swin_unetr_brats21_segmentation_3d.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Az8DGB2AeMQ2",
        "outputId": "b3f80cfc-5b46-44c6-9e65-8aa95c9500ea"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "_VALRmqWfAlt"
      },
      "outputs": [],
      "source": [
        "!mkdir TrainingData"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "5-aTUbvrcUBR"
      },
      "outputs": [],
      "source": [
        "!tar -xf /content/drive/MyDrive/Data/Brats2021/BraTS2021_Training_Data.tar -C TrainingData"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HE9moA5Xytdo"
      },
      "source": [
        "# 3D Brain Tumor Segmentation with Swin UNETR (BraTS 21 Challenge)\n",
        "\n",
        "\n",
        "This tutorial uses the [Swin UNETR](https://arxiv.org/pdf/2201.01266.pdf) [1,2] model for the task of brain tumor segmentation using the [BraTS 21](http://braintumorsegmentation.org/) challenge dataset [3,4,5,6]. Swin UNETR ranked among top-performing models in the BraTS 21 validation phase. The architecture of Swin UNETR is demonstrated below\n",
        "\n",
        "![swin_brats](https://github.com/Project-MONAI/tutorials/blob/main/figures/swin_brats21.png?raw=1)\n",
        "\n",
        "The following features are included in this tutorial:\n",
        "1. Transforms for dictionary format data.\n",
        "1. Define a new transform according to MONAI transform API.\n",
        "1. Load Nifti image with metadata, load a list of images and stack them.\n",
        "1. Randomly rotate across each axes for data augmentation.\n",
        "1. Randomly adjust the intensity for data augmentation.\n",
        "1. Cache IO and transforms to accelerate training and validation.\n",
        "1. Swin UNETR model, Dice loss function, Mean Dice metric for brain tumor segmentation task.\n",
        "\n",
        "For more information access to pre-trained models and distributed training, please refer to Swin UNETR BraTS 21 official repository:\n",
        "\n",
        "https://github.com/Project-MONAI/research-contributions/tree/main/SwinUNETR/BRATS21\n",
        "\n",
        "## Data Description\n",
        "\n",
        "Modality: MRI\n",
        "Size: 1470 3D volumes (1251 Training + 219 Validation)  \n",
        "Challenge: RSNA-ASNR-MICCAI Brain Tumor Segmentation (BraTS) Challenge\n",
        "\n",
        "The dataset needs to be downloaded from the official BraTS 21 challenge portal as in the following\n",
        "\n",
        "https://www.synapse.org/#!Synapse:syn27046444/wiki/616992\n",
        "\n",
        "The JSON file containing training and validation sets (internal split) needs to be downloaded from this [link](https://drive.google.com/file/d/1i-BXYe-wZ8R9Vp3GXoajGyqaJ65Jybg1/view?usp=sharing) and placed in the same folder as the dataset. As discussed in the following, this tutorial uses fold 1 for training a Swin UNETR model on the BraTS 21 challenge.\n",
        "\n",
        "### Tumor Characteristics\n",
        "\n",
        "The sub-regions considered for evaluation in the BraTS 21 challenge are the \"enhancing tumor\" (ET), the \"tumor core\" (TC), and the \"whole tumor\" (WT). The ET is described by areas that show hyper-intensity in T1Gd when compared to T1, but also when compared to “healthy” white matter in T1Gd. The TC describes the bulk of the tumor, which is what is typically resected. The TC entails the ET, as well as the necrotic (NCR) parts of the tumor. The appearance of NCR is typically hypo-intense in T1-Gd when compared to T1. The WT describes the complete extent of the disease, as it entails the TC and the peritumoral edematous/invaded tissue (ED), which is typically depicted by the hyper-intense signal in FLAIR [[BraTS 21]](http://braintumorsegmentation.org/).\n",
        "\n",
        "The provided segmentation labels have values of 1 for NCR, 2 for ED, 4 for ET, and 0 for everything else.\n",
        "\n",
        "![image](https://github.com/Project-MONAI/tutorials/blob/main/figures/fig_brats21.png?raw=1)\n",
        "\n",
        "Figure from [Baid et al.](https://arxiv.org/pdf/2107.02314v1.pdf) [3]\n",
        "\n",
        "\n",
        "\n",
        "## References\n",
        "\n",
        "\n",
        "If you find this tutorial helpful, please consider citing [1] and [2]:\n",
        "\n",
        "[1]: Hatamizadeh, A., Nath, V., Tang, Y., Yang, D., Roth, H. and Xu, D., 2022. Swin UNETR: Swin Transformers for Semantic Segmentation of Brain Tumors in MRI Images. arXiv preprint arXiv:2201.01266.\n",
        "\n",
        "[2]: Tang, Y., Yang, D., Li, W., Roth, H.R., Landman, B., Xu, D., Nath, V. and Hatamizadeh, A., 2022. Self-supervised pre-training of swin transformers for 3d medical image analysis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 20730-20740).\n",
        "\n",
        "\n",
        "### BraTS Dataset References\n",
        "\n",
        "[3] U.Baid, et al., The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification, arXiv:2107.02314, 2021.\n",
        "\n",
        "[4] B. H. Menze, A. Jakab, S. Bauer, J. Kalpathy-Cramer, K. Farahani, J. Kirby, et al. \"The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)\", IEEE Transactions on Medical Imaging 34(10), 1993-2024 (2015) DOI: 10.1109/TMI.2014.2377694\n",
        "\n",
        "[5] S. Bakas, H. Akbari, A. Sotiras, M. Bilello, M. Rozycki, J.S. Kirby, et al., \"Advancing The Cancer Genome Atlas glioma MRI collections with expert segmentation labels and radiomic features\", Nature Scientific Data, 4:170117 (2017) DOI: 10.1038/sdata.2017.117\n",
        "\n",
        "[6] S. Bakas, H. Akbari, A. Sotiras, M. Bilello, M. Rozycki, J. Kirby, et al., \"Segmentation Labels and Radiomic Features for the Pre-operative Scans of the TCGA-GBM collection\", The Cancer Imaging Archive, 2017. DOI: 10.7937/K9/TCIA.2017.KLXWJJ1Q\n",
        "\n",
        "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/Project-MONAI/tutorials/blob/main/3d_segmentation/swin_unetr_brats21_segmentation_3d.ipynb)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Uim7HW13ytdr"
      },
      "source": [
        "## Swin UNETR Model\n",
        "\n",
        "The inputs to [Swin UNETR](https://arxiv.org/pdf/2201.01266.pdf) are 3D multi-modal MRI images with 4 channels.\n",
        "The patch partition block creates non-overlapping patches of the input data and projects them into embedding tokens with a resolution of 128x128x128.\n",
        "The projected tokens are then encoded by using a 3D [Swin Transformer](https://openaccess.thecvf.com/content/ICCV2021/papers/Liu_Swin_Transformer_Hierarchical_Vision_Transformer_Using_Shifted_Windows_ICCV_2021_paper.pdf) in which the self-attention is computed within local windows.\n",
        "The interaction between different windows is obtained by using 3D window shifting as illustrated below.\n",
        "\n",
        "![image](https://github.com/Project-MONAI/tutorials/blob/main/figures/shift_patch.png?raw=1)\n",
        "\n",
        "The transformer-based encoder is connected to a CNN-decoder via skip connection at multiple resolutions.\n",
        "The segmentation output consists of 3 output channels corresponding to ET, WT, and TC sub-regions and is computed by using a 1x1x1 convolutional layer followed by Sigmoid activation function.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Gr7OtgGkytds"
      },
      "source": [
        " ## Download dataset and json file\n",
        "\n",
        "- Register and download the official BraTS 21 dataset from the link below and place them into \"TrainingData\" in the dataset folder:\n",
        "\n",
        "  https://www.synapse.org/#!Synapse:syn27046444/wiki/616992\n",
        "  \n",
        "  For example, the address of a single file is as follows:\n",
        "  \n",
        "  \"TrainingData/BraTS2021_01146/BraTS2021_01146_flair.nii.gz\"\n",
        "  \n",
        "\n",
        "- Download the json file from this [link](https://drive.google.com/file/d/1i-BXYe-wZ8R9Vp3GXoajGyqaJ65Jybg1/view?usp=sharing) and placed in the same folder as the dataset.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PPs8JPUZytds"
      },
      "source": [
        "## Setup environment"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nf-ldIfRytds",
        "outputId": "d17a58f0-0b9e-4149-8b9a-f7b9d6953119"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Traceback (most recent call last):\n",
            "  File \"<string>\", line 1, in <module>\n",
            "ModuleNotFoundError: No module named 'monai'\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.4/1.4 MB\u001b[0m \u001b[31m17.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m44.6/44.6 kB\u001b[0m \u001b[31m6.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m23.7/23.7 MB\u001b[0m \u001b[31m70.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m823.6/823.6 kB\u001b[0m \u001b[31m67.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m14.1/14.1 MB\u001b[0m \u001b[31m88.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m731.7/731.7 MB\u001b[0m \u001b[31m2.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m410.6/410.6 MB\u001b[0m \u001b[31m2.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m121.6/121.6 MB\u001b[0m \u001b[31m13.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m56.5/56.5 MB\u001b[0m \u001b[31m30.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m124.2/124.2 MB\u001b[0m \u001b[31m7.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m196.0/196.0 MB\u001b[0m \u001b[31m3.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m166.0/166.0 MB\u001b[0m \u001b[31m10.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m99.1/99.1 kB\u001b[0m \u001b[31m14.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m21.1/21.1 MB\u001b[0m \u001b[31m77.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h"
          ]
        }
      ],
      "source": [
        "!python -c \"import monai\" || pip install -q \"monai-weekly[nibabel, einops]\"\n",
        "!python -c \"import matplotlib\" || pip install -q matplotlib\n",
        "%matplotlib inline"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "isF4U30Yytdt"
      },
      "source": [
        "## Setup imports"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GNKICEqKytdt",
        "outputId": "6a4f524d-e7a3-441a-8a98-f25671054151"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "MONAI version: 1.4.dev2412\n",
            "Numpy version: 1.25.2\n",
            "Pytorch version: 2.2.1+cu121\n",
            "MONAI flags: HAS_EXT = False, USE_COMPILED = False, USE_META_DICT = False\n",
            "MONAI rev id: 66af634f24d71a71a281a3325a33df9418627320\n",
            "MONAI __file__: /usr/local/lib/python3.10/dist-packages/monai/__init__.py\n",
            "\n",
            "Optional dependencies:\n",
            "Pytorch Ignite version: NOT INSTALLED or UNKNOWN VERSION.\n",
            "ITK version: NOT INSTALLED or UNKNOWN VERSION.\n",
            "Nibabel version: 4.0.2\n",
            "scikit-image version: 0.19.3\n",
            "scipy version: 1.11.4\n",
            "Pillow version: 9.4.0\n",
            "Tensorboard version: 2.15.2\n",
            "gdown version: 4.7.3\n",
            "TorchVision version: 0.17.1+cu121\n",
            "tqdm version: 4.66.2\n",
            "lmdb version: NOT INSTALLED or UNKNOWN VERSION.\n",
            "psutil version: 5.9.5\n",
            "pandas version: 1.5.3\n",
            "einops version: 0.7.0\n",
            "transformers version: 4.38.2\n",
            "mlflow version: NOT INSTALLED or UNKNOWN VERSION.\n",
            "pynrrd version: NOT INSTALLED or UNKNOWN VERSION.\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 json\n",
        "import shutil\n",
        "import tempfile\n",
        "import time\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import nibabel as nib\n",
        "\n",
        "from monai.losses import DiceLoss\n",
        "from monai.inferers import sliding_window_inference\n",
        "from monai import transforms\n",
        "from monai.transforms import (\n",
        "    AsDiscrete,\n",
        "    Activations,\n",
        ")\n",
        "\n",
        "from monai.config import print_config\n",
        "from monai.metrics import DiceMetric\n",
        "from monai.utils.enums import MetricReduction\n",
        "from monai.networks.nets import SwinUNETR\n",
        "from monai import data\n",
        "from monai.data import decollate_batch\n",
        "from functools import partial\n",
        "\n",
        "import torch\n",
        "import monai.utils\n",
        "\n",
        "\n",
        "print_config()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V7EDQSDRytdu"
      },
      "source": [
        "## Setup data directory\n",
        "\n",
        "You can specify a directory with the `MONAI_DATA_DIRECTORY` environment variable.  \n",
        "This allows you to save results and reuse downloads.  \n",
        "If not specified a temporary directory will be used."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uUmN9RUdytdu",
        "outputId": "c0ee02b1-7dde-4398-da77-2ea1951d9a19"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/tmp/tmpl4fmmacz\n"
          ]
        }
      ],
      "source": [
        "directory = os.environ.get(\"MONAI_DATA_DIRECTORY\")\n",
        "root_dir = tempfile.mkdtemp() if directory is None else directory\n",
        "print(root_dir)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "EN2QVYMvVaCL"
      },
      "outputs": [],
      "source": [
        "model_name = \"Attention_Swin_UNETR\"\n",
        "base_dir = \"/content/drive/MyDrive/Model/\"+model_name+\"/\"\n",
        "if not os.path.exists(base_dir):\n",
        "    os.makedirs(base_dir)\n",
        "log_dir = \"/content/drive/MyDrive/Model/\"+model_name+\"/logs/\"\n",
        "if not os.path.exists(log_dir):\n",
        "    os.makedirs(log_dir)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cSPSUjzhytdu"
      },
      "source": [
        "## Setup average meter, fold reader, checkpoint saver"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "id": "6oJ34N6cytdu"
      },
      "outputs": [],
      "source": [
        "class AverageMeter(object):\n",
        "    def __init__(self):\n",
        "        self.reset()\n",
        "\n",
        "    def reset(self):\n",
        "        self.val = 0\n",
        "        self.avg = 0\n",
        "        self.sum = 0\n",
        "        self.count = 0\n",
        "\n",
        "    def update(self, val, n=1):\n",
        "        self.val = val\n",
        "        self.sum += val * n\n",
        "        self.count += n\n",
        "        self.avg = np.where(self.count > 0, self.sum / self.count, self.sum)\n",
        "\n",
        "\n",
        "def datafold_read(datalist, basedir, fold=0, key=\"training\"):\n",
        "    with open(datalist) as f:\n",
        "        json_data = json.load(f)\n",
        "\n",
        "    json_data = json_data[key]\n",
        "\n",
        "    for d in json_data:\n",
        "        for k in d:\n",
        "            if isinstance(d[k], list):\n",
        "                d[k] = [os.path.join(basedir, iv) for iv in d[k]]\n",
        "            elif isinstance(d[k], str):\n",
        "                d[k] = os.path.join(basedir, d[k]) if len(d[k]) > 0 else d[k]\n",
        "\n",
        "    tr = []\n",
        "    val = []\n",
        "    for d in json_data:\n",
        "        if \"fold\" in d and d[\"fold\"] == fold:\n",
        "            val.append(d)\n",
        "        else:\n",
        "            tr.append(d)\n",
        "\n",
        "    return tr, val\n",
        "\n",
        "\n",
        "def save_checkpoint(model, optimizer, scheduler, epoch, model_name=model_name, best_acc=0, dir_add=base_dir, is_best=True):\n",
        "    state_dict = model.state_dict()\n",
        "    optimizer_state_dict = optimizer.state_dict()\n",
        "    scheduler_state_dict = scheduler.state_dict()\n",
        "    save_dict = {\"epoch\": epoch, \"best_acc\": best_acc, \"state_dict\": state_dict,\"optimizer_state_dict\": optimizer_state_dict, \"scheduler_state_dict\": scheduler_state_dict}\n",
        "    if is_best:\n",
        "        filename = \"best_\"+model_name+\".pt\"\n",
        "    else:\n",
        "        filename = \"last_\"+model_name+\".pt\"\n",
        "    filename = os.path.join(dir_add, filename)\n",
        "    torch.save(save_dict, filename)\n",
        "    print(\"Saving checkpoint\", filename)\n",
        "\n",
        "\n",
        "def save_logs(data, train=True, dir_add = base_dir):\n",
        "    base_dir = dir_add+\"logs/\"\n",
        "    if train:\n",
        "        filename = \"train_logs.txt\"\n",
        "    else:\n",
        "        filename = \"val_logs.txt\"\n",
        "    filename = os.path.join(base_dir, filename)\n",
        "    with open(filename, \"a+\") as f:\n",
        "        f.write(data +'\\n')\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4ikfU60Rytdv"
      },
      "source": [
        "## Setup dataloader"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "j5wKhJt2ytdv"
      },
      "outputs": [],
      "source": [
        "def get_loader(batch_size, data_dir, json_list, fold, roi):\n",
        "    data_dir = data_dir\n",
        "    datalist_json = json_list\n",
        "    train_files, validation_files = datafold_read(datalist=datalist_json, basedir=data_dir, fold=fold)\n",
        "    train_transform = transforms.Compose(\n",
        "        [\n",
        "            transforms.LoadImaged(keys=[\"image\", \"label\"]),\n",
        "            transforms.ConvertToMultiChannelBasedOnBratsClassesd(keys=\"label\"),\n",
        "            transforms.CropForegroundd(\n",
        "                keys=[\"image\", \"label\"],\n",
        "                source_key=\"image\",\n",
        "                k_divisible=[roi[0], roi[1], roi[2]],\n",
        "            ),\n",
        "            transforms.RandSpatialCropd(\n",
        "                keys=[\"image\", \"label\"],\n",
        "                roi_size=[roi[0], roi[1], roi[2]],\n",
        "                random_size=False,\n",
        "            ),\n",
        "            transforms.RandFlipd(keys=[\"image\", \"label\"], prob=0.5, spatial_axis=0),\n",
        "            transforms.RandFlipd(keys=[\"image\", \"label\"], prob=0.5, spatial_axis=1),\n",
        "            transforms.RandFlipd(keys=[\"image\", \"label\"], prob=0.5, spatial_axis=2),\n",
        "            transforms.NormalizeIntensityd(keys=\"image\", nonzero=True, channel_wise=True),\n",
        "            transforms.RandScaleIntensityd(keys=\"image\", factors=0.1, prob=1.0),\n",
        "            transforms.RandShiftIntensityd(keys=\"image\", offsets=0.1, prob=1.0),\n",
        "        ]\n",
        "    )\n",
        "    val_transform = transforms.Compose(\n",
        "        [\n",
        "            transforms.LoadImaged(keys=[\"image\", \"label\"]),\n",
        "            transforms.ConvertToMultiChannelBasedOnBratsClassesd(keys=\"label\"),\n",
        "            transforms.NormalizeIntensityd(keys=\"image\", nonzero=True, channel_wise=True),\n",
        "        ]\n",
        "    )\n",
        "\n",
        "    train_ds = data.Dataset(data=train_files, transform=train_transform)\n",
        "\n",
        "    train_loader = data.DataLoader(\n",
        "        train_ds,\n",
        "        batch_size=batch_size,\n",
        "        shuffle=True,\n",
        "        num_workers=4,\n",
        "        pin_memory=True,\n",
        "    )\n",
        "    val_ds = data.Dataset(data=validation_files, transform=val_transform)\n",
        "    val_loader = data.DataLoader(\n",
        "        val_ds,\n",
        "        batch_size=1,\n",
        "        shuffle=False,\n",
        "        num_workers=4,\n",
        "        pin_memory=True,\n",
        "    )\n",
        "\n",
        "    return train_loader, val_loader"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CB4fl4UAytdv"
      },
      "source": [
        "## Set dataset root directory and hyper-parameters\n",
        "\n",
        "The following hyper-parameters are set for the purpose of this tutorial. However, additional changes, as described below, maybe beneficial.\n",
        "\n",
        "If GPU memory is not sufficient, reduce sw_batch_size to 2 or batch_size to 1.\n",
        "\n",
        "Decrease val_every (validation frequency) to 1 for obtaining more accurate checkpoints."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Sj0nLhSMytdv",
        "outputId": "2ba3ba80-b061-41c2-d651-c6c9a76e38ce"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/monai/utils/deprecate_utils.py:321: FutureWarning: monai.transforms.croppad.dictionary CropForegroundd.__init__:allow_smaller: Current default value of argument `allow_smaller=True` has been deprecated since version 1.2. It will be changed to `allow_smaller=False` in version 1.5.\n",
            "  warn_deprecated(argname, msg, warning_category)\n"
          ]
        }
      ],
      "source": [
        "data_dir = \"\"\n",
        "json_list = \"./brats21_folds_small.json\"\n",
        "roi = (128, 128, 128)\n",
        "batch_size = 1\n",
        "sw_batch_size = 2\n",
        "fold = 1\n",
        "infer_overlap = 0.5\n",
        "max_epochs = 180\n",
        "val_every = 5\n",
        "monai.utils.set_determinism(seed=42)\n",
        "train_loader, val_loader = get_loader(batch_size, data_dir, json_list, fold, roi)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aXVBLym1ytdv"
      },
      "source": [
        "## Check data shape and visualize"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 294
        },
        "id": "q0cbREipytdv",
        "outputId": "53f70f37-2a76-4a5d-c4dd-d60edab4b67f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "image shape: (240, 240, 155), label shape: (240, 240, 155)\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1800x600 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "img_add = os.path.join(data_dir, \"TrainingData/BraTS2021_00006/BraTS2021_00006_flair.nii.gz\")\n",
        "label_add = os.path.join(data_dir, \"TrainingData/BraTS2021_00006/BraTS2021_00006_seg.nii.gz\")\n",
        "img = nib.load(img_add).get_fdata()\n",
        "label = nib.load(label_add).get_fdata()\n",
        "print(f\"image shape: {img.shape}, label shape: {label.shape}\")\n",
        "plt.figure(\"image\", (18, 6))\n",
        "plt.subplot(1, 2, 1)\n",
        "plt.title(\"image\")\n",
        "plt.imshow(img[:, :, 78], cmap=\"gray\")\n",
        "plt.subplot(1, 2, 2)\n",
        "plt.title(\"label\")\n",
        "plt.imshow(label[:, :, 78])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1yt1pleDytdv"
      },
      "source": [
        "## Create Swin UNETR model\n",
        "\n",
        "In this scetion, we create Swin UNETR model for the 3-class brain tumor semantic segmentation. We use a feature size of 48. We also use gradient checkpointing (use_checkpoint) for more memory-efficient training. However, use_checkpoint for faster training if enough GPU memory is available.  "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "CJ3XfIA4zHoX"
      },
      "outputs": [],
      "source": [
        "from attention_swin_unetr import SwinUNETR"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "JPsjGR6wytdv"
      },
      "outputs": [],
      "source": [
        "os.environ[\"CUDA_DEVICE_ORDER\"] = \"PCI_BUS_ID\"\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
        "\n",
        "model = SwinUNETR(\n",
        "    img_size=roi,\n",
        "    in_channels=4,\n",
        "    out_channels=3,\n",
        "    feature_size=48,\n",
        "    drop_rate=0.0,\n",
        "    attn_drop_rate=0.0,\n",
        "    dropout_path_rate=0.0,\n",
        "    use_checkpoint=True,\n",
        ")\n",
        "if torch.cuda.device_count() > 1:\n",
        "    model = torch.nn.DataParallel(model)\n",
        "model = model.to(device)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3E-ft0_MDW_d",
        "outputId": "c012e2fb-6d5d-4a56-b46d-3f9c7f43bc9a"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SwinUNETR(\n",
              "  (swinViT): SwinTransformer(\n",
              "    (patch_embed): PatchEmbed(\n",
              "      (proj): Conv3d(4, 48, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n",
              "    )\n",
              "    (pos_drop): Dropout(p=0.0, inplace=False)\n",
              "    (layers1): ModuleList(\n",
              "      (0): BasicLayer(\n",
              "        (blocks): ModuleList(\n",
              "          (0-1): 2 x SwinTransformerBlock(\n",
              "            (norm1): LayerNorm((48,), eps=1e-05, elementwise_affine=True)\n",
              "            (attn): WindowAttention(\n",
              "              (qkv): Linear(in_features=48, out_features=144, bias=True)\n",
              "              (attn_drop): Dropout(p=0.0, inplace=False)\n",
              "              (proj): Linear(in_features=48, out_features=48, bias=True)\n",
              "              (proj_drop): Dropout(p=0.0, inplace=False)\n",
              "              (softmax): Softmax(dim=-1)\n",
              "            )\n",
              "            (drop_path): Identity()\n",
              "            (norm2): LayerNorm((48,), eps=1e-05, elementwise_affine=True)\n",
              "            (mlp): MLPBlock(\n",
              "              (linear1): Linear(in_features=48, out_features=192, bias=True)\n",
              "              (linear2): Linear(in_features=192, out_features=48, bias=True)\n",
              "              (fn): GELU(approximate='none')\n",
              "              (drop1): Dropout(p=0.0, inplace=False)\n",
              "              (drop2): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "        )\n",
              "        (downsample): PatchMerging(\n",
              "          (reduction): Linear(in_features=384, out_features=96, bias=False)\n",
              "          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)\n",
              "        )\n",
              "      )\n",
              "    )\n",
              "    (layers2): ModuleList(\n",
              "      (0): BasicLayer(\n",
              "        (blocks): ModuleList(\n",
              "          (0-1): 2 x SwinTransformerBlock(\n",
              "            (norm1): LayerNorm((96,), eps=1e-05, elementwise_affine=True)\n",
              "            (attn): WindowAttention(\n",
              "              (qkv): Linear(in_features=96, out_features=288, bias=True)\n",
              "              (attn_drop): Dropout(p=0.0, inplace=False)\n",
              "              (proj): Linear(in_features=96, out_features=96, bias=True)\n",
              "              (proj_drop): Dropout(p=0.0, inplace=False)\n",
              "              (softmax): Softmax(dim=-1)\n",
              "            )\n",
              "            (drop_path): Identity()\n",
              "            (norm2): LayerNorm((96,), eps=1e-05, elementwise_affine=True)\n",
              "            (mlp): MLPBlock(\n",
              "              (linear1): Linear(in_features=96, out_features=384, bias=True)\n",
              "              (linear2): Linear(in_features=384, out_features=96, bias=True)\n",
              "              (fn): GELU(approximate='none')\n",
              "              (drop1): Dropout(p=0.0, inplace=False)\n",
              "              (drop2): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "        )\n",
              "        (downsample): PatchMerging(\n",
              "          (reduction): Linear(in_features=768, out_features=192, bias=False)\n",
              "          (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
              "        )\n",
              "      )\n",
              "    )\n",
              "    (layers3): ModuleList(\n",
              "      (0): BasicLayer(\n",
              "        (blocks): ModuleList(\n",
              "          (0-1): 2 x SwinTransformerBlock(\n",
              "            (norm1): LayerNorm((192,), eps=1e-05, elementwise_affine=True)\n",
              "            (attn): WindowAttention(\n",
              "              (qkv): Linear(in_features=192, out_features=576, bias=True)\n",
              "              (attn_drop): Dropout(p=0.0, inplace=False)\n",
              "              (proj): Linear(in_features=192, out_features=192, bias=True)\n",
              "              (proj_drop): Dropout(p=0.0, inplace=False)\n",
              "              (softmax): Softmax(dim=-1)\n",
              "            )\n",
              "            (drop_path): Identity()\n",
              "            (norm2): LayerNorm((192,), eps=1e-05, elementwise_affine=True)\n",
              "            (mlp): MLPBlock(\n",
              "              (linear1): Linear(in_features=192, out_features=768, bias=True)\n",
              "              (linear2): Linear(in_features=768, out_features=192, bias=True)\n",
              "              (fn): GELU(approximate='none')\n",
              "              (drop1): Dropout(p=0.0, inplace=False)\n",
              "              (drop2): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "        )\n",
              "        (downsample): PatchMerging(\n",
              "          (reduction): Linear(in_features=1536, out_features=384, bias=False)\n",
              "          (norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)\n",
              "        )\n",
              "      )\n",
              "    )\n",
              "    (layers4): ModuleList(\n",
              "      (0): BasicLayer(\n",
              "        (blocks): ModuleList(\n",
              "          (0-1): 2 x SwinTransformerBlock(\n",
              "            (norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)\n",
              "            (attn): WindowAttention(\n",
              "              (qkv): Linear(in_features=384, out_features=1152, bias=True)\n",
              "              (attn_drop): Dropout(p=0.0, inplace=False)\n",
              "              (proj): Linear(in_features=384, out_features=384, bias=True)\n",
              "              (proj_drop): Dropout(p=0.0, inplace=False)\n",
              "              (softmax): Softmax(dim=-1)\n",
              "            )\n",
              "            (drop_path): Identity()\n",
              "            (norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)\n",
              "            (mlp): MLPBlock(\n",
              "              (linear1): Linear(in_features=384, out_features=1536, bias=True)\n",
              "              (linear2): Linear(in_features=1536, out_features=384, bias=True)\n",
              "              (fn): GELU(approximate='none')\n",
              "              (drop1): Dropout(p=0.0, inplace=False)\n",
              "              (drop2): Dropout(p=0.0, inplace=False)\n",
              "            )\n",
              "          )\n",
              "        )\n",
              "        (downsample): PatchMerging(\n",
              "          (reduction): Linear(in_features=3072, out_features=768, bias=False)\n",
              "          (norm): LayerNorm((3072,), eps=1e-05, elementwise_affine=True)\n",
              "        )\n",
              "      )\n",
              "    )\n",
              "  )\n",
              "  (encoder1): UnetrBasicBlock(\n",
              "    (layer): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(4, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(48, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (conv3): Convolution(\n",
              "        (conv): Conv3d(4, 48, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (norm3): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (encoder2): UnetrBasicBlock(\n",
              "    (layer): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(48, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(48, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (encoder3): UnetrBasicBlock(\n",
              "    (layer): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(96, 96, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(96, 96, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(96, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(96, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (encoder4): UnetrBasicBlock(\n",
              "    (layer): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(192, 192, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(192, 192, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(192, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(192, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (encoder10): UnetrBasicBlock(\n",
              "    (layer): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(768, 768, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(768, 768, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(768, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(768, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (decoder5): UnetrUpBlock(\n",
              "    (transp_conv): Convolution(\n",
              "      (conv): ConvTranspose3d(768, 384, kernel_size=(2, 2, 2), stride=(2, 2, 2), bias=False)\n",
              "    )\n",
              "    (conv_block): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(768, 384, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(384, 384, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(384, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(384, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (scse_layer): ChannelSELayer(\n",
              "        (avg_pool): AdaptiveAvgPool3d(output_size=1)\n",
              "        (fc): Sequential(\n",
              "          (0): Linear(in_features=384, out_features=192, bias=True)\n",
              "          (1): ReLU(inplace=True)\n",
              "          (2): Linear(in_features=192, out_features=384, bias=True)\n",
              "          (3): Sigmoid()\n",
              "        )\n",
              "      )\n",
              "      (conv3): Convolution(\n",
              "        (conv): Conv3d(768, 384, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (norm3): InstanceNorm3d(384, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (decoder4): UnetrUpBlock(\n",
              "    (transp_conv): Convolution(\n",
              "      (conv): ConvTranspose3d(384, 192, kernel_size=(2, 2, 2), stride=(2, 2, 2), bias=False)\n",
              "    )\n",
              "    (conv_block): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(384, 192, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(192, 192, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(192, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(192, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (scse_layer): ChannelSELayer(\n",
              "        (avg_pool): AdaptiveAvgPool3d(output_size=1)\n",
              "        (fc): Sequential(\n",
              "          (0): Linear(in_features=192, out_features=96, bias=True)\n",
              "          (1): ReLU(inplace=True)\n",
              "          (2): Linear(in_features=96, out_features=192, bias=True)\n",
              "          (3): Sigmoid()\n",
              "        )\n",
              "      )\n",
              "      (conv3): Convolution(\n",
              "        (conv): Conv3d(384, 192, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (norm3): InstanceNorm3d(192, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (decoder3): UnetrUpBlock(\n",
              "    (transp_conv): Convolution(\n",
              "      (conv): ConvTranspose3d(192, 96, kernel_size=(2, 2, 2), stride=(2, 2, 2), bias=False)\n",
              "    )\n",
              "    (conv_block): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(192, 96, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(96, 96, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(96, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(96, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (scse_layer): ChannelSELayer(\n",
              "        (avg_pool): AdaptiveAvgPool3d(output_size=1)\n",
              "        (fc): Sequential(\n",
              "          (0): Linear(in_features=96, out_features=48, bias=True)\n",
              "          (1): ReLU(inplace=True)\n",
              "          (2): Linear(in_features=48, out_features=96, bias=True)\n",
              "          (3): Sigmoid()\n",
              "        )\n",
              "      )\n",
              "      (conv3): Convolution(\n",
              "        (conv): Conv3d(192, 96, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (norm3): InstanceNorm3d(96, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (decoder2): UnetrUpBlock(\n",
              "    (transp_conv): Convolution(\n",
              "      (conv): ConvTranspose3d(96, 48, kernel_size=(2, 2, 2), stride=(2, 2, 2), bias=False)\n",
              "    )\n",
              "    (conv_block): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(96, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(48, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (scse_layer): ChannelSELayer(\n",
              "        (avg_pool): AdaptiveAvgPool3d(output_size=1)\n",
              "        (fc): Sequential(\n",
              "          (0): Linear(in_features=48, out_features=24, bias=True)\n",
              "          (1): ReLU(inplace=True)\n",
              "          (2): Linear(in_features=24, out_features=48, bias=True)\n",
              "          (3): Sigmoid()\n",
              "        )\n",
              "      )\n",
              "      (conv3): Convolution(\n",
              "        (conv): Conv3d(96, 48, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (norm3): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (decoder1): UnetrUpBlock(\n",
              "    (transp_conv): Convolution(\n",
              "      (conv): ConvTranspose3d(48, 48, kernel_size=(2, 2, 2), stride=(2, 2, 2), bias=False)\n",
              "    )\n",
              "    (conv_block): UnetResBlock(\n",
              "      (conv1): Convolution(\n",
              "        (conv): Conv3d(96, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (conv2): Convolution(\n",
              "        (conv): Conv3d(48, 48, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (lrelu): LeakyReLU(negative_slope=0.01, inplace=True)\n",
              "      (norm1): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (norm2): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "      (scse_layer): ChannelSELayer(\n",
              "        (avg_pool): AdaptiveAvgPool3d(output_size=1)\n",
              "        (fc): Sequential(\n",
              "          (0): Linear(in_features=48, out_features=24, bias=True)\n",
              "          (1): ReLU(inplace=True)\n",
              "          (2): Linear(in_features=24, out_features=48, bias=True)\n",
              "          (3): Sigmoid()\n",
              "        )\n",
              "      )\n",
              "      (conv3): Convolution(\n",
              "        (conv): Conv3d(96, 48, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n",
              "      )\n",
              "      (norm3): InstanceNorm3d(48, eps=1e-05, momentum=0.1, affine=False, track_running_stats=False)\n",
              "    )\n",
              "  )\n",
              "  (out): UnetOutBlock(\n",
              "    (conv): Convolution(\n",
              "      (conv): Conv3d(48, 3, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n",
              "    )\n",
              "  )\n",
              ")"
            ]
          },
          "metadata": {},
          "execution_count": 16
        }
      ],
      "source": [
        "model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EbFUEcbYRXLv",
        "outputId": "252d1232-66a6-4137-f842-3181c93f9b7c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "----------------------------------------------------------------\n",
            "        Layer (type)               Output Shape         Param #\n",
            "================================================================\n",
            "            Conv3d-1       [-1, 48, 64, 64, 64]           1,584\n",
            "        PatchEmbed-2       [-1, 48, 64, 64, 64]               0\n",
            "           Dropout-3       [-1, 48, 64, 64, 64]               0\n",
            "         LayerNorm-4       [-1, 64, 64, 64, 48]              96\n",
            "            Linear-5             [-1, 343, 144]           7,056\n",
            "           Softmax-6          [-1, 3, 343, 343]               0\n",
            "           Dropout-7          [-1, 3, 343, 343]               0\n",
            "            Linear-8              [-1, 343, 48]           2,352\n",
            "           Dropout-9              [-1, 343, 48]               0\n",
            "  WindowAttention-10              [-1, 343, 48]               0\n",
            "         Identity-11       [-1, 64, 64, 64, 48]               0\n",
            "        LayerNorm-12       [-1, 64, 64, 64, 48]              96\n",
            "           Linear-13      [-1, 64, 64, 64, 192]           9,408\n",
            "             GELU-14      [-1, 64, 64, 64, 192]               0\n",
            "          Dropout-15      [-1, 64, 64, 64, 192]               0\n",
            "           Linear-16       [-1, 64, 64, 64, 48]           9,264\n",
            "          Dropout-17       [-1, 64, 64, 64, 48]               0\n",
            "         MLPBlock-18       [-1, 64, 64, 64, 48]               0\n",
            "         Identity-19       [-1, 64, 64, 64, 48]               0\n",
            "SwinTransformerBlock-20       [-1, 64, 64, 64, 48]               0\n",
            "        LayerNorm-21       [-1, 64, 64, 64, 48]              96\n",
            "           Linear-22             [-1, 343, 144]           7,056\n",
            "          Softmax-23          [-1, 3, 343, 343]               0\n",
            "          Dropout-24          [-1, 3, 343, 343]               0\n",
            "           Linear-25              [-1, 343, 48]           2,352\n",
            "          Dropout-26              [-1, 343, 48]               0\n",
            "  WindowAttention-27              [-1, 343, 48]               0\n",
            "         Identity-28       [-1, 64, 64, 64, 48]               0\n",
            "        LayerNorm-29       [-1, 64, 64, 64, 48]              96\n",
            "           Linear-30      [-1, 64, 64, 64, 192]           9,408\n",
            "             GELU-31      [-1, 64, 64, 64, 192]               0\n",
            "          Dropout-32      [-1, 64, 64, 64, 192]               0\n",
            "           Linear-33       [-1, 64, 64, 64, 48]           9,264\n",
            "          Dropout-34       [-1, 64, 64, 64, 48]               0\n",
            "         MLPBlock-35       [-1, 64, 64, 64, 48]               0\n",
            "         Identity-36       [-1, 64, 64, 64, 48]               0\n",
            "SwinTransformerBlock-37       [-1, 64, 64, 64, 48]               0\n",
            "        LayerNorm-38      [-1, 32, 32, 32, 384]             768\n",
            "           Linear-39       [-1, 32, 32, 32, 96]          36,864\n",
            "     PatchMerging-40       [-1, 32, 32, 32, 96]               0\n",
            "       BasicLayer-41       [-1, 96, 32, 32, 32]               0\n",
            "        LayerNorm-42       [-1, 32, 32, 32, 96]             192\n",
            "           Linear-43             [-1, 343, 288]          27,936\n",
            "          Softmax-44          [-1, 6, 343, 343]               0\n",
            "          Dropout-45          [-1, 6, 343, 343]               0\n",
            "           Linear-46              [-1, 343, 96]           9,312\n",
            "          Dropout-47              [-1, 343, 96]               0\n",
            "  WindowAttention-48              [-1, 343, 96]               0\n",
            "         Identity-49       [-1, 32, 32, 32, 96]               0\n",
            "        LayerNorm-50       [-1, 32, 32, 32, 96]             192\n",
            "           Linear-51      [-1, 32, 32, 32, 384]          37,248\n",
            "             GELU-52      [-1, 32, 32, 32, 384]               0\n",
            "          Dropout-53      [-1, 32, 32, 32, 384]               0\n",
            "           Linear-54       [-1, 32, 32, 32, 96]          36,960\n",
            "          Dropout-55       [-1, 32, 32, 32, 96]               0\n",
            "         MLPBlock-56       [-1, 32, 32, 32, 96]               0\n",
            "         Identity-57       [-1, 32, 32, 32, 96]               0\n",
            "SwinTransformerBlock-58       [-1, 32, 32, 32, 96]               0\n",
            "        LayerNorm-59       [-1, 32, 32, 32, 96]             192\n",
            "           Linear-60             [-1, 343, 288]          27,936\n",
            "          Softmax-61          [-1, 6, 343, 343]               0\n",
            "          Dropout-62          [-1, 6, 343, 343]               0\n",
            "           Linear-63              [-1, 343, 96]           9,312\n",
            "          Dropout-64              [-1, 343, 96]               0\n",
            "  WindowAttention-65              [-1, 343, 96]               0\n",
            "         Identity-66       [-1, 32, 32, 32, 96]               0\n",
            "        LayerNorm-67       [-1, 32, 32, 32, 96]             192\n",
            "           Linear-68      [-1, 32, 32, 32, 384]          37,248\n",
            "             GELU-69      [-1, 32, 32, 32, 384]               0\n",
            "          Dropout-70      [-1, 32, 32, 32, 384]               0\n",
            "           Linear-71       [-1, 32, 32, 32, 96]          36,960\n",
            "          Dropout-72       [-1, 32, 32, 32, 96]               0\n",
            "         MLPBlock-73       [-1, 32, 32, 32, 96]               0\n",
            "         Identity-74       [-1, 32, 32, 32, 96]               0\n",
            "SwinTransformerBlock-75       [-1, 32, 32, 32, 96]               0\n",
            "        LayerNorm-76      [-1, 16, 16, 16, 768]           1,536\n",
            "           Linear-77      [-1, 16, 16, 16, 192]         147,456\n",
            "     PatchMerging-78      [-1, 16, 16, 16, 192]               0\n",
            "       BasicLayer-79      [-1, 192, 16, 16, 16]               0\n",
            "        LayerNorm-80      [-1, 16, 16, 16, 192]             384\n",
            "           Linear-81             [-1, 343, 576]         111,168\n",
            "          Softmax-82         [-1, 12, 343, 343]               0\n",
            "          Dropout-83         [-1, 12, 343, 343]               0\n",
            "           Linear-84             [-1, 343, 192]          37,056\n",
            "          Dropout-85             [-1, 343, 192]               0\n",
            "  WindowAttention-86             [-1, 343, 192]               0\n",
            "         Identity-87      [-1, 16, 16, 16, 192]               0\n",
            "        LayerNorm-88      [-1, 16, 16, 16, 192]             384\n",
            "           Linear-89      [-1, 16, 16, 16, 768]         148,224\n",
            "             GELU-90      [-1, 16, 16, 16, 768]               0\n",
            "          Dropout-91      [-1, 16, 16, 16, 768]               0\n",
            "           Linear-92      [-1, 16, 16, 16, 192]         147,648\n",
            "          Dropout-93      [-1, 16, 16, 16, 192]               0\n",
            "         MLPBlock-94      [-1, 16, 16, 16, 192]               0\n",
            "         Identity-95      [-1, 16, 16, 16, 192]               0\n",
            "SwinTransformerBlock-96      [-1, 16, 16, 16, 192]               0\n",
            "        LayerNorm-97      [-1, 16, 16, 16, 192]             384\n",
            "           Linear-98             [-1, 343, 576]         111,168\n",
            "          Softmax-99         [-1, 12, 343, 343]               0\n",
            "         Dropout-100         [-1, 12, 343, 343]               0\n",
            "          Linear-101             [-1, 343, 192]          37,056\n",
            "         Dropout-102             [-1, 343, 192]               0\n",
            " WindowAttention-103             [-1, 343, 192]               0\n",
            "        Identity-104      [-1, 16, 16, 16, 192]               0\n",
            "       LayerNorm-105      [-1, 16, 16, 16, 192]             384\n",
            "          Linear-106      [-1, 16, 16, 16, 768]         148,224\n",
            "            GELU-107      [-1, 16, 16, 16, 768]               0\n",
            "         Dropout-108      [-1, 16, 16, 16, 768]               0\n",
            "          Linear-109      [-1, 16, 16, 16, 192]         147,648\n",
            "         Dropout-110      [-1, 16, 16, 16, 192]               0\n",
            "        MLPBlock-111      [-1, 16, 16, 16, 192]               0\n",
            "        Identity-112      [-1, 16, 16, 16, 192]               0\n",
            "SwinTransformerBlock-113      [-1, 16, 16, 16, 192]               0\n",
            "       LayerNorm-114        [-1, 8, 8, 8, 1536]           3,072\n",
            "          Linear-115         [-1, 8, 8, 8, 384]         589,824\n",
            "    PatchMerging-116         [-1, 8, 8, 8, 384]               0\n",
            "      BasicLayer-117         [-1, 384, 8, 8, 8]               0\n",
            "       LayerNorm-118         [-1, 8, 8, 8, 384]             768\n",
            "          Linear-119            [-1, 343, 1152]         443,520\n",
            "         Softmax-120         [-1, 24, 343, 343]               0\n",
            "         Dropout-121         [-1, 24, 343, 343]               0\n",
            "          Linear-122             [-1, 343, 384]         147,840\n",
            "         Dropout-123             [-1, 343, 384]               0\n",
            " WindowAttention-124             [-1, 343, 384]               0\n",
            "        Identity-125         [-1, 8, 8, 8, 384]               0\n",
            "       LayerNorm-126         [-1, 8, 8, 8, 384]             768\n",
            "          Linear-127        [-1, 8, 8, 8, 1536]         591,360\n",
            "            GELU-128        [-1, 8, 8, 8, 1536]               0\n",
            "         Dropout-129        [-1, 8, 8, 8, 1536]               0\n",
            "          Linear-130         [-1, 8, 8, 8, 384]         590,208\n",
            "         Dropout-131         [-1, 8, 8, 8, 384]               0\n",
            "        MLPBlock-132         [-1, 8, 8, 8, 384]               0\n",
            "        Identity-133         [-1, 8, 8, 8, 384]               0\n",
            "SwinTransformerBlock-134         [-1, 8, 8, 8, 384]               0\n",
            "       LayerNorm-135         [-1, 8, 8, 8, 384]             768\n",
            "          Linear-136            [-1, 343, 1152]         443,520\n",
            "         Softmax-137         [-1, 24, 343, 343]               0\n",
            "         Dropout-138         [-1, 24, 343, 343]               0\n",
            "          Linear-139             [-1, 343, 384]         147,840\n",
            "         Dropout-140             [-1, 343, 384]               0\n",
            " WindowAttention-141             [-1, 343, 384]               0\n",
            "        Identity-142         [-1, 8, 8, 8, 384]               0\n",
            "       LayerNorm-143         [-1, 8, 8, 8, 384]             768\n",
            "          Linear-144        [-1, 8, 8, 8, 1536]         591,360\n",
            "            GELU-145        [-1, 8, 8, 8, 1536]               0\n",
            "         Dropout-146        [-1, 8, 8, 8, 1536]               0\n",
            "          Linear-147         [-1, 8, 8, 8, 384]         590,208\n",
            "         Dropout-148         [-1, 8, 8, 8, 384]               0\n",
            "        MLPBlock-149         [-1, 8, 8, 8, 384]               0\n",
            "        Identity-150         [-1, 8, 8, 8, 384]               0\n",
            "SwinTransformerBlock-151         [-1, 8, 8, 8, 384]               0\n",
            "       LayerNorm-152        [-1, 4, 4, 4, 3072]           6,144\n",
            "          Linear-153         [-1, 4, 4, 4, 768]       2,359,296\n",
            "    PatchMerging-154         [-1, 4, 4, 4, 768]               0\n",
            "      BasicLayer-155         [-1, 768, 4, 4, 4]               0\n",
            " SwinTransformer-156  [[-1, 48, 64, 64, 64], [-1, 96, 32, 32, 32], [-1, 192, 16, 16, 16], [-1, 384, 8, 8, 8], [-1, 768, 4, 4, 4]]               0\n",
            "          Conv3d-157    [-1, 48, 128, 128, 128]           5,184\n",
            "  InstanceNorm3d-158    [-1, 48, 128, 128, 128]               0\n",
            "       LeakyReLU-159    [-1, 48, 128, 128, 128]               0\n",
            "          Conv3d-160    [-1, 48, 128, 128, 128]          62,208\n",
            "  InstanceNorm3d-161    [-1, 48, 128, 128, 128]               0\n",
            "          Conv3d-162    [-1, 48, 128, 128, 128]             192\n",
            "  InstanceNorm3d-163    [-1, 48, 128, 128, 128]               0\n",
            "       LeakyReLU-164    [-1, 48, 128, 128, 128]               0\n",
            "    UnetResBlock-165    [-1, 48, 128, 128, 128]               0\n",
            " UnetrBasicBlock-166    [-1, 48, 128, 128, 128]               0\n",
            "          Conv3d-167       [-1, 48, 64, 64, 64]          62,208\n",
            "  InstanceNorm3d-168       [-1, 48, 64, 64, 64]               0\n",
            "       LeakyReLU-169       [-1, 48, 64, 64, 64]               0\n",
            "          Conv3d-170       [-1, 48, 64, 64, 64]          62,208\n",
            "  InstanceNorm3d-171       [-1, 48, 64, 64, 64]               0\n",
            "       LeakyReLU-172       [-1, 48, 64, 64, 64]               0\n",
            "    UnetResBlock-173       [-1, 48, 64, 64, 64]               0\n",
            " UnetrBasicBlock-174       [-1, 48, 64, 64, 64]               0\n",
            "          Conv3d-175       [-1, 96, 32, 32, 32]         248,832\n",
            "  InstanceNorm3d-176       [-1, 96, 32, 32, 32]               0\n",
            "       LeakyReLU-177       [-1, 96, 32, 32, 32]               0\n",
            "          Conv3d-178       [-1, 96, 32, 32, 32]         248,832\n",
            "  InstanceNorm3d-179       [-1, 96, 32, 32, 32]               0\n",
            "       LeakyReLU-180       [-1, 96, 32, 32, 32]               0\n",
            "    UnetResBlock-181       [-1, 96, 32, 32, 32]               0\n",
            " UnetrBasicBlock-182       [-1, 96, 32, 32, 32]               0\n",
            "          Conv3d-183      [-1, 192, 16, 16, 16]         995,328\n",
            "  InstanceNorm3d-184      [-1, 192, 16, 16, 16]               0\n",
            "       LeakyReLU-185      [-1, 192, 16, 16, 16]               0\n",
            "          Conv3d-186      [-1, 192, 16, 16, 16]         995,328\n",
            "  InstanceNorm3d-187      [-1, 192, 16, 16, 16]               0\n",
            "       LeakyReLU-188      [-1, 192, 16, 16, 16]               0\n",
            "    UnetResBlock-189      [-1, 192, 16, 16, 16]               0\n",
            " UnetrBasicBlock-190      [-1, 192, 16, 16, 16]               0\n",
            "          Conv3d-191         [-1, 768, 4, 4, 4]      15,925,248\n",
            "  InstanceNorm3d-192         [-1, 768, 4, 4, 4]               0\n",
            "       LeakyReLU-193         [-1, 768, 4, 4, 4]               0\n",
            "          Conv3d-194         [-1, 768, 4, 4, 4]      15,925,248\n",
            "  InstanceNorm3d-195         [-1, 768, 4, 4, 4]               0\n",
            "       LeakyReLU-196         [-1, 768, 4, 4, 4]               0\n",
            "    UnetResBlock-197         [-1, 768, 4, 4, 4]               0\n",
            " UnetrBasicBlock-198         [-1, 768, 4, 4, 4]               0\n",
            " ConvTranspose3d-199         [-1, 384, 8, 8, 8]       2,359,296\n",
            "          Conv3d-200         [-1, 384, 8, 8, 8]       7,962,624\n",
            "  InstanceNorm3d-201         [-1, 384, 8, 8, 8]               0\n",
            "       LeakyReLU-202         [-1, 384, 8, 8, 8]               0\n",
            "          Conv3d-203         [-1, 384, 8, 8, 8]       3,981,312\n",
            "  InstanceNorm3d-204         [-1, 384, 8, 8, 8]               0\n",
            "          Conv3d-205         [-1, 384, 8, 8, 8]         294,912\n",
            "  InstanceNorm3d-206         [-1, 384, 8, 8, 8]               0\n",
            "AdaptiveAvgPool3d-207         [-1, 384, 1, 1, 1]               0\n",
            "          Linear-208                  [-1, 192]          73,920\n",
            "            ReLU-209                  [-1, 192]               0\n",
            "          Linear-210                  [-1, 384]          74,112\n",
            "         Sigmoid-211                  [-1, 384]               0\n",
            "  ChannelSELayer-212         [-1, 384, 8, 8, 8]               0\n",
            "       LeakyReLU-213         [-1, 384, 8, 8, 8]               0\n",
            "    UnetResBlock-214         [-1, 384, 8, 8, 8]               0\n",
            "    UnetrUpBlock-215         [-1, 384, 8, 8, 8]               0\n",
            " ConvTranspose3d-216      [-1, 192, 16, 16, 16]         589,824\n",
            "          Conv3d-217      [-1, 192, 16, 16, 16]       1,990,656\n",
            "  InstanceNorm3d-218      [-1, 192, 16, 16, 16]               0\n",
            "       LeakyReLU-219      [-1, 192, 16, 16, 16]               0\n",
            "          Conv3d-220      [-1, 192, 16, 16, 16]         995,328\n",
            "  InstanceNorm3d-221      [-1, 192, 16, 16, 16]               0\n",
            "          Conv3d-222      [-1, 192, 16, 16, 16]          73,728\n",
            "  InstanceNorm3d-223      [-1, 192, 16, 16, 16]               0\n",
            "AdaptiveAvgPool3d-224         [-1, 192, 1, 1, 1]               0\n",
            "          Linear-225                   [-1, 96]          18,528\n",
            "            ReLU-226                   [-1, 96]               0\n",
            "          Linear-227                  [-1, 192]          18,624\n",
            "         Sigmoid-228                  [-1, 192]               0\n",
            "  ChannelSELayer-229      [-1, 192, 16, 16, 16]               0\n",
            "       LeakyReLU-230      [-1, 192, 16, 16, 16]               0\n",
            "    UnetResBlock-231      [-1, 192, 16, 16, 16]               0\n",
            "    UnetrUpBlock-232      [-1, 192, 16, 16, 16]               0\n",
            " ConvTranspose3d-233       [-1, 96, 32, 32, 32]         147,456\n",
            "          Conv3d-234       [-1, 96, 32, 32, 32]         497,664\n",
            "  InstanceNorm3d-235       [-1, 96, 32, 32, 32]               0\n",
            "       LeakyReLU-236       [-1, 96, 32, 32, 32]               0\n",
            "          Conv3d-237       [-1, 96, 32, 32, 32]         248,832\n",
            "  InstanceNorm3d-238       [-1, 96, 32, 32, 32]               0\n",
            "          Conv3d-239       [-1, 96, 32, 32, 32]          18,432\n",
            "  InstanceNorm3d-240       [-1, 96, 32, 32, 32]               0\n",
            "AdaptiveAvgPool3d-241          [-1, 96, 1, 1, 1]               0\n",
            "          Linear-242                   [-1, 48]           4,656\n",
            "            ReLU-243                   [-1, 48]               0\n",
            "          Linear-244                   [-1, 96]           4,704\n",
            "         Sigmoid-245                   [-1, 96]               0\n",
            "  ChannelSELayer-246       [-1, 96, 32, 32, 32]               0\n",
            "       LeakyReLU-247       [-1, 96, 32, 32, 32]               0\n",
            "    UnetResBlock-248       [-1, 96, 32, 32, 32]               0\n",
            "    UnetrUpBlock-249       [-1, 96, 32, 32, 32]               0\n",
            " ConvTranspose3d-250       [-1, 48, 64, 64, 64]          36,864\n",
            "          Conv3d-251       [-1, 48, 64, 64, 64]         124,416\n",
            "  InstanceNorm3d-252       [-1, 48, 64, 64, 64]               0\n",
            "       LeakyReLU-253       [-1, 48, 64, 64, 64]               0\n",
            "          Conv3d-254       [-1, 48, 64, 64, 64]          62,208\n",
            "  InstanceNorm3d-255       [-1, 48, 64, 64, 64]               0\n",
            "          Conv3d-256       [-1, 48, 64, 64, 64]           4,608\n",
            "  InstanceNorm3d-257       [-1, 48, 64, 64, 64]               0\n",
            "AdaptiveAvgPool3d-258          [-1, 48, 1, 1, 1]               0\n",
            "          Linear-259                   [-1, 24]           1,176\n",
            "            ReLU-260                   [-1, 24]               0\n",
            "          Linear-261                   [-1, 48]           1,200\n",
            "         Sigmoid-262                   [-1, 48]               0\n",
            "  ChannelSELayer-263       [-1, 48, 64, 64, 64]               0\n",
            "       LeakyReLU-264       [-1, 48, 64, 64, 64]               0\n",
            "    UnetResBlock-265       [-1, 48, 64, 64, 64]               0\n",
            "    UnetrUpBlock-266       [-1, 48, 64, 64, 64]               0\n",
            " ConvTranspose3d-267    [-1, 48, 128, 128, 128]          18,432\n",
            "          Conv3d-268    [-1, 48, 128, 128, 128]         124,416\n",
            "  InstanceNorm3d-269    [-1, 48, 128, 128, 128]               0\n",
            "       LeakyReLU-270    [-1, 48, 128, 128, 128]               0\n",
            "          Conv3d-271    [-1, 48, 128, 128, 128]          62,208\n",
            "  InstanceNorm3d-272    [-1, 48, 128, 128, 128]               0\n",
            "          Conv3d-273    [-1, 48, 128, 128, 128]           4,608\n",
            "  InstanceNorm3d-274    [-1, 48, 128, 128, 128]               0\n",
            "AdaptiveAvgPool3d-275          [-1, 48, 1, 1, 1]               0\n",
            "          Linear-276                   [-1, 24]           1,176\n",
            "            ReLU-277                   [-1, 24]               0\n",
            "          Linear-278                   [-1, 48]           1,200\n",
            "         Sigmoid-279                   [-1, 48]               0\n",
            "  ChannelSELayer-280    [-1, 48, 128, 128, 128]               0\n",
            "       LeakyReLU-281    [-1, 48, 128, 128, 128]               0\n",
            "    UnetResBlock-282    [-1, 48, 128, 128, 128]               0\n",
            "    UnetrUpBlock-283    [-1, 48, 128, 128, 128]               0\n",
            "          Conv3d-284     [-1, 3, 128, 128, 128]             147\n",
            "    UnetOutBlock-285     [-1, 3, 128, 128, 128]               0\n",
            "================================================================\n",
            "Total params: 62,193,507\n",
            "Trainable params: 62,193,507\n",
            "Non-trainable params: 0\n",
            "----------------------------------------------------------------\n",
            "Input size (MB): 32.00\n",
            "Forward/backward pass size (MB): 25328.32\n",
            "Params size (MB): 237.25\n",
            "Estimated Total Size (MB): 25597.57\n",
            "----------------------------------------------------------------\n"
          ]
        }
      ],
      "source": [
        "from torchsummary import summary\n",
        "summary(model, (4, 128, 128, 128))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4VGuWAUtytdv"
      },
      "source": [
        "## Optimizer and loss function"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "id": "Fu56VNj_ytdv"
      },
      "outputs": [],
      "source": [
        "torch.backends.cudnn.benchmark = True\n",
        "dice_loss = DiceLoss(to_onehot_y=False, sigmoid=True)\n",
        "post_sigmoid = Activations(sigmoid=True)\n",
        "post_pred = AsDiscrete(argmax=False, threshold=0.5)\n",
        "dice_acc = DiceMetric(include_background=True, reduction=MetricReduction.MEAN_BATCH, get_not_nans=True)\n",
        "model_inferer = partial(\n",
        "    sliding_window_inference,\n",
        "    roi_size=[roi[0], roi[1], roi[2]],\n",
        "    sw_batch_size=sw_batch_size,\n",
        "    predictor=model,\n",
        "    overlap=infer_overlap,\n",
        ")\n",
        "\n",
        "optimizer = torch.optim.AdamW(model.parameters(), lr=1e-4, weight_decay=1e-5)\n",
        "scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max=max_epochs)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2phuZ80wytdw"
      },
      "source": [
        "## Define Train and Validation Epoch"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "B4WguQBFytdw",
        "scrolled": true
      },
      "outputs": [],
      "source": [
        "def train_epoch(model, loader, optimizer, epoch, loss_func):\n",
        "    model.train()\n",
        "    start_time = time.time()\n",
        "    run_loss = AverageMeter()\n",
        "    for idx, batch_data in enumerate(loader):\n",
        "        data, target = batch_data[\"image\"].to(device), batch_data[\"label\"].to(device)\n",
        "        logits = model(data)\n",
        "        loss = loss_func(logits, target)\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        run_loss.update(loss.item(), n=batch_size)\n",
        "        print(\n",
        "            \"Epoch {}/{} {}/{}\".format(epoch, max_epochs, idx, len(loader)),\n",
        "            \"loss: {:.4f}\".format(run_loss.avg),\n",
        "            \"time {:.2f}s\".format(time.time() - start_time),\n",
        "        )\n",
        "        start_time = time.time()\n",
        "    return run_loss.avg\n",
        "\n",
        "\n",
        "def val_epoch(\n",
        "    model,\n",
        "    loader,\n",
        "    epoch,\n",
        "    acc_func,\n",
        "    model_inferer=None,\n",
        "    post_sigmoid=None,\n",
        "    post_pred=None,\n",
        "):\n",
        "    model.eval()\n",
        "    start_time = time.time()\n",
        "    run_acc = AverageMeter()\n",
        "    run_loss = AverageMeter()\n",
        "\n",
        "    with torch.no_grad():\n",
        "        for idx, batch_data in enumerate(loader):\n",
        "            data, target = batch_data[\"image\"].to(device), batch_data[\"label\"].to(device)\n",
        "            logits = model_inferer(data)\n",
        "            loss = dice_loss(logits, target)\n",
        "            run_loss.update(loss.item(), n=batch_size)\n",
        "            val_labels_list = decollate_batch(target)\n",
        "            val_outputs_list = decollate_batch(logits)\n",
        "            val_output_convert = [post_pred(post_sigmoid(val_pred_tensor)) for val_pred_tensor in val_outputs_list]\n",
        "            acc_func.reset()\n",
        "            acc_func(y_pred=val_output_convert, y=val_labels_list)\n",
        "            acc, not_nans = acc_func.aggregate()\n",
        "            run_acc.update(acc.cpu().numpy(), n=not_nans.cpu().numpy())\n",
        "            dice_tc = run_acc.avg[0]\n",
        "            dice_wt = run_acc.avg[1]\n",
        "            dice_et = run_acc.avg[2]\n",
        "            print(\n",
        "                \"Val {}/{} {}/{}\".format(epoch, max_epochs, idx, len(loader)),\n",
        "                \", dice_tc:\",\n",
        "                dice_tc,\n",
        "                \", dice_wt:\",\n",
        "                dice_wt,\n",
        "                \", dice_et:\",\n",
        "                dice_et,\n",
        "                \", loss: {:.4f}\".format(run_loss.avg),\n",
        "                \", time {:.2f}s\".format(time.time() - start_time),\n",
        "            )\n",
        "            start_time = time.time()\n",
        "    return run_acc.avg, run_loss.avg"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EfOmr_38MUuG"
      },
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "i1OLSb1zVHPV"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "id": "asAHwYfdEhSa"
      },
      "outputs": [],
      "source": [
        "filename = os.path.join(base_dir, model_name+\"_last.pt\")\n",
        "if os.path.exists(filename):\n",
        "  load_dict = torch.load(filename)\n",
        "else:\n",
        "    load_dict = None\n",
        "    start_epoch = 0\n",
        "    val_acc_best = 0.0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "id": "TqyGo7A3SrtE"
      },
      "outputs": [],
      "source": [
        "load_dict"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "id": "9S1d7eV_MVs8"
      },
      "outputs": [],
      "source": [
        "if load_dict is not None:\n",
        "  model.load_state_dict(load_dict[\"state_dict\"])\n",
        "  if \"optimizer_state_dict\" in load_dict:\n",
        "    optimizer.load_state_dict(load_dict[\"optimizer_state_dict\"])\n",
        "  if \"scheduler\" in load_dict:\n",
        "    scheduler.load_state_dict(load_dict[\"scheduler_state_dict\"])\n",
        "  if \"epoch\" in load_dict:\n",
        "    start_epoch = load_dict[\"epoch\"] + 1\n",
        "  else:\n",
        "    start_epoch = 0\n",
        "  if \"best_acc\" in load_dict:\n",
        "    val_acc_best = load_dict[\"best_acc\"]\n",
        "  else:\n",
        "    val_acc_best = 0.0\n",
        "  print(\"Loaded model from epoch {}\".format(start_epoch))\n",
        "  print(\"Best Accuracy: \", val_acc_best)\n",
        "  model.to(device)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4UHQb6l3ytdw"
      },
      "source": [
        "## Define Trainer"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "id": "XXzkn7Woytdw"
      },
      "outputs": [],
      "source": [
        "def trainer(\n",
        "    model,\n",
        "    train_loader,\n",
        "    val_loader,\n",
        "    optimizer,\n",
        "    loss_func,\n",
        "    acc_func,\n",
        "    scheduler,\n",
        "    model_inferer=None,\n",
        "    start_epoch=0,\n",
        "    post_sigmoid=None,\n",
        "    post_pred=None,\n",
        "):\n",
        "    val_acc_max = val_acc_best\n",
        "    dices_tc = []\n",
        "    dices_wt = []\n",
        "    dices_et = []\n",
        "    dices_avg = []\n",
        "    loss_epochs = []\n",
        "    val_loss_epochs = []\n",
        "    trains_epoch = []\n",
        "    for epoch in range(start_epoch, max_epochs):\n",
        "        print(time.ctime(), \"Epoch:\", epoch)\n",
        "        epoch_time = time.time()\n",
        "        train_loss = train_epoch(\n",
        "            model,\n",
        "            train_loader,\n",
        "            optimizer,\n",
        "            epoch=epoch,\n",
        "            loss_func=loss_func,\n",
        "        )\n",
        "        print(\n",
        "            \"Final training  {}/{}\".format(epoch, max_epochs - 1),\n",
        "            \"loss: {:.4f}\".format(train_loss),\n",
        "            \"time {:.2f}s\".format(time.time() - epoch_time),\n",
        "        )\n",
        "        stats = \"Epoch:  {}/{}\\t loss: {:.4f}\".format(epoch+1, max_epochs, train_loss)\n",
        "        save_logs(stats, True)\n",
        "        if (epoch + 1) % val_every == 0 or epoch == 0:\n",
        "            loss_epochs.append(train_loss)\n",
        "            trains_epoch.append(int(epoch))\n",
        "            epoch_time = time.time()\n",
        "            val_acc, val_loss = val_epoch(\n",
        "                model,\n",
        "                val_loader,\n",
        "                epoch=epoch,\n",
        "                acc_func=acc_func,\n",
        "                model_inferer=model_inferer,\n",
        "                post_sigmoid=post_sigmoid,\n",
        "                post_pred=post_pred,\n",
        "            )\n",
        "            dice_tc = val_acc[0]\n",
        "            dice_wt = val_acc[1]\n",
        "            dice_et = val_acc[2]\n",
        "            val_avg_acc = np.mean(val_acc)\n",
        "            print(\n",
        "                \"Final validation stats {}/{}\".format(epoch, max_epochs - 1),\n",
        "                \", dice_tc:\",\n",
        "                dice_tc,\n",
        "                \", dice_wt:\",\n",
        "                dice_wt,\n",
        "                \", dice_et:\",\n",
        "                dice_et,\n",
        "                \", Dice_Avg:\",\n",
        "                val_avg_acc,\n",
        "                \", time {:.2f}s\".format(time.time() - epoch_time),\n",
        "            )\n",
        "            stats = \"Epoch:  {}/{}\\t loss: {:.4f}\".format(epoch+1, max_epochs, val_loss)\n",
        "            stats += \"\\t dice_tc: {:.4f}\".format(dice_tc)\n",
        "            stats += \"\\t dice_wt: {:.4f}\".format(dice_wt)\n",
        "            stats += \"\\t dice_et: {:.4f}\".format(dice_et)\n",
        "            stats += \"\\t Dice_Avg: {:.4f}\".format(val_avg_acc)\n",
        "            save_logs(stats, False)\n",
        "            dices_tc.append(dice_tc)\n",
        "            dices_wt.append(dice_wt)\n",
        "            dices_et.append(dice_et)\n",
        "            dices_avg.append(val_avg_acc)\n",
        "            val_loss_epochs.append(val_loss)\n",
        "            save_checkpoint(\n",
        "                    model,\n",
        "                    optimizer,\n",
        "                    scheduler,\n",
        "                    epoch,\n",
        "                    best_acc=max(val_acc_max,val_avg_acc),\n",
        "                    is_best=False,\n",
        "                )\n",
        "            if val_avg_acc > val_acc_max:\n",
        "                print(\"new best ({:.6f} --> {:.6f}). \".format(val_acc_max, val_avg_acc))\n",
        "                val_acc_max = val_avg_acc\n",
        "                save_checkpoint(\n",
        "                    model,\n",
        "                    optimizer,\n",
        "                    scheduler,\n",
        "                    epoch,\n",
        "                    best_acc=val_acc_max,\n",
        "                )\n",
        "            scheduler.step()\n",
        "    print(\"Training Finished !, Best Accuracy: \", val_acc_max)\n",
        "    return (\n",
        "        val_acc_max,\n",
        "        dices_tc,\n",
        "        dices_wt,\n",
        "        dices_et,\n",
        "        dices_avg,\n",
        "        loss_epochs,\n",
        "        trains_epoch,\n",
        "        val_loss_epochs,\n",
        "    )"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MziSUobbytdw"
      },
      "source": [
        "## Execute training"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "isCjF7T6ytdw",
        "outputId": "97889694-6fc3-4d8b-8ed1-83c885b10a24",
        "scrolled": true
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mon Mar 25 21:02:59 2024 Epoch: 0\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 0/80 loss: 0.9768 time 145.60s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 1/80 loss: 0.9884 time 3.22s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 2/80 loss: 0.9891 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 3/80 loss: 0.9817 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 4/80 loss: 0.9672 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 5/80 loss: 0.9500 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 6/80 loss: 0.9571 time 3.39s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 7/80 loss: 0.9625 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 8/80 loss: 0.9560 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 9/80 loss: 0.9501 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 10/80 loss: 0.9416 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 11/80 loss: 0.9364 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 12/80 loss: 0.9412 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 13/80 loss: 0.9389 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 14/80 loss: 0.9386 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 15/80 loss: 0.9417 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 16/80 loss: 0.9447 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 17/80 loss: 0.9474 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 18/80 loss: 0.9500 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 19/80 loss: 0.9524 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 20/80 loss: 0.9523 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 21/80 loss: 0.9511 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 22/80 loss: 0.9474 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 23/80 loss: 0.9464 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 24/80 loss: 0.9468 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 25/80 loss: 0.9455 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 26/80 loss: 0.9473 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 27/80 loss: 0.9490 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 28/80 loss: 0.9501 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 29/80 loss: 0.9516 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 30/80 loss: 0.9521 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 31/80 loss: 0.9501 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 32/80 loss: 0.9502 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 33/80 loss: 0.9515 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 34/80 loss: 0.9491 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 35/80 loss: 0.9465 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 36/80 loss: 0.9460 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 37/80 loss: 0.9469 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 38/80 loss: 0.9462 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 39/80 loss: 0.9475 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 40/80 loss: 0.9442 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 41/80 loss: 0.9452 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 42/80 loss: 0.9465 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 43/80 loss: 0.9470 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 44/80 loss: 0.9456 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 45/80 loss: 0.9468 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 46/80 loss: 0.9474 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 47/80 loss: 0.9479 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 48/80 loss: 0.9486 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 49/80 loss: 0.9482 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 50/80 loss: 0.9486 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 51/80 loss: 0.9485 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 52/80 loss: 0.9487 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 53/80 loss: 0.9482 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 54/80 loss: 0.9473 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 55/80 loss: 0.9477 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 56/80 loss: 0.9479 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 57/80 loss: 0.9487 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 58/80 loss: 0.9488 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 59/80 loss: 0.9485 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 60/80 loss: 0.9492 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 61/80 loss: 0.9498 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 62/80 loss: 0.9502 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 63/80 loss: 0.9497 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 64/80 loss: 0.9498 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 65/80 loss: 0.9487 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 66/80 loss: 0.9494 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 67/80 loss: 0.9496 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 68/80 loss: 0.9492 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 69/80 loss: 0.9489 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 70/80 loss: 0.9488 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 71/80 loss: 0.9493 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 72/80 loss: 0.9498 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 73/80 loss: 0.9483 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 74/80 loss: 0.9481 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 75/80 loss: 0.9477 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 76/80 loss: 0.9481 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 77/80 loss: 0.9475 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 78/80 loss: 0.9474 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 0/180 79/80 loss: 0.9468 time 3.15s\n",
            "Final training  0/179 loss: 0.9468 time 394.38s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 0/21 , dice_tc: 0.06277787 , dice_wt: 0.42334035 , dice_et: 0.023552475 , loss: 0.9650 , time 19.94s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 1/21 , dice_tc: 0.1455796 , dice_wt: 0.37013727 , dice_et: 0.04300438 , loss: 0.9528 , time 17.34s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 2/21 , dice_tc: 0.121952005 , dice_wt: 0.29940724 , dice_et: 0.039030466 , loss: 0.9595 , time 17.35s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 3/21 , dice_tc: 0.15599996 , dice_wt: 0.3154586 , dice_et: 0.058510587 , loss: 0.9554 , time 17.30s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 4/21 , dice_tc: 0.12707105 , dice_wt: 0.2566557 , dice_et: 0.048664745 , loss: 0.9635 , time 17.28s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 5/21 , dice_tc: 0.13339366 , dice_wt: 0.263048 , dice_et: 0.05224532 , loss: 0.9629 , time 17.27s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 6/21 , dice_tc: 0.115062766 , dice_wt: 0.23250426 , dice_et: 0.045142304 , loss: 0.9676 , time 17.27s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 7/21 , dice_tc: 0.10217626 , dice_wt: 0.22406046 , dice_et: 0.040507127 , loss: 0.9700 , time 17.27s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 8/21 , dice_tc: 0.118258834 , dice_wt: 0.23406498 , dice_et: 0.057901163 , loss: 0.9679 , time 17.30s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 9/21 , dice_tc: 0.10890593 , dice_wt: 0.23223023 , dice_et: 0.05349735 , loss: 0.9687 , time 17.31s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 10/21 , dice_tc: 0.10993123 , dice_wt: 0.22697477 , dice_et: 0.049310923 , loss: 0.9692 , time 17.29s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 11/21 , dice_tc: 0.10173107 , dice_wt: 0.2113309 , dice_et: 0.045865845 , loss: 0.9714 , time 17.30s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 12/21 , dice_tc: 0.096190214 , dice_wt: 0.20484217 , dice_et: 0.04343011 , loss: 0.9726 , time 17.33s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 13/21 , dice_tc: 0.09282617 , dice_wt: 0.21709159 , dice_et: 0.04206555 , loss: 0.9721 , time 17.32s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 14/21 , dice_tc: 0.09360257 , dice_wt: 0.2190275 , dice_et: 0.04328053 , loss: 0.9715 , time 17.30s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 15/21 , dice_tc: 0.08925091 , dice_wt: 0.21321402 , dice_et: 0.041426264 , loss: 0.9726 , time 17.31s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 16/21 , dice_tc: 0.09626486 , dice_wt: 0.21522693 , dice_et: 0.043332644 , loss: 0.9711 , time 17.31s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 17/21 , dice_tc: 0.09327113 , dice_wt: 0.21346027 , dice_et: 0.042436916 , loss: 0.9716 , time 17.31s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 18/21 , dice_tc: 0.09738043 , dice_wt: 0.21553355 , dice_et: 0.043080002 , loss: 0.9711 , time 17.33s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 19/21 , dice_tc: 0.09916608 , dice_wt: 0.21781456 , dice_et: 0.044412114 , loss: 0.9708 , time 17.32s\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "torch.Size([2, 768, 8, 8, 8])\n",
            "torch.Size([2, 384, 16, 16, 16])\n",
            "torch.Size([2, 192, 32, 32, 32])\n",
            "torch.Size([2, 96, 64, 64, 64])\n",
            "torch.Size([2, 96, 128, 128, 128])\n",
            "Val 0/180 20/21 , dice_tc: 0.097974576 , dice_wt: 0.2152801 , dice_et: 0.044412114 , loss: 0.9711 , time 17.30s\n",
            "Final validation stats 0/179 , dice_tc: 0.097974576 , dice_wt: 0.2152801 , dice_et: 0.044412114 , Dice_Avg: 0.11922226 , time 366.12s\n",
            "Saving checkpoint /content/drive/MyDrive/Model/Attention_Swin_UNETR/last_Attention_Swin_UNETR.pt\n",
            "new best (0.000000 --> 0.119222). \n",
            "Saving checkpoint /content/drive/MyDrive/Model/Attention_Swin_UNETR/best_Attention_Swin_UNETR.pt\n",
            "Mon Mar 25 21:15:46 2024 Epoch: 1\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 0/80 loss: 0.8667 time 4.95s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 1/80 loss: 0.9333 time 3.12s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 2/80 loss: 0.9366 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 3/80 loss: 0.9324 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 4/80 loss: 0.9450 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 5/80 loss: 0.9394 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 6/80 loss: 0.9448 time 3.17s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 7/80 loss: 0.9199 time 3.19s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 8/80 loss: 0.9202 time 3.19s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 9/80 loss: 0.9282 time 3.19s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 10/80 loss: 0.9261 time 3.17s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 11/80 loss: 0.9215 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 12/80 loss: 0.9261 time 3.17s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 13/80 loss: 0.9265 time 3.16s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 14/80 loss: 0.9285 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 15/80 loss: 0.9298 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 16/80 loss: 0.9329 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 17/80 loss: 0.9358 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 18/80 loss: 0.9386 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 19/80 loss: 0.9384 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 20/80 loss: 0.9396 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 21/80 loss: 0.9397 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 22/80 loss: 0.9396 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 23/80 loss: 0.9411 time 3.12s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 24/80 loss: 0.9397 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 25/80 loss: 0.9372 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 26/80 loss: 0.9335 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 27/80 loss: 0.9345 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 28/80 loss: 0.9364 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 29/80 loss: 0.9381 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 30/80 loss: 0.9373 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 31/80 loss: 0.9382 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 32/80 loss: 0.9401 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 33/80 loss: 0.9417 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 34/80 loss: 0.9394 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 35/80 loss: 0.9401 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 36/80 loss: 0.9372 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 37/80 loss: 0.9386 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 38/80 loss: 0.9398 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 39/80 loss: 0.9374 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 40/80 loss: 0.9360 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 41/80 loss: 0.9375 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 42/80 loss: 0.9376 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 43/80 loss: 0.9339 time 3.15s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 44/80 loss: 0.9335 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 45/80 loss: 0.9347 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 46/80 loss: 0.9351 time 3.13s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n",
            "Epoch 1/180 47/80 loss: 0.9364 time 3.14s\n",
            "torch.Size([1, 768, 8, 8, 8])\n",
            "torch.Size([1, 384, 16, 16, 16])\n",
            "torch.Size([1, 192, 32, 32, 32])\n",
            "torch.Size([1, 96, 64, 64, 64])\n",
            "torch.Size([1, 96, 128, 128, 128])\n"
          ]
        }
      ],
      "source": [
        "start_epoch = start_epoch\n",
        "\n",
        "(\n",
        "    val_acc_max,\n",
        "    dices_tc,\n",
        "    dices_wt,\n",
        "    dices_et,\n",
        "    dices_avg,\n",
        "    loss_epochs,\n",
        "    trains_epoch,\n",
        "    val_loss_epochs,\n",
        ") = trainer(\n",
        "    model=model,\n",
        "    train_loader=train_loader,\n",
        "    val_loader=val_loader,\n",
        "    optimizer=optimizer,\n",
        "    loss_func=dice_loss,\n",
        "    acc_func=dice_acc,\n",
        "    scheduler=scheduler,\n",
        "    model_inferer=model_inferer,\n",
        "    start_epoch=start_epoch,\n",
        "    post_sigmoid=post_sigmoid,\n",
        "    post_pred=post_pred,\n",
        ")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "hlJDbUWPgYLB"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Qt0xYCVCytdw"
      },
      "outputs": [],
      "source": [
        "print(f\"train completed, best average dice: {val_acc_max:.4f} \")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "yHsE1o6zMVyb"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0oBzp7Peytdw"
      },
      "source": [
        "### Plot the loss and Dice metric"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "X-fNZMgBytdw"
      },
      "outputs": [],
      "source": [
        "plt.figure(\"train\", (12, 6))\n",
        "plt.subplot(1, 4, 1)\n",
        "plt.title(\"Epoch Average Loss\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.plot(trains_epoch, loss_epochs, color=\"red\")\n",
        "plt.subplot(1, 4, 2)\n",
        "plt.title(\"Epoch Average Val Loss\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.plot(trains_epoch, val_loss_epochs, color=\"blue\")\n",
        "plt.subplot(1, 4, 3)\n",
        "plt.title(\"Epoch Average Loss Comparison\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.plot(trains_epoch, loss_epochs, color=\"red\", label=\"Training Loss\")\n",
        "plt.plot(trains_epoch, val_loss_epochs, color=\"blue\", label=\"Validation Loss\")\n",
        "plt.legend()\n",
        "plt.subplot(1, 4, 4)\n",
        "plt.title(\"Val Mean Dice\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.plot(trains_epoch, dices_avg, color=\"green\")\n",
        "plt.show()\n",
        "plt.figure(\"train\", (18, 6))\n",
        "plt.subplot(1, 3, 1)\n",
        "plt.title(\"Val Mean Dice TC\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.plot(trains_epoch, dices_tc, color=\"blue\")\n",
        "plt.subplot(1, 3, 2)\n",
        "plt.title(\"Val Mean Dice WT\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.plot(trains_epoch, dices_wt, color=\"brown\")\n",
        "plt.subplot(1, 3, 3)\n",
        "plt.title(\"Val Mean Dice ET\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.plot(trains_epoch, dices_et, color=\"purple\")\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 456
        },
        "id": "bWZcmzDsnNYw",
        "outputId": "00fffcd6-c8f0-4a79-bd04-c694bde7a280"
      },
      "outputs": [
        {
          "ename": "FileNotFoundError",
          "evalue": "[Errno 2] No such file or directory: '/content/drive/MyDrive/Model/Attention_Swin_UNETR/best_swin_unetr_model.pt'",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mFileNotFoundError\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-27-76b463651898>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload_state_dict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjoin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbase_dir\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\"best_swin_unetr_model.pt\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"state_dict\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msave\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstate_dict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjoin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mroot_dir\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m'swin_unetr.pt'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/torch/serialization.py\u001b[0m in \u001b[0;36mload\u001b[0;34m(f, map_location, pickle_module, weights_only, mmap, **pickle_load_args)\u001b[0m\n\u001b[1;32m    996\u001b[0m         \u001b[0mpickle_load_args\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'encoding'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m'utf-8'\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    997\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 998\u001b[0;31m     \u001b[0;32mwith\u001b[0m \u001b[0m_open_file_like\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'rb'\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mopened_file\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    999\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0m_is_zipfile\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mopened_file\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1000\u001b[0m             \u001b[0;31m# The zipfile reader is going to advance the current file position.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/torch/serialization.py\u001b[0m in \u001b[0;36m_open_file_like\u001b[0;34m(name_or_buffer, mode)\u001b[0m\n\u001b[1;32m    443\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0m_open_file_like\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname_or_buffer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmode\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    444\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0m_is_path\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname_or_buffer\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 445\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0m_open_file\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname_or_buffer\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmode\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    446\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    447\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0;34m'w'\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mmode\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/torch/serialization.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, name, mode)\u001b[0m\n\u001b[1;32m    424\u001b[0m \u001b[0;32mclass\u001b[0m \u001b[0m_open_file\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_opener\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    425\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__init__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmode\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 426\u001b[0;31m         \u001b[0msuper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__init__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mopen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmode\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    427\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    428\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__exit__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: '/content/drive/MyDrive/Model/Attention_Swin_UNETR/best_swin_unetr_model.pt'"
          ]
        }
      ],
      "source": [
        "model.load_state_dict(torch.load(os.path.join(base_dir,\"best_\"+model_name+\".pt\"))[\"state_dict\"])\n",
        "torch.save(model.state_dict(), os.path.join(root_dir,'swin_unetr.pt'))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0QxAzZdfytdw"
      },
      "source": [
        "## Create test set dataloader"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KWzsgggxytdx"
      },
      "outputs": [],
      "source": [
        "case_num = \"01619\"\n",
        "\n",
        "test_files = [\n",
        "    {\n",
        "        \"image\": [\n",
        "            os.path.join(\n",
        "                data_dir,\n",
        "                \"TrainingData/BraTS2021_\" + case_num + \"/BraTS2021_\" + case_num + \"_flair.nii.gz\",\n",
        "            ),\n",
        "            os.path.join(\n",
        "                data_dir,\n",
        "                \"TrainingData/BraTS2021_\" + case_num + \"/BraTS2021_\" + case_num + \"_t1ce.nii.gz\",\n",
        "            ),\n",
        "            os.path.join(\n",
        "                data_dir,\n",
        "                \"TrainingData/BraTS2021_\" + case_num + \"/BraTS2021_\" + case_num + \"_t1.nii.gz\",\n",
        "            ),\n",
        "            os.path.join(\n",
        "                data_dir,\n",
        "                \"TrainingData/BraTS2021_\" + case_num + \"/BraTS2021_\" + case_num + \"_t2.nii.gz\",\n",
        "            ),\n",
        "        ],\n",
        "        \"label\": os.path.join(\n",
        "            data_dir,\n",
        "            \"TrainingData/BraTS2021_\" + case_num + \"/BraTS2021_\" + case_num + \"_seg.nii.gz\",\n",
        "        ),\n",
        "    }\n",
        "]\n",
        "\n",
        "test_transform = transforms.Compose(\n",
        "    [\n",
        "        transforms.LoadImaged(keys=[\"image\", \"label\"]),\n",
        "        transforms.ConvertToMultiChannelBasedOnBratsClassesd(keys=\"label\"),\n",
        "        transforms.NormalizeIntensityd(keys=\"image\", nonzero=True, channel_wise=True),\n",
        "    ]\n",
        ")\n",
        "\n",
        "test_ds = data.Dataset(data=test_files, transform=test_transform)\n",
        "\n",
        "test_loader = data.DataLoader(\n",
        "    test_ds,\n",
        "    batch_size=1,\n",
        "    shuffle=False,\n",
        "    num_workers=8,\n",
        "    pin_memory=True,\n",
        ")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cha4hOmzytdx"
      },
      "source": [
        "## Load the best saved checkpoint and perform inference\n",
        "\n",
        "We select a single case from the validation set and perform inference to compare the model segmentation output with the corresponding label."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ipAfvyXcytdx"
      },
      "outputs": [],
      "source": [
        "model.load_state_dict(torch.load(os.path.join(root_dir,'swin_unetr.pt')))\n",
        "model.to(device)\n",
        "model.eval()\n",
        "\n",
        "model_inferer_test = partial(\n",
        "    sliding_window_inference,\n",
        "    roi_size=[roi[0], roi[1], roi[2]],\n",
        "    sw_batch_size=1,\n",
        "    predictor=model,\n",
        "    overlap=0.6,\n",
        ")\n",
        "\n",
        "\n",
        "with torch.no_grad():\n",
        "    for batch_data in test_loader:\n",
        "        image = batch_data[\"image\"].cuda()\n",
        "        prob = torch.sigmoid(model_inferer_test(image))\n",
        "        seg = prob[0].detach().cpu().numpy()\n",
        "        seg = (seg > 0.5).astype(np.int8)\n",
        "        seg_out = np.zeros((seg.shape[1], seg.shape[2], seg.shape[3]))\n",
        "        seg_out[seg[1] == 1] = 2\n",
        "        seg_out[seg[0] == 1] = 1\n",
        "        seg_out[seg[2] == 1] = 4"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GhQDRV1yytdx"
      },
      "source": [
        "## Visualize segmentation output and compare with label"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 223
        },
        "id": "MfW_elwIytdx",
        "outputId": "8c71bcbe-98b0-4c7c-9584-308a5416ca7e"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1800x600 with 3 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "slice_num = 67\n",
        "img_add = os.path.join(\n",
        "    data_dir,\n",
        "    \"TrainingData/BraTS2021_\" + case_num + \"/BraTS2021_\" + case_num + \"_t1ce.nii.gz\",\n",
        ")\n",
        "label_add = os.path.join(\n",
        "    data_dir,\n",
        "    \"TrainingData/BraTS2021_\" + case_num + \"/BraTS2021_\" + case_num + \"_seg.nii.gz\",\n",
        ")\n",
        "img = nib.load(img_add).get_fdata()\n",
        "label = nib.load(label_add).get_fdata()\n",
        "plt.figure(\"image\", (18, 6))\n",
        "plt.subplot(1, 3, 1)\n",
        "plt.title(\"image\")\n",
        "plt.imshow(img[:, :, slice_num], cmap=\"gray\")\n",
        "plt.subplot(1, 3, 2)\n",
        "plt.title(\"label\")\n",
        "plt.imshow(label[:, :, slice_num])\n",
        "plt.subplot(1, 3, 3)\n",
        "plt.title(\"segmentation\")\n",
        "plt.imshow(seg_out[:, :, slice_num])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cP2AwHnFFXyX",
        "outputId": "adaaa58a-269f-4758-a3d8-e855acc818ab"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "dice_tc: 0.5513648 , dice_wt: 0.8542162 , dice_et: 0.8136095\n"
          ]
        }
      ],
      "source": [
        "model.eval()\n",
        "run_acc = AverageMeter()\n",
        "with torch.no_grad():\n",
        "        for idx, batch_data in enumerate(test_loader):\n",
        "            data, target = batch_data[\"image\"].to(device), batch_data[\"label\"].to(device)\n",
        "            logits = model_inferer(data)\n",
        "            val_labels_list = decollate_batch(target)\n",
        "            val_outputs_list = decollate_batch(logits)\n",
        "            val_output_convert = [post_pred(post_sigmoid(val_pred_tensor)) for val_pred_tensor in val_outputs_list]\n",
        "            dice_acc.reset()\n",
        "            dice_acc(y_pred=val_output_convert, y=val_labels_list)\n",
        "            acc, not_nans = dice_acc.aggregate()\n",
        "            run_acc.update(acc.cpu().numpy(), n=not_nans.cpu().numpy())\n",
        "            dice_tc = run_acc.avg[0]\n",
        "            dice_wt = run_acc.avg[1]\n",
        "            dice_et = run_acc.avg[2]\n",
        "            print(\n",
        "                \"dice_tc:\",\n",
        "                dice_tc,\n",
        "                \", dice_wt:\",\n",
        "                dice_wt,\n",
        "                \", dice_et:\",\n",
        "                dice_et\n",
        "            )\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cy6NrhvPHMfj"
      },
      "source": [
        "Swin-unetr - (4e-4), 128*128*128, 100 epochs -"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BLG_InbXytdx"
      },
      "source": [
        "## Cleanup data directory\n",
        "\n",
        "Remove directory if a temporary was used."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "btRf0fvmytdx"
      },
      "outputs": [],
      "source": [
        "if directory is None:\n",
        "    shutil.rmtree(root_dir)"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "gpuType": "T4",
      "machine_shape": "hm",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.13"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}