualisation = torch.cat(\n",
    "    [check_data[\"image\"][0, 0], check_data[\"image\"][1, 0], check_data[\"image\"][2, 0], check_data[\"image\"][3, 0]], dim=1\n",
    ")\n",
    "plt.figure(\"training images\", (12, 6))\n",
    "plt.imshow(image_visualisation, vmin=0, vmax=1, cmap=\"gray\")\n",
    "plt.axis(\"off\")\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fa96ffaa-c768-472f-87e2-4ee4ff7568e4",
   "metadata": {},
   "source": [
    "## Define network, scheduler, optimizer, and inferer\n",
    "\n",
    "We are using the original DDPM scheduler containing `1000` timesteps in its Markov chain, and a 2D UNET with attention mechanisms in the 3rd level, each with 1 attention head (`num_head_channels=64`).\n",
    "\n",
    "In order to pass conditioning variables with dimension of 1 (just specifying the modality of the image), we use `with_conditioning=True, cross_attention_dim=1`.\n",
    "\n",
    "We will also use `DDIMScheduler` to enable faster sampling."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "f1007c39-1fd8-4c3c-bf23-5589c693f8ea",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda\")\n",
    "\n",
    "model = DiffusionModelUNet(\n",
    "    spatial_dims=2,\n",
    "    in_channels=1,\n",
    "    out_channels=1,\n",
    "    num_channels=(64, 64, 64),\n",
    "    attention_levels=(False, False, True),\n",
    "    num_res_blocks=1,\n",
    "    num_head_channels=(0, 0, 64),\n",
    "    with_conditioning=True,\n",
    "    cross_attention_dim=1,\n",
    ")\n",
    "model.to(device)\n",
    "\n",
    "scheduler = DDPMScheduler(num_train_timesteps=1000)\n",
    "\n",
    "optimizer = torch.optim.Adam(params=model.parameters(), lr=2.5e-5)\n",
    "\n",
    "inferer = DiffusionInferer(scheduler)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1e6bbe56-09b9-4d18-a7ca-95f753edf124",
   "metadata": {},
   "source": [
    "## Model training\n",
    "\n",
    "Here, we are training our model for 75 epochs (training time: ~50 minutes).\n",
    "\n",
    "For the purpose of tutorial, we would like to skip the training and use a pre-trained model instead, set `use_pretrained=True`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "0024a723-ad06-4260-bbdb-5dfc0f7babbb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Download start!\n",
      "Download Complete.\n"
     ]
    }
   ],
   "source": [
    "use_pretrained = True\n",
    "\n",
    "if use_pretrained:\n",
    "    model = DiffusionModelUNet(\n",
    "        spatial_dims=2,\n",
    "        in_channels=1,\n",
    "        out_channels=1,\n",
    "        num_channels=(64, 64, 64),\n",
    "        attention_levels=(False, False, True),\n",
    "        num_res_blocks=1,\n",
    "        num_head_channels=(0, 0, 64),\n",
    "        with_conditioning=True,\n",
    "        cross_attention_dim=1,\n",
    "    )\n",
    "    os.makedirs('./pretrained/', exist_ok=True)\n",
    "    import urllib.request\n",
    "    url = 'https://raw.githubusercontent.com/vios-s/ISBI-2024-DiMEDIA-MONAI-Tutorial/main/pretrained/classifier_free.pt'\n",
    "    print (\"Download start!\")\n",
    "    filename, headers = urllib.request.urlretrieve(url, filename=\"./pretrained/classifier_free.pt\")\n",
    "    print(\"Download Complete.\")\n",
    "    model.load_state_dict(torch.load('./pretrained/classifier_free.pt'))\n",
    "    model.to(device)\n",
    "else:\n",
    "    n_epochs = 75\n",
    "    val_interval = 5\n",
    "    epoch_loss_list = []\n",
    "    val_epoch_loss_list = []\n",
    "    \n",
    "    scaler = GradScaler()\n",
    "    total_start = time.time()\n",
    "    for epoch in range(n_epochs):\n",
    "        model.train()\n",
    "        epoch_loss = 0\n",
    "        progress_bar = tqdm(enumerate(train_loader), total=len(train_loader), ncols=70)\n",
    "        progress_bar.set_description(f\"Epoch {epoch}\")\n",
    "        for step, batch in progress_bar:\n",
    "            images = batch[\"image\"].to(device)\n",
    "            classes = batch[\"class\"].to(device)\n",
    "            optimizer.zero_grad(set_to_none=True)\n",
    "    \n",
    "            with autocast(enabled=True):\n",
    "                # Generate random noise\n",
    "                noise = torch.randn_like(images).to(device)\n",
    "                # Create timesteps\n",
    "                timesteps = torch.randint(\n",
    "                    0, \n",
    "                    inferer.scheduler.num_train_timesteps, \n",
    "                    (images.shape[0],), \n",
    "                    device=images.device\n",
    "                ).long()\n",
    "                # Get model prediction\n",
    "                noise_pred = inferer(inputs=images, diffusion_model=model, noise=noise, condition=classes, timesteps=timesteps)\n",
    "    \n",
    "                loss = F.mse_loss(noise_pred.float(), noise.float())\n",
    "    \n",
    "            scaler.scale(loss).backward()\n",
    "            scaler.step(optimizer)\n",
    "            scaler.update()\n",
    "    \n",
    "            epoch_loss += loss.item()\n",
    "    \n",
    "            progress_bar.set_postfix({\"loss\": epoch_loss / (step + 1)})\n",
    "        epoch_loss_list.append(epoch_loss / (step + 1))\n",
    "    \n",
    "        if (epoch + 1) % val_interval == 0:\n",
    "            model.eval()\n",
    "            val_epoch_loss = 0\n",
    "            for step, batch in enumerate(val_loader):\n",
    "                images = batch[\"image\"].to(device)\n",
    "                classes = batch[\"class\"].to(device)\n",
    "                timesteps = torch.randint(\n",
    "                            0, inferer.scheduler.num_train_timesteps, (images.shape[0],), device=images.device\n",
    "                        ).long()\n",
    "                with torch.no_grad():\n",
    "                    with autocast(enabled=True):\n",
    "                        noise = torch.randn_like(images).to(device)\n",
    "                        noise_pred = inferer(inputs=images, diffusion_model=model, noise=noise, condition=classes, timesteps=timesteps)\n",
    "                        val_loss = F.mse_loss(noise_pred.float(), noise.float())\n",
    "    \n",
    "                val_epoch_loss += val_loss.item()\n",
    "                progress_bar.set_postfix({\"val_loss\": val_epoch_loss / (step + 1)})\n",
    "            val_epoch_loss_list.append(val_epoch_loss / (step + 1))\n",
    "\n",
    "            # Sampling image during training\n",
    "            noise = torch.randn((1, 1, 64, 64))\n",
    "            noise = noise.to(device)\n",
    "            condition = (2 * torch.ones((1, 1, 1))).to(device)\n",
    "            scheduler.set_timesteps(num_inference_steps=1000)\n",
    "            with autocast(enabled=True):\n",
    "                image = inferer.sample(input_noise=noise, diffusion_model=model, scheduler=scheduler, conditioning=condition)\n",
    "\n",
    "            plt.figure(figsize=(2, 2))\n",
    "            plt.imshow(image[0, 0].cpu(), vmin=0, vmax=1, cmap=\"gray\")\n",
    "            plt.tight_layout()\n",
    "            plt.axis(\"off\")\n",
    "            plt.show()\n",
    "    \n",
    "    total_time = time.time() - total_start\n",
    "    print(f\"train completed, total time: {total_time}.\")\n",
    "    torch.save(model.state_dict(), 'pretrained/classifier_free.pt')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c92e6adc-f8ba-4c07-940a-9ac15af166f9",
   "metadata": {},
   "source": [
    "## Sampling process with classifier-free guidance\n",
    "\n",
    "In order to sample using classifier-free guidance, for each step of the process we need to have 2 elements, one generated conditioned in the desired class (here we want to condition on Hands `=1`) and one using the unconditional class (`=-1`). \n",
    "\n",
    "Instead using directly the predicted class in every step, we use the unconditional plus the direction vector pointing to the condition that we want (`noise_pred_text - noise_pred_uncond`). The effect of the condition is defined by the `guidance_scale` defining the influence of our direction vector.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "34fa9c68-9c33-4319-93b5-96db440a2a42",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████| 1000/1000 [00:21<00:00, 47.09it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model.eval()\n",
    "guidance_scale = 7.0\n",
    "conditioning = torch.cat([-1 * torch.ones(1, 1, 1).float(), torch.ones(1, 1, 1).float()], dim=0).to(device)\n",
    "\n",
    "noise = torch.randn((1, 1, 64, 64))\n",
    "noise = noise.to(device)\n",
    "scheduler.set_timesteps(num_inference_steps=1000)\n",
    "progress_bar = tqdm(scheduler.timesteps)\n",
    "for t in progress_bar:\n",
    "    with autocast(enabled=True):\n",
    "        with torch.no_grad():\n",
    "            noise_input = torch.cat([noise] * 2)\n",
    "            model_output = model(noise_input, timesteps=torch.Tensor((t,)).to(noise.device), context=conditioning)\n",
    "            noise_pred_uncond, noise_pred_text = model_output.chunk(2)\n",
    "            noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)\n",
    "\n",
    "    noise, _ = scheduler.step(noise_pred, t, noise)\n",
    "\n",
    "plt.style.use(\"default\")\n",
    "plt.imshow(noise[0, 0].cpu(), vmin=0, vmax=1, cmap=\"gray\")\n",
    "plt.tight_layout()\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "0d9b7496-5333-45b9-889a-b16937ad19c0",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████| 250/250 [00:05<00:00, 47.99it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbsAAAG7CAYAAABaaTseAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAABCV0lEQVR4nO3dZ7Be5XX//SVUjgqq56j3XlFHCKSoIAsQRhgFghE2IpgWmyTgJGPHSTwO9sSxTXUgHpsSmikDAmEQCGyEei+oo957PypHXXrePHqG+f+v34/7HPO0a76fl2uxjva97733xT2zrrUrXbhw4UIAAJCxS/7fPgAAAP7vxmIHAMgeix0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHssdgCA7FUp9D+sVKmSzJWUlCTjx48flzUNGzZMxrdu3SprioqKZO78+fPJ+JkzZ2SNU6dOnWS8atWqsubAgQPJeJcuXWTN6tWry3dgoc93RMT+/ftlTh17Rc5RrVq1ZM597xXhznmVKulL+Ny5c7JGXStnz54t34H9GapXry5zJ0+eTMbdPahmQ6jr+MuOYd++fcl49+7dZc2KFStkTt3vnTp1kjWzZ88ud02NGjVkbsmSJcl4//79Zc3ChQtlTt3XmzZtkjVNmjRJxrds2SJr2rVrJ3OHDx9Oxg8ePChrHHVff9X3dIMGDWTuyJEjybi7PwuZjcIvOwBA9ljsAADZY7EDAGSPxQ4AkD0WOwBA9ljsAADZq1To++xc23NFatQ/W1xcLGtOnz4tc0ePHk3G27dvX6G/p7Y5qO0FERGHDh1KxmvWrClrunXrJnOqrdg5deqUzKm27LVr18qa0aNHJ+OlpaUVOoZ58+Yl45deeqmsOXbsmMypFmbV6h4RsWbNmmS8ItdrhL7G3DYQt3Vj165d5T6Gpk2bJuNuC4baXhAR0bdv32R82bJlsqZx48Yyp76nlStXyhp1Xt1961r41fnr3LmzrFHXSkW1adMmGXfPCLd9ZO7cueX+e2VlZTJXEZUrV07G3bXntr2o70mdu4jCtnDxyw4AkD0WOwBA9ljsAADZY7EDAGSPxQ4AkL2CB0FXhOrSiYjo0aNHMq6GtX4Z1bHkBpi6waLq2F23qMq5QcuqgzNCdzO5v+c6rVwnqbJ8+fJkvGvXrrKmUaNGMqc6TF3XW0UGabtBuLVr107GVUfvl1HXiutYdTmlWbNmMqeu5b1798oa1xG3Y8eOZNxde65bbtGiRcm4u58U153oOlaHDx+ejE+dOlXWuONT91O9evVkjcq5Z5E7BjUI3Q3NdznFDdhW38eJEydkjRvurgZp/7mdsfyyAwBkj8UOAJA9FjsAQPZY7AAA2WOxAwBkj8UOAJC9r2QQdIcOHZLxnTt3yhrVwuxam2+88UaZ++STT5Jx14p/xRVXyJzasuDaqydOnJiMDxkyRNa41nDVVt+zZ09Zo7YKRPi2bEV9ty1atKjQv6MGPm/btk3WrFixQub69OmTjLdq1UrWqDZ4Nzx33bp1Mqe2iKhzF6Fb+yMizp8/n4y7YdnqGNxQbrUFI0Jvc3CDoDt16iRzajvK9OnTZY06PtcG79r+mzRpkoxv3bpV1pw8eVLm1JYT992qLVfunnHDstXnVfdFRMSUKVNkTn1P7pyr4dvqfEfoweAREbt3707GK/IigC/ilx0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHsFd2NecoleF1u2bJmMu+6xVatWJeOuc9ENiVbdPa4b03VCqk5N12nVv3//ZPzdd9+VNU2bNpW5Xbt2yZziOuy6d++ejC9evFjWjB49Ohl3HWK1atWSOTXM9dixYxX6e2oIrTt3qubIkSOyxnVqumtMcd2iajh4/fr1y30MbiC261xU3ZiuM9Z1N8+bNy8ZHzNmjKzZt29fMj5z5kxZ07lzZ5lT196IESNkzeTJk2WuefPmyfiwYcNkTbVq1ZLxF154QdZURLt27WRu48aNMqeuCTc8Wj1z1EDnCN09HKE7/11HqBv4fxG/7AAA2WOxAwBkj8UOAJA9FjsAQPZY7AAA2WOxAwBkr+CtByUlJTKn2n3Xrl0ra1SrtNuu4LYKqMHE27dvlzWuVVoNCz579qys6dKlSzLuBs2qbRsREdWrV0/GXZvt+vXrZU4dn2p1j4jYs2dPMu4Gubp2d9XS7tr33TBjtSXGHZ+7jhQ1nDlCb0vo2rWrrHFbN9Q94I5b3Z+uFV9t/4nQ19jAgQNljdvuoVrXly5dKmvUPa0GBUdENGzYUObU4Ga3BcNttVCt8CdOnJA1ahvNNddcI2vc80MN5nZbD9QWjIiIxo0bJ+NVqlSRNeq8uuHM7pmjjl1tGYqI+Oyzz2TuIn7ZAQCyx2IHAMgeix0AIHssdgCA7LHYAQCyx2IHAMhewVsP3OT5unXrJuOu/VW9PcAdTtWqVWVOtem7NmU3aV+1f7vp96pFuHfv3rJGtfZH6OnfJ0+elDVuMr76nnbu3Clr1PaMDz/8sELHoLYeuO0KFWlhvu6662TN/Pnzk3G3vcBdK506dUrGXWu/22KjtmG441PctefeIqLa/t1zwJ0jdb+r7TAREatXr07G3QT+M2fOyJzaYuCuPfdWEnWNuWNQ0/7dlqZrr71W5tSbBVyb/oQJE2RObYlx96B6Jqo3PET4Nxioa2LKlCmyppBljF92AIDssdgBALLHYgcAyB6LHQAgeyx2AIDsFdyN6TohVefPkSNHZI3rPlJ69Oghc/v27UvGXbfjV011ETVq1EjWTJ8+XeZcx57iOuJUJ50bLK106NBB5twwaqVv374yd+DAAZlT15i79lRHnOu0dYOqVVee6wR2w3jdIGFFDRR358F1VroOXUUN5Y7QHbquw7R27drJuBuM7AZBt2rVKhl335PqtI2IWLduXTLu7vfFixcn465zccCAATJ3+PDhZLxJkyayxnUqq8HcCxYskDXqvLrvSXXaur+nPmtERGlpqcxdxC87AED2WOwAANljsQMAZI/FDgCQPRY7AED2WOwAANmrUuh/6LYKqFz79u1lTeXKlZNxNzx3xYoVMqcGDA8cOFDWzJ07V+ZUC7Pa4hChh7JWZJtFhG4fdq34bvj2xo0bk3HXKq2OvaSkRNa4wbrq33LDrd05V1sC3DYVtTXCDad1Ww+WL1+ejKvhxxER/fr1kzk13NptI9i2bVsyrgY6R0Rs375d5lTL/f79+2WN+97V1iU1wD0i4oYbbkjG3SBjd6+dOnUqGXfn1Q0f7tatWzKuthdE6OvSPdvcc0p9Xvfd7t27V+bU1iV17iL0c1494yP8964GlLt7phD8sgMAZI/FDgCQPRY7AED2WOwAANljsQMAZI/FDgCQvYLfelCpUiWZU+3kbnr70aNHC/lnC6beOLBr1y5ZU1xcLHOqTb8iXBu8e7OBajnu1auXrHHTv1WL8OTJk2XNoEGDknHV6h4RUaWK3tHyVZ7XCD0h3W0jUNelm9LevXt3mVOt3I0bN5Y1rtVcXZfuDQFq64a7B+vWrStz6jpy94zbEqMMGzZM5lQbvNum4q5Lda+5LSK9e/eWOcW1/RcVFSXj7m0l7n5S59xtOXFt/yrnnivqnKutGRER119/vcypN8G4N0PMmDFD5i7ilx0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHsFD4JWg2EjItauXZuMN2/eXNbUq1cvGe/QoYOscUNZVeeb67RSnXwRejisG4SrqM8aETFz5sxy1y1durTcxxARsXnz5mRcdYhFRDRo0CAZnzVrlqxp2bJluY4rwnduueHgqtvWfbeq69J1sKnh0RF6GK8bYO2G5KoOStdhqr7DWrVqyRp3LQ8fPjwZd/dgRSxatEjm1DlSQ6UjIpo2bSpz6lnQuXNnWbNs2TKZU92drtvRPY+Unj17ylybNm2S8d27d8sa12E6fvz4ZFxdDxERe/bsScZdV6rrUFf34T/8wz/ImkLwyw4AkD0WOwBA9ljsAADZY7EDAGSPxQ4AkD0WOwBA9r6SQdBqWPCGDRtkjWrPdW27bsirGpLrWpHPnTtX7pxrHVZt3m7Iq2srVq27bhCuG+570003JeNuiOq6deuS8VOnTska18Kv2sm3bNkia9SQ7wj9fbjtHuo6cu3QbqvAzp07k3G1bSMiYs2aNTKntvmoLQ4up4a0R0QsXLhQ5tQ5d9+TGjQe4QdfK+o6qmhLuzqG2rVryxq3JUZdE7Nnz5Y1t912WzL+zjvvyJrTp0/L3ODBg5NxdU1G+GHs6hnmhpqr78Md9+jRo2VObaNxa8Mrr7wicxfxyw4AkD0WOwBA9ljsAADZY7EDAGSPxQ4AkL2vpBtTdeq4rsGjR48m467zTg0yjtDdmK1bt5Y1aoBphO7GVMNfI/Tn7dq1q6xxn8l1VFXEkCFDknH3PSllZWUy5zrv1HVUv359WXPw4EGZa9KkSTLuPtPhw4eTcTeE3B2D6thr2LChrHGDtFUXp7tV1XXpOu/69esnc5988kky7u7PM2fOyJy6Xlz38OrVq5Nx9z25710NvnY1rrNYdSG6Qcvq39q0aZOscZ3AzZo1S8bdteLuNTWEfOXKlbKmZs2ayXiVKvo9A+5aUc8I14154MABmbuIX3YAgOyx2AEAssdiBwDIHosdACB7LHYAgOyx2AEAsveVbD2oVq1aMu7a9FVObUmI8O3uaoBpnz59ZM2cOXNkTrX0qtbciIhdu3Yl424Yrzuv6jOpfydCfxcREYcOHUrG3TDq48ePJ+OuVXrMmDEyp4YwL168WNa0a9dO5tauXStzitoq41raly9fXu5/58iRIzKntkxEROzevTsZ79+/v6xR360bxu6uFTXE1x23G3ytBna7NnjVpu/umXHjxsncr371q2TcPQLd51XPsPXr18sapU2bNjKnvtuIiNLS0mTcDcBXg8Yj9FYVN8xePafccV9yif6d1bx582RcbRmK8M/E/+vf/NL/AgCA/59jsQMAZI/FDgCQPRY7AED2WOwAANljsQMAZK/grQeqFT+isLbP/5Wanu4mkKsp4xERZ8+eTcZd239RUZHMqUnjqs02QrfGui0Trj1dvXnBtRW7rRGqfXjAgAGyZunSpcm4ak2P8BPN1UT4JUuWyJoePXrInNqqoibcR+itB+4NAZ07d5a5K664Ihnft2+frHFbN771rW8l4+741P3krr1JkybJ3NChQ5Nx11Y/YcIEmSsuLk7G3dYD9W+5Z5GbjK9ybmvLZ599JnNqO4V7M4R6k4PbeqDOXURESUlJMq7exBHhn7Hvv/9+Mq62w0ToN8u4LSLubS/qWX7VVVfJGvcWkYv4ZQcAyB6LHQAgeyx2AIDssdgBALLHYgcAyN5XMghaDRatXbu2rFm0aFEy7oYSf/755zKnupIOHjwoa1RXXkRE1apVk3HXCTlz5sxkvG7durJGHXdERI0aNZJx95U1bNhQ5lTn4ooVK2SNGqTtOsTWrVsnc2pwc82aNWWNG1isOmDd0GT199xwWnct33PPPcn4sGHDZI3rFlUde8OHD5c1TzzxRDLuBiMPGTJE5lQHrOvk69Wrl8y98sorybgbFq86ol1XquvcVdf5oEGDZI3qiI6ImDt3rswpqkty1apV5f5bjhv27J456t51Xe2q69jdt66zXnHX67Rp0760nl92AIDssdgBALLHYgcAyB6LHQAgeyx2AIDssdgBALJX8NaDli1bypwaiNq3b19Zc+rUqWTctcG7Ia8tWrRIxktLS2WNam2O0MOR3VYG1RLtWrJVK36Ebk93n8m13KsW4W3btskadXl07dpV1jRo0EDm1Pfk2qGffvppmVNDdzt06CBr1Dly2ynUto2IiLvvvjsZd8Np1fDoiIjp06cn4zt37pQ1apvDhx9+KGtGjRolc1dffXUyfvPNN8sa94xQ11iVKlVkjRoS/dxzz8kad38q7rniWu7V8R04cEDWqK0Wbpi+u98V92xT26oi/NYSRQ2xbtWqlaxxW2/UeuLOQyHLGL/sAADZY7EDAGSPxQ4AkD0WOwBA9ljsAADZY7EDAGSv4K0HrtVctau69uoFCxakD8i8XcEd6vnz55PxQ4cOyZq2bdvKXJ06dZJxN61769atyXjr1q1lzYYNG2RO6dy5s8ytWbNG5lQ7frt27WTNH//4x2TcvaVAta1HRDRv3jwZHzp0qKwpKiqSOfUdujdkqJbowYMHy5q9e/fK3Pz585Nx9faCiIgmTZrI3ObNm5PxLVu2yJqf/vSnybh6G0KEv47Onj2bjKv7IsK/cUC147s3ZKhztHjxYlnj7k+3ZUFxzyP1BhTXvq+2RrhnRK1atWROnVd3Px05ckTmPv7442TcPf/VM8e9McJR96fb0vTZZ5996d/llx0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHsFd2Nef/31MlevXr1k/PXXX9f/sOhycofjBs3WrVs3GT9x4oSscYNKly9fnoyXlZXJGtWdVZGOywjdHVi7dm1Zs3Tp0nL/O6rzLkJ3Yc2bN0/WuG60e+65Jxl3Q4ndOb/pppuScTVwNyJiyZIlyfiiRYtkzSeffCJzI0aMSMb/8z//U9b07NlT5tSQdDfAV3U+Hz58WNa4ocmqo1bd6xF+ALIaQn78+HFZ8+CDDybje/bskTWrVq2SuW7duiXj//iP/yhr1LUSoTtTV65cKWtU57MaIh+hjztCd+6qQfYREVOnTpU51fHorhU1xN117rrPO2bMmGTcdVxu2rRJ5i7ilx0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHssdgCA7FUp9D/ctm2bzE2aNCkZd+3kqsZx7a+KO243CPov//Ivk3HXXr1ixYrCD+z/NGzYMJnr379/Ml5cXCxrqlWrJnNqy4IbGlulSvoSmT17tqwZPXq0zKnhyPv375c1DzzwgMzNmjUrGXetyGo7hRp6HeEHAqt2d9de7c65+nu7d++WNepec4OR3fGp7SNuy8Tjjz8uc2PHjk3G3RYRtfVAbfGJiPj2t78tc2oo/HvvvSdrfvOb38icui7dkHR1zhs2bChr3HYKxW13aty4scypZ4R79rZo0SIZV9u3vsyECROS8V69elXo713ELzsAQPZY7AAA2WOxAwBkj8UOAJA9FjsAQPYKHgTtOqDUMNejR4/KmurVqyfjvXv3ljVu+LDqCHJDjl2H0Z133pmMu2HBu3btSsbV8NcIP/haDWV1g2HdANhLLkn/v40bsKq6x5o3by5r3N9T3+GZM2dkzbvvvitzatis66xUQ8NdJ6vrHlaDfwcNGiRr3DBv1fHratasWZOM169fX9bMmTNH5po1a5aMHzt2TNa4Qe3XXHNNMu46Qt9///1kvHv37rKmY8eOMqeGJn/3u9+VNW4A/m9/+9tk/JFHHpE1ajD3kSNHZE379u3LnXPXv7uOVCepe5afO3cuGVcDzb+MGmo+f/58WXP+/Pkv/bv8sgMAZI/FDgCQPRY7AED2WOwAANljsQMAZI/FDgCQvYK3HtSqVUvmVNu4aydXbcpuq4BrZVXbCKpWrSpr3PGpgbducK1qA7777rsrdAxqq0CjRo1kTY8ePWSurKwsGf/DH/4gay677LJkXA10jogYPny4zKntGWpIb0TEkCFDZO7WW29Nxt1g3YEDBybjr776qqxR7dUREfXq1UvG3bXitmeoodiuDb60tDQZV1sIIvzgcvV9uMHl69evlznVGq6O21FbciL0AGuXc9eK2wqltjmoLQ4REU899VQyvnXrVlnjqGHxbouUul4j9Day06dPy5oOHTok424Iv1t21L9Vo0YNWaOebV/ELzsAQPZY7AAA2WOxAwBkj8UOAJA9FjsAQPZY7AAA2atS6H/o2qhVK7BrT1dtqZ07d5Y1arK7oyZoR0Ts3LlT5pYtW5aMu+ObPXt2Mr5p0yZZ8+STT8qcqhsxYoSsURP4I/TbEsaOHStrlNdee03mbrnlFplTLcJuy8R1110nc6r12rVXq9Zm1UIdEbFlyxaZa9OmTTLu2sndNdG3b99kXE3Mj4hYuHBhMn7jjTfKGvdGBLXtZcaMGbJGvXkkIqJy5crJeKtWrcp9DEuXLpU17p5W2zDUGxkiIkpKSmRObc9Q12SE/m7dteLenrFjx45kvE+fPrJm8eLFMqeo7yJCvxHBvSnHvXFG3Z8VfYvCRfyyAwBkj8UOAJA9FjsAQPZY7AAA2WOxAwBkr+BuTDW4OUIPlO3UqZOs6devXzL+3nvvyZpq1arJnOrqmjlzpqxxnVaK6yJS3MDdX/3qVzJ30003JeNuIOqjjz4qc//zP/8jc8qkSZOS8Z///Oey5u///u9lbuTIkcm46taLiHj22Wdl7je/+U0yfuedd8qaZ555JhmvXr26rFm3bp3MFRcXJ+OuI/Taa6+VuVmzZiXj7vjuv//+ZPyHP/yhrHFdrlWqpB8N7tpr27atzL377rvJ+FVXXSVr3nrrrWTcdYa7z6T+3nPPPSdr3Ge67777kvG5c+fKGtUt6joXVbdjhH4eufupZs2aMqeu5RMnTsiaoqKiZNx1Dx85ckTmBg8eXO6/Vwh+2QEAssdiBwDIHosdACB7LHYAgOyx2AEAssdiBwDIXqULFy5cKOQ/bNy4sczVqVMnGXetovv370/GXbt2aWmpzKmtB2rgdETE+fPnZU5x7d8qV6lSJVnjhsb+6U9/SsZVu31ExL/927/JnGrHf+yxx2SNOvbXX39d1uzZs0fm1MDniRMnypq/+7u/k7kFCxYk44sWLZI1x48fT8ZdS/vLL78sc2rwtWuvdoN/1fkbMmSIrFH3mto6EuEHQattPq5tXQ3wjYho3rx5Mr5v3z5Z44ZEK+q7jYhYv359Mq4GpEdETJ48WeZ+8YtfJONq2HOEHqjsvqdf/vKXMqe2GJw7d07WOL17907GT548KWvUM3vz5s0VOoYzZ84k4wcOHJA1hSxj/LIDAGSPxQ4AkD0WOwBA9ljsAADZY7EDAGSPxQ4AkL2C33pQVlYmc6odWW0viNDT76dNmyZrXHupao112wvcNgL199q0aSNrtmzZkow3a9ZM1mzcuFHm1ER4N4HftX//+te/TsbdhPRHHnkkGXef6YYbbpA5dR25ye5uS8CSJUuS8dmzZ8sadexuwr2bpr906dJkvGrVqrLGvT1ATZh3rdfqvLr2fXcMCxcuTMbHjBkja9z99NlnnyXjbtq/2hrh7hn1xogI/R0+8MADsmb48OEyN2fOnGR8w4YNsmb06NHJ+LBhw2SNu9cefPDBZLxXr16y5vPPP5c5tV3AbZ9avXp1Mu7eerN27VqZq127djLungOF4JcdACB7LHYAgOyx2AEAssdiBwDIHosdACB7BQ+C7tq1q8ypLhnV0RURMXTo0GTcdWOqmoiIlStXJuNq8HBExIoVK2RODTd1g2Z37dqVjBcVFcma9u3by9w999yTjKvupwg9RDVCd/m5rinVYerOgxvc3KJFi2S8ZcuWssZ1Nd5+++3JuOroiojo3r17Mu6GPd96660y99FHHyXjFen2jdBdl126dJE1ahC068519/SMGTOS8bp168qayy+/XOaeeuqpZNwNLL7tttuScdeNvHPnTplTnanuWunXr5/MNWrUKBl3zz11nd98883lronwnZoV0aBBg2T84MGD5f5b7tjc9X/o0KFk3A0ud53KF/HLDgCQPRY7AED2WOwAANljsQMAZI/FDgCQPRY7AED2Ct564AaBqjZ9N2BYtfTWqVNH1owfP17mVHu/G+7bpEkTmZs6dWoyroYzR+gBq6qVNsK3k6uhsa4FV20viIhYs2ZNMu7atWfOnJmMDx48WNa4z/STn/wkGW/YsKGscVs3mjdvnoy746tVq1YyvnXrVlnjHDt2LBmvUkXPWd+xY4fMlZaWJuPuWlbt8x07dpQ1roVfbZs4deqUrDl79qzMqcG/e/fulTV/8zd/k4y77TXumfP6668n42oLQYT/vFdeeWUy7obPq4HPbij3tddeK3Pjxo1LxqdMmSJr3DB7tVXFbf9Rbf9qm1GEvmci9Hriri833PoiftkBALLHYgcAyB6LHQAgeyx2AIDssdgBALJXcDfmqFGjZE51sakhwhG6u8cNJXbdbep18q7T0A3q7dmzZzLuhlFPmjQpGd+wYYOs+cY3viFzf/rTn5Jx193pOsFUB+yiRYtkzcCBA5Px6dOnyxp3SVWrVi0Z3717t6xxn0kNy3YDkFWHnTvuMWPGyNyzzz6bjLtu35o1a8qcGpK7adMmWaOGW6vh5BERHTp0kDnVFb19+3ZZ46hnRFlZmaxRQ9zvv/9+WfPhhx/K3LJly5Jx1fUcEfHcc8/JnOoOVIPxI3TXseui/v3vfy9z6pnz/e9/X9a4rkb1/FXdtBH6GTF37lxZM2LECJlTXcJuaHgh1yW/7AAA2WOxAwBkj8UOAJA9FjsAQPZY7AAA2WOxAwBkT/fy/y9WrVolc+3atUvG1XaACL31wLWXHjx4UObU8Fw3wNq1hqvP645Bca2+rq1+woQJyfinn34qa+bNmydzqn34W9/6lqxRbfVukLFrd1cDi9Vw5oiI9evXy5waOr1v3z5Zs3jx4mTcbW1ZunSpzKlBuG5ouPsO//qv/zoZd+foiiuuSMZLSkpkjdvCsnz58mS8V69esmb+/Pkyp7YeuIHAajuK+27dIPlvfvObyfiMGTNkjdsatGLFimRcXQ8Reoi123rz6KOPytyQIUOScbdV7P3335c596xS1BYDd/1Pmzat3MfgvttC8MsOAJA9FjsAQPZY7AAA2WOxAwBkj8UOAJA9FjsAQPYKfuvB7bffLnMff/xxMu4mebuc4qa0q3ZVN137gw8+kDnVCty/f39Zs3DhwmRcbbOIiLjzzjtlTm3p+Jd/+RdZ46i2/x/84AeyRk3nb968uawpKiqSuUGDBiXjbiuDexOGmpA+fvx4WTNr1qxk/OjRo7LGfV51jl5//XVZ07RpU5nr1q1bMt6qVStZU69evWTctbS7NvNLLkn/f7A7bvVGiwj9HbZt21bWqLd+rFmzRta4t3Go6989Vxo0aCBz6i0K6u0iEfoNKO5NCU888YTMFRcXJ+MdO3aUNS1btpQ5pXPnzjI3e/bsZNxtEXHbaLp27ZqMr1y5Uta4t2dcxC87AED2WOwAANljsQMAZI/FDgCQPRY7AED2Cu7GVF0/EXqYq+tkUl1iarBvhB52GxExcODAZHznzp2yZu/evTI3fPjwZHzTpk2yRg1udsNue/bsKXP169dPxt2AbXXcEbqjynXRqWHB7du3lzWqmzAiYvXq1cl4o0aNZI3rQvzpT3+ajM+ZM0fWqG4vNaw4wnePvfrqq8m4G0I+ZswYmVPDxt2g5cGDByfjCxYskDUnT56UOXUduXvGfV7Vzeq66D7//PNk3D1X3ODmm266KRm/+eabZY0bQq4GSx8+fFjWbNu2LRkfMGCArHGdkGowtzq2CN91rFSuXFnmVOenute/jOrcded1z549X/p3+WUHAMgeix0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHtVCv0PS0pKZE61Sqt4RMSwYcOScdcy3rhxY5lTbeNTpkyRNc7vfve7ZNwNllat127g7vHjx2VODU12bf9usK4a4uuG0E6cODEZd4Ob3daDt99+OxlXW0ci/NYD1d7cvXt3WaMGQX/961+XNe47VIN/V61aJWvcd6i2t6jtBRH6+3D/TpMmTcr999xOpTNnzsic+p7ctTd06NBk/KWXXpI1ixcvljnVuu62NDlqC4Qblt2sWbNk3B33k08+KXNqO4p7DqgB1hF60L0ach+hv9s6derIGjd8W52LNm3ayJpC8MsOAJA9FjsAQPZY7AAA2WOxAwBkj8UOAJA9FjsAQPYKfuvBDTfcIHMrV65Mxvv06SNr1BsC1KT/CD05PSJi48aNyXjt2rVljZuUrVq2N2zYIGsq4vvf/77MnT9/Phl3k93VBPIIvQ3jvffekzUrVqxIxtWbAyIi3nzzTZl74403knH3VgH3hgA11f/RRx+VNaWlpcm4a4OvWbOmzLVo0SIZd1Pf1fT7CN3m7VqvW7dunYxPmzZN1rgtHVdddVUyXlRUJGsuuUT/v7M6f65Fft26dcm4e1uDe3OFeruHu5anTp0qc6dPn07Gq1evLmvUPe22gXzve9+TOfVGBHcPPv744zKnvkP3HHVvYakItS3BvYGikGWMX3YAgOyx2AEAssdiBwDIHosdACB7LHYAgOwVPAjadS6qzrJTp07Jmnbt2iXjd9xxh6x59dVXZU4N6nVdRK77TnXsVa5cWdYUFxcn42PHjpU1TzzxhMzdcsstyfikSZNkTe/evWXu0KFDyXjLli1ljeqIU4O8IyIWLVokcz179kzG9+/fL2vcEFo1hPm2226TNU899VQyrroqIyI++OADmbv77ruTcTcs23XsKefOnZM51XXphhIfOXJE5lQXrhuaPH78eJlTHXbunladhu654obFz5s3Lxl3XanuGXHXXXcl424g9scff5yM161bV9aoofQREa+99loyvnz5clnj/i11zbprRT0LXCerGpof4Yeu/zn4ZQcAyB6LHQAgeyx2AIDssdgBALLHYgcAyB6LHQAgewVvPdi1a5fMqTZSV6PaVR9++GFZ49r+Vdu4Gx7qNGzYMBl37d+qzdu16atW5IiIOXPmJOOXXXaZrHGDtG+99dZkvEGDBrJGDc912wvUdoCIiPnz5yfjrkXeDepdunRpMu4GN/fo0SMZd9+F257xySefJONucPmMGTNkTnFt9Wr7jzsPu3fvljnV/r1p0yZZ487RgQMHknE33Prqq69Oxt12BffMUc8IN4TZDZZWW2LWrl0ra9R1vmXLFlmjhnJH6O1Ebii32/bitk0oNWrUSMa7dOkiaw4ePChzzZo1S8b/3C0J/LIDAGSPxQ4AkD0WOwBA9ljsAADZY7EDAGSv4G5MN9RWdew5y5YtK3fNyJEjy13Tt29fmXOdVseOHUvGXefiD37wg2RcdcpF+G5R1X3kOuzceVUDYN33N3v27GTcdYg5p0+fTsZdZ6AaGh4R0atXr2TcDfCtU6dOMv6v//qvskZ1kUboTrqysjJZ4zrLVLeo69hznW+K6+odMGBAMr548WJZ07Fjx3Ifg7v+9+7dm4y7a88dn+rUdJ2BrqtRdWy7QcuNGjVKxt3Aetf5rM65ux7UPe2O4+jRo7Jm8uTJybi61yMihg8fLnOK65otBL/sAADZY7EDAGSPxQ4AkD0WOwBA9ljsAADZY7EDAGSv4K0Hql07IuK6665Lxl0bsGpL7dmzp6xxLb1Dhw5Nxt3Q2NLSUplTbd4PPfSQrBk1alQy/tJLL8kaNUQ1Qrenv/XWW7Lm8OHDMqcG9d5///2y5vLLL5e5ihg8eHAyvnHjRllTVFQkc5999lkyvmbNGlmjBgJ/8MEHssYNAO/evXsyvm/fPlnjto9cuHAhGVfbYSL09hG3xcG1pz/33HPJuGuRr1JFP07UAGR3v0+aNCkZVy3/EX4bzTXXXJOMT5w4Uda4wdKqHd9te/noo4+ScXc9uAH4Y8eOTcbVNRTht1wpVatWlTl17O6+dcPiH3300cIPrBz4ZQcAyB6LHQAgeyx2AIDssdgBALLHYgcAyB6LHQAgewVvPVDTuiN0m76baK64ttg77rhD5r73ve8l4507d5Y1rk3/zjvvTMZ/97vfyZpHHnkkGf/1r38ta+6++26Ze/bZZ5Nx9xYF18L/V3/1V8n4K6+8Imu2b9+ejM+ZM0fWvPfeezKnzrmbkO706dMnGXdvSlBbN6ZNmyZrnnnmGZn78Y9/nIy7dnJ1XiP0Z9q8ebOsUdP5hwwZImvc8an2/quvvlrWuOn8qoW/uLhY1tx3333JuLv+X3vtNZlTWwW+853vyJp///d/l7mdO3cm461atZI16hnh3rzgnlPq3/rZz34ma5zGjRsn43v27JE16vjUFp+Iir31pnr16uWu+SJ+2QEAssdiBwDIHosdACB7LHYAgOyx2AEAslfpgpsY+sX/sFKlcv/xkSNHypzqqHLDZN0gXPUxDhw4IGtct9CZM2eS8UOHDska1VHlBsMeP35c5tRQ21OnTsmaBx98UOZuv/32ZHzAgAGy5ic/+UkyXq9ePVkzf/58mXvxxReT8d27d8uasrIymevatavMKW+88UYy7o57xowZMtegQYNk3HV3qsHlEXpI7scffyxrVOfzkSNHZI0bwqyGpA8cOFDWuOHb48aNS8Yff/xxWdO+fftkfNWqVbLGdeGqe011skZENG/eXObUs+DkyZOyZu/evcn4P/3TP8majh07ypzqknz66adlzZIlS2RODfx3z2X3kgDFdVauXr06GXeDpd05v4hfdgCA7LHYAQCyx2IHAMgeix0AIHssdgCA7LHYAQCyV/Ag6FGjRsncp59+moy7Fle1lUG15kZENG3aVObUoNnKlSvLmh07dsicGpLbtm1bWaPa4M+ePStrSkpKZG7lypXJ+Llz52TN22+/LXNqCK36/iIiFixYkIyPGTNG1rhW6X379iXjbgD4X/zFX8jcpk2bknE35FgNWn7sscdkzbBhw2RODUB2betu8K/aqtKtWzdZo7bl3HvvvbLmgw8+kLnevXsn47fccousUYORI/TWDfeZateunYy7679Zs2Yyt27dumTcPXPcvaueYfXr15c1rVu3ljlFfRcReji4G5btjk89L92gdnVPu8+qhnK7Y3BbRArBLzsAQPZY7AAA2WOxAwBkj8UOAJA9FjsAQPZY7AAA2Sv4rQeu7V+1iroW1xUrVpTrb0VUbOK6a5l1E+HVsbu3Hqh294YNG8oa18qtpvOfP39e1qgtGBERzz//fDKupoxH6JbxGjVqyJr+/fvLnNrKcNttt8mae+65R+bU9aKm9kdE1K1bNxlXWzMiIn7xi1/InGqfd1sP3Bs81Fst2rRpI2uqVq2ajLv274kTJ8qcOkdu+r1rDVf3hnpbQ4Secu/up6VLl8qcOnZ3/d9///0y99prryXjbmuE+ntua9egQYNkTn3eH/3oR7LGvY1DXUfuzS3umagMHz5c5tS2HPfdFrKM8csOAJA9FjsAQPZY7AAA2WOxAwBkj8UOAJC9ggdBd+jQQebUwGfXwakGttapU0fWzJs3T+ZGjBiRjJ84cULW7N69W+bWrl2bjLsuIjUQ2HXEtWzZUubU+Vu+fLmsUR1sERF33313Mn7XXXfJmrlz5ybjrovUdc2q43PdaGfOnJE51Y3pOuJUF25ZWZms+fnPfy5zTz75ZDKuBuRG+G5W1fl28uRJWaM+06xZs8r970REtGjRIhl3HXFuSHqXLl2ScXfO1fe+fft2WdOjRw+ZUwPA//CHP8ga17nYvXv3ZLxRo0aypnHjxsn4yJEjZY3qyo6IePTRR5Nx94xw1H3jOtdVF657lk+ZMqV8BxYRffv2LXfNF/HLDgCQPRY7AED2WOwAANljsQMAZI/FDgCQPRY7AED2Ct560KRJE5lTg3DdYFg1cNQNZXU6d+6cjK9cuVLWuIHKihu0rAa2ujZbNwj3G9/4RjLujvvAgQMyt2XLlmT8008/lTVDhgxJxt02kNtvv13mfvazn5X777388ssyp7Y5fO1rX5M16ju8+eabZY0b3Ky2TcyePVvW1KtXT+bGjRuXjE+dOlXWqK0MnTp1kjVu68HGjRuTcTVwPcLfG4oa+huhB1+74e7O3r17k3G3Nejzzz+XOXXtua0CatvE5s2bZc3MmTNlTg3Ud1vF3DNHbd1Qz/gIfa3Mnz9f1jhq24vbclIIftkBALLHYgcAyB6LHQAgeyx2AIDssdgBALJXcDeme/W66rpp166drHEDW5WjR4/K3DPPPJOM9+7dW9aUlJTIXMeOHZPxSZMmyZp169Yl467rzQ2CVoObDx8+LGtee+01mfvOd76TjKvO2IiIyy67LBl3HWJXX321zKmOqiuuuELWuNzYsWOT8erVq8saNdxadYF9WU4NNVfdtBERL774osxNnz49GXfDfYuLi5Nx1SEcEfHZZ5/JnBpifcMNN8gadf1H6G6+s2fPypqaNWsm4/369ZM1jhr87q5/N3z7lVdeScbdc+W///u/k/Fq1arJmgsXLsic+kyu49INVFbH4Z7/6v7csWOHrHHD3Y8fP17uYygEv+wAANljsQMAZI/FDgCQPRY7AED2WOwAANljsQMAZK/SBdfX+gVuWKpqhT9y5Ei5/54b8uqG8bZv377cf+/cuXMypwbe1qpVS9Y0aNAgGd+2bZuscQNbH3vssWTctQ67Fnl1/t566y1Z8+GHHybjrgX9zjvvlDnFDY3dv3+/zKlh4+7v9ezZMxlXA50jIoqKimRu8uTJyfhHH30ka1xruPq33DGoVm637cUNAL/xxhuTcdXqHhFx+eWXy9yGDRuScTWcOUJ/7+4zucHNasD2m2++KWu6du0qc+o+XLp0qax56qmnkvF//ud/ljVbt26VOdWm/+Mf/1jWPPTQQzJXqVKlZNw9/z/++ONk3A1jX7JkSbmPwV3/7hl7Eb/sAADZY7EDAGSPxQ4AkD0WOwBA9ljsAADZY7EDAGSv4LcenD9/XubUFgPVFh6h2+CHDx8ua1avXi1zjRo1SsZnzZola9yuC7UlwLW/qi0YzZs3lzXqLQAREe+8804y7ib6u+Nr2LBhMv7+++/LmokTJybj3/3ud8tdExHRqlWrZPyFF16QNe7tAWpqvjp3Efr83XTTTbLm+eefl7mXX345GXdv/XBbRNR9c+rUKVlTv379ZHzevHmyxt1rqpXbteKrbSoREX369EnG3XYF9YxwW5rcm1HUveGuL3fO1fPjBz/4gax56aWXknG1benLTJ06NRl317L7TIp7+4Oyc+dOmWvTpo3MnThxIhl3b3spBL/sAADZY7EDAGSPxQ4AkD0WOwBA9ljsAADZK7gbs3HjxjKnOnXc0GTVcTN+/HhZ47rbVIdW1apVZY3r7tm4cWMy7obQqk5D9VkjfJeT6qT77W9/K2vWrVsnc//xH/+RjDdp0kTWdOvWLRl3x+3O0RtvvJGMq8HDEb77Tg3xfeCBB2TN22+/nYz36NFD1qxcuVLmysrKyvXvROjB5RERV155ZTKuhv5G6G4+d99+85vflLmHH344GV+1apWsuffee2WuWrVqybi7N9Q537Fjh6wZNGiQzKmB3a5b2nWUq+v80KFDsqZOnTrJ+I9+9CNZU69ePZlTzxw1eDsi4qqrrpI5dY2pwfgREddcc00yrs53hB/C371792Tcfe+F4JcdACB7LHYAgOyx2AEAssdiBwDIHosdACB7LHYAgOxVuuCmIX+Ba2GuWbNmMu7aS1u2bJmMq+GvEb4Ffc+ePcm423rgWu7VgOYqVfRuDXV8rqV9xYoVMqcGFqvzHRExdOhQmVPDm++66y5Zo7YEqBbqiIiPP/5Y5pYsWZKM/9d//ZescS3Hqt192rRpsmbOnDnJuBqUHaG3okREfP7558n4gQMHZI37TF//+teTcTVMOSJi69atyfj8+fNlzZgxY2ROfb/uvBYXF8tcSUlJMr5s2TJZc9lllyXju3fvljW9e/eWuYEDBybjmzdvljVuYLHaCrV27VpZc8cddyTj6vkVETFq1CiZW758eTLuBsx36dJF5tR2LLftRQ3fdufODaxXS5K7Z9y6cRG/7AAA2WOxAwBkj8UOAJA9FjsAQPZY7AAA2WOxAwBkr+CtB5UqVZK5Fi1aJOP79++XNaqF//Tp07LGtbKq1thTp07JGjdxXW1ZcG9K+KqpVm43pb1fv34yp1qOf/jDH8oateXETYN3rc1TpkxJxkeOHClrVGtzRMSDDz6YjN92222yRvnoo49k7vHHH5e50tLSZNxtvenUqZPMTZw4MRl392D9+vWT8b59+8oat31EvXGjUaNGsuZv//ZvZU5972o7QETEzp07k/FmzZrJGvX2h4iInj17JuN79+6VNeqtHxER48aNS8bd2y7UmxKuu+46WeO28qiWe/dYb926tcxt2bIlGXdvXlDPRPVGhgi9VSYiokaNGsm4e14Xsozxyw4AkD0WOwBA9ljsAADZY7EDAGSPxQ4AkD091bgcKleunIy7bjQ1dNd1+bkBq2qwdFlZmaxxOdUR5Dz00EPJ+HPPPSdr3ABT1WG3Zs0aWeO60X7/+9+Xu0YNqL3qqqtkjerkcxYtWiRz7poYO3ZsMq6GPUdE/PKXv0zGXfdY06ZNZU51SU6YMEHWfPvb35a5a665Jhlv27atrFFdfmpQcIQeYB2hOw3VffZlVFej6/Lr379/Mu66E2vVqiVz6vtduXKlrLn//vtlbtKkScm4G/zesWPHZHz8+PGyxhk8eHAy7gbgq6HcEXp4+ZVXXilrZsyYkYyfPHlS1qgh2hF+0Pefg192AIDssdgBALLHYgcAyB6LHQAgeyx2AIDssdgBALJX8CBo18p69uzZZNy1KR86dCgZP3/+vKxxf2/fvn3JeLVq1WSNy9WuXTsZV8NpI/RncufuzJkzMqfa3d0g1/nz58ucMnr0aJl7//33k3H3PamW8YiI1atXJ+NuC4YbgDxkyJBkfOPGjbJGtYa77RQvvviizPXu3TsZf/rpp2WN+7wVGZqs/l737t1ljRsEvWTJkmRcfdaIim3lcYPVVeu626bStWtXmVMDmletWiVrZs6cKXPqM61YsULWqPPnBuCreyZCPxPVdrAI/9xTz/LNmzfLmg4dOiTj7rjr1q0rc+o6cgPA3fFdxC87AED2WOwAANljsQMAZI/FDgCQPRY7AED2Cu7GdANba9asmYyroaIRuhPMDXJ1nZCqM8p1DbpONTXM+J577pE1auBzly5dZI3jupkU1y3Xp0+fZHz79u2yRp2HI0eOyBo3WPrgwYMyp7Rp00bmVBeW6zTs2bNnMu4G1951110ypzo/K9JpGxExdOjQZHzq1KmyZuvWrcn4iRMnZM3evXtl7vrrr0/G3XPg6NGjMrd06dJkfNeuXbJG3dPr16+XNa5TU3Uhuk7Ifv36yZz6TLfccouseeONN2ROcdeKO39fJXXPROjnlDuvRUVFMnf55Zcn4+rlARER77zzjsxdxC87AED2WOwAANljsQMAZI/FDgCQPRY7AED2WOwAANkreOtBq1atZG7Hjh3J+ODBg2XN9OnTk/FGjRrJGjfkVR2Dikf4gailpaXJ+H333Sdr1BBmNVQ3IqJbt24yp1qsXQuuG7DqBt4qasirGnod4bd0qHZ316bshgVXqVKl3DXq33JbRNS2DefSSy+VuXXr1smc2mLw8MMPy5q33nqr3MfgzJ07Nxl3WzBeeOEFmVPXhBq47o7BccOy1dYldwxu+5R6Rpw6dUrWqPvdDY921PVfXFwsa9q2bStz6r5WnzUioqSkJBl3zym1VSYiYsOGDcn4yJEjZc0f//hHmbuIX3YAgOyx2AEAssdiBwDIHosdACB7LHYAgOyx2AEAslfw1oPOnTvLnJo879rJW7ZsmYxv27atkMMpWPv27WXOHZ9rH1aaNWuWjLutB641XL0BYs+ePbLGbREpKytLxs+dOydrrr322mT8k08+kTWXXKL/H0qdI/fGAXXcEfrtC27rwY033piML168WNa4N0Pceuutyfjs2bNljZrsHhExYcIEmSuvsWPHytybb74pc1WrVk3Gq1evLmvcdaTeiKC2tkTorTfuGnfbidQ2JPccaNKkicypN6rs3r1b1qg3IsyYMUPWtGjRQubU1gO3bcN9h+redVsZ1HaFHj16yBr3toYtW7Yk4+57VzVfxC87AED2WOwAANljsQMAZI/FDgCQPRY7AED2Cu7GrFSpUrn/eL9+/WRu48aNybgbMDxq1CiZmzRpUjLuuqkc1bF08OBBWdOzZ89kXH3WCD2c1mncuLHMuU5NpU2bNjKnOs727dsna9zxHTt2LBmvV6+erDlz5ozMFdKFVajevXvLnBswrIaau2vZdaMpqvPO5dwwXtf5rK5/1zXrqEHQK1euLPffcoObXU7da6rrOSLCPR7VOXLdw+p+Ut2vEf76V3U1a9aUNa6zUnU3u+e/6nw+fvy4rHHPnEGDBiXjbqj/lClTZO4iftkBALLHYgcAyB6LHQAgeyx2AIDssdgBALLHYgcAyJ7uZS4H1d7vBiC3bt06GXcDmCsyJNoNZXWt5mpwrRtyvGzZsmRctfNWlNte0L9/f5lTQ6dda7Nq93XfkxoMHqEHQRcVFckat81Buffee2XuhRdeSMbd97RgwQKZW7VqVTKurvEvo86R2rYRoY/dDc9195MaqFzRrQebNm0qd43auuT+Vp06dWRObcNwn8ltZahbt24yPnnyZFmjDBgwQOZmzpwpc40aNUrG1fMrImL16tUypz5TaWmprFHUdRzhnxEdO3ZMxqdOnVruY/giftkBALLHYgcAyB6LHQAgeyx2AIDssdgBALLHYgcAyF7Bbz1wLbiqJVq1sUZEtGzZMhlfsWJFIYfzv1FTtN1bBbp06SJzahtBu3btZI1qiS7wFP9v1NsD3NYDtzVCtfRu375d1nTo0CEZd1PL3XYP9QYIN51/7dq1Mnf27Nlk3LXcnz59Ohl37donTpyQOTXJ3rVeu7ceVOR6URP43Rso3Bsj1JsA3CR7dQzuOL7Kt1ZERDRt2lTm1Pd+4MABWePemqKu8x49esgada+5txSoazxCbzlxb3JwW2w6d+6cjLttKmpbgnsW9e3bV+bU/e6Ou5B7hl92AIDssdgBALLHYgcAyB6LHQAgeyx2AIDsFTwI2g3+rV+/fjJ+6NAhWaM6aypVqiRrevXqJXN79+5NxlUHVkRE1apVZa5Tp07JuOvYa9u2bblrXLec6gA8fPiwrHGfV3W+uUG4qvPTdeWtX79e5lq0aJGMu86tESNGyJy6xrZu3SprysrKknHVrRrhhzCrY3DfRZUq+tZzg7mVkpKSch+Do7pj3ffuhjCre81d/yqnOqUj/HWkui7d8PSKDLB2nbbq+eaeRa5TsyIdpk5FuuGLi4vLfQyLFy+WuZ49eybjrtO2EPyyAwBkj8UOAJA9FjsAQPZY7AAA2WOxAwBkj8UOAJC9grceXHrppTJXkfbmBg0aJONFRUWyxg0jVW2urVu3ljVuS4AaRuoGw6rtGfv27ZM1brhpef+dCD+Mt3fv3sm4a9NXbfBue4EaJhuhB4q74dauHVq1jbvWfjUk113jmzdvljm1FaRbt26yZv/+/TJXr169cv07EXrAsBuw7a5l9b27Fnk3LF5dL25gtxqwXa1aNVmzY8cOmVPfu7tnxo0bJ3NPPPFEMu62sNSoUSMZd9spXAu/On/q3EX4Z47acuXa/tX2DPecuvzyy2VOXWNuC0Yh+GUHAMgeix0AIHssdgCA7LHYAQCyx2IHAMhewd2YbqizGtjqhsaqzrJz587JGtWlFhHRqFGjZFwNP47wXVht2rRJxt3xqW7RgQMHypq5c+fKnOokdV15lStXlrn58+cn465zS1HdtBG+c2vq1KnJeJcuXcp9DBERCxcuTMb79esnaxYtWpSMl5aWypoOHTrInPo+KjJEOELfT27Iseryc92J6p6J0PeuG1Jdkc5i142pBr+77lx37an7aebMmbJmw4YNMqe6bS+77DJZo64xNUQ+IuLgwYMyp7qOXaet6xZV94bqTo+IuOqqq5Lx2267Tda0b99e5mbPnp2Muy7cQvDLDgCQPRY7AED2WOwAANljsQMAZI/FDgCQPRY7AED2Kl24cOFCQf9hpUoy16JFi2T87NmzskYNN3WtzY5q13bDeN0A66VLlybjffv2lTVlZWXJ+MaNG2WNa73eu3dvMu7Oq9tOoVrX3ZBj1XJfp04dWbN48WKZU1SbeYT+LiIiBgwYkIyrbRYRET169EjGXdv6unXrZM6dP6V+/foypwbhuiHp6n5S1+SXUa3hrhXftYZXZDuF2jY0atQoWTNp0iSZa9asWTLungNuYHdFqO0ebsuQG76tvveSkhJZ4wbgq20JbmuEum9efPFFWeO2sqn1pLi4WNYsWbJE5i7ilx0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHssdgCA7H0lWw8U18qtWpvdBHJHTc1fvXp1hf7e/1MuvfRSmTt27Fgy7tr03ZT7irRRq60b7u0PtWvXlrndu3cn4/v27ZM1qm09IkJdvuoNFI5rnXe3ifoO3dsk3L+lJuO3bNlS1uzZsycZd23rbvL8+vXrk3F33GqrTITeIuLeNKEm+rvnyvLly2VOnaOKUm9RcG31J06cSMabN28ua9zWCLWdwr0Z4oorrpA5tdXo+eeflzXqDQvqXo/Q238i9Ntt3JYm94adi/hlBwDIHosdACB7LHYAgOyx2AEAssdiBwDI3lfSjVmjRo1y16gONtfR5TrBVMeS6laK8N1CqpPODfB1XViKOncRuhvNDXKtWbOmzKmhwA0aNJA1avjwrl27ZI0bQqsG3rqB3Tt37pS57du3J+MNGzaUNWrQrOsEdp2Q6vjcsGw3oFl9724Yr7omXNesOncReqD4yZMnZc1XrVOnTsm4O68LFy6UOdXld+TIEVnjhg+rDmJ1fTnuu3BUR6gaoh3hnxFqUHVFhp33799f5tz39LWvfS0Z7927t6x55JFHvvR4+GUHAMgeix0AIHssdgCA7LHYAQCyx2IHAMgeix0AIHvpHueEAncoAADw/zn8sgMAZI/FDgCQPRY7AED2WOwAANljsQMAZI/FDgCQPRY7AED2WOwAANljsQMAZO//ALS1iDUnXRluAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model.eval()\n",
    "guidance_scale = 7.0\n",
    "conditioning = torch.cat([-1 * torch.ones(1, 1, 1).float(), 2*torch.ones(1, 1, 1).float()], dim=0).to(device)\n",
    "\n",
    "noise = torch.randn((1, 1, 64, 64))\n",
    "noise = noise.to(device)\n",
    "\n",
    "scheduler_ddim = DDIMScheduler(\n",
    "    num_train_timesteps=1000, schedule=\"scaled_linear_beta\", beta_start=0.0005, beta_end=0.0195, clip_sample=False\n",
    ")\n",
    "scheduler_ddim.set_timesteps(num_inference_steps=250)\n",
    "progress_bar = tqdm(scheduler_ddim.timesteps)\n",
    "for t in progress_bar:\n",
    "    with autocast(enabled=True):\n",
    "        with torch.no_grad():\n",
    "            noise_input = torch.cat([noise] * 2)\n",
    "            model_output = model(noise_input, timesteps=torch.Tensor((t,)).to(noise.device), context=conditioning)\n",
    "            noise_pred_uncond, noise_pred_text = model_output.chunk(2)\n",
    "            noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)\n",
    "\n",
    "    noise, _ = scheduler_ddim.step(noise_pred, t, noise)\n",
    "\n",
    "plt.style.use(\"default\")\n",
    "plt.imshow(noise[0, 0].cpu(), vmin=0, vmax=1, cmap=\"gray\")\n",
    "plt.tight_layout()\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "32256aed-b4ad-4e6b-a547-71b64678e224",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}