06_NADE.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "06_CF-NADE.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyOyaiE0XSxjL8l0/DQakKKz",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "a884df26186043a4b19cf6639d4ba334": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_9001ecebd0b346edac1e4474683b6c11",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_e2a5737a63b4433b9fdbbb91f154a8d7",
              "IPY_MODEL_9c315ea33eba48e9868158f6da47cce9"
            ]
          }
        },
        "9001ecebd0b346edac1e4474683b6c11": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "e2a5737a63b4433b9fdbbb91f154a8d7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_85491d3120424aefa5e342775072695f",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 604,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 604,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_89529e4329334ccd9581d97d6bbe29e5"
          }
        },
        "9c315ea33eba48e9868158f6da47cce9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_f0456d5f98d547cebb9fd5467218f044",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 604/604 [00:00&lt;00:00, 2730.71it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_9aa212296ef64bc6adf11e2e4e4842fc"
          }
        },
        "85491d3120424aefa5e342775072695f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "89529e4329334ccd9581d97d6bbe29e5": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "f0456d5f98d547cebb9fd5467218f044": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "9aa212296ef64bc6adf11e2e4e4842fc": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "1a60e654f1444d02a1e14953e9067791": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_76703a17c3a649a0a7c14a6831e163ed",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_590a428858d34eb2bd3a67b559776fa6",
              "IPY_MODEL_976a64acd3914d4ebdbbc3d6fa3d3903"
            ]
          }
        },
        "76703a17c3a649a0a7c14a6831e163ed": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "590a428858d34eb2bd3a67b559776fa6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_f263d2672c224b4fbbf0035406566d3c",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 604,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 604,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_cf4dbed5d8424f71b9aa7135b608ca40"
          }
        },
        "976a64acd3914d4ebdbbc3d6fa3d3903": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_f71f5fba2340430798c254fa2d11f813",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 604/604 [00:03&lt;00:00, 153.81it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_797de20f949544108d613e6f9e26b186"
          }
        },
        "f263d2672c224b4fbbf0035406566d3c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "cf4dbed5d8424f71b9aa7135b608ca40": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "f71f5fba2340430798c254fa2d11f813": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "797de20f949544108d613e6f9e26b186": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "b7293a184ac14f28951ef12bca4ac2c7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_8a3e493cb72e4e56bb1534ce64c5f866",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_e2bd5a9c4df941ce8f07647e01c915fc",
              "IPY_MODEL_9180ec9018ca4228ad2136fba88af727"
            ]
          }
        },
        "8a3e493cb72e4e56bb1534ce64c5f866": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "e2bd5a9c4df941ce8f07647e01c915fc": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_69d85fb761b745cd80d31959794107f2",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 604,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 604,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_91ce92a2775040efa3b0e5738fc2db81"
          }
        },
        "9180ec9018ca4228ad2136fba88af727": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_1d19a1761d7d48e482cf1f138ebe42da",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 604/604 [00:00&lt;00:00, 3228.63it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_0311652fa38a4ea58aa545b0ba9af1e5"
          }
        },
        "69d85fb761b745cd80d31959794107f2": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "91ce92a2775040efa3b0e5738fc2db81": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "1d19a1761d7d48e482cf1f138ebe42da": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "0311652fa38a4ea58aa545b0ba9af1e5": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "292aff7af4464aea8f8a015534075ef0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_238fa20115f84500acfe2ac966573d0a",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_4c6e7efb5f36483aaef4d91ec080a74c",
              "IPY_MODEL_9a19ca7430424feb9ccef9aeea20c759"
            ]
          }
        },
        "238fa20115f84500acfe2ac966573d0a": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "4c6e7efb5f36483aaef4d91ec080a74c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_b1b7fbcb709842c481b02c2c08ffa71c",
            "_dom_classes": [],
            "description": "100%",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 604,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 604,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_794ff38cb9f54cc8a2593bce688c00d4"
          }
        },
        "9a19ca7430424feb9ccef9aeea20c759": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_ed882353954c430eb479a7c2634d9c50",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 604/604 [00:01&lt;00:00, 383.14it/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_fd7cd0718b60427f8dd035b38ce2a593"
          }
        },
        "b1b7fbcb709842c481b02c2c08ffa71c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "initial",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "794ff38cb9f54cc8a2593bce688c00d4": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "ed882353954c430eb479a7c2634d9c50": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "fd7cd0718b60427f8dd035b38ce2a593": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/ivoryRabbit/RecSys/blob/master/06_NADE.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IgX4n-LEi8Dk"
      },
      "source": [
        "# NADE\r\n",
        "\r\n",
        "- [A Neural Autoregressive Approach to Collaborative Filtering](https://arxiv.org/pdf/1605.09477.pdf)\r\n",
        "- [Neural Autoregressive Collaborative Filtering for Implicit Feedback](https://arxiv.org/pdf/1606.07674.pdf)\r\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qHa-HURmijB4"
      },
      "source": [
        "import glob\r\n",
        "import numpy as np\r\n",
        "import pandas as pd\r\n",
        "from typing import Callable, Tuple, List\r\n",
        "from tqdm.notebook import tqdm\r\n",
        "\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "%matplotlib inline\r\n",
        "plt.style.use('ggplot')\r\n",
        "\r\n",
        "from sklearn.model_selection import train_test_split\r\n",
        "\r\n",
        "import tensorflow as tf\r\n",
        "import tensorflow.keras.backend as K\r\n",
        "\r\n",
        "from tensorflow.keras.layers import Input, Dense\r\n",
        "from tensorflow.keras.models import Model, load_model\r\n",
        "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\r\n",
        "from tensorflow.keras.optimizers import Adam\r\n",
        "from tensorflow.keras.regularizers import l2\r\n",
        "\r\n",
        "from tensorflow.keras.utils import get_file\r\n",
        "import zipfile\r\n",
        "\r\n",
        "np.random.seed(777)\r\n",
        "tf.random.set_seed(777)"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r7-Ipa5Limcz"
      },
      "source": [
        "def load_data(data_size : str) -> pd.DataFrame:\r\n",
        "    ''' load Movie Lens data '''\r\n",
        "\r\n",
        "    if data_size == '1m':\r\n",
        "        fname = 'ml-1m.zip'\r\n",
        "        data = 'ml-1m/ratings.dat'\r\n",
        "    elif data_size == '10m':\r\n",
        "        fname = 'ml-10m.zip'\r\n",
        "        data = 'ml-10M100K/ratings.dat'\r\n",
        "    elif data_size == '20m':\r\n",
        "        fname = 'ml-20m.zip'\r\n",
        "        data = 'ml-20m/ratings.csv'\r\n",
        "    elif data_size == '25m':\r\n",
        "        fname = 'ml-25m.zip'\r\n",
        "        data = 'ml-25m/ratings.csv'\r\n",
        "    if not glob.glob(data):\r\n",
        "        origin = f'http://files.grouplens.org/datasets/movielens/{fname}'\r\n",
        "        file = get_file(fname, origin)\r\n",
        "        zip_ref = zipfile.ZipFile(file, 'r')\r\n",
        "        zip_ref.extractall()\r\n",
        "\r\n",
        "    col_names = ['user_id', 'movie_id', 'rating', 'timestamp']\r\n",
        "    if data_size in ['20m', '25m']:\r\n",
        "        ratings = pd.read_csv(data, names = col_names, engine = 'python')\r\n",
        "    else:\r\n",
        "        ratings = pd.read_csv(data, sep = '|', delimiter = '::', names = col_names, engine = 'python')\r\n",
        "    print(ratings.shape)\r\n",
        "    return ratings"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5ir_ByEOimUs",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 213
        },
        "outputId": "01847da6-2639-4a85-aa0d-115aea91169c"
      },
      "source": [
        "ratings = load_data('1m')\r\n",
        "ratings.head()"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(1000209, 4)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>user_id</th>\n",
              "      <th>movie_id</th>\n",
              "      <th>rating</th>\n",
              "      <th>timestamp</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>1193</td>\n",
              "      <td>5</td>\n",
              "      <td>978300760</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>661</td>\n",
              "      <td>3</td>\n",
              "      <td>978302109</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1</td>\n",
              "      <td>914</td>\n",
              "      <td>3</td>\n",
              "      <td>978301968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1</td>\n",
              "      <td>3408</td>\n",
              "      <td>4</td>\n",
              "      <td>978300275</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1</td>\n",
              "      <td>2355</td>\n",
              "      <td>5</td>\n",
              "      <td>978824291</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   user_id  movie_id  rating  timestamp\n",
              "0        1      1193       5  978300760\n",
              "1        1       661       3  978302109\n",
              "2        1       914       3  978301968\n",
              "3        1      3408       4  978300275\n",
              "4        1      2355       5  978824291"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q7NxLKRuimOY",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f0bb7af4-2f5a-46b4-dc6e-9b833f15f610"
      },
      "source": [
        "user_list = ratings.user_id.unique()\r\n",
        "user_idx_map = {e: i for i, e in enumerate(user_list)}\r\n",
        "n_user = len(user_list)\r\n",
        "print(f'# of users = {n_user}')\r\n",
        "\r\n",
        "item_list = ratings.movie_id.unique()\r\n",
        "item_idx_map = {e: i for i, e in enumerate(item_list)}\r\n",
        "n_item = len(item_list)\r\n",
        "print(f'# of items = {n_item}')"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "# of users = 6040\n",
            "# of items = 3706\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b5i143z_3sw8"
      },
      "source": [
        "def binarizer(df: pd.DataFrame, threshold = 4) -> pd.DataFrame:\r\n",
        "    df = df.assign(rating = np.where(df.rating >= threshold, 1, 0))\r\n",
        "    return df[df.rating > 0.0].reset_index(drop = True)\r\n",
        "\r\n",
        "def make_warm(df: pd.DataFrame, threshold = 5) -> pd.DataFrame: # remove cold starters\r\n",
        "    positive = df.groupby('user_id')['movie_id'].count()\r\n",
        "    positive = positive.index[positive >= threshold]\r\n",
        "    return df[df.user_id.isin(positive)].reset_index(drop = True)\r\n",
        "\r\n",
        "def train_valid_test_split(df: pd.DataFrame, size: float) -> pd.DataFrame:\r\n",
        "    train_user, test_user = train_test_split(df.user_id.unique(), test_size = 2 * size, random_state = 777)\r\n",
        "    valid_user, test_user = train_test_split(test_user, test_size = 0.5, random_state = 777)\r\n",
        "    train, valid, test = map(lambda x: df[df.user_id.isin(x)], (train_user, valid_user, test_user))\r\n",
        "    train, valid, test = map(lambda df: df.reset_index(drop = True), (train, valid, test))\r\n",
        "    return train, valid, test\r\n",
        "\r\n",
        "def query_relev_split(df: pd.DataFrame, size: float) -> pd.DataFrame:\r\n",
        "    timeorder = df.groupby('user_id')['timestamp'].rank(method = 'first', ascending = True)\r\n",
        "    seen_cnts = df.groupby('user_id')['movie_id'].transform('count')\r\n",
        "    df = df.assign(seen_cnts = seen_cnts, timeorder = timeorder)\r\n",
        "    query = df[df.timeorder < df.seen_cnts * (1-size)]\r\n",
        "    relev = df[df.timeorder >= df.seen_cnts * (1-size)]\r\n",
        "    relev = relev[relev.user_id.isin(query.user_id.unique())]\r\n",
        "    query, relev = map(lambda df: df.drop(columns = ['timeorder', 'seen_cnts']), (query, relev))\r\n",
        "    query, relev = map(lambda df: df.reset_index(drop = True), (query, relev))\r\n",
        "    return query, relev\r\n",
        "\r\n",
        "def list_agg(df: pd.DataFrame) -> pd.DataFrame:\r\n",
        "    return df.groupby('user_id', as_index = False)[['movie_id']].agg(list)"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3tFezRJKitQ_",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 195
        },
        "outputId": "9cadd55a-2bd4-41c0-fc29-da96db531d4a"
      },
      "source": [
        "data = ratings.assign(user_id = lambda df : df.user_id.map(user_idx_map),\r\n",
        "                      movie_id = lambda df: df.movie_id.map(item_idx_map))\r\n",
        "data = binarizer(data)\r\n",
        "data = make_warm(data)\r\n",
        "data.head()"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>user_id</th>\n",
              "      <th>movie_id</th>\n",
              "      <th>rating</th>\n",
              "      <th>timestamp</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>978300760</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>3</td>\n",
              "      <td>1</td>\n",
              "      <td>978300275</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>978824291</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>978302039</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>7</td>\n",
              "      <td>1</td>\n",
              "      <td>978300719</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   user_id  movie_id  rating  timestamp\n",
              "0        0         0       1  978300760\n",
              "1        0         3       1  978300275\n",
              "2        0         4       1  978824291\n",
              "3        0         6       1  978302039\n",
              "4        0         7       1  978300719"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sA9gEBFmvoCW"
      },
      "source": [
        "train, valid, test = train_valid_test_split(data, size = 0.1)\r\n",
        "valid_q, valid_r = query_relev_split(valid, size = 0.2)\r\n",
        "test_q, test_r = query_relev_split(test, size = 0.2)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1sKFhqmivs9U"
      },
      "source": [
        "def train_generator(train: pd.DataFrame, \r\n",
        "                    n_item: int, \r\n",
        "                    batch_size: int) -> np.ndarray:\r\n",
        "\r\n",
        "    train = list_agg(train)\r\n",
        "    \r\n",
        "    n_data = train.index.size\r\n",
        "    ids = np.arange(n_data)\r\n",
        "    profile = train['movie_id']\r\n",
        "\r\n",
        "    n_batch = int(np.ceil(n_data / batch_size))\r\n",
        "    while True:\r\n",
        "        np.random.shuffle(ids)\r\n",
        "        for batch_step in range(n_batch):     \r\n",
        "            lower = batch_size * batch_step\r\n",
        "            upper = batch_size + lower\r\n",
        "            \r\n",
        "            batch_id = ids[lower: upper]\r\n",
        "            batch = np.zeros(shape = (batch_id.size, n_item), dtype = np.float32)\r\n",
        "            for i, idx in enumerate(batch_id):\r\n",
        "                batch[i, profile[idx]] = 1.0\r\n",
        "            \r\n",
        "            yield batch"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "raJ3gC-5vujr"
      },
      "source": [
        "def valid_generator(query: pd.DataFrame,\r\n",
        "                    relev: pd.DataFrame,\r\n",
        "                    n_item: int,\r\n",
        "                    batch_size: int) -> Tuple[np.ndarray]:\r\n",
        "\r\n",
        "    query = list_agg(query)\r\n",
        "    relev = list_agg(relev)\r\n",
        "\r\n",
        "    n_data = query.index.size\r\n",
        "    ids = np.arange(n_data)\r\n",
        "    profile_q = query['movie_id']\r\n",
        "    profile_r = relev['movie_id']\r\n",
        "\r\n",
        "    n_batch = int(np.ceil(n_data / batch_size))\r\n",
        "    while True:\r\n",
        "        np.random.shuffle(ids)\r\n",
        "        for batch_step in range(n_batch):\r\n",
        "            lower = batch_size * batch_step\r\n",
        "            upper = batch_size + lower\r\n",
        "            \r\n",
        "            batch_id = ids[lower: upper]\r\n",
        "            batch_q = np.zeros(shape = (batch_id.size, n_item), dtype = np.float32)\r\n",
        "            batch_r = np.zeros(shape = (batch_id.size, n_item), dtype = np.float32)\r\n",
        "            for i, idx in enumerate(batch_id):\r\n",
        "                batch_q[i, profile_q[idx]] = 1.0 # seen\r\n",
        "                batch_r[i, profile_r[idx]] = 1.0 # unseen\r\n",
        "\r\n",
        "            yield batch_q, batch_r"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0eD6RLN7vwmE"
      },
      "source": [
        "def binary_cross_entropy(true: tf.Tensor, pred: tf.Tensor) -> tf.Tensor:\r\n",
        "    ce = K.sum(K.binary_crossentropy(true, pred), axis = 1)\r\n",
        "    return K.mean(ce, axis = 0)\r\n",
        "\r\n",
        "def multinomial(true: tf.Tensor, pred: tf.Tensor) -> tf.Tensor:\r\n",
        "    ll = -K.sum(true * K.log(pred), axis = 1) #K.categorical_crossentropy(true, pred) # -K.sum(true * K.log(pred), axis = 1)\r\n",
        "    return K.mean(ll, axis = 0)\r\n",
        "\r\n",
        "def batch_ndcg(query: tf.Tensor, \r\n",
        "               relev: tf.Tensor,\r\n",
        "               pred: tf.Tensor, \r\n",
        "               weight: tf.Tensor,\r\n",
        "               k = 100) -> tf.Tensor: # only for validation\r\n",
        "\r\n",
        "    rec = tf.where(query == 1.0, -1.0, pred) # leave only unseen-data\r\n",
        "    rec = tf.argsort(rec, axis = 1, direction = 'DESCENDING')[:, :k]\r\n",
        "    shape = tf.shape(rec)\r\n",
        "\r\n",
        "    row = tf.repeat(tf.range(shape[0]), shape[1])\r\n",
        "    row = tf.expand_dims(row, axis = 1)\r\n",
        "    col = tf.reshape(rec, shape = [-1])\r\n",
        "    col = tf.expand_dims(col, axis = 1)\r\n",
        "    loc = tf.concat([row, col], axis = 1)\r\n",
        "\r\n",
        "    top_k = tf.reshape(tf.gather_nd(relev, loc), shape = shape)\r\n",
        "    ideal = tf.sort(relev, axis = 1, direction = 'DESCENDING')[:, :k]\r\n",
        "    weight = weight[:shape[1]]\r\n",
        "\r\n",
        "    dcg = tf.reduce_sum(top_k * weight, axis = 1)\r\n",
        "    idcg = tf.reduce_sum(ideal * weight, axis = 1)\r\n",
        "    return tf.reduce_mean(dcg / idcg, axis = 0)"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HWWJlFN3wepv"
      },
      "source": [
        "class Encoder(tf.keras.layers.Layer):\r\n",
        "    def __init__(self, n_unit):\r\n",
        "        super(Encoder, self).__init__()\r\n",
        "        self.n_unit = n_unit\r\n",
        "\r\n",
        "    def build(self, input_shape):\r\n",
        "        self.W = self.add_weight(\r\n",
        "            shape = (input_shape[-1], self.n_unit),\r\n",
        "            initializer = 'glorot_uniform',\r\n",
        "            regularizer = l2(1e-3),\r\n",
        "            trainable = True,\r\n",
        "            name = 'W'\r\n",
        "        )\r\n",
        "\r\n",
        "    def call(self, inputs):\r\n",
        "        r = tf.einsum('nm,mh->nmh', inputs, self.W)\r\n",
        "        logit_h = tf.math.cumsum(r, axis = 1, exclusive = True)\r\n",
        "        return tf.nn.sigmoid(logit_h)"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bmHxbKBZxebb"
      },
      "source": [
        "class Decoder(tf.keras.layers.Layer):\r\n",
        "    def __init__(self, n_unit, likelihood):\r\n",
        "        super(Decoder, self).__init__()\r\n",
        "        self.n_unit = n_unit\r\n",
        "        self.likelihood = likelihood\r\n",
        "\r\n",
        "    def build(self, input_shape):\r\n",
        "        self.V = self.add_weight(\r\n",
        "            shape = (input_shape[-1], self.n_unit),\r\n",
        "            initializer = 'glorot_uniform',\r\n",
        "            regularizer = l2(1e-3),\r\n",
        "            trainable = True,\r\n",
        "            name = 'V'\r\n",
        "        )\r\n",
        "        self.c = self.add_weight(\r\n",
        "            shape = (1, self.n_unit),\r\n",
        "            initializer = 'zeros',\r\n",
        "            trainable = True,\r\n",
        "            name = 'c'\r\n",
        "        )\r\n",
        "        \r\n",
        "    def call(self, inputs):\r\n",
        "        logit_o = tf.einsum('nmh,hm->nm', inputs, self.V) + self.c\r\n",
        "        if self.likelihood == 'multinomial':\r\n",
        "            return tf.nn.softmax(logit_o)\r\n",
        "        return tf.nn.sigmoid(logit_o)"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7zK_FFNtyKAE"
      },
      "source": [
        "class NADE(tf.keras.models.Model):\r\n",
        "    def __init__(self, input_dim, latent_dim, likelihood = 'logistic'):\r\n",
        "        super(NADE, self).__init__()\r\n",
        "\r\n",
        "        self.input_dim = input_dim\r\n",
        "        self.latent_dim = latent_dim\r\n",
        "        self.likelihood = likelihood\r\n",
        "\r\n",
        "        self.encoder = Encoder(self.latent_dim)\r\n",
        "        self.decoder = Decoder(self.input_dim, self.likelihood)\r\n",
        "\r\n",
        "        _ = self.call(Input(shape = (self.input_dim, )))\r\n",
        "        self.build(input_shape = (None, self.input_dim))\r\n",
        "\r\n",
        "    def call(self, inputs):\r\n",
        "        h = self.encoder(inputs)\r\n",
        "        return self.decoder(h)\r\n",
        "\r\n",
        "    def train_step(self, inputs):\r\n",
        "        with tf.GradientTape() as tape:\r\n",
        "            outputs = self(inputs)\r\n",
        "            loss = self.loss(inputs, outputs)\r\n",
        "        grad = tape.gradient(loss, self.trainable_variables)\r\n",
        "        self.optimizer.apply_gradients(zip(grad, self.trainable_variables))\r\n",
        "        return {'loss': loss}\r\n",
        "\r\n",
        "    def test_step(self, data):\r\n",
        "        query, relev = data\r\n",
        "        pred = self(query)\r\n",
        "\r\n",
        "        weight = tf.range(2, self.input_dim, dtype = 'float32')\r\n",
        "        weight = tf.math.reciprocal(tf.math.log(weight))\r\n",
        "        ndcg = batch_ndcg(query, relev, pred, weight)\r\n",
        "        return {'ndcg': ndcg}"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ATaH4QrqMEIo"
      },
      "source": [
        "def show_history(hist, loss: str, metric: str) -> None:\r\n",
        "    fig, (ax1, ax2) = plt.subplots(2, 1, figsize = (8, 6))\r\n",
        "\r\n",
        "    ax1.plot(hist.history[f'{loss}'])\r\n",
        "    ax1.set_title(loss, fontsize = 20)\r\n",
        "    ax1.set_ylabel(loss)\r\n",
        "    ax1.set_xlabel('epoch')\r\n",
        "\r\n",
        "    ax2.plot(hist.history[f'val_{metric}'])\r\n",
        "    ax2.set_title(metric, fontsize = 20)\r\n",
        "    ax2.set_ylabel(metric)\r\n",
        "    ax2.set_xlabel('epoch')\r\n",
        "    plt.tight_layout()"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bqGhj7lOwPNE"
      },
      "source": [
        "def test_predictor(query: pd.DataFrame, \r\n",
        "                   model: Callable,\r\n",
        "                   n_item: int, \r\n",
        "                   batch_size: int, \r\n",
        "                   N = 100) -> float:\r\n",
        "\r\n",
        "    query = list_agg(query)\r\n",
        "\r\n",
        "    n_data = query.index.size\r\n",
        "    ids = np.arange(n_data)\r\n",
        "    profile = query['movie_id']\r\n",
        "\r\n",
        "    pred = []\r\n",
        "    n_batch = int(np.ceil(n_data / batch_size))\r\n",
        "    for batch_step in range(n_batch):\r\n",
        "        lower = batch_size * batch_step\r\n",
        "        upper = batch_size + lower\r\n",
        "\r\n",
        "        batch_id = ids[lower: upper]\r\n",
        "        batch = np.zeros(shape = (batch_id.size, n_item))\r\n",
        "        for i, idx in enumerate(batch_id):\r\n",
        "            batch[i, profile[idx]] = 1.0\r\n",
        "\r\n",
        "        pred_batch = model.predict(batch, verbose = 0)\r\n",
        "        \r\n",
        "        rec = np.argsort(np.where(batch == 1.0, -1.0, pred_batch), axis = 1)[:, :-N-1:-1]\r\n",
        "        pred.append(rec)\r\n",
        "    pred = np.concatenate(pred)\r\n",
        "    pred = [{'user_id': query.at[i, 'user_id'], 'movie_id': pred[i]} for i in range(n_data)]\r\n",
        "    return pd.DataFrame(pred)"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fLM1CfikCjiu"
      },
      "source": [
        "class evaluate:\r\n",
        "    def __init__(self, true: pd.DataFrame, pred: pd.DataFrame):\r\n",
        "        self.true = true\r\n",
        "        self.pred = pred\r\n",
        "        self.max_K = 10000\r\n",
        "        self.idcg = np.cumsum([1.0 / np.log2(i+2) for i in range(self.max_K)])\r\n",
        "\r\n",
        "    def _recall(self, gt: List, rec: List, K = None) -> float: # Recall\r\n",
        "        res = [r for r in rec[:K] if r in gt]\r\n",
        "        return len(res) / np.min([K, len(gt)])\r\n",
        "    \r\n",
        "    def _precision(self, gt: List, rec: List, K = None) -> float: # Precision\r\n",
        "        res = [r for r in rec[:K] if r in gt]\r\n",
        "        return len(res) / len(rec[:K])\r\n",
        "\r\n",
        "    def _AP(self, gt: List, rec: List, K = None) -> float: # Average Precision\r\n",
        "        res = 0.0\r\n",
        "        for i, r in enumerate(rec[:K]):\r\n",
        "            if r in gt:\r\n",
        "                res += self._precision(gt, rec[:K], i+1)\r\n",
        "        return res / np.min([K, len(gt)])\r\n",
        "\r\n",
        "    def _HR(self, gt: List, rec: List, K = None) -> float: # Hit Rate\r\n",
        "        for i, r in enumerate(rec[:K]):\r\n",
        "            if r in gt:\r\n",
        "                return  1.0\r\n",
        "        return 0\r\n",
        "\r\n",
        "    def _RR(self, gt: List, rec: List, K = None) -> float: # Reciprocal Rank\r\n",
        "        for i, r in enumerate(rec[:K]):\r\n",
        "            if r in gt:\r\n",
        "                return  1.0 / (i+1)\r\n",
        "        return 0\r\n",
        "\r\n",
        "    def _nDCG(self, gt: List, rec: List, K = None) -> float: # normalized Discounted Cumulative Gain\r\n",
        "        dcg = 0.0\r\n",
        "        for i, r in enumerate(rec[:K]):\r\n",
        "            if r in gt:\r\n",
        "                dcg += 1.0 / np.log2(i+2)\r\n",
        "        idcg = self.idcg[min([len(gt), K])-1]\r\n",
        "        return dcg / idcg\r\n",
        "\r\n",
        "    def _get_item(self, rec: List, K = None) -> float:\r\n",
        "        for r in rec[:K]:\r\n",
        "            self.uniq_item[r] = self.uniq_item.get(r, 0) + 1\r\n",
        "\r\n",
        "    def __call__(self, K = None):\r\n",
        "        self.K = K if K else self.max_K\r\n",
        "        self.recall = 0.0\r\n",
        "        self.precision = 0.0\r\n",
        "        self.MAP = 0.0\r\n",
        "        self.HR = 0.0\r\n",
        "        self.MRR = 0.0\r\n",
        "        self.nDCG = 0.0\r\n",
        "        self.uniq_item = {}\r\n",
        "        n = self.true.index.size\r\n",
        "        for gt, rec in zip(tqdm(self.true.movie_id), self.pred.movie_id):\r\n",
        "            self.recall += self._recall(gt, rec, K) / n\r\n",
        "            self.precision += self._precision(gt, rec, K) / n\r\n",
        "            self.MAP += self._AP(gt, rec, K) / n\r\n",
        "            self.HR += self._HR(gt, rec, K) / n\r\n",
        "            self.MRR += self._RR(gt, rec, K) / n\r\n",
        "            self.nDCG += self._nDCG(gt, rec, K) / n\r\n",
        "            self._get_item(rec, K)\r\n",
        "        self.CO = len(self.uniq_item) / (n*K) # Coverage\r\n",
        "        self.ED = -np.sum([self.uniq_item[i] * np.log(self.uniq_item[i] / (n*K)) for i in self.uniq_item]) / (n*K) # Entropy-Diversity\r\n",
        "\r\n",
        "    def print_all(self):\r\n",
        "        K = '@' + str(self.K) if self.K else ''\r\n",
        "        print(f'{\"Recall\":>12}{K} : {self.recall:.5f}',\r\n",
        "              f'\\n{\"Precision\":>12}{K} : {self.precision:.5f}',\r\n",
        "              f'\\n{\"MAP\":>12}{K} : {self.MAP:.5f}',\r\n",
        "              f'\\n{\"HR\":>12}{K} : {self.HR:.5f}', \r\n",
        "              f'\\n{\"MRR\":>12}{K} : {self.MRR:.5f}',\r\n",
        "              f'\\n{\"nDCG\":>12}{K} : {self.nDCG:.5f}',\r\n",
        "              f'\\n{\"CO\":>12}{K} : {self.CO:.5f}',\r\n",
        "              f'\\n{\"ED\":>12}{K} : {self.ED:.5f}')"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mV7QWG3L0la9"
      },
      "source": [
        "epochs = 100\r\n",
        "batch_size = 128\r\n",
        "\r\n",
        "train_gen = train_generator(train, n_item, batch_size)\r\n",
        "valid_gen = valid_generator(valid_q, valid_r, n_item, batch_size)\r\n",
        "\r\n",
        "steps_per_epoch = train.user_id.nunique() // batch_size + 1\r\n",
        "validation_steps = valid.user_id.nunique() // batch_size + 1"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WpOEoTiNr0I5"
      },
      "source": [
        "## 1. Bernoulli"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "u1Ob-gxj0M_-",
        "outputId": "8d370f91-3b5b-4e11-b3d9-77f13935933e"
      },
      "source": [
        "model = NADE(n_item, 256)\r\n",
        "model.compile(optimizer = Adam(lr = 1e-3), loss = binary_cross_entropy)\r\n",
        "\r\n",
        "model.summary()"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"nade\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "encoder (Encoder)            (None, 3706, 256)         948736    \n",
            "_________________________________________________________________\n",
            "decoder (Decoder)            (None, 3706)              952442    \n",
            "=================================================================\n",
            "Total params: 1,901,178\n",
            "Trainable params: 1,901,178\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "rSOFDHkR0ZYT",
        "outputId": "26fa4f8d-8fd5-43e4-d0de-b2a5b9f75f90"
      },
      "source": [
        "%%time\r\n",
        "early_stopping = EarlyStopping(monitor = 'val_ndcg', mode = 'max', verbose = 1, patience = epochs // 5)\r\n",
        "model_checkpoint = ModelCheckpoint('NADE-CF.h5', monitor = 'val_ndcg', mode = 'max', save_best_only = True)\r\n",
        "\r\n",
        "hist = model.fit(x = train_gen, validation_data = valid_gen, epochs = epochs,\r\n",
        "                 steps_per_epoch = steps_per_epoch, validation_steps = validation_steps, \r\n",
        "                 verbose = 1, callbacks = [early_stopping, model_checkpoint])\r\n",
        "show_history(hist, 'loss', 'ndcg')"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "38/38 [==============================] - 3s 69ms/step - loss: 822.7874 - val_ndcg: 0.1232\n",
            "Epoch 2/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 404.8673 - val_ndcg: 0.1485\n",
            "Epoch 3/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 397.0951 - val_ndcg: 0.1463\n",
            "Epoch 4/100\n",
            "38/38 [==============================] - 2s 58ms/step - loss: 388.5630 - val_ndcg: 0.1473\n",
            "Epoch 5/100\n",
            "38/38 [==============================] - 2s 58ms/step - loss: 375.8088 - val_ndcg: 0.1505\n",
            "Epoch 6/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 356.2937 - val_ndcg: 0.1499\n",
            "Epoch 7/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 342.5567 - val_ndcg: 0.1274\n",
            "Epoch 8/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 329.4057 - val_ndcg: 0.1581\n",
            "Epoch 9/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 320.4754 - val_ndcg: 0.1404\n",
            "Epoch 10/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 315.4363 - val_ndcg: 0.1544\n",
            "Epoch 11/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 312.9611 - val_ndcg: 0.1526\n",
            "Epoch 12/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 307.7131 - val_ndcg: 0.1564\n",
            "Epoch 13/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 304.0927 - val_ndcg: 0.1406\n",
            "Epoch 14/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 301.6692 - val_ndcg: 0.1493\n",
            "Epoch 15/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 298.1800 - val_ndcg: 0.1556\n",
            "Epoch 16/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 294.2759 - val_ndcg: 0.1621\n",
            "Epoch 17/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 290.1502 - val_ndcg: 0.1431\n",
            "Epoch 18/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 288.0469 - val_ndcg: 0.1705\n",
            "Epoch 19/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 284.3662 - val_ndcg: 0.1609\n",
            "Epoch 20/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 282.1423 - val_ndcg: 0.1584\n",
            "Epoch 21/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 280.6033 - val_ndcg: 0.1458\n",
            "Epoch 22/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 277.9857 - val_ndcg: 0.1972\n",
            "Epoch 23/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 276.2709 - val_ndcg: 0.1541\n",
            "Epoch 24/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 274.5542 - val_ndcg: 0.1880\n",
            "Epoch 25/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 271.5531 - val_ndcg: 0.1825\n",
            "Epoch 26/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 269.7580 - val_ndcg: 0.1955\n",
            "Epoch 27/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 267.6841 - val_ndcg: 0.1824\n",
            "Epoch 28/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 266.8851 - val_ndcg: 0.2003\n",
            "Epoch 29/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 264.1829 - val_ndcg: 0.1676\n",
            "Epoch 30/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 263.3634 - val_ndcg: 0.1984\n",
            "Epoch 31/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 262.6982 - val_ndcg: 0.1872\n",
            "Epoch 32/100\n",
            "38/38 [==============================] - 2s 58ms/step - loss: 261.0434 - val_ndcg: 0.1837\n",
            "Epoch 33/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 258.8518 - val_ndcg: 0.1792\n",
            "Epoch 34/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 257.5662 - val_ndcg: 0.1781\n",
            "Epoch 35/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 256.2377 - val_ndcg: 0.1885\n",
            "Epoch 36/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 255.7482 - val_ndcg: 0.1671\n",
            "Epoch 37/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 254.2704 - val_ndcg: 0.1952\n",
            "Epoch 38/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 252.9704 - val_ndcg: 0.1956\n",
            "Epoch 39/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 252.4469 - val_ndcg: 0.1943\n",
            "Epoch 40/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 250.6049 - val_ndcg: 0.1722\n",
            "Epoch 41/100\n",
            "38/38 [==============================] - 2s 58ms/step - loss: 250.6052 - val_ndcg: 0.1805\n",
            "Epoch 42/100\n",
            "38/38 [==============================] - 2s 58ms/step - loss: 249.6136 - val_ndcg: 0.1927\n",
            "Epoch 43/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 248.3698 - val_ndcg: 0.1919\n",
            "Epoch 44/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 247.2648 - val_ndcg: 0.1874\n",
            "Epoch 45/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 246.8781 - val_ndcg: 0.1981\n",
            "Epoch 46/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 246.6306 - val_ndcg: 0.1954\n",
            "Epoch 47/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 246.0485 - val_ndcg: 0.1959\n",
            "Epoch 48/100\n",
            "38/38 [==============================] - 2s 61ms/step - loss: 243.9130 - val_ndcg: 0.1881\n",
            "Epoch 00048: early stopping\n",
            "CPU times: user 46.2 s, sys: 24.9 s, total: 1min 11s\n",
            "Wall time: 1min 48s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 195
        },
        "id": "mvyvxj-ECqvg",
        "outputId": "f1812e7c-3845-4a6c-8c92-449b671b7f6d"
      },
      "source": [
        "true = list_agg(test_r)\r\n",
        "true.head(5)"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>user_id</th>\n",
              "      <th>movie_id</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2</td>\n",
              "      <td>[177, 179, 183, 84, 4, 198, 199, 207]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>8</td>\n",
              "      <td>[59, 438, 93, 123, 463, 525, 30, 288, 534, 322...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>41</td>\n",
              "      <td>[175, 182, 563, 184, 1176, 116, 128, 1037, 104...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>78</td>\n",
              "      <td>[1258, 305, 500, 2230]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>91</td>\n",
              "      <td>[558, 1176, 195, 587, 592, 74, 1904, 186, 647,...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   user_id                                           movie_id\n",
              "0        2              [177, 179, 183, 84, 4, 198, 199, 207]\n",
              "1        8  [59, 438, 93, 123, 463, 525, 30, 288, 534, 322...\n",
              "2       41  [175, 182, 563, 184, 1176, 116, 128, 1037, 104...\n",
              "3       78                             [1258, 305, 500, 2230]\n",
              "4       91  [558, 1176, 195, 587, 592, 74, 1904, 186, 647,..."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9FKhv5m328J2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 195
        },
        "outputId": "5a3de64d-9e4f-45f0-ec82-ce7e400a22de"
      },
      "source": [
        "pred = test_predictor(test_q, model, n_item, batch_size, N = 100)\r\n",
        "pred.head(5)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>user_id</th>\n",
              "      <th>movie_id</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2</td>\n",
              "      <td>[210, 273, 711, 209, 213, 217, 385, 189, 214, ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>8</td>\n",
              "      <td>[244, 184, 134, 68, 5, 315, 92, 44, 243, 256, ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>41</td>\n",
              "      <td>[132, 189, 160, 62, 775, 155, 92, 711, 758, 85...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>78</td>\n",
              "      <td>[5, 0, 104, 23, 44, 38, 4, 48, 524, 3, 669, 51...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>91</td>\n",
              "      <td>[44, 9, 596, 23, 396, 13, 116, 64, 381, 39, 60...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   user_id                                           movie_id\n",
              "0        2  [210, 273, 711, 209, 213, 217, 385, 189, 214, ...\n",
              "1        8  [244, 184, 134, 68, 5, 315, 92, 44, 243, 256, ...\n",
              "2       41  [132, 189, 160, 62, 775, 155, 92, 711, 758, 85...\n",
              "3       78  [5, 0, 104, 23, 44, 38, 4, 48, 524, 3, 669, 51...\n",
              "4       91  [44, 9, 596, 23, 396, 13, 116, 64, 381, 39, 60..."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MwxqO7y728C_",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 398,
          "referenced_widgets": [
            "a884df26186043a4b19cf6639d4ba334",
            "9001ecebd0b346edac1e4474683b6c11",
            "e2a5737a63b4433b9fdbbb91f154a8d7",
            "9c315ea33eba48e9868158f6da47cce9",
            "85491d3120424aefa5e342775072695f",
            "89529e4329334ccd9581d97d6bbe29e5",
            "f0456d5f98d547cebb9fd5467218f044",
            "9aa212296ef64bc6adf11e2e4e4842fc",
            "1a60e654f1444d02a1e14953e9067791",
            "76703a17c3a649a0a7c14a6831e163ed",
            "590a428858d34eb2bd3a67b559776fa6",
            "976a64acd3914d4ebdbbc3d6fa3d3903",
            "f263d2672c224b4fbbf0035406566d3c",
            "cf4dbed5d8424f71b9aa7135b608ca40",
            "f71f5fba2340430798c254fa2d11f813",
            "797de20f949544108d613e6f9e26b186"
          ]
        },
        "outputId": "0a62283a-499e-421b-9e18-ad3786bae21f"
      },
      "source": [
        "scores = evaluate(true, pred)\r\n",
        "\r\n",
        "scores(K = 10)\r\n",
        "scores.print_all()\r\n",
        "\r\n",
        "scores(K = 100)\r\n",
        "scores.print_all()"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "a884df26186043a4b19cf6639d4ba334",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=604.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "      Recall@10 : 0.09643 \n",
            "   Precision@10 : 0.08328 \n",
            "         MAP@10 : 0.04118 \n",
            "          HR@10 : 0.47848 \n",
            "         MRR@10 : 0.19985 \n",
            "        nDCG@10 : 0.09416 \n",
            "          CO@10 : 0.07152 \n",
            "          ED@10 : 4.87943\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "1a60e654f1444d02a1e14953e9067791",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=604.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "      Recall@100 : 0.35006 \n",
            "   Precision@100 : 0.05394 \n",
            "         MAP@100 : 0.04932 \n",
            "          HR@100 : 0.92550 \n",
            "         MRR@100 : 0.21889 \n",
            "        nDCG@100 : 0.19187 \n",
            "          CO@100 : 0.02051 \n",
            "          ED@100 : 6.17745\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0bB91ASHrnHY"
      },
      "source": [
        "## 2. Multinomial"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XMU7a_AjrnOv",
        "outputId": "717d7f05-f756-440f-b798-3e4045ac27a0"
      },
      "source": [
        "model = NADE(n_item, latent_dim = 256, likelihood = 'multinomial')\r\n",
        "model.compile(optimizer = Adam(lr = 1e-4), loss = multinomial)\r\n",
        "\r\n",
        "model.summary()"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"nade_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "encoder_1 (Encoder)          (None, 3706, 256)         948736    \n",
            "_________________________________________________________________\n",
            "decoder_1 (Decoder)          (None, 3706)              952442    \n",
            "=================================================================\n",
            "Total params: 1,901,178\n",
            "Trainable params: 1,901,178\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "wVZQCkVcrnOz",
        "outputId": "213027fe-ff0c-4d92-e0f7-294ca0d24341"
      },
      "source": [
        "%%time\r\n",
        "early_stopping = EarlyStopping(monitor = 'val_ndcg', mode = 'max', verbose = 1, patience = epochs // 5)\r\n",
        "model_checkpoint = ModelCheckpoint('mult-NADE-CF.h5', monitor = 'val_ndcg', mode = 'max', save_best_only = True)\r\n",
        "\r\n",
        "hist = model.fit(x = train_gen, validation_data = valid_gen, epochs = epochs,\r\n",
        "                 steps_per_epoch = steps_per_epoch, validation_steps = validation_steps, \r\n",
        "                 verbose = 1, callbacks = [early_stopping, model_checkpoint])\r\n",
        "show_history(hist, 'loss', 'ndcg')"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "38/38 [==============================] - 3s 62ms/step - loss: 768.0152 - val_ndcg: 0.0900\n",
            "Epoch 2/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 730.5489 - val_ndcg: 0.1139\n",
            "Epoch 3/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 710.6059 - val_ndcg: 0.1290\n",
            "Epoch 4/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 697.8722 - val_ndcg: 0.1340\n",
            "Epoch 5/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 703.5203 - val_ndcg: 0.1350\n",
            "Epoch 6/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 688.8900 - val_ndcg: 0.1610\n",
            "Epoch 7/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 693.5176 - val_ndcg: 0.1546\n",
            "Epoch 8/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 700.5817 - val_ndcg: 0.1531\n",
            "Epoch 9/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 697.7477 - val_ndcg: 0.1482\n",
            "Epoch 10/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 696.0366 - val_ndcg: 0.1417\n",
            "Epoch 11/100\n",
            "38/38 [==============================] - 2s 62ms/step - loss: 697.3469 - val_ndcg: 0.1537\n",
            "Epoch 12/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 693.7525 - val_ndcg: 0.1475\n",
            "Epoch 13/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 696.5686 - val_ndcg: 0.1359\n",
            "Epoch 14/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 700.3203 - val_ndcg: 0.1331\n",
            "Epoch 15/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 690.8731 - val_ndcg: 0.1517\n",
            "Epoch 16/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 686.8125 - val_ndcg: 0.1406\n",
            "Epoch 17/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 685.6095 - val_ndcg: 0.1636\n",
            "Epoch 18/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 689.9434 - val_ndcg: 0.1418\n",
            "Epoch 19/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 693.2881 - val_ndcg: 0.1628\n",
            "Epoch 20/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 684.2415 - val_ndcg: 0.1487\n",
            "Epoch 21/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 688.6358 - val_ndcg: 0.1492\n",
            "Epoch 22/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 687.7422 - val_ndcg: 0.1546\n",
            "Epoch 23/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 684.9156 - val_ndcg: 0.1601\n",
            "Epoch 24/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 680.2334 - val_ndcg: 0.1589\n",
            "Epoch 25/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 684.4926 - val_ndcg: 0.1645\n",
            "Epoch 26/100\n",
            "38/38 [==============================] - 2s 62ms/step - loss: 678.1012 - val_ndcg: 0.1580\n",
            "Epoch 27/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 681.4152 - val_ndcg: 0.1651\n",
            "Epoch 28/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 676.6204 - val_ndcg: 0.1385\n",
            "Epoch 29/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 678.8749 - val_ndcg: 0.1478\n",
            "Epoch 30/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 674.6748 - val_ndcg: 0.1633\n",
            "Epoch 31/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 676.3364 - val_ndcg: 0.1730\n",
            "Epoch 32/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 671.6476 - val_ndcg: 0.1346\n",
            "Epoch 33/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 673.8775 - val_ndcg: 0.1491\n",
            "Epoch 34/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 675.1299 - val_ndcg: 0.1775\n",
            "Epoch 35/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 670.9230 - val_ndcg: 0.1748\n",
            "Epoch 36/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 675.4945 - val_ndcg: 0.1632\n",
            "Epoch 37/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 668.6983 - val_ndcg: 0.1625\n",
            "Epoch 38/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 677.6058 - val_ndcg: 0.1768\n",
            "Epoch 39/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 662.8358 - val_ndcg: 0.1775\n",
            "Epoch 40/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 682.0251 - val_ndcg: 0.1656\n",
            "Epoch 41/100\n",
            "38/38 [==============================] - 2s 62ms/step - loss: 665.9118 - val_ndcg: 0.1791\n",
            "Epoch 42/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 669.8043 - val_ndcg: 0.1616\n",
            "Epoch 43/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 669.2010 - val_ndcg: 0.1731\n",
            "Epoch 44/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 662.2617 - val_ndcg: 0.1655\n",
            "Epoch 45/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 663.4209 - val_ndcg: 0.1631\n",
            "Epoch 46/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 674.2023 - val_ndcg: 0.1771\n",
            "Epoch 47/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 662.6279 - val_ndcg: 0.1763\n",
            "Epoch 48/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 665.1552 - val_ndcg: 0.1670\n",
            "Epoch 49/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 659.4077 - val_ndcg: 0.1537\n",
            "Epoch 50/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 661.7720 - val_ndcg: 0.1441\n",
            "Epoch 51/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 660.8852 - val_ndcg: 0.1619\n",
            "Epoch 52/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 662.8141 - val_ndcg: 0.1912\n",
            "Epoch 53/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 665.3347 - val_ndcg: 0.1675\n",
            "Epoch 54/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 659.0980 - val_ndcg: 0.1781\n",
            "Epoch 55/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 659.8811 - val_ndcg: 0.1947\n",
            "Epoch 56/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 667.2930 - val_ndcg: 0.1529\n",
            "Epoch 57/100\n",
            "38/38 [==============================] - 2s 62ms/step - loss: 660.4504 - val_ndcg: 0.1743\n",
            "Epoch 58/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 658.4382 - val_ndcg: 0.1741\n",
            "Epoch 59/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 657.6028 - val_ndcg: 0.1633\n",
            "Epoch 60/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 659.5257 - val_ndcg: 0.1810\n",
            "Epoch 61/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 659.2927 - val_ndcg: 0.1795\n",
            "Epoch 62/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 656.3065 - val_ndcg: 0.1757\n",
            "Epoch 63/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 666.4012 - val_ndcg: 0.1712\n",
            "Epoch 64/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 653.0677 - val_ndcg: 0.1850\n",
            "Epoch 65/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 654.3431 - val_ndcg: 0.1744\n",
            "Epoch 66/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 657.1117 - val_ndcg: 0.1767\n",
            "Epoch 67/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 653.2043 - val_ndcg: 0.1749\n",
            "Epoch 68/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 659.3682 - val_ndcg: 0.1784\n",
            "Epoch 69/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 653.3728 - val_ndcg: 0.1801\n",
            "Epoch 70/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 653.0872 - val_ndcg: 0.1815\n",
            "Epoch 71/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 653.3841 - val_ndcg: 0.1825\n",
            "Epoch 72/100\n",
            "38/38 [==============================] - 2s 62ms/step - loss: 664.1051 - val_ndcg: 0.1789\n",
            "Epoch 73/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 655.7795 - val_ndcg: 0.1847\n",
            "Epoch 74/100\n",
            "38/38 [==============================] - 2s 60ms/step - loss: 648.0564 - val_ndcg: 0.1744\n",
            "Epoch 75/100\n",
            "38/38 [==============================] - 2s 59ms/step - loss: 658.4414 - val_ndcg: 0.1892\n",
            "Epoch 00075: early stopping\n",
            "CPU times: user 1min 11s, sys: 39.1 s, total: 1min 50s\n",
            "Wall time: 2min 49s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 195
        },
        "id": "RD78EWrXrnO1",
        "outputId": "46556774-14e1-42d1-a9b6-bd5869998bf0"
      },
      "source": [
        "true = list_agg(test_r)\r\n",
        "true.head(5)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>user_id</th>\n",
              "      <th>movie_id</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2</td>\n",
              "      <td>[177, 179, 183, 84, 4, 198, 199, 207]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>8</td>\n",
              "      <td>[59, 438, 93, 123, 463, 525, 30, 288, 534, 322...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>41</td>\n",
              "      <td>[175, 182, 563, 184, 1176, 116, 128, 1037, 104...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>78</td>\n",
              "      <td>[1258, 305, 500, 2230]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>91</td>\n",
              "      <td>[558, 1176, 195, 587, 592, 74, 1904, 186, 647,...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   user_id                                           movie_id\n",
              "0        2              [177, 179, 183, 84, 4, 198, 199, 207]\n",
              "1        8  [59, 438, 93, 123, 463, 525, 30, 288, 534, 322...\n",
              "2       41  [175, 182, 563, 184, 1176, 116, 128, 1037, 104...\n",
              "3       78                             [1258, 305, 500, 2230]\n",
              "4       91  [558, 1176, 195, 587, 592, 74, 1904, 186, 647,..."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 195
        },
        "id": "hQfVqNQ0rnO2",
        "outputId": "516e278f-df36-4646-c80a-1aa9d2a82aad"
      },
      "source": [
        "pred = test_predictor(test_q, model, n_item, batch_size, N = 100)\r\n",
        "pred.head(5)"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>user_id</th>\n",
              "      <th>movie_id</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2</td>\n",
              "      <td>[132, 48, 38, 97, 128, 92, 23, 51, 134, 189, 2...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>8</td>\n",
              "      <td>[184, 134, 244, 315, 524, 499, 788, 831, 171, ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>41</td>\n",
              "      <td>[132, 189, 856, 160, 92, 48, 775, 155, 38, 216...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>78</td>\n",
              "      <td>[104, 44, 48, 38, 124, 23, 128, 64, 51, 184, 9...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>91</td>\n",
              "      <td>[124, 733, 396, 381, 127, 596, 741, 44, 189, 1...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   user_id                                           movie_id\n",
              "0        2  [132, 48, 38, 97, 128, 92, 23, 51, 134, 189, 2...\n",
              "1        8  [184, 134, 244, 315, 524, 499, 788, 831, 171, ...\n",
              "2       41  [132, 189, 856, 160, 92, 48, 775, 155, 38, 216...\n",
              "3       78  [104, 44, 48, 38, 124, 23, 128, 64, 51, 184, 9...\n",
              "4       91  [124, 733, 396, 381, 127, 596, 741, 44, 189, 1..."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 398,
          "referenced_widgets": [
            "b7293a184ac14f28951ef12bca4ac2c7",
            "8a3e493cb72e4e56bb1534ce64c5f866",
            "e2bd5a9c4df941ce8f07647e01c915fc",
            "9180ec9018ca4228ad2136fba88af727",
            "69d85fb761b745cd80d31959794107f2",
            "91ce92a2775040efa3b0e5738fc2db81",
            "1d19a1761d7d48e482cf1f138ebe42da",
            "0311652fa38a4ea58aa545b0ba9af1e5",
            "292aff7af4464aea8f8a015534075ef0",
            "238fa20115f84500acfe2ac966573d0a",
            "4c6e7efb5f36483aaef4d91ec080a74c",
            "9a19ca7430424feb9ccef9aeea20c759",
            "b1b7fbcb709842c481b02c2c08ffa71c",
            "794ff38cb9f54cc8a2593bce688c00d4",
            "ed882353954c430eb479a7c2634d9c50",
            "fd7cd0718b60427f8dd035b38ce2a593"
          ]
        },
        "id": "OhkjVW6grnO2",
        "outputId": "c4c6f0ca-dd8f-4288-b910-b9dc779e66c8"
      },
      "source": [
        "scores = evaluate(true, pred)\r\n",
        "\r\n",
        "scores(K = 10)\r\n",
        "scores.print_all()\r\n",
        "\r\n",
        "scores(K = 100)\r\n",
        "scores.print_all()"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "b7293a184ac14f28951ef12bca4ac2c7",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=604.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "      Recall@10 : 0.09249 \n",
            "   Precision@10 : 0.07881 \n",
            "         MAP@10 : 0.03929 \n",
            "          HR@10 : 0.44702 \n",
            "         MRR@10 : 0.19184 \n",
            "        nDCG@10 : 0.09010 \n",
            "          CO@10 : 0.05762 \n",
            "          ED@10 : 4.22650\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "292aff7af4464aea8f8a015534075ef0",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "HBox(children=(FloatProgress(value=0.0, max=604.0), HTML(value='')))"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "      Recall@100 : 0.32731 \n",
            "   Precision@100 : 0.05232 \n",
            "         MAP@100 : 0.04605 \n",
            "          HR@100 : 0.89238 \n",
            "         MRR@100 : 0.21031 \n",
            "        nDCG@100 : 0.18106 \n",
            "          CO@100 : 0.01742 \n",
            "          ED@100 : 5.77019\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}