train_MiCE_.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Αντίγραφο train_MiCE.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "8d5955c02d1c460cbf07f3bc15d2527e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_384cb904311241faa2d2dade79050b7b",
              "IPY_MODEL_50be5d9e386e44a79bc4c80e9e92d019",
              "IPY_MODEL_e089bda735c143b5a9b4a9e059a4ad14"
            ],
            "layout": "IPY_MODEL_b6cc7203d541453ba72364da792532c8"
          }
        },
        "384cb904311241faa2d2dade79050b7b": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_28b2b78b4ba449698b880e5f28f8253d",
            "placeholder": "​",
            "style": "IPY_MODEL_7d7b9f5f81e145bda1ae6433186aff4d",
            "value": ""
          }
        },
        "50be5d9e386e44a79bc4c80e9e92d019": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_f1b00a70bf6343e680dae099eaba7d6a",
            "max": 170498071,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_8861d23735d14bba889931a51d3b851c",
            "value": 170498071
          }
        },
        "e089bda735c143b5a9b4a9e059a4ad14": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_ca1e48ac6d2d414f8b2bbdf2d2312dcb",
            "placeholder": "​",
            "style": "IPY_MODEL_6aa16ba20bd5468e897c70419835231f",
            "value": " 170499072/? [00:02&lt;00:00, 84056870.85it/s]"
          }
        },
        "b6cc7203d541453ba72364da792532c8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "28b2b78b4ba449698b880e5f28f8253d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "7d7b9f5f81e145bda1ae6433186aff4d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "f1b00a70bf6343e680dae099eaba7d6a": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "8861d23735d14bba889931a51d3b851c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "ca1e48ac6d2d414f8b2bbdf2d2312dcb": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "6aa16ba20bd5468e897c70419835231f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "83b3a8a5183745d1b3e1f92b1e553121": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_1f1ce47020dc42dd876fd6b36feba28c",
              "IPY_MODEL_b4e5978fd88541729496a092041ce105",
              "IPY_MODEL_6ada64656f0e4e74b75447c8a7a24607"
            ],
            "layout": "IPY_MODEL_aabce2b8337a428f99dc60cd04fb263c"
          }
        },
        "1f1ce47020dc42dd876fd6b36feba28c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_122bbaef86c44a8ea48514edb5faf5ed",
            "placeholder": "​",
            "style": "IPY_MODEL_e972f96a1995468eb1fa35be2ddd3078",
            "value": ""
          }
        },
        "b4e5978fd88541729496a092041ce105": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_4b6253e5d35c454ba5e462c746aeaa01",
            "max": 170498071,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_2a7cf12f3dd94fa2840910889cf50ddb",
            "value": 170498071
          }
        },
        "6ada64656f0e4e74b75447c8a7a24607": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_0d6e41f06f6c49d3a29268deb5ab118d",
            "placeholder": "​",
            "style": "IPY_MODEL_60f6848d0d0d4953afb21823d77727d2",
            "value": " 170499072/? [00:02&lt;00:00, 74278593.65it/s]"
          }
        },
        "aabce2b8337a428f99dc60cd04fb263c": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "122bbaef86c44a8ea48514edb5faf5ed": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "e972f96a1995468eb1fa35be2ddd3078": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "4b6253e5d35c454ba5e462c746aeaa01": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "2a7cf12f3dd94fa2840910889cf50ddb": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "0d6e41f06f6c49d3a29268deb5ab118d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "60f6848d0d0d4953afb21823d77727d2": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BBK2a8LcYZLc",
        "outputId": "cbb6e532-dee6-4ba3-f45a-96f1305b1096"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import math\n",
        "import numpy as np\n",
        "import torch.utils.model_zoo as model_zoo\n",
        "\n",
        "__all__ = ['ResNet', 'resnet18', 'resnet34', 'resnet50', 'resnet101',\n",
        "           'resnet152']\n",
        "\n",
        "model_urls = {\n",
        "    'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',\n",
        "    'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',\n",
        "    'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',\n",
        "    'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',\n",
        "    'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',\n",
        "}\n",
        "\n",
        "\n",
        "def conv3x3(in_planes, out_planes, stride=1):\n",
        "    \"\"\"3x3 convolution with padding\"\"\"\n",
        "    return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,\n",
        "                     padding=1, bias=False)\n",
        "\n",
        "\n",
        "class Normalize(nn.Module):\n",
        "\n",
        "    def __init__(self, power=2):\n",
        "        super(Normalize, self).__init__()\n",
        "        self.power = power\n",
        "\n",
        "    def forward(self, x):\n",
        "        norm = x.pow(self.power).sum(1, keepdim=True).pow(1. / self.power)\n",
        "        out = x.div(norm)\n",
        "        return out\n",
        "\n",
        "\n",
        "class BasicBlock(nn.Module):\n",
        "    expansion = 1\n",
        "\n",
        "    def __init__(self, inplanes, planes, stride=1, downsample=None):\n",
        "        super(BasicBlock, self).__init__()\n",
        "        self.conv1 = conv3x3(inplanes, planes, stride)\n",
        "        self.bn1 = nn.BatchNorm2d(planes)\n",
        "        self.relu = nn.ReLU(inplace=True)\n",
        "        self.conv2 = conv3x3(planes, planes)\n",
        "        self.bn2 = nn.BatchNorm2d(planes)\n",
        "        self.downsample = downsample\n",
        "        self.stride = stride\n",
        "\n",
        "    def forward(self, x):\n",
        "        residual = x\n",
        "\n",
        "        out = self.conv1(x)\n",
        "        out = self.bn1(out)\n",
        "        out = self.relu(out)\n",
        "\n",
        "        out = self.conv2(out)\n",
        "        out = self.bn2(out)\n",
        "\n",
        "        if self.downsample is not None:\n",
        "            residual = self.downsample(x)\n",
        "\n",
        "        out += residual\n",
        "        out = self.relu(out)\n",
        "\n",
        "        return out\n",
        "\n",
        "\n",
        "class Bottleneck(nn.Module):\n",
        "    expansion = 4\n",
        "\n",
        "    def __init__(self, inplanes, planes, stride=1, downsample=None):\n",
        "        super(Bottleneck, self).__init__()\n",
        "        self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)\n",
        "        self.bn1 = nn.BatchNorm2d(planes)\n",
        "        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride,\n",
        "                               padding=1, bias=False)\n",
        "        self.bn2 = nn.BatchNorm2d(planes)\n",
        "        self.conv3 = nn.Conv2d(planes, planes * 4, kernel_size=1, bias=False)\n",
        "        self.bn3 = nn.BatchNorm2d(planes * 4)\n",
        "        self.relu = nn.ReLU(inplace=True)\n",
        "        self.downsample = downsample\n",
        "        self.stride = stride\n",
        "\n",
        "    def forward(self, x):\n",
        "        residual = x\n",
        "\n",
        "        out = self.conv1(x)\n",
        "        out = self.bn1(out)\n",
        "        out = self.relu(out)\n",
        "\n",
        "        out = self.conv2(out)\n",
        "        out = self.bn2(out)\n",
        "        out = self.relu(out)\n",
        "\n",
        "        out = self.conv3(out)\n",
        "        out = self.bn3(out)\n",
        "\n",
        "        if self.downsample is not None:\n",
        "            residual = self.downsample(x)\n",
        "\n",
        "        out += residual\n",
        "        out = self.relu(out)\n",
        "\n",
        "        return out\n",
        "\n",
        "\n",
        "class ResNet(nn.Module):\n",
        "\n",
        "    def __init__(self, block, layers, low_dim=128, in_channel=3, width=1, n_label=10):\n",
        "        self.inplanes = 64\n",
        "        super(ResNet, self).__init__()\n",
        "        self.conv1 = nn.Conv2d(in_channel, 64, kernel_size=7, stride=2, padding=3,\n",
        "                               bias=False)\n",
        "        self.bn1 = nn.BatchNorm2d(64)\n",
        "        self.relu = nn.ReLU(inplace=True)\n",
        "\n",
        "        self.base = int(64 * width)\n",
        "\n",
        "        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)\n",
        "        self.layer1 = self._make_layer(block, self.base, layers[0])\n",
        "        self.layer2 = self._make_layer(block, self.base * 2, layers[1], stride=2)\n",
        "        self.layer3 = self._make_layer(block, self.base * 4, layers[2], stride=2)\n",
        "        self.layer4 = self._make_layer(block, self.base * 8, layers[3], stride=2)\n",
        "        self.avgpool = nn.AdaptiveAvgPool2d((1, 1)) #nn.AvgPool2d(7, stride=1)\n",
        "        self.l2norm = Normalize(2)\n",
        "        self.n_label = n_label\n",
        "\n",
        "        print(\"Resnet with Linear classifier\")\n",
        "        self.linear_K = nn.ModuleList(\n",
        "            [\n",
        "                nn.Linear(self.base * 8 * block.expansion, low_dim)\n",
        "                for  i in range(n_label)\n",
        "            ]\n",
        "        )  # create n_label mlp head\n",
        "\n",
        "\n",
        "        self.classifier = nn.Linear(self.base * 8 * block.expansion, low_dim)\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "        for m in self.modules():\n",
        "            if isinstance(m, nn.Conv2d):\n",
        "                n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels\n",
        "                m.weight.data.normal_(0, math.sqrt(2. / n))\n",
        "            elif isinstance(m, nn.BatchNorm2d):\n",
        "                m.weight.data.fill_(1)\n",
        "                m.bias.data.zero_()\n",
        "\n",
        "\n",
        "    def _make_layer(self, block, planes, blocks, stride=1):\n",
        "        downsample = None\n",
        "        if stride != 1 or self.inplanes != planes * block.expansion:\n",
        "            downsample = nn.Sequential(\n",
        "                nn.Conv2d(self.inplanes, planes * block.expansion,\n",
        "                          kernel_size=1, stride=stride, bias=False),\n",
        "                nn.BatchNorm2d(planes * block.expansion),\n",
        "            )\n",
        "\n",
        "        layers = []\n",
        "        layers.append(block(self.inplanes, planes, stride, downsample))\n",
        "        self.inplanes = planes * block.expansion\n",
        "        for i in range(1, blocks):\n",
        "            layers.append(block(self.inplanes, planes))\n",
        "\n",
        "        return nn.Sequential(*layers)\n",
        "\n",
        "    def forward(self, x, gating=False):\n",
        "        x = self.conv1(x)\n",
        "        x = self.bn1(x)\n",
        "        x = self.relu(x)\n",
        "        x = self.maxpool(x)\n",
        "        x = self.layer1(x)\n",
        "        x = self.layer2(x)\n",
        "        x = self.layer3(x)\n",
        "        x = self.layer4(x)\n",
        "        x = self.avgpool(x)\n",
        "        x = x.view(x.size(0), -1)\n",
        "\n",
        "        gating_logits = self.classifier(x)\n",
        "        gating_logits = self.l2norm(gating_logits)\n",
        "\n",
        "        x_list = [self.l2norm( self.linear_K[i](x) ) for i in range(self.n_label)]\n",
        "        if not gating:\n",
        "            return torch.stack(x_list, dim=1)\n",
        "        \n",
        "        return torch.stack(x_list, dim=1), gating_logits\n",
        "\n",
        "\n",
        "def resnet18(pretrained=False, **kwargs):\n",
        "    \"\"\"Constructs a ResNet-18 model.\n",
        "    Args:\n",
        "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
        "    \"\"\"\n",
        "    model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)\n",
        "    if pretrained:\n",
        "        model.load_state_dict(model_zoo.load_url(model_urls['resnet18']))\n",
        "    return model\n",
        "\n",
        "\n",
        "def resnet34(pretrained=False, **kwargs):\n",
        "    \"\"\"Constructs a ResNet-34 model.\n",
        "    Args:\n",
        "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
        "    \"\"\"\n",
        "    model = ResNet(BasicBlock, [3, 4, 6, 3], **kwargs)\n",
        "    if pretrained:\n",
        "        model.load_state_dict(model_zoo.load_url(model_urls['resnet34']))\n",
        "    return model\n",
        "\n",
        "\n",
        "def resnet50(pretrained=False, **kwargs):\n",
        "    \"\"\"Constructs a ResNet-50 model.\n",
        "    Args:\n",
        "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
        "    \"\"\"\n",
        "    model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)\n",
        "    if pretrained:\n",
        "        model.load_state_dict(model_zoo.load_url(model_urls['resnet50']))\n",
        "    return model\n",
        "\n",
        "\n",
        "def resnet101(pretrained=False, **kwargs):\n",
        "    \"\"\"Constructs a ResNet-101 model.\n",
        "    Args:\n",
        "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
        "    \"\"\"\n",
        "    model = ResNet(Bottleneck, [3, 4, 23, 3], **kwargs)\n",
        "    if pretrained:\n",
        "        model.load_state_dict(model_zoo.load_url(model_urls['resnet101']))\n",
        "    return model\n",
        "\n",
        "\n",
        "def resnet152(pretrained=False, **kwargs):\n",
        "    \"\"\"Constructs a ResNet-152 model.\n",
        "    Args:\n",
        "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
        "    \"\"\"\n",
        "    model = ResNet(Bottleneck, [3, 8, 36, 3], **kwargs)\n",
        "    if pretrained:\n",
        "        model.load_state_dict(model_zoo.load_url(model_urls['resnet152']))\n",
        "    return model\n",
        "\n",
        "\n",
        "class ResNet34(nn.Module):\n",
        "    \"\"\"Encoder for MiCE\"\"\"\n",
        "    def __init__(self, n_label=10):\n",
        "        super(ResNet34, self).__init__()\n",
        "        self.encoder = resnet34(n_label=n_label)\n",
        "        self.encoder = nn.DataParallel(self.encoder)\n",
        "\n",
        "    def forward(self, x, gating=False):\n",
        "        return self.encoder(x, gating=gating)"
      ],
      "metadata": {
        "id": "xvXFYNE9SjmR"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "'''ResNet in PyTorch.\n",
        "For Pre-activation ResNet, see 'preact_resnet.py'.\n",
        "Reference:\n",
        "[1] Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun\n",
        "    Deep Residual Learning for Image Recognition. arXiv:1512.03385\n",
        "'''\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "# from models.normalize import Normalize\n",
        "import math\n",
        "from torch.autograd import Variable\n",
        "\n",
        "import torch\n",
        "from torch.autograd import Variable\n",
        "from torch import nn\n",
        "\n",
        "\n",
        "class Normalize(nn.Module):\n",
        "\n",
        "    def __init__(self, power=2):\n",
        "        super(Normalize, self).__init__()\n",
        "        self.power = power\n",
        "\n",
        "    def forward(self, x):\n",
        "        norm = x.pow(self.power).sum(1, keepdim=True).pow(1. / self.power)\n",
        "        out = x.div(norm)\n",
        "        return out\n",
        "\n",
        "\n",
        "class BasicBlock(nn.Module):\n",
        "    expansion = 1\n",
        "\n",
        "    def __init__(self, in_planes, planes, stride=1):\n",
        "        super(BasicBlock, self).__init__()\n",
        "        self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)\n",
        "        self.bn1 = nn.BatchNorm2d(planes)\n",
        "        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False)\n",
        "        self.bn2 = nn.BatchNorm2d(planes)\n",
        "\n",
        "        self.shortcut = nn.Sequential()\n",
        "        if stride != 1 or in_planes != self.expansion*planes:\n",
        "            self.shortcut = nn.Sequential(\n",
        "                nn.Conv2d(in_planes, self.expansion*planes, kernel_size=1, stride=stride, bias=False),\n",
        "                nn.BatchNorm2d(self.expansion*planes)\n",
        "            )\n",
        "\n",
        "    def forward(self, x):\n",
        "        out = F.relu(self.bn1(self.conv1(x)))\n",
        "        out = self.bn2(self.conv2(out))\n",
        "        out += self.shortcut(x)\n",
        "        out = F.relu(out)\n",
        "        return out\n",
        "\n",
        "\n",
        "class Bottleneck(nn.Module):\n",
        "    expansion = 4\n",
        "\n",
        "    def __init__(self, in_planes, planes, stride=1):\n",
        "        super(Bottleneck, self).__init__()\n",
        "        self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=1, bias=False)\n",
        "        self.bn1 = nn.BatchNorm2d(planes)\n",
        "        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)\n",
        "        self.bn2 = nn.BatchNorm2d(planes)\n",
        "        self.conv3 = nn.Conv2d(planes, self.expansion*planes, kernel_size=1, bias=False)\n",
        "        self.bn3 = nn.BatchNorm2d(self.expansion*planes)\n",
        "\n",
        "        self.shortcut = nn.Sequential()\n",
        "        if stride != 1 or in_planes != self.expansion*planes:\n",
        "            self.shortcut = nn.Sequential(\n",
        "                nn.Conv2d(in_planes, self.expansion*planes, kernel_size=1, stride=stride, bias=False),\n",
        "                nn.BatchNorm2d(self.expansion*planes)\n",
        "            )\n",
        "\n",
        "    def forward(self, x):\n",
        "        out = F.relu(self.bn1(self.conv1(x)))\n",
        "        out = F.relu(self.bn2(self.conv2(out)))\n",
        "        out = self.bn3(self.conv3(out))\n",
        "        out += self.shortcut(x)\n",
        "        out = F.relu(out)\n",
        "        return out\n",
        "\n",
        "\n",
        "class ResNet(nn.Module):\n",
        "    def __init__(self, block, num_blocks, pool_len=4, low_dim=128, width=1, n_label=10):\n",
        "        super(ResNet, self).__init__()\n",
        "        self.in_planes = 64\n",
        "\n",
        "        self.conv1 = nn.Conv2d(3, 64, kernel_size=3, stride=1, padding=1, bias=False)\n",
        "        self.bn1 = nn.BatchNorm2d(64)\n",
        "\n",
        "        self.base = int(64 * width)\n",
        "        self.layer1 = self._make_layer(block, self.base, num_blocks[0], stride=1)\n",
        "        self.layer2 = self._make_layer(block, self.base * 2, num_blocks[1], stride=2)\n",
        "        self.layer3 = self._make_layer(block, self.base * 4, num_blocks[2], stride=2)\n",
        "        self.layer4 = self._make_layer(block, self.base * 8, num_blocks[3], stride=2)\n",
        "\n",
        "        print(\"ResNet MOE with Linear classifier \")\n",
        "        self.linear_K = nn.ModuleList(\n",
        "            [\n",
        "                nn.Linear(self.base * 8 * block.expansion, low_dim)\n",
        "                for  i in range(n_label)\n",
        "            ]\n",
        "        )\n",
        "\n",
        "        self.n_label = n_label\n",
        "        self.classifier = nn.Linear(self.base * 8 * block.expansion, low_dim)\n",
        "\n",
        "\n",
        "        self.softmax = nn.Softmax(dim=1)\n",
        "\n",
        "        self.l2norm = Normalize(2)\n",
        "        self.pool_len = pool_len\n",
        "\n",
        "        for m in self.modules():\n",
        "            if isinstance(m, nn.Conv2d):\n",
        "                n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels\n",
        "                m.weight.data.normal_(0, math.sqrt(2. / n))\n",
        "            elif isinstance(m, nn.BatchNorm2d):\n",
        "                m.weight.data.fill_(1)\n",
        "                m.bias.data.zero_()\n",
        "\n",
        "\n",
        "\n",
        "    def _make_layer(self, block, planes, num_blocks, stride):\n",
        "        strides = [stride] + [1]*(num_blocks-1)\n",
        "        layers = []\n",
        "        for stride in strides:\n",
        "            layers.append(block(self.in_planes, planes, stride))\n",
        "            self.in_planes = planes * block.expansion\n",
        "        return nn.Sequential(*layers)\n",
        "\n",
        "    def forward(self, x, gating=False):\n",
        "        out = F.relu(self.bn1(self.conv1(x)))\n",
        "        out = self.layer1(out)\n",
        "        out = self.layer2(out)\n",
        "        out = self.layer3(out)\n",
        "        out = self.layer4(out)\n",
        "        out = F.avg_pool2d(out, self.pool_len)\n",
        "        x = out.view(out.size(0), -1)\n",
        "\n",
        "        # if layer == 6:\n",
        "        #     return x\n",
        "\n",
        "        gating_logits = self.classifier(x)\n",
        "        gating_logits = self.l2norm(gating_logits)\n",
        "\n",
        "        # x = self.mlp(x)\n",
        "        # x = self.l2norm(x)\n",
        "\n",
        "        x_list = [self.l2norm( self.linear_K[i](x) ) for i in range(self.n_label)]\n",
        "        if not gating:\n",
        "            return torch.stack(x_list, dim=1)\n",
        "\n",
        "        return torch.stack(x_list, dim=1), gating_logits\n",
        "\n",
        "\n",
        "def ResNet18(pool_len = 4, low_dim=128):\n",
        "    return ResNet(BasicBlock, [2,2,2,2], pool_len, low_dim)\n",
        "\n",
        "def ResNet34(pool_len = 4, low_dim=128, n_label=10):\n",
        "    return ResNet(BasicBlock, [3,4,6,3], pool_len, low_dim, width=1, n_label=n_label)\n",
        "\n",
        "def ResNet50(pool_len = 4, low_dim=128, **kwargs):\n",
        "    return ResNet(Bottleneck, [3,4,6,3], pool_len, low_dim, **kwargs)\n",
        "\n",
        "def ResNet101(pool_len = 4, low_dim=128):\n",
        "    return ResNet(Bottleneck, [3,4,23,3], pool_len, low_dim)\n",
        "\n",
        "def ResNet152(pool_len = 4, low_dim=128):\n",
        "    return ResNet(Bottleneck, [3,8,36,3], pool_len, low_dim)\n",
        "\n",
        "\n",
        "class ResNet34_cifar(nn.Module):\n",
        "    \"\"\"Encoder for MiCE\"\"\"\n",
        "    def __init__(self, width=1, low_dim=128, n_label=10):\n",
        "        super(ResNet34_cifar, self).__init__()\n",
        "        self.encoder = ResNet34(low_dim=low_dim, n_label=n_label)\n",
        "        self.encoder = nn.DataParallel(self.encoder)\n",
        "\n",
        "    def forward(self, x, gating=False):\n",
        "        return self.encoder(x, gating=gating)"
      ],
      "metadata": {
        "id": "4qLYr0TaSlE3"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import torch\n",
        "from torch import nn\n",
        "import math\n",
        "import numpy as np\n",
        "import torch.nn.functional as F\n",
        "\n",
        "class MiCE_ELBO(nn.Module):\n",
        "    def __init__(self, inputSize, outputSize, nu, tau=1.0, n_class=10):\n",
        "        super(MiCE_ELBO, self).__init__()\n",
        "\n",
        "        print(\"--------------------Initializing -----------------------------------\")\n",
        "        self.outputSize = outputSize\n",
        "        self.inputSize = inputSize\n",
        "        self.queueSize = nu\n",
        "        self.tau = tau\n",
        "        self.kappa = self.tau\n",
        "        self.index = 0\n",
        "        self.n_class = n_class\n",
        "\n",
        "        stdv = 1. / math.sqrt(inputSize / 3)\n",
        "        self.register_buffer('queue', torch.rand(self.queueSize, self.n_class, inputSize).mul_(2 * stdv).add_(-stdv))\n",
        "        print('using queue shape: ({},{},{})'.format(self.queueSize, self.n_class, inputSize))\n",
        "\n",
        "        if n_class==10:\n",
        "            cluster_file = np.loadtxt(\"/content/drive/MyDrive/project/DL/kernel_paras/meanvar1_featuredim128_class10.mat\")\n",
        "        elif n_class == 20:\n",
        "            cluster_file = np.loadtxt(\"/content/drive/MyDrive/project/DL/kernel_paras/meanvar1_featuredim128_class20.mat\")\n",
        "        elif n_class ==15:\n",
        "            cluster_file = np.loadtxt(\"/content/drive/MyDrive/project/DL/kernel_paras/meanvar1_featuredim128_class15.mat\")\n",
        "\n",
        "        self.register_buffer('gating_prototype', torch.from_numpy(cluster_file).type(torch.FloatTensor))\n",
        "        self.expert_prototype = nn.Parameter(torch.Tensor(self.n_class, inputSize))\n",
        "\n",
        "        self.logSoftmax = torch.nn.LogSoftmax(dim=1)\n",
        "        self.logSoftmax2 = torch.nn.LogSoftmax(dim=2)\n",
        "        print(\"--------------------Initialization Ends-----------------------------------\")\n",
        "\n",
        "    def update_cluster(self, new_center):\n",
        "\n",
        "        with torch.no_grad():\n",
        "            new_center = F.normalize(new_center, dim=1)\n",
        "            out_ids = torch.arange(self.n_class).cuda()\n",
        "            out_ids = out_ids.long()  # BS x 1\n",
        "            self.expert_prototype.index_copy_(0, out_ids, new_center) \n",
        "\n",
        "    def forward(self, f, v, g,):\n",
        "        batchSize = f.shape[0] \n",
        "        v = v.detach()\n",
        "\n",
        "        pi_logit_student = torch.div(torch.einsum('kd,bd->bk', [self.gating_prototype.detach().clone(), g]),\n",
        "                                         self.kappa)  # K x D  vs B x D ---> BS x K\n",
        "        log_pi = self.logSoftmax(pi_logit_student + 1e-18)\n",
        "        pi = torch.exp(log_pi)  # p(z | x)\n",
        "\n",
        "        # positive sample\n",
        "        v_f = torch.einsum(\"bkd,bkd->bk\", [f, v]).unsqueeze(-1) # BS x K x D --> BS x K x 1\n",
        "        v_mu = torch.einsum(\"bkd,kd->bk\", [v, F.normalize(self.expert_prototype, dim=1)]).unsqueeze(-1)  # BS x K x 1\n",
        "\n",
        "        l_pos = (v_f + v_mu)  # BS x K x 1\n",
        "        l_pos = torch.einsum('bki->kbi', [l_pos]) # BS x K x 1 --> K x BS x 1\n",
        "\n",
        "        # Negative sample\n",
        "        queue = self.queue.detach().clone()  # nu x D x K\n",
        "        queue_f = torch.einsum('vkd,bkd->kbv', [queue, f])  # K x BS x nu\n",
        "        queue_mu = torch.einsum('vkd,kd->kv', [queue, F.normalize(self.expert_prototype, dim=1)]).unsqueeze(1)  # K x 1 x nu\n",
        "        del queue\n",
        "\n",
        "        l_neg = queue_f + queue_mu  # K x BS x nu\n",
        "        out = torch.div(torch.cat([l_pos, l_neg], dim=2), self.tau) # K x BS x (nu + 1)\n",
        "        del l_pos, l_neg\n",
        "\n",
        "        log_phi = self.logSoftmax2(out + 1e-18)\n",
        "        normalized_phi = torch.exp(log_phi) # p(y | x, z)\n",
        "\n",
        "        log_phi_pos = log_phi[:, :, 0]\n",
        "        normalized_phi_pos = normalized_phi[:, :, 0]\n",
        "        del normalized_phi\n",
        "\n",
        "        normalized_phi_pos = normalized_phi_pos.transpose(1, 0) * pi                # BS x K\n",
        "        posterior = torch.div(normalized_phi_pos, normalized_phi_pos.sum(1).view(-1, 1)).squeeze().contiguous()  # BS x K: posterior -> each row = p(z | v_i, x_i)  # BS x Classes, probability\n",
        "        log_phi_pos = log_phi_pos.transpose(1, 0).squeeze().contiguous() # K x BS -> BS x K\n",
        "        loss = -torch.sum(posterior * (log_pi + log_phi_pos - torch.log(posterior + 1e-18))) / float(batchSize)\n",
        "\n",
        "        # update queue using EMA predictions\n",
        "        with torch.no_grad():\n",
        "            out_ids = torch.arange(batchSize).cuda()\n",
        "            out_ids += self.index\n",
        "            out_ids = torch.fmod(out_ids, self.queueSize)\n",
        "            out_ids = out_ids.long() \n",
        "\n",
        "            self.queue.index_copy_(0, out_ids, v)\n",
        "            self.index = (self.index + batchSize) % self.queueSize  \n",
        "        return loss, loss, posterior, log_pi "
      ],
      "metadata": {
        "id": "W3G9mlURStWs"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from __future__ import print_function\n",
        "import torch\n",
        "import torchvision.datasets as datasets\n",
        "from PIL import Image\n",
        "import numpy as np\n",
        "\n",
        "\n",
        "class ThreeCropsTransform:\n",
        "    \"\"\"Take two random crops of one image as the query and key.\"\"\"\n",
        "\n",
        "    def __init__(self, base_transform):\n",
        "        self.base_transform = base_transform\n",
        "    def __call__(self, x):\n",
        "        x1 = self.base_transform(x)\n",
        "        x2 = self.base_transform(x)\n",
        "        x3 = self.base_transform(x)\n",
        "        return [x1, x2, x3]\n",
        "\n",
        "\n",
        "def get_dataset_stat(dataset):\n",
        "\n",
        "    if dataset == 'cifar10':\n",
        "        image_size = 32\n",
        "        mean = [0.4914, 0.4822, 0.4465]\n",
        "        std = [0.2470, 0.2435, 0.2616]\n",
        "        n_class = 10\n",
        "    elif dataset == 'cifar100' or dataset == 'cifar20':\n",
        "        image_size = 32\n",
        "        mean = [0.5071, 0.4867, 0.4408]\n",
        "        std = [0.2675, 0.2565, 0.2761]\n",
        "        if dataset == 'cifar100':\n",
        "            n_class = 100\n",
        "        else:\n",
        "            n_class = 20\n",
        "    elif dataset == 'stl10':\n",
        "        image_size = 96\n",
        "        mean = [0.4409, 0.4279, 0.3868]\n",
        "        std = [0.2683, 0.2610, 0.2687]\n",
        "        n_class = 10\n",
        "\n",
        "    return image_size, mean, std, n_class\n",
        "\n",
        "\n",
        "def create_dataset(dataset, train_transform, test_transform):\n",
        "    print(\"Create dataset with tripple transform\")\n",
        "    train_transform = ThreeCropsTransform(train_transform)\n",
        "    test_transform = ThreeCropsTransform(test_transform)\n",
        "\n",
        "    if dataset == 'cifar10':\n",
        "        train_dataset = datasets.CIFAR10(root='./data', train=True, download=True, transform=train_transform,)\n",
        "        test_dataset = datasets.CIFAR10(root='./data', train=False, download=True, transform=test_transform,)\n",
        "\n",
        "    elif dataset == 'cifar100':\n",
        "        train_dataset = datasets.CIFAR100(root='./data', train=True, download=True, transform=train_transform, )\n",
        "        test_dataset = datasets.CIFAR100(root='./data', train=False, download=True, transform=test_transform, )\n",
        "\n",
        "    elif dataset == 'cifar20':\n",
        "        train_dataset = datasets.CIFAR100(root='./data', train=True, download=True, transform=train_transform, target_transform=_cifar100_to_cifar20)\n",
        "        test_dataset = datasets.CIFAR100(root='./data', train=False, download=True, transform=test_transform, target_transform=_cifar100_to_cifar20)\n",
        "\n",
        "    elif dataset == 'stl10':\n",
        "        train_dataset = datasets.STL10(root='./data', split='train', download=True, transform=train_transform, )\n",
        "        test_dataset = datasets.STL10(root='./data', split='test', download=True, transform=test_transform, )\n",
        "\n",
        "    return train_dataset, test_dataset\n",
        "\n",
        "\n",
        "class ImageFolderTripple(datasets.ImageFolder):\n",
        "    \"\"\"Folder datasets which returns the index of the image as well\n",
        "    \"\"\"\n",
        "    def __init__(self, root, transform=None, target_transform=None, two_crop=False):\n",
        "        super(ImageFolderTripple, self).__init__(root, transform, target_transform)\n",
        "        self.two_crop = two_crop\n",
        "\n",
        "    def __getitem__(self, index):\n",
        "        \"\"\"\n",
        "        Args:\n",
        "            index (int): Index\n",
        "        Returns:\n",
        "            tuple: (image, target, index) where target is class_index of the target class.\n",
        "        \"\"\"\n",
        "        path, target = self.imgs[index]\n",
        "        image = self.loader(path)\n",
        "        if self.transform is not None:\n",
        "            img = self.transform(image)\n",
        "        if self.target_transform is not None:\n",
        "            target = self.target_transform(target)\n",
        "\n",
        "        img2 = self.transform(image)\n",
        "        img3 = self.transform(image)\n",
        "\n",
        "        return [img, img2, img3], target\n",
        "\n",
        "\n",
        "def _cifar100_to_cifar20(target):\n",
        "    _dict = {\n",
        "         0: 4,\n",
        "         1: 1,\n",
        "         2: 14,\n",
        "         3: 8,\n",
        "         4: 0,\n",
        "         5: 6,\n",
        "         6: 7,\n",
        "         7: 7,\n",
        "         8: 18,\n",
        "         9: 3,\n",
        "         10: 3,\n",
        "         11: 14,\n",
        "         12: 9,\n",
        "         13: 18,\n",
        "         14: 7,\n",
        "         15: 11,\n",
        "         16: 3,\n",
        "         17: 9,\n",
        "         18: 7,\n",
        "         19: 11,\n",
        "         20: 6,\n",
        "         21: 11,\n",
        "         22: 5,\n",
        "         23: 10,\n",
        "         24: 7,\n",
        "         25: 6,\n",
        "         26: 13,\n",
        "         27: 15,\n",
        "         28: 3,\n",
        "         29: 15,\n",
        "         30: 0,\n",
        "         31: 11,\n",
        "         32: 1,\n",
        "         33: 10,\n",
        "         34: 12,\n",
        "         35: 14,\n",
        "         36: 16,\n",
        "         37: 9,\n",
        "         38: 11,\n",
        "         39: 5,\n",
        "         40: 5,\n",
        "         41: 19,\n",
        "         42: 8,\n",
        "         43: 8,\n",
        "         44: 15,\n",
        "         45: 13,\n",
        "         46: 14,\n",
        "         47: 17,\n",
        "         48: 18,\n",
        "         49: 10,\n",
        "         50: 16,\n",
        "         51: 4,\n",
        "         52: 17,\n",
        "         53: 4,\n",
        "         54: 2,\n",
        "         55: 0,\n",
        "         56: 17,\n",
        "         57: 4,\n",
        "         58: 18,\n",
        "         59: 17,\n",
        "         60: 10,\n",
        "         61: 3,\n",
        "         62: 2,\n",
        "         63: 12,\n",
        "         64: 12,\n",
        "         65: 16,\n",
        "         66: 12,\n",
        "         67: 1,\n",
        "         68: 9,\n",
        "         69: 19,\n",
        "         70: 2,\n",
        "         71: 10,\n",
        "         72: 0,\n",
        "         73: 1,\n",
        "         74: 16,\n",
        "         75: 12,\n",
        "         76: 9,\n",
        "         77: 13,\n",
        "         78: 15,\n",
        "         79: 13,\n",
        "         80: 16,\n",
        "         81: 19,\n",
        "         82: 2,\n",
        "         83: 4,\n",
        "         84: 6,\n",
        "         85: 19,\n",
        "         86: 5,\n",
        "         87: 5,\n",
        "         88: 8,\n",
        "         89: 19,\n",
        "         90: 18,\n",
        "         91: 1,\n",
        "         92: 2,\n",
        "         93: 15,\n",
        "         94: 6,\n",
        "         95: 0,\n",
        "         96: 17,\n",
        "         97: 8,\n",
        "         98: 14,\n",
        "         99: 13\n",
        "    }\n",
        "\n",
        "    return _dict[target]"
      ],
      "metadata": {
        "id": "nmao7cDWSzov"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from __future__ import print_function\n",
        "import torch\n",
        "from sklearn.metrics import confusion_matrix\n",
        "#from models.resnet_MiCE import ResNet34\n",
        "#from models.resnet_cifar_MiCE import ResNet34_cifar\n",
        "import numpy as np\n",
        "\n",
        "from sklearn.metrics import normalized_mutual_info_score, adjusted_rand_score\n",
        "!pip install scikit-learn==0.22.2\n",
        "from sklearn.utils.linear_assignment_ import linear_assignment\n",
        "#from scipy.optimize import linear_sum_assignment as linear_assignment #https://stackoverflow.com/questions/62390517/no-module-named-sklearn-utils-linear-assignment\n",
        "\n",
        "nmi = normalized_mutual_info_score\n",
        "ari = adjusted_rand_score\n",
        "\n",
        "\n",
        "def create_model(model_name, n_label=10):\n",
        "    if model_name == 'resnet34':\n",
        "        model = ResNet34(n_label=n_label)\n",
        "        model_ema = ResNet34(n_label=n_label)\n",
        "    elif model_name == 'resnet34_cifar':\n",
        "        model = ResNet34_cifar(n_label=n_label)\n",
        "        model_ema = ResNet34_cifar(n_label=n_label)\n",
        "    else:\n",
        "        raise NotImplementedError('model not supported {}'.format(model_name))\n",
        "\n",
        "    return model, model_ema\n",
        "\n",
        "\n",
        "def adjust_learning_rate(epoch, opt, optimizer):\n",
        "    steps = np.sum(epoch > np.asarray(opt.lr_decay_epochs))\n",
        "    if epoch <= opt.lr_warmup:  # warm up epoch = 10\n",
        "        print(\"Linear warm up\")\n",
        "        decay_weight = linear_rampup(epoch, opt.lr_warmup)\n",
        "\n",
        "        for param_group in optimizer.param_groups:\n",
        "            param_group['lr'] =  opt.learning_rate * decay_weight\n",
        "        return\n",
        "\n",
        "    if steps > 0:\n",
        "        decay_weight = (opt.lr_decay_rate ** steps)\n",
        "        for param_group in optimizer.param_groups:\n",
        "            param_group['lr'] =  opt.learning_rate * decay_weight\n",
        "\n",
        "\n",
        "def linear_rampup(current, rampup_length):\n",
        "    \"\"\"Linear rampup\"\"\"\n",
        "    assert current >= 0 and rampup_length >= 0\n",
        "    if current >= rampup_length:\n",
        "        return 1.0\n",
        "    else:\n",
        "        return current / rampup_length\n",
        "\n",
        "\n",
        "class AverageMeter(object):\n",
        "    \"\"\"Computes and stores the average and current value\"\"\"\n",
        "    def __init__(self):\n",
        "        self.val = 0\n",
        "        self.avg = 0\n",
        "        self.sum = 0\n",
        "        self.count = 0\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 = self.sum / self.count\n",
        "\n",
        "\n",
        "def accuracy(output, target, topk=(1,)):\n",
        "    \"\"\"Computes the accuracy over the k top predictions for the specified values of k\"\"\"\n",
        "    with torch.no_grad():\n",
        "        maxk = max(topk)\n",
        "        batch_size = target.size(0)\n",
        "\n",
        "        _, pred = output.topk(maxk, 1, True, True)\n",
        "        pred = pred.t()\n",
        "        correct = pred.eq(target.view(1, -1).expand_as(pred))\n",
        "\n",
        "        res = []\n",
        "        for k in topk:\n",
        "            correct_k = correct[:k].view(-1).float().sum(0, keepdim=True)\n",
        "            res.append(correct_k.mul_(100.0 / batch_size))\n",
        "        return res\n",
        "\n",
        "\n",
        "def moment_update(model, model_ema, m):\n",
        "    \"\"\" model_ema = m * model_ema + (1 - m) model \"\"\"\n",
        "    for p1, p2 in zip(model.parameters(), model_ema.parameters()):\n",
        "        p2.data.mul_(m).add_(1-m, p1.detach().data)\n",
        "\n",
        "\n",
        "def get_shuffle_ids(bsz):\n",
        "    \"\"\"generate shuffle ids for ShuffleBN\"\"\"\n",
        "    forward_inds = torch.randperm(bsz).long().cuda()\n",
        "    backward_inds = torch.zeros(bsz).long().cuda()\n",
        "    value = torch.arange(bsz).long().cuda()\n",
        "    backward_inds.index_copy_(0, forward_inds, value)\n",
        "    return forward_inds, backward_inds\n",
        "\n",
        "\n",
        "def set_bn_train(m):\n",
        "    classname = m.__class__.__name__\n",
        "    if classname.find('BatchNorm') != -1:\n",
        "        m.train()\n",
        "\n",
        "def _make_cost_m(cm):\n",
        "    s = np.max(cm)\n",
        "    return (- cm + s)\n",
        "!pip install scikit-learn==0.22.2\n",
        "import seaborn as sns; sns.set()\n",
        "def acc(y_true, y_pred):\n",
        "    \"\"\"\n",
        "    https://github.com/XifengGuo/DEC-keras/blob/master/metrics.py\n",
        "    Calculate clustering accuracy. Require scikit-learn installed\n",
        "    # Arguments\n",
        "        y: true labels, numpy.array with shape `(n_samples,)`\n",
        "        y_pred: predicted labels, numpy.array with shape `(n_samples,)`\n",
        "    # Return\n",
        "        accuracy, in [0,1]\n",
        "    \"\"\"\n",
        "    y_true = y_true.astype(np.int64)\n",
        "    assert y_pred.size == y_true.size\n",
        "    D = max(y_pred.max(), y_true.max()) + 1\n",
        "    w = np.zeros((D, D), dtype=np.int64)\n",
        "    for i in range(y_pred.size):\n",
        "        w[y_pred[i], y_true[i]] += 1\n",
        "    ind = linear_assignment(w.max() - w)\n",
        "\n",
        "    cm = confusion_matrix(y_true, y_pred)\n",
        "    indexes = linear_assignment(_make_cost_m(cm))\n",
        "    js = [e[1] for e in sorted(indexes, key=lambda x: x[0])]\n",
        "    cm2 = cm[:, js]\n",
        "    fig, ax = plt.subplots(figsize=(8,8)) \n",
        "    ax = sns.heatmap(cm2,linewidths=.5, annot=True, fmt=\"d\", cmap=\"Blues\")\n",
        "    #ax = sns.heatmap(cm2, annot=True, fmt=\"d\", cmap=\"Blues\")\n",
        "\n",
        "    return sum([w[i, j] for i, j in ind]) * 1.0 / y_pred.size , cm2\n",
        "\n",
        "\n",
        "if __name__ == '__main__':\n",
        "    print()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 506
        },
        "id": "9XN3D6S5S5Ek",
        "outputId": "cf774c70-1e34-4a55-f7da-432a27b8370e"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting scikit-learn==0.22.2\n",
            "  Downloading scikit_learn-0.22.2-cp37-cp37m-manylinux1_x86_64.whl (7.1 MB)\n",
            "\u001b[K     |████████████████████████████████| 7.1 MB 4.5 MB/s \n",
            "\u001b[?25hRequirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.22.2) (1.1.0)\n",
            "Requirement already satisfied: scipy>=0.17.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.22.2) (1.4.1)\n",
            "Requirement already satisfied: numpy>=1.11.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.22.2) (1.21.6)\n",
            "Installing collected packages: scikit-learn\n",
            "  Attempting uninstall: scikit-learn\n",
            "    Found existing installation: scikit-learn 1.0.2\n",
            "    Uninstalling scikit-learn-1.0.2:\n",
            "      Successfully uninstalled scikit-learn-1.0.2\n",
            "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
            "yellowbrick 1.4 requires scikit-learn>=1.0.0, but you have scikit-learn 0.22.2 which is incompatible.\n",
            "imbalanced-learn 0.8.1 requires scikit-learn>=0.24, but you have scikit-learn 0.22.2 which is incompatible.\u001b[0m\n",
            "Successfully installed scikit-learn-0.22.2\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.colab-display-data+json": {
              "pip_warning": {
                "packages": [
                  "sklearn"
                ]
              }
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/utils/linear_assignment_.py:22: FutureWarning: The linear_assignment_ module is deprecated in 0.21 and will be removed from 0.23. Use scipy.optimize.linear_sum_assignment instead.\n",
            "  FutureWarning)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: scikit-learn==0.22.2 in /usr/local/lib/python3.7/dist-packages (0.22.2)\n",
            "Requirement already satisfied: scipy>=0.17.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.22.2) (1.4.1)\n",
            "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.22.2) (1.1.0)\n",
            "Requirement already satisfied: numpy>=1.11.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.22.2) (1.21.6)\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# from sklearn.metrics import confusion_matrix\n",
        "# from scipy.optimize import linear_sum_assignment as linear_assignment\n",
        "# #from sklearn.utils.linear_assignment_ import linear_assignment\n",
        "# def acc(true_row_labels, predicted_row_labels):\n",
        "#     \"\"\"Get the best accuracy.\n",
        "\n",
        "#     Parameters\n",
        "#     ----------\n",
        "#     true_row_labels: array-like\n",
        "#         The true row labels, given as external information\n",
        "#     predicted_row_labels: array-like\n",
        "#         The row labels predicted by the model\n",
        "\n",
        "#     Returns\n",
        "#     -------\n",
        "#     float\n",
        "#         Best value of accuracy\n",
        "#     \"\"\"\n",
        "\n",
        "#     cm = confusion_matrix(true_row_labels, predicted_row_labels)\n",
        "#     indexes = linear_assignment(_make_cost_m(cm))\n",
        "#     total = 0\n",
        "#     for row, column in indexes:\n",
        "#         value = cm[row][column]\n",
        "#         total += value\n",
        "\n",
        "#     return (total * 1. / np.sum(cm))"
      ],
      "metadata": {
        "id": "F7g5FXcAJv4J"
      },
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install tensorboard_logger"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TRoG3jgXTkJ7",
        "outputId": "55fd13bd-b94a-49b5-de0e-9610ead44a38"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting tensorboard_logger\n",
            "  Downloading tensorboard_logger-0.1.0-py2.py3-none-any.whl (17 kB)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from tensorboard_logger) (1.21.6)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from tensorboard_logger) (1.15.0)\n",
            "Requirement already satisfied: pillow>=4.1.1 in /usr/local/lib/python3.7/dist-packages (from tensorboard_logger) (7.1.2)\n",
            "Requirement already satisfied: protobuf in /usr/local/lib/python3.7/dist-packages (from tensorboard_logger) (3.17.3)\n",
            "Requirement already satisfied: scipy>=0.19.1 in /usr/local/lib/python3.7/dist-packages (from tensorboard_logger) (1.4.1)\n",
            "Installing collected packages: tensorboard-logger\n",
            "Successfully installed tensorboard-logger-0.1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import sys\n",
        "sys.argv=['']\n",
        "del sys"
      ],
      "metadata": {
        "id": "XmTlNCzcTxko"
      },
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from __future__ import print_function\n",
        "import os\n",
        "import sys\n",
        "import time\n",
        "import torch\n",
        "import torch.backends.cudnn as cudnn\n",
        "import argparse\n",
        "import numpy as np\n",
        "import tensorboard_logger as tb_logger\n",
        "from torchvision import transforms, datasets\n",
        "#from util import *\n",
        "#from ELBO import MiCE_ELBO\n",
        "#from dataset import get_dataset_stat, create_dataset\n",
        "from torch.utils.data import ConcatDataset\n",
        "\n",
        "\n",
        "def parse_option():\n",
        "    parser = argparse.ArgumentParser('argument for training')\n",
        "\n",
        "    parser.add_argument('--print_freq', type=int, default=50, help='print frequency')\n",
        "    parser.add_argument('--save_freq', type=int, default=100, help='save frequency')\n",
        "    parser.add_argument('--batch_size', type=int, default=256, help='batch_size')\n",
        "    parser.add_argument('--num_workers', type=int, default=32, help='num of workers to use')\n",
        "    parser.add_argument('--epochs', type=int, default=100, help='number of training epochs')\n",
        "\n",
        "    # optimization\n",
        "    parser.add_argument('--learning_rate', type=float, default=1.0, help='learning rate')\n",
        "    parser.add_argument('--lr_decay_epochs', type=str, default='480,640,800', help='where to decay lr, can be a list')\n",
        "    parser.add_argument('--lr_decay_rate', type=float, default=0.1, help='decay rate for learning rate')\n",
        "    parser.add_argument('--weight_decay', type=float, default=1e-4, help='weight decay')\n",
        "    parser.add_argument('--momentum', type=float, default=0.9, help='momentum')\n",
        "\n",
        "    # dataset\n",
        "    parser.add_argument('--dataset', type=str, default='cifar10', choices=['cifar10', 'cifar20', 'stl10'])\n",
        "\n",
        "    # resume\n",
        "    parser.add_argument('--resume', default='/content/drive/MyDrive/project/DL/cifar10_nu16384_resnet34_cifar_lr1/ckpt_epoch_0.pth', type=str, metavar='PATH',\n",
        "                        help='path to latest checkpoint (default: none)')\n",
        "\n",
        "    # model definition\n",
        "    parser.add_argument('--model', type=str, default='resnet34_cifar', choices=['resnet34', 'resnet34_cifar'])\n",
        "    parser.add_argument('--low_dim', type=int, default=128, help='Dimension of each embedding')\n",
        "\n",
        "    # loss function\n",
        "    parser.add_argument('--nu', type=int, default=16384, help='Dimension of the queue')\n",
        "    parser.add_argument('--tau', type=float, default=1.0, help='temperature')\n",
        "\n",
        "    # EMA setting\n",
        "    parser.add_argument('--m', type=float, default=0.999, help='exponential moving average weight')\n",
        "\n",
        "    # GPU setting\n",
        "    parser.add_argument('--gpu', default=None, type=int, help='GPU id to use.')\n",
        "\n",
        "    parser.add_argument('--lr_warmup', type=float, default=10, help='Linear warm-up cycle')\n",
        "    parser.add_argument('--suffix', type=str, default=None, help='suffix to add on the model name')\n",
        "    parser.add_argument('--data_folder', type=str, default=\"./\", help='path to data')\n",
        "    parser.add_argument('--model_path', type=str, default=\"/content/drive/MyDrive/project/DL\", help='path to save model')\n",
        "    parser.add_argument('--tb_path', type=str, default=\"./tensorboard\", help='path to tensorboard')\n",
        "\n",
        "    opt = parser.parse_args()\n",
        "\n",
        "    iterations = opt.lr_decay_epochs.split(',')\n",
        "    opt.lr_decay_epochs = list([])\n",
        "    for it in iterations:\n",
        "        opt.lr_decay_epochs.append(int(it))\n",
        "\n",
        "    opt.model_name = '{}_nu{}_{}_lr{}_bsz{}_epoch{}_tau{}'.format(opt.dataset, opt.nu, opt.model,\n",
        "                                                        opt.learning_rate,opt.batch_size, opt.epochs, opt.tau)\n",
        "    opt.model_name += str(opt.suffix)\n",
        "\n",
        "    opt.model_folder = os.path.join(opt.model_path, opt.model_name)\n",
        "    if not os.path.isdir(opt.model_folder):\n",
        "        os.makedirs(opt.model_folder)\n",
        "\n",
        "    opt.tb_folder = os.path.join(opt.tb_path, opt.model_name)\n",
        "    if not os.path.isdir(opt.tb_folder):\n",
        "        os.makedirs(opt.tb_folder)\n",
        "\n",
        "    return opt\n",
        "\n",
        "def main():\n",
        "\n",
        "    args = parse_option()\n",
        "    if args.gpu is not None:\n",
        "        print(\"Use GPU: {} for training\".formfat(args.gpu))\n",
        "    logger = tb_logger.Logger(logdir=args.tb_folder, flush_secs=5)\n",
        "\n",
        "    #----------------------------------------------prepare dataset--------------------------------------------------------\n",
        "    image_size, mean, std, n_class = get_dataset_stat(args.dataset)\n",
        "    args.n_class = n_class\n",
        "    normalize = transforms.Normalize(mean=mean, std=std)\n",
        "\n",
        "    train_transform = transforms.Compose([\n",
        "        transforms.RandomResizedCrop(image_size, scale=(0.2, 1.)),\n",
        "        transforms.RandomGrayscale(p=0.2),\n",
        "        transforms.ColorJitter(0.4, 0.4, 0.4, 0.4),\n",
        "        transforms.RandomHorizontalFlip(),\n",
        "        transforms.ToTensor(),\n",
        "        normalize\n",
        "    ])\n",
        "\n",
        "    train_dataset, test_dataset = create_dataset(args.dataset, train_transform, train_transform)\n",
        "    full_dataset = ConcatDataset([train_dataset, test_dataset])\n",
        "    full_loader = torch.utils.data.DataLoader(full_dataset,\n",
        "                                              batch_size=args.batch_size,\n",
        "                                              shuffle=True,\n",
        "                                              num_workers=args.num_workers,\n",
        "                                              pin_memory=True,\n",
        "                                              drop_last=True)\n",
        "    n_full_data = len(full_dataset)\n",
        "\n",
        "\n",
        "    #----------------------------------------------prepare model/loss/optimizer--------------------------------------------------------\n",
        "    model, model_ema = create_model(args.model, args.n_class)\n",
        "    moment_update(model, model_ema, 0)\n",
        "\n",
        "    elbo = MiCE_ELBO(args.low_dim, n_full_data, args.nu, tau=args.tau, n_class=args.n_class).cuda(args.gpu)\n",
        "\n",
        "    model = model.cuda()\n",
        "    model_ema = model_ema.cuda()\n",
        "    weight_decay = args.weight_decay\n",
        "\n",
        "    parameters = [{'params': model.parameters(), 'weight_decay': weight_decay},\n",
        "                  {'params': elbo.parameters(), 'weight_decay': weight_decay}]\n",
        "    print(\"Update elbo parameters:\", list(parameters[1]['params']), parameters[1]['weight_decay'],)\n",
        "    optimizer = torch.optim.SGD(parameters,\n",
        "                                lr=args.learning_rate,\n",
        "                                momentum=args.momentum,\n",
        "                                weight_decay=0.0)\n",
        "    cudnn.benchmark = True\n",
        "\n",
        "    # optionally resume from a checkpoint\n",
        "    args.start_epoch = 0\n",
        "    if args.resume:\n",
        "        if os.path.isfile(args.resume):\n",
        "            print(\"=> loading checkpoint '{}'\".format(args.resume))\n",
        "            checkpoint = torch.load(args.resume, map_location='cpu')\n",
        "\n",
        "            args.start_epoch = checkpoint['epoch'] + 1\n",
        "            model.load_state_dict(checkpoint['model'])\n",
        "\n",
        "            elbo.load_state_dict(checkpoint['elbo'])\n",
        "            optimizer.load_state_dict(checkpoint['optimizer'])\n",
        "\n",
        "            for param_group in optimizer.param_groups:\n",
        "                print(param_group['lr'], param_group['weight_decay'])\n",
        "\n",
        "            model_ema.load_state_dict(checkpoint['model_ema'])\n",
        "            print(\"=> loaded successfully '{}' (epoch {})\"\n",
        "                  .format(args.resume, checkpoint['epoch']))\n",
        "            del checkpoint\n",
        "            torch.cuda.empty_cache()\n",
        "        else:\n",
        "            print(\"=> no checkpoint found at '{}'\".format(args.resume))\n",
        "\n",
        "    #----------------------------------------------start training-------------------------------------------------------\n",
        "    for epoch in range(args.start_epoch, args.epochs + 1):\n",
        "        adjust_learning_rate(epoch, args, optimizer)\n",
        "        print(\"==> training...\")\n",
        "\n",
        "        time1 = time.time()\n",
        "        log_dict = train_MiCE(epoch, full_loader, model, model_ema, elbo, optimizer, args)\n",
        "        time2 = time.time()\n",
        "        print('epoch {}, total time {:.2f}'.format(epoch, time2 - time1))\n",
        "\n",
        "        # tensorboard logger\n",
        "        for key, val in log_dict.items():\n",
        "            logger.log_value(key, val, epoch)\n",
        "        logger.log_value('learning_rate', optimizer.param_groups[0]['lr'], epoch)\n",
        "\n",
        "        # saving the model\n",
        "        if epoch % 5 == 0:\n",
        "            print('==> Saving to current...')\n",
        "            state = {\n",
        "                    'opt': args,\n",
        "                    'model': model.state_dict(),\n",
        "                    'elbo': elbo.state_dict(),\n",
        "                    'optimizer': optimizer.state_dict(),\n",
        "                    'epoch': epoch,\n",
        "            }\n",
        "            state['model_ema'] = model_ema.state_dict()\n",
        "            save_file = os.path.join(args.model_folder, 'current.pth')\n",
        "            torch.save(state, save_file)\n",
        "            if epoch % args.save_freq == 0:\n",
        "                save_file = os.path.join(args.model_folder, 'ckpt_epoch_{epoch}.pth'.format(epoch=epoch))\n",
        "                torch.save(state, save_file)\n",
        "            del state\n",
        "            torch.cuda.empty_cache()\n",
        "\n",
        "    return\n",
        "\n",
        "\n",
        "def train_MiCE(epoch, train_loader, model, model_ema, elbo, optimizer, opt):\n",
        "    \"\"\"\n",
        "    one epoch training for MiCE\n",
        "    \"\"\"\n",
        "    model.train()\n",
        "    model_ema.eval()\n",
        "    model_ema.apply(set_bn_train)\n",
        "\n",
        "    batch_time = AverageMeter()\n",
        "    data_time = AverageMeter()\n",
        "    loss_meter = AverageMeter()\n",
        "\n",
        "    all_cluster_labels = []\n",
        "    all_true_labels = []\n",
        "    all_pi_labels = []\n",
        "\n",
        "    new_center_v = torch.zeros((opt.n_class, opt.low_dim)).cuda()\n",
        "\n",
        "    end = time.time()\n",
        "    for idx, (inputs, target) in enumerate(train_loader):\n",
        "        data_time.update(time.time() - end)\n",
        "\n",
        "        bsz = inputs[0].size(0)\n",
        "        # ===================forward=====================\n",
        "        x1, x2, x3 = inputs[0].cuda(), inputs[1].cuda(), inputs[2].cuda()\n",
        "        shuffle_ids, reverse_ids = get_shuffle_ids(bsz)\n",
        "        f = model(x1)\n",
        "        with torch.no_grad():\n",
        "            x2 = x2[shuffle_ids]\n",
        "            v = model_ema(x2)\n",
        "            v = v[reverse_ids]\n",
        "\n",
        "        _, g = model(x3, True)\n",
        "\n",
        "        loss, out, posterior, pi_logit = elbo(f, v, g)\n",
        "\n",
        "        # For cluster label\n",
        "        feat_pseudo_label = torch.argmax(posterior, dim=1)  # batch x 1\n",
        "        pi_label = torch.argmax(pi_logit, dim=1)\n",
        "        all_cluster_labels.extend(list(feat_pseudo_label.data.cpu().numpy()))\n",
        "        all_true_labels.extend(list(target.data.cpu().numpy()))\n",
        "        all_pi_labels.extend(list(pi_label.data.cpu().numpy()))\n",
        "\n",
        "        # ===================backward=====================\n",
        "        optimizer.zero_grad()\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "\n",
        "        # ===================meters=====================\n",
        "        loss_meter.update(loss.item(), bsz)\n",
        "        moment_update(model, model_ema, opt.m)\n",
        "\n",
        "        torch.cuda.synchronize()\n",
        "        batch_time.update(time.time() - end)\n",
        "        end = time.time()\n",
        "\n",
        "        if (idx + 1) % opt.print_freq == 0:\n",
        "            print('Train: [{0}][{1}/{2}]\\t'\n",
        "                  'BT {batch_time.val:.3f} ({batch_time.avg:.3f})\\t'\n",
        "                  'DT {data_time.val:.3f} ({data_time.avg:.3f})\\t'\n",
        "                  'loss {loss.val:.3f} ({loss.avg:.3f})'.format(\n",
        "                   epoch, idx + 1, len(train_loader), batch_time=batch_time,\n",
        "                   data_time=data_time, loss=loss_meter))\n",
        "            sys.stdout.flush()\n",
        "\n",
        "        # Aggregate the teacher's results for the analytical update (Eq.13)\n",
        "        with torch.no_grad():\n",
        "            one_hot_pseudo = torch.nn.functional.one_hot(feat_pseudo_label, num_classes=opt.n_class).float()\n",
        "            new_center_v += torch.einsum(\"bk,bkd->kd\", [one_hot_pseudo, v])\n",
        "\n",
        "    elbo.update_cluster(new_center_v)\n",
        "\n",
        "    all_true_labels = np.array(all_true_labels)\n",
        "    all_cluster_labels = np.array(all_cluster_labels)\n",
        "    all_pi_labels = np.array(all_pi_labels)\n",
        "    cluster_acc = acc(all_true_labels, all_cluster_labels)\n",
        "    pi_cluster_acc = acc(all_true_labels, all_pi_labels)\n",
        "\n",
        "    log_dict = {}\n",
        "    log_dict['loss'] = loss_meter.avg\n",
        "\n",
        "    log_dict['Train_cluster_acc'] = cluster_acc\n",
        "    log_dict['Train_pi_cluster_acc'] = pi_cluster_acc\n",
        "    print(np.bincount(all_cluster_labels))\n",
        "    print(\"Cluster ACC:\", cluster_acc, \"PI Cluster ACC:\", pi_cluster_acc)\n",
        "\n",
        "    return log_dict\n",
        "\n",
        "\n",
        "if __name__ == '__main__':\n",
        "    main()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 430,
          "referenced_widgets": [
            "8d5955c02d1c460cbf07f3bc15d2527e",
            "384cb904311241faa2d2dade79050b7b",
            "50be5d9e386e44a79bc4c80e9e92d019",
            "e089bda735c143b5a9b4a9e059a4ad14",
            "b6cc7203d541453ba72364da792532c8",
            "28b2b78b4ba449698b880e5f28f8253d",
            "7d7b9f5f81e145bda1ae6433186aff4d",
            "f1b00a70bf6343e680dae099eaba7d6a",
            "8861d23735d14bba889931a51d3b851c",
            "ca1e48ac6d2d414f8b2bbdf2d2312dcb",
            "6aa16ba20bd5468e897c70419835231f"
          ]
        },
        "id": "RQW9tnyiTQwL",
        "outputId": "8eaf7ba2-c370-4f2b-fbc3-84c98745436e"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Create dataset with tripple transform\n",
            "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "  0%|          | 0/170498071 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "8d5955c02d1c460cbf07f3bc15d2527e"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting ./data/cifar-10-python.tar.gz to ./data\n"
          ]
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-10-a4d6c9c950aa>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m    281\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    282\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0m__name__\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'__main__'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 283\u001b[0;31m     \u001b[0mmain\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[0;32m<ipython-input-10-a4d6c9c950aa>\u001b[0m in \u001b[0;36mmain\u001b[0;34m()\u001b[0m\n\u001b[1;32m    100\u001b[0m     ])\n\u001b[1;32m    101\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 102\u001b[0;31m     \u001b[0mtrain_dataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtest_dataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcreate_dataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_transform\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_transform\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    103\u001b[0m     \u001b[0mfull_dataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mConcatDataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mtrain_dataset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtest_dataset\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    104\u001b[0m     full_loader = torch.utils.data.DataLoader(full_dataset,\n",
            "\u001b[0;32m<ipython-input-5-791e10b9177d>\u001b[0m in \u001b[0;36mcreate_dataset\u001b[0;34m(dataset, train_transform, test_transform)\u001b[0m\n\u001b[1;32m     49\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mdataset\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'cifar10'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     50\u001b[0m         \u001b[0mtrain_dataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdatasets\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mCIFAR10\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mroot\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'./data'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdownload\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtransform\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtrain_transform\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---> 51\u001b[0;31m         \u001b[0mtest_dataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdatasets\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mCIFAR10\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mroot\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'./data'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdownload\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtransform\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtest_transform\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     52\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     53\u001b[0m     \u001b[0;32melif\u001b[0m \u001b[0mdataset\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'cifar100'\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/python3.7/dist-packages/torchvision/datasets/cifar.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, root, train, transform, target_transform, download)\u001b[0m\n\u001b[1;32m     63\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     64\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mdownload\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 65\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdownload\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     66\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     67\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_check_integrity\u001b[0m\u001b[0;34m(\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/python3.7/dist-packages/torchvision/datasets/cifar.py\u001b[0m in \u001b[0;36mdownload\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    136\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    137\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mdownload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 138\u001b[0;31m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_check_integrity\u001b[0m\u001b[0;34m(\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    139\u001b[0m             \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Files already downloaded and verified\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    140\u001b[0m             \u001b[0;32mreturn\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/torchvision/datasets/cifar.py\u001b[0m in \u001b[0;36m_check_integrity\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    131\u001b[0m             \u001b[0mfilename\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmd5\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfentry\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfentry\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    132\u001b[0m             \u001b[0mfpath\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\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbase_folder\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfilename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 133\u001b[0;31m             \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mcheck_integrity\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfpath\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmd5\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    134\u001b[0m                 \u001b[0;32mreturn\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    135\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/torchvision/datasets/utils.py\u001b[0m in \u001b[0;36mcheck_integrity\u001b[0;34m(fpath, md5)\u001b[0m\n\u001b[1;32m     69\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mmd5\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     70\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 71\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mcheck_md5\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfpath\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmd5\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     72\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     73\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/torchvision/datasets/utils.py\u001b[0m in \u001b[0;36mcheck_md5\u001b[0;34m(fpath, md5, **kwargs)\u001b[0m\n\u001b[1;32m     61\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     62\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mcheck_md5\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfpath\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mstr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmd5\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mstr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mAny\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0mbool\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 63\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mmd5\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0mcalculate_md5\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfpath\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\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     64\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     65\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/torchvision/datasets/utils.py\u001b[0m in \u001b[0;36mcalculate_md5\u001b[0;34m(fpath, chunk_size)\u001b[0m\n\u001b[1;32m     56\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mopen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfpath\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"rb\"\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     57\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mchunk\u001b[0m \u001b[0;32min\u001b[0m \u001b[0miter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mlambda\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mread\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mchunk_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34mb\"\"\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---> 58\u001b[0;31m             \u001b[0mmd5\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mupdate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mchunk\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     59\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mmd5\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhexdigest\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     60\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import itertools\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import matplotlib.pyplot as plt\n",
        "import itertools\n",
        "from sklearn.metrics import confusion_matrix\n",
        "\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize=False,\n",
        "                          title='Confusion matrix',\n",
        "                          cmap=plt.cm.Blues):\n",
        "    \"\"\"\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize=True`.\n",
        "    \"\"\"\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
        "        #print(\"Normalized confusion matrix\")\n",
        "\n",
        "    #print(cm)\n",
        "\n",
        "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation=45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    fmt = '.2f' if normalize else 'd'\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, format(cm[i, j], fmt),\n",
        "                 horizontalalignment=\"center\",\n",
        "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
        "\n",
        "    #plt.tight_layout()\n",
        "    plt.ylabel('True label')\n",
        "    plt.xlabel('Predicted label')"
      ],
      "metadata": {
        "id": "LGv9sD-tvmDs"
      },
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from __future__ import print_function\n",
        "import os\n",
        "import torch\n",
        "import torch.backends.cudnn as cudnn\n",
        "import argparse\n",
        "import time\n",
        "from torchvision import transforms, datasets\n",
        "from torch.utils.data import ConcatDataset\n",
        "from sklearn import metrics\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "def parse_option():\n",
        "    parser = argparse.ArgumentParser('argument for eval')\n",
        "\n",
        "    parser.add_argument('--print_freq', type=int, default=10, help='print frequency')\n",
        "    parser.add_argument('--tb_freq', type=int, default=500, help='tb frequency')\n",
        "    parser.add_argument('--save_freq', type=int, default=100, help='save frequency')\n",
        "    parser.add_argument('--batch_size', type=int, default=100, help='batch_size')\n",
        "    parser.add_argument('--num_workers', type=int, default=32, help='num of workers to use')\n",
        "\n",
        "    # dataset\n",
        "    parser.add_argument('--dataset', type=str, default='cifar10', choices=['cifar10', 'cifar20', 'stl10'])\n",
        "\n",
        "    # resume\n",
        "    parser.add_argument('--resume', default='', type=str, metavar='PATH',\n",
        "                        help='path to latest checkpoint (default: none)')\n",
        "\n",
        "    # model definition\n",
        "    parser.add_argument('--model', type=str, default='resnet34_cifar', choices=['resnet34', 'resnet34_cifar'])\n",
        "    parser.add_argument('--low_dim', type=int, default=128, help='Dimension of each embedding')\n",
        "    parser.add_argument('--test_path', type=str, default='/content/drive/MyDrive/project/DL/ckpt_epoch_1000.pth', help='the model to test')\n",
        "\n",
        "    # loss function\n",
        "    parser.add_argument('--nu', type=int, default=16384, help='Dimension of the queue')\n",
        "    parser.add_argument('--tau', type=float, default=1.0, help='temperature')\n",
        "\n",
        "    # EMA setting\n",
        "    parser.add_argument('--m', type=float, default=0.999, help='exponential moving average weight')\n",
        "\n",
        "    # GPU setting\n",
        "    parser.add_argument('--gpu', default=None, type=int, help='GPU id to use.')\n",
        "    parser.add_argument('--suffix', type=str, default=None, help='suffix to add on the model name')\n",
        "    parser.add_argument('--data_folder', type=str, default=\"./\", help='path to data')\n",
        "    parser.add_argument('--save_path', type=str, default=\"./model_save/eval\", help='path to save model')\n",
        "    parser.add_argument('--tb_path', type=str, default=\"./tensorboard/eval\", help='path to tensorboard')\n",
        "\n",
        "    opt = parser.parse_args()\n",
        "\n",
        "    opt.model_name = opt.test_path.split('/')[-2]\n",
        "    opt.tb_folder = os.path.join(opt.tb_path, opt.model_name)\n",
        "    if not os.path.isdir(opt.tb_folder):\n",
        "        os.makedirs(opt.tb_folder)\n",
        "\n",
        "    opt.save_folder = os.path.join(opt.save_path, opt.model_name)\n",
        "    if not os.path.isdir(opt.save_folder):\n",
        "        os.makedirs(opt.save_folder)\n",
        "\n",
        "    if opt.dataset == 'cifar10' or opt.dataset == 'stl10':\n",
        "        opt.n_class = 10\n",
        "    elif opt.dataset == 'cifar20':\n",
        "        opt.n_class = 20\n",
        "    elif opt.dataset == 'imagenet_dog':\n",
        "        opt.n_class = 15\n",
        "\n",
        "    return opt\n",
        "\n",
        "\n",
        "def get_MiCE_performance(model, model_ema, elbo, dataloader, ndata, n_class, batchSize=100):\n",
        "    model.eval()\n",
        "    model_ema.eval()\n",
        "    all_cluster_labels = []\n",
        "    all_true_labels = []\n",
        "    all_pi_labels = []\n",
        "\n",
        "    with torch.no_grad():\n",
        "        for batch_idx, (inputs, targets) in enumerate(dataloader):\n",
        "            x1, x2, x3 = inputs[0].cuda(), inputs[1].cuda(), inputs[2].cuda()\n",
        "            targets = targets.cuda()\n",
        "\n",
        "            with torch.no_grad():\n",
        "                f = model(x1)\n",
        "                v = model_ema(x2)\n",
        "            _, g = model(x3, True)\n",
        "\n",
        "            loss, out, posterior, pi_logit = elbo(f, v, g)\n",
        "\n",
        "            feat_pseudo_label = torch.argmax(posterior, dim=1)  # batch x 1\n",
        "            pi_label = torch.argmax(pi_logit, dim=1)\n",
        "            #print (feat_pseudo_label)\n",
        "            #print (pi_label)\n",
        "            all_cluster_labels.extend(list(feat_pseudo_label.data.cpu().numpy()))\n",
        "            all_true_labels.extend(list(targets.data.cpu().numpy()))\n",
        "            all_pi_labels.extend(list(pi_label.data.cpu().numpy()))\n",
        "\n",
        "    all_true_labels = np.array(all_true_labels)\n",
        "    all_cluster_labels = np.array(all_cluster_labels)\n",
        "    all_pi_labels = np.array(all_pi_labels)\n",
        "    print (\"True label stat:\", all_true_labels)\n",
        "    print (len(all_true_labels))\n",
        "    print (\"Cluster label stat:\", all_cluster_labels)\n",
        "    print (len(all_cluster_labels))\n",
        "    print(\"True label stat:\", np.bincount(all_true_labels.astype(int)))\n",
        "    print(\"Cluster label stat:\", np.bincount(all_cluster_labels.astype(int)))\n",
        "    \n",
        "    y=np.arange(0,10)\n",
        "    plt.bar(y,np.bincount(all_cluster_labels.astype(int)))\n",
        "    plt.show()\n",
        "\n",
        "    class_names=[0,1,2,3,4,5,6,7,8,9]\n",
        "\n",
        "    mice_acc = acc(all_true_labels, all_cluster_labels)\n",
        "    pi_cluster_acc = acc(all_true_labels, all_pi_labels)\n",
        "\n",
        "    nmi = metrics.normalized_mutual_info_score(labels_true=all_true_labels, labels_pred=all_cluster_labels)\n",
        "    ari = metrics.adjusted_rand_score(labels_true=all_true_labels, labels_pred=all_cluster_labels)\n",
        "    return mice_acc, pi_cluster_acc, nmi, ari\n",
        "\n",
        "\n",
        "def main():\n",
        "    args = parse_option()\n",
        "    if args.gpu is not None:\n",
        "        print(\"Use GPU: {} for training\".format(args.gpu))\n",
        "\n",
        "    image_size, mean, std, n_class = get_dataset_stat(args.dataset)\n",
        "    normalize = transforms.Normalize(mean=mean, std=std)\n",
        "\n",
        "    train_transform = transforms.Compose([\n",
        "        transforms.Resize((image_size, image_size)),\n",
        "        transforms.ToTensor(),\n",
        "        normalize,\n",
        "    ])\n",
        "\n",
        "    train_dataset, test_dataset = create_dataset(args.dataset, train_transform, train_transform)\n",
        "    full_dataset = ConcatDataset([train_dataset, test_dataset])\n",
        "    n_full_data = len(full_dataset)\n",
        "    print(n_full_data)\n",
        "    full_loader = torch.utils.data.DataLoader(full_dataset,\n",
        "                                              batch_size=args.batch_size, shuffle=False,\n",
        "                                              num_workers=args.num_workers, pin_memory=True)\n",
        "    model, model_ema = create_model(args.model, n_class)\n",
        "    elbo = MiCE_ELBO(args.low_dim, n_full_data, args.nu, tau=args.tau, n_class=n_class).cuda(\n",
        "        args.gpu)\n",
        "\n",
        "    ckpt = torch.load(args.test_path)\n",
        "    model.load_state_dict(ckpt['model'])\n",
        "\n",
        "    model_ema.load_state_dict(ckpt['model_ema'])\n",
        "    elbo.load_state_dict(ckpt['elbo'])\n",
        "\n",
        "    print(\"==> loaded checkpoint '{}' (epoch {})\".format(args.test_path, ckpt['epoch']))\n",
        "    print('==> done')\n",
        "\n",
        "    model = model.cuda()\n",
        "    model_ema = model_ema.cuda()\n",
        "    model.eval()\n",
        "    model_ema.eval()\n",
        "    cudnn.benchmark = True\n",
        "\n",
        "    n_data = len(full_dataset)\n",
        "    mice_acc, mice_pi_acc, mice_nmi, mice_ari = get_MiCE_performance(\n",
        "                                                model, model_ema, elbo, full_loader, n_data, n_class\n",
        "    )\n",
        "    print(\"CMoE average:\", \" NMI:\", mice_nmi, \"| Cluster ACC:\", mice_acc, \"| ARI:\", mice_ari)\n",
        "\n",
        "if __name__ == '__main__':\n",
        "    main()\n"
      ],
      "metadata": {
        "id": "xlImsHRhlStf",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000,
          "referenced_widgets": [
            "83b3a8a5183745d1b3e1f92b1e553121",
            "1f1ce47020dc42dd876fd6b36feba28c",
            "b4e5978fd88541729496a092041ce105",
            "6ada64656f0e4e74b75447c8a7a24607",
            "aabce2b8337a428f99dc60cd04fb263c",
            "122bbaef86c44a8ea48514edb5faf5ed",
            "e972f96a1995468eb1fa35be2ddd3078",
            "4b6253e5d35c454ba5e462c746aeaa01",
            "2a7cf12f3dd94fa2840910889cf50ddb",
            "0d6e41f06f6c49d3a29268deb5ab118d",
            "60f6848d0d0d4953afb21823d77727d2"
          ]
        },
        "outputId": "370948e2-2828-4725-98fe-2fe32ff8e3ac"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Create dataset with tripple transform\n",
            "Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "  0%|          | 0/170498071 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "83b3a8a5183745d1b3e1f92b1e553121"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting ./data/cifar-10-python.tar.gz to ./data\n",
            "Files already downloaded and verified\n",
            "60000\n",
            "ResNet MOE with Linear classifier \n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/torch/utils/data/dataloader.py:490: UserWarning: This DataLoader will create 32 worker processes in total. Our suggested max number of worker in current system is 4, which is smaller than what this DataLoader is going to create. Please be aware that excessive worker creation might get DataLoader running slow or even freeze, lower the worker number to avoid potential slowness/freeze if necessary.\n",
            "  cpuset_checked))\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "ResNet MOE with Linear classifier \n",
            "--------------------Initializing -----------------------------------\n",
            "using queue shape: (16384,10,128)\n",
            "--------------------Initialization Ends-----------------------------------\n",
            "==> loaded checkpoint '/content/drive/MyDrive/project/DL/ckpt_epoch_1000.pth' (epoch 1000)\n",
            "==> done\n",
            "True label stat: [6 9 9 ... 5 1 7]\n",
            "60000\n",
            "Cluster label stat: [8 7 7 ... 6 5 9]\n",
            "60000\n",
            "True label stat: [6000 6000 6000 6000 6000 6000 6000 6000 6000 6000]\n",
            "Cluster label stat: [6095 6107 6795 6737 4599 6025 6338 5962 6028 5314]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/utils/linear_assignment_.py:128: FutureWarning: The linear_assignment function is deprecated in 0.21 and will be removed from 0.23. Use scipy.optimize.linear_sum_assignment instead.\n",
            "  FutureWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/utils/linear_assignment_.py:128: FutureWarning: The linear_assignment function is deprecated in 0.21 and will be removed from 0.23. Use scipy.optimize.linear_sum_assignment instead.\n",
            "  FutureWarning)\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/utils/linear_assignment_.py:128: FutureWarning: The linear_assignment function is deprecated in 0.21 and will be removed from 0.23. Use scipy.optimize.linear_sum_assignment instead.\n",
            "  FutureWarning)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "CMoE average:  NMI: 0.7388670720549555 | Cluster ACC: (0.83665, array([[5537,   15,   92,   80,   31,   37,   19,    6,  146,   37],\n",
            "       [  18, 5839,    6,    9,    1,    1,    0,    1,   29,   96],\n",
            "       [ 268,    1, 3835,  359, 1052,  193,  230,   31,   27,    4],\n",
            "       [  68,    6,  143, 3906,  158, 1452,  206,   21,   14,   26],\n",
            "       [  36,    0,  166,  239, 5152,  104,  159,  136,    8,    0],\n",
            "       [  17,    0,   60, 1476,  203, 4042,   50,  144,    4,    4],\n",
            "       [  35,    2,  190,  339,   24,   53, 5341,    2,   14,    0],\n",
            "       [  17,    1,   80,  361,  107,  437,   13, 4970,   12,    2],\n",
            "       [  58,   27,   16,   12,    6,   16,    6,    1, 5821,   37],\n",
            "       [  41,  134,   11,   14,    3,    3,    4,    2,   32, 5756]])) | ARI: 0.700505838363436\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc=[32.14,34.29,36.74,38.57,39.69,40.71,42.95,45.68,46.83,49.36,50.67,51.62,52.83,54.15,56.02,57.17,57.45,59.65,60.19,61.22,62.08,63.3,64.42,64.83,65.67,66.35,66.33,66.91,66.92,67.80,68.99,69.08,69.03,69.52,69.74,70.87,70.64,70.23,71.84,71.07,71.37,71.98,72.1,71.08,72.45,72.61,72.36,72.52,72.31,72.99,73.22,73.27,72.86,74.27,73.94,73.81,74.38,74.18,74.92,74.66,74.92,74.35,74.77,74.59,74.92,75.21,75.38,75.46,75.28,75.92,75.19,75.63,75.55,76.38,75.87,76.51,76.81,76.99,76.14,76.77,76.96,77.31,76.82,76.34,77.03,77.31,77.33,77.24,77.2,77.34,77.27,77.66,77.82,77.71,77.22,78.05,77.74,77.42,78.17,78.35,78.36,78.25,78.60,78.84,78.36,78.78,78.75,78.8,79.07,78.81,78.84,78.94,78.96,79.60,78.9,79.25,79.93,78.97,79.41,79.88,79.7,80.0,80.08,79.75,80.11,79.86,80.36,80.5,80.54,80.66,80.62,80.47,80.85,80.62,80.81,81.08,80.32,80.60,80.7,80.63,80.46,81.0,81.17,81.19,80.56,80.78,81.17,81.28,81.47,81.51,80.94,81.38,81.26,81.35,81.36,81.39,81.01,81.83,81.49,81.87,81.67,81.53,81.67,82.03,81.58,82.14,81.73,81.96,81.95,81.67,82.03,82.21,81.96,82.09,82.13,81.83,82.12,81.96,82.48,82.28,82.14,82.21,82.34,82.47,82.26,82.36,82.16,82.14,82.39,82.45,82.55,82.44,82.28,82.33,82.38,82.43,82.41,82.39,82.36,82.32]\n",
        "epoch=np.arange(1,201)\n",
        "print (epoch)\n",
        "\n",
        "plt.title(\"Accuracy of MOCO\")\n",
        "plt.xlabel(\"epoch\")\n",
        "plt.ylabel(\"Clustering Accuracy\")\n",
        "\n",
        "plt.plot(epoch, acc)\n",
        "plt.grid()\n",
        "plt.show()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 509
        },
        "id": "5cnwkxjGOG7c",
        "outputId": "0506f2c4-564c-48ed-f3d1-d33f0b521abe"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[  1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18\n",
            "  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36\n",
            "  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54\n",
            "  55  56  57  58  59  60  61  62  63  64  65  66  67  68  69  70  71  72\n",
            "  73  74  75  76  77  78  79  80  81  82  83  84  85  86  87  88  89  90\n",
            "  91  92  93  94  95  96  97  98  99 100 101 102 103 104 105 106 107 108\n",
            " 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126\n",
            " 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144\n",
            " 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162\n",
            " 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180\n",
            " 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198\n",
            " 199 200]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    }
  ]
}