tripplet_loss.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "tripplet_loss.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "fca3c9896c004cbeb304f8be8a18d5ac": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_a4c889e6a1604ea9902222a9e615e4ca",
              "IPY_MODEL_339c86ebc30c4ff5a00a645c42dc1a56",
              "IPY_MODEL_ad872588905b47918db72109040d54af"
            ],
            "layout": "IPY_MODEL_6104556d8d264a5c911a587cd1a724c7"
          }
        },
        "a4c889e6a1604ea9902222a9e615e4ca": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_ba5395fcad28422e97c19aab9c4b7def",
            "placeholder": "​",
            "style": "IPY_MODEL_6084b858bde44fd3ac7f00188702a5de",
            "value": ""
          }
        },
        "339c86ebc30c4ff5a00a645c42dc1a56": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_47f6d2b2f5bb41309aa813cf98ea8494",
            "max": 9912422,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_37da9c38792149ad872f9be73833827b",
            "value": 9912422
          }
        },
        "ad872588905b47918db72109040d54af": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_552dfbd46a7647089bd5c8b1fb950b1f",
            "placeholder": "​",
            "style": "IPY_MODEL_038177e488774cc198f43b8bf86f4f53",
            "value": " 9913344/? [00:00&lt;00:00, 41112146.71it/s]"
          }
        },
        "6104556d8d264a5c911a587cd1a724c7": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "ba5395fcad28422e97c19aab9c4b7def": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "6084b858bde44fd3ac7f00188702a5de": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "47f6d2b2f5bb41309aa813cf98ea8494": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "37da9c38792149ad872f9be73833827b": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "552dfbd46a7647089bd5c8b1fb950b1f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "038177e488774cc198f43b8bf86f4f53": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "9e6d6a7723414997835983351f427e2d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_bb354b13d4d647aaba3303049549289d",
              "IPY_MODEL_094a1714d18b41eab44e143f8d3bbc4d",
              "IPY_MODEL_a0bcffc03df446b7b27e9e875daf9e2b"
            ],
            "layout": "IPY_MODEL_f60fc11a6df2468396788c077920f517"
          }
        },
        "bb354b13d4d647aaba3303049549289d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_fe1df494d12d4ac4b20e7d990b4a7b50",
            "placeholder": "​",
            "style": "IPY_MODEL_ea1fe41b14264de893dbb016f04439d7",
            "value": ""
          }
        },
        "094a1714d18b41eab44e143f8d3bbc4d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_45f661870158445cb7af2e004ad36bc1",
            "max": 28881,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_3a6f74c98de148dba8ac778673a5df32",
            "value": 28881
          }
        },
        "a0bcffc03df446b7b27e9e875daf9e2b": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_dc535f683122479a8a0311ac01e1f488",
            "placeholder": "​",
            "style": "IPY_MODEL_8876637dfc5541b285523fba33f2cba4",
            "value": " 29696/? [00:00&lt;00:00, 9276.15it/s]"
          }
        },
        "f60fc11a6df2468396788c077920f517": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "fe1df494d12d4ac4b20e7d990b4a7b50": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "ea1fe41b14264de893dbb016f04439d7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "45f661870158445cb7af2e004ad36bc1": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "3a6f74c98de148dba8ac778673a5df32": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "dc535f683122479a8a0311ac01e1f488": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "8876637dfc5541b285523fba33f2cba4": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "48f7434299f6426bad2a38abd628bc60": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_328a8dfdae4c4f67a20832a8d29d3937",
              "IPY_MODEL_f7a6a5ee16e046a08c4befb62a46f9f6",
              "IPY_MODEL_45b2d1cc706543c1b5ec78dc0a01e1d7"
            ],
            "layout": "IPY_MODEL_b5ee181d6774474796ecffe18be35359"
          }
        },
        "328a8dfdae4c4f67a20832a8d29d3937": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_17ee0a4495e242ea8db87c26d422e415",
            "placeholder": "​",
            "style": "IPY_MODEL_2498b42f6004417c8ee280ce05b31e12",
            "value": ""
          }
        },
        "f7a6a5ee16e046a08c4befb62a46f9f6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_afa896cba14946288bff446b619ae92f",
            "max": 1648877,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_64e73011ffdd4a12a3eda5f38c1132a5",
            "value": 1648877
          }
        },
        "45b2d1cc706543c1b5ec78dc0a01e1d7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_6663b968af46485dbce792ab21b2a524",
            "placeholder": "​",
            "style": "IPY_MODEL_c70010a96a6041afb8dd7803c7e8834c",
            "value": " 1649664/? [00:00&lt;00:00, 6057036.43it/s]"
          }
        },
        "b5ee181d6774474796ecffe18be35359": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "17ee0a4495e242ea8db87c26d422e415": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "2498b42f6004417c8ee280ce05b31e12": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "afa896cba14946288bff446b619ae92f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "64e73011ffdd4a12a3eda5f38c1132a5": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "6663b968af46485dbce792ab21b2a524": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "c70010a96a6041afb8dd7803c7e8834c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "1cad630c8086498d91f63334e391e422": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_e25353bf08934ff6af630b7d219c8fcf",
              "IPY_MODEL_b6bb708623724cc390dfe4fbd18247af",
              "IPY_MODEL_9bef743aabb240448c1d4b3b8d671562"
            ],
            "layout": "IPY_MODEL_e8656608361f45aeb6e317036a8ccf05"
          }
        },
        "e25353bf08934ff6af630b7d219c8fcf": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_954db9d001ea451992aef8c9b73fa25d",
            "placeholder": "​",
            "style": "IPY_MODEL_4e55202eeff942099c292aea5d04085c",
            "value": ""
          }
        },
        "b6bb708623724cc390dfe4fbd18247af": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_acc0eef891f34aa08eafc05cd397274e",
            "max": 4542,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_4f9f7aa2f8a8452488926e8a32eb9a33",
            "value": 4542
          }
        },
        "9bef743aabb240448c1d4b3b8d671562": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_0a351356377544f8acd5795c3718d6cb",
            "placeholder": "​",
            "style": "IPY_MODEL_fd34d5911abe47c58250892f9685062c",
            "value": " 5120/? [00:00&lt;00:00, 53950.23it/s]"
          }
        },
        "e8656608361f45aeb6e317036a8ccf05": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "954db9d001ea451992aef8c9b73fa25d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "4e55202eeff942099c292aea5d04085c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "acc0eef891f34aa08eafc05cd397274e": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "4f9f7aa2f8a8452488926e8a32eb9a33": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "0a351356377544f8acd5795c3718d6cb": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "fd34d5911abe47c58250892f9685062c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "KTgSefFDENKQ"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "\n",
        "from torch.utils.data import DataLoader\n",
        "from torchvision.datasets import MNIST\n",
        "from torchvision.transforms import Compose, ToTensor, Normalize\n",
        "\n",
        "import numpy as np\n",
        "from sklearn.datasets import make_blobs\n",
        "from sklearn.manifold import TSNE"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import seaborn as sns\n",
        "import matplotlib.pyplot as plt\n",
        "import matplotlib.patheffects as PathEffects\n",
        "import os\n",
        "\n",
        "# Define our own plot function\n",
        "def scatter(x, labels, root='plot', subtitle=None, dataset='MNIST'):\n",
        "    \n",
        "    num_classes = len(set(labels)) # Calculate the number of classes\n",
        "    palette = np.array(sns.color_palette(\"hls\", num_classes)) # Choosing color\n",
        "\n",
        "    ## Create a seaborn scatter plot ##\n",
        "    f = plt.figure(figsize=(8, 8))\n",
        "    ax = plt.subplot(aspect='equal')\n",
        "    sc = ax.scatter(x[:,0], x[:,1], lw=0, s=40,\n",
        "                    c=palette[labels.astype(np.int)])\n",
        "    plt.xlim(-25, 25)\n",
        "    plt.ylim(-25, 25)\n",
        "    ax.axis('off')\n",
        "    ax.axis('tight')\n",
        "    ## ---------------------------- ##\n",
        "    \n",
        "    ## Add label on top of each cluster ##\n",
        "    if dataset=='MNIST':\n",
        "        idx2name = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']\n",
        "    elif dataset=='CIFAR10':\n",
        "        idx2name = ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']\n",
        "    else:\n",
        "        raise Exception(\"Please specify the dataset\")\n",
        "        \n",
        "    txts = []\n",
        "    for i in range(num_classes):\n",
        "        # Position of each label.\n",
        "        xtext, ytext = np.median(x[labels == i, :], axis=0)\n",
        "        txt = ax.text(xtext, ytext, idx2name[i], fontsize=24)\n",
        "        txt.set_path_effects([\n",
        "            PathEffects.Stroke(linewidth=5, foreground=\"w\"),\n",
        "            PathEffects.Normal()])\n",
        "        txts.append(txt)        \n",
        "        \n",
        "    ## ---------------------------- ##    \n",
        "        \n",
        "    if subtitle != None:\n",
        "        plt.suptitle(subtitle)\n",
        "        \n",
        "    if not os.path.exists(root):\n",
        "        os.makedirs(root)        \n",
        "    plt.savefig(os.path.join(root, str(subtitle)))"
      ],
      "metadata": {
        "id": "m2u8mB6crMov"
      },
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "device = torch.device('cuda:0')\n",
        "tsne = TSNE(random_state=0)\n",
        "batch_size_train = 16 # When the batch size is too huge the model collapse, but why?\n",
        "vis_batch_size_train = 256"
      ],
      "metadata": {
        "id": "kE3Tua8lE5eI"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_loader = DataLoader(MNIST('./MNIST/', train=True, download=True,\n",
        "                                transform=Compose([ToTensor(),Normalize((0.1307,), (0.3081,))])\n",
        "                               ),batch_size=vis_batch_size_train, shuffle=True)\n",
        "test_loader = DataLoader(MNIST('./MNIST/', train=False, download=True,\n",
        "                                transform=Compose([ToTensor(),Normalize((0.1307,), (0.3081,))])\n",
        "                               ),batch_size=vis_batch_size_train, shuffle=False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423,
          "referenced_widgets": [
            "fca3c9896c004cbeb304f8be8a18d5ac",
            "a4c889e6a1604ea9902222a9e615e4ca",
            "339c86ebc30c4ff5a00a645c42dc1a56",
            "ad872588905b47918db72109040d54af",
            "6104556d8d264a5c911a587cd1a724c7",
            "ba5395fcad28422e97c19aab9c4b7def",
            "6084b858bde44fd3ac7f00188702a5de",
            "47f6d2b2f5bb41309aa813cf98ea8494",
            "37da9c38792149ad872f9be73833827b",
            "552dfbd46a7647089bd5c8b1fb950b1f",
            "038177e488774cc198f43b8bf86f4f53",
            "9e6d6a7723414997835983351f427e2d",
            "bb354b13d4d647aaba3303049549289d",
            "094a1714d18b41eab44e143f8d3bbc4d",
            "a0bcffc03df446b7b27e9e875daf9e2b",
            "f60fc11a6df2468396788c077920f517",
            "fe1df494d12d4ac4b20e7d990b4a7b50",
            "ea1fe41b14264de893dbb016f04439d7",
            "45f661870158445cb7af2e004ad36bc1",
            "3a6f74c98de148dba8ac778673a5df32",
            "dc535f683122479a8a0311ac01e1f488",
            "8876637dfc5541b285523fba33f2cba4",
            "48f7434299f6426bad2a38abd628bc60",
            "328a8dfdae4c4f67a20832a8d29d3937",
            "f7a6a5ee16e046a08c4befb62a46f9f6",
            "45b2d1cc706543c1b5ec78dc0a01e1d7",
            "b5ee181d6774474796ecffe18be35359",
            "17ee0a4495e242ea8db87c26d422e415",
            "2498b42f6004417c8ee280ce05b31e12",
            "afa896cba14946288bff446b619ae92f",
            "64e73011ffdd4a12a3eda5f38c1132a5",
            "6663b968af46485dbce792ab21b2a524",
            "c70010a96a6041afb8dd7803c7e8834c",
            "1cad630c8086498d91f63334e391e422",
            "e25353bf08934ff6af630b7d219c8fcf",
            "b6bb708623724cc390dfe4fbd18247af",
            "9bef743aabb240448c1d4b3b8d671562",
            "e8656608361f45aeb6e317036a8ccf05",
            "954db9d001ea451992aef8c9b73fa25d",
            "4e55202eeff942099c292aea5d04085c",
            "acc0eef891f34aa08eafc05cd397274e",
            "4f9f7aa2f8a8452488926e8a32eb9a33",
            "0a351356377544f8acd5795c3718d6cb",
            "fd34d5911abe47c58250892f9685062c"
          ]
        },
        "id": "68SpeUgME7eq",
        "outputId": "ae465cfc-952a-49cd-bafa-2e4c3efdde18"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz\n",
            "Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to ./MNIST/MNIST/raw/train-images-idx3-ubyte.gz\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "  0%|          | 0/9912422 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "fca3c9896c004cbeb304f8be8a18d5ac"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting ./MNIST/MNIST/raw/train-images-idx3-ubyte.gz to ./MNIST/MNIST/raw\n",
            "\n",
            "Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz\n",
            "Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to ./MNIST/MNIST/raw/train-labels-idx1-ubyte.gz\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "  0%|          | 0/28881 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "9e6d6a7723414997835983351f427e2d"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting ./MNIST/MNIST/raw/train-labels-idx1-ubyte.gz to ./MNIST/MNIST/raw\n",
            "\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to ./MNIST/MNIST/raw/t10k-images-idx3-ubyte.gz\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "  0%|          | 0/1648877 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "48f7434299f6426bad2a38abd628bc60"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting ./MNIST/MNIST/raw/t10k-images-idx3-ubyte.gz to ./MNIST/MNIST/raw\n",
            "\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to ./MNIST/MNIST/raw/t10k-labels-idx1-ubyte.gz\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "  0%|          | 0/4542 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "1cad630c8086498d91f63334e391e422"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting ./MNIST/MNIST/raw/t10k-labels-idx1-ubyte.gz to ./MNIST/MNIST/raw\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "x_test, y_test = next(iter(test_loader))\n",
        "x_test = x_test.to(device)\n",
        "y_test = y_test.to(device)"
      ],
      "metadata": {
        "id": "A2pFDkMtE9ZJ"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train, y_train = next(iter(train_loader))\n",
        "x_train = x_train.to(device)\n",
        "y_train = y_train.to(device)"
      ],
      "metadata": {
        "id": "DGYW2s6yE-0P"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_loader = DataLoader(MNIST('./MNIST/', train=True, download=True,\n",
        "                                transform=Compose([ToTensor(),Normalize((0.1307,), (0.3081,))])\n",
        "                               ),batch_size=batch_size_train, shuffle=True)\n",
        "test_loader = DataLoader(MNIST('./MNIST/', train=False, download=True,\n",
        "                                transform=Compose([ToTensor(),Normalize((0.1307,), (0.3081,))])\n",
        "                               ),batch_size=batch_size_train, shuffle=False)"
      ],
      "metadata": {
        "id": "_ISybtcjFFqv"
      },
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_tsne_embeds = tsne.fit_transform(x_train.flatten(1).cpu().detach().numpy())\n",
        "test_tsne_embeds = tsne.fit_transform(x_test.flatten(1).cpu().detach().numpy())\n",
        "\n",
        "scatter(train_tsne_embeds, y_train.cpu().numpy(), subtitle=f'online hard Original MNIST distribution (train set)')\n",
        "scatter(test_tsne_embeds, y_test.cpu().numpy(), subtitle=f'online hard Original MNIST distribution (test set)')\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "a-HaghVkFHZX",
        "outputId": "7acc1c09-9536-425f-a2b2-cd363cd16d0c"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:783: FutureWarning: The default initialization in TSNE will change from 'random' to 'pca' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:793: FutureWarning: The default learning rate in TSNE will change from 200.0 to 'auto' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:783: FutureWarning: The default initialization in TSNE will change from 'random' to 'pca' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:793: FutureWarning: The default learning rate in TSNE will change from 200.0 to 'auto' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:17: DeprecationWarning: `np.int` is a deprecated alias for the builtin `int`. To silence this warning, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.\n",
            "Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "\n",
        "class TNN(nn.Module):\n",
        "    def __init__(self, input_shape=(), output_size=4):\n",
        "        super(TNN, self).__init__()\n",
        "        self.cnn1 = nn.Conv2d(input_shape[0], 128, (7,7))\n",
        "        self.cnn2 = nn.Conv2d(128, 256, (5,5))\n",
        "        self.pooling = nn.MaxPool2d((2,2), (2,2))\n",
        "        self.CNN_outshape = self._get_conv_output(input_shape)\n",
        "        self.linear = nn.Linear(self.CNN_outshape, output_size)\n",
        "             \n",
        "    def _get_conv_output(self, shape):\n",
        "        bs = 1\n",
        "        dummy_x = torch.empty(bs, *shape)\n",
        "        x = self._forward_features(dummy_x)\n",
        "        CNN_outshape = x.flatten(1).size(1)\n",
        "        return CNN_outshape\n",
        "    \n",
        "    def _forward_features(self, x):\n",
        "        x = F.relu(self.cnn1(x))\n",
        "        x = self.pooling(x)\n",
        "        x = F.relu(self.cnn2(x))\n",
        "        x = self.pooling(x)\n",
        "        return x     \n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self._forward_features(x)\n",
        "        x = self.linear(x.flatten(1))\n",
        "        return x\n",
        "\n",
        "    \n",
        "def AddGaussianNoise(tensor, mean=0., std=1.):      \n",
        "    return tensor + torch.randn(tensor.size(), device=tensor.device) * std + mean    \n",
        "    \n",
        "class SimpleTNN(nn.Module):\n",
        "    def __init__(self, input_shape=(), output_size=4):\n",
        "        super(SimpleTNN, self).__init__()\n",
        "        self.CNN_outshape = self._get_output(input_shape)\n",
        "        self.linear1 = nn.Linear(self.CNN_outshape, 512)\n",
        "        self.linear2 = nn.Linear(512, output_size)\n",
        "        self.drop = nn.Dropout(p=0.1)\n",
        "             \n",
        "    def _get_output(self, shape):\n",
        "        bs = 1\n",
        "        dummy_x = torch.empty(bs, *shape)\n",
        "        output_shape = dummy_x.flatten(1).size(1)\n",
        "        return output_shape\n",
        "     \n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = torch.relu(self.linear1(x.flatten(1)))\n",
        "        x = self.drop(x)\n",
        "        x = self.linear2(AddGaussianNoise(x, mean=0, std=1e-2))\n",
        "        x = self.drop(x)\n",
        "        return x \n",
        "    \n",
        "class TNN_CIFAR10(nn.Module):\n",
        "    def __init__(self, input_shape=(), output_size=4):\n",
        "        super(TNN_CIFAR10, self).__init__()\n",
        "        self.cnn1a = nn.Conv2d(input_shape[0], 16, kernel_size=3)\n",
        "        self.cnn1b = nn.Conv2d(16, 16, kernel_size=3)\n",
        "        self.cnn2a = nn.Conv2d(16, 32, kernel_size=3)\n",
        "        self.cnn2b = nn.Conv2d(32, 32, kernel_size=3)\n",
        "        self.cnn3a = nn.Conv2d(32, 64, kernel_size=3)\n",
        "        self.cnn3b = nn.Conv2d(64, 64, kernel_size=3)\n",
        "        self.pooling = nn.MaxPool2d((3,3), (1,1))\n",
        "        self.CNN_outshape = self._get_conv_output(input_shape)\n",
        "        self.linear = nn.Linear(self.CNN_outshape, output_size, bias=False)\n",
        "             \n",
        "    def _get_conv_output(self, shape):\n",
        "        bs = 1\n",
        "        dummy_x = torch.empty(bs, *shape)\n",
        "        x = self._forward_features(dummy_x)\n",
        "        CNN_outshape = x.flatten(1).size(1)\n",
        "        return CNN_outshape\n",
        "    \n",
        "    def _forward_features(self, x):\n",
        "        x = F.relu(self.cnn1a(x))\n",
        "        x = self.pooling(x)\n",
        "        x = F.relu(self.cnn1b(x))\n",
        "        x = self.pooling(x)        \n",
        "        \n",
        "        x = F.relu(self.cnn2a(x))\n",
        "        x = self.pooling(x)\n",
        "        x = F.relu(self.cnn2b(x))\n",
        "        x = self.pooling(x)\n",
        "        \n",
        "        x = F.relu(self.cnn3a(x))\n",
        "        x = self.pooling(x)        \n",
        "        x = F.relu(self.cnn3b(x))\n",
        "        x = self.pooling(x)    \n",
        "        return x     \n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self._forward_features(x)\n",
        "        x = self.linear(x.flatten(1))\n",
        "        return x\n",
        "    \n",
        "class TNN_CIFAR10_Drop(nn.Module):\n",
        "    def __init__(self, input_shape=(), output_size=4):\n",
        "        super(TNN_CIFAR10_Drop, self).__init__()\n",
        "        self.cnn1a = nn.Conv2d(input_shape[0], 16, kernel_size=5)\n",
        "        self.cnn2a = nn.Conv2d(16, 32, kernel_size=5)\n",
        "        self.cnn3a = nn.Conv2d(32, 64, kernel_size=5)\n",
        "        self.pooling = nn.MaxPool2d((3,3), (1,1))\n",
        "        self.drop2d = nn.Dropout(p=0.2)\n",
        "        self.CNN_outshape = self._get_conv_output(input_shape)\n",
        "        self.linear = nn.Linear(self.CNN_outshape, output_size)\n",
        "             \n",
        "    def _get_conv_output(self, shape):\n",
        "        bs = 1\n",
        "        dummy_x = torch.empty(bs, *shape)\n",
        "        x = self._forward_features(dummy_x)\n",
        "        CNN_outshape = x.flatten(1).size(1)\n",
        "        return CNN_outshape\n",
        "    \n",
        "    def _forward_features(self, x):\n",
        "        x = F.relu(self.drop2d(self.cnn1a(x)))\n",
        "        x = self.pooling(x)        \n",
        "        \n",
        "        x = F.relu(self.drop2d(self.cnn2a(x)))\n",
        "        x = self.pooling(x)\n",
        "        \n",
        "        x = F.relu(self.drop2d(self.cnn3a(x)))\n",
        "        x = self.pooling(x)        \n",
        "  \n",
        "        return x     \n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self._forward_features(x)\n",
        "        x = self.linear(x.flatten(1))\n",
        "        return x    \n",
        "    \n",
        "    \n",
        "class TNN_CIFAR10_Dropv2(nn.Module):\n",
        "    def __init__(self, input_shape=(), output_size=4):\n",
        "        super(TNN_CIFAR10_Dropv2, self).__init__()\n",
        "        self.cnn1a = nn.Conv2d(input_shape[0], 16, kernel_size=5)\n",
        "        self.cnn1b = nn.Conv2d(16, 16, kernel_size=3)\n",
        "        self.cnn2a = nn.Conv2d(16, 32, kernel_size=5)\n",
        "        self.cnn2b = nn.Conv2d(32, 32, kernel_size=3)\n",
        "        self.cnn3a = nn.Conv2d(32, 64, kernel_size=5)\n",
        "        self.cnn3b = nn.Conv2d(64, 64, kernel_size=3)\n",
        "        self.pooling = nn.MaxPool2d((3,3), (1,1))\n",
        "        self.drop = nn.Dropout(p=0.2)\n",
        "        self.CNN_outshape = self._get_conv_output(input_shape)\n",
        "        self.linear = nn.Linear(self.CNN_outshape, 512)\n",
        "        self.linear_out = nn.Linear(512, output_size)    \n",
        "    def _get_conv_output(self, shape):\n",
        "        bs = 1\n",
        "        dummy_x = torch.empty(bs, *shape)\n",
        "        x = self._forward_features(dummy_x)\n",
        "        CNN_outshape = x.flatten(1).size(1)\n",
        "        return CNN_outshape\n",
        "    \n",
        "    def _forward_features(self, x):\n",
        "        x = F.relu(self.cnn1a(x))\n",
        "        x = F.relu(self.cnn1b(x))\n",
        "        x = self.pooling(x)        \n",
        "        \n",
        "        x = F.relu(self.cnn2a(x))\n",
        "        x = F.relu(self.cnn2b(x))\n",
        "        x = self.pooling(x)\n",
        "        \n",
        "        x = F.relu(self.cnn3a(x))\n",
        "        x = F.relu(self.cnn3b(x))\n",
        "        x = self.pooling(x)        \n",
        "  \n",
        "        return x     \n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self._forward_features(x)\n",
        "        x = self.drop(self.linear(x.flatten(1)))\n",
        "        return x    \n",
        "    \n",
        "    \n",
        "class TNN_CIFAR10_Drop2d(nn.Module):\n",
        "    def __init__(self, input_shape=(), output_size=4):\n",
        "        super(TNN_CIFAR10_Drop2d, self).__init__()\n",
        "        self.cnn1a = nn.Conv2d(input_shape[0], 16, kernel_size=5)\n",
        "        self.cnn2a = nn.Conv2d(16, 32, kernel_size=5)\n",
        "        self.cnn3a = nn.Conv2d(32, 64, kernel_size=5)\n",
        "        self.pooling = nn.MaxPool2d((3,3), (1,1))\n",
        "        self.drop2da = nn.Dropout2d(p=0.15)\n",
        "        self.drop2db = nn.Dropout2d(p=0.1)\n",
        "        self.drop1d = nn.Dropout(p=0.1)\n",
        "        self.CNN_outshape = self._get_conv_output(input_shape)\n",
        "        self.linear = nn.Linear(self.CNN_outshape, output_size)\n",
        "             \n",
        "    def _get_conv_output(self, shape):\n",
        "        bs = 1\n",
        "        dummy_x = torch.empty(bs, *shape)\n",
        "        x = self._forward_features(dummy_x)\n",
        "        CNN_outshape = x.flatten(1).size(1)\n",
        "        return CNN_outshape\n",
        "    \n",
        "    def _forward_features(self, x):\n",
        "        x = F.relu(self.drop2da(self.cnn1a(x)))\n",
        "        x = self.pooling(x)        \n",
        "        \n",
        "        x = F.relu(self.drop2db(self.cnn2a(x)))\n",
        "        x = self.pooling(x)\n",
        "        \n",
        "        x = F.relu(self.drop2db(self.cnn3a(x)))\n",
        "        x = self.pooling(x)        \n",
        "  \n",
        "        return x     \n",
        "        \n",
        "    def forward(self, x):\n",
        "        x = self._forward_features(x)\n",
        "        x = F.tanh(self.linear(x.flatten(1)))\n",
        "        return x        \n",
        "    \n",
        "    \n",
        "cfg = {\n",
        "    'VGG11': [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n",
        "    'VGG13': [64, 64, 'M', 128, 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n",
        "    'VGG16': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512, 'M'],\n",
        "    'VGG19': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 256, 'M', 512, 512, 512, 512, 'M', 512, 512, 512, 512, 'M'],\n",
        "}\n",
        "    \n",
        "    \n",
        "class VGG(nn.Module):\n",
        "    def __init__(self, input_shape, vgg_name, output_size):\n",
        "        super(VGG, self).__init__()\n",
        "        self.features = self._make_layers(cfg[vgg_name])\n",
        "        self.CNN_outshape = self._get_conv_output(input_shape)\n",
        "        self.classifier = nn.Linear(self.CNN_outshape, output_size)\n",
        "\n",
        "    def forward(self, x):\n",
        "        out = self.features(x)\n",
        "        out = out.view(out.size(0), -1)\n",
        "        out = self.classifier(out)\n",
        "        return out\n",
        "    \n",
        "    def _get_conv_output(self, shape):\n",
        "        bs = 1\n",
        "        dummy_x = torch.empty(bs, *shape)\n",
        "        x = self._forward_features(dummy_x)\n",
        "        CNN_outshape = x.flatten(1).size(1)\n",
        "        return CNN_outshape\n",
        "    \n",
        "    def _forward_features(self, x):\n",
        "        x = self.features(x)        \n",
        "        return x      \n",
        "\n",
        "    def _make_layers(self, cfg):\n",
        "        layers = []\n",
        "        in_channels = 3\n",
        "        for x in cfg:\n",
        "            if x == 'M':\n",
        "                layers += [nn.MaxPool2d(kernel_size=2, stride=2)]\n",
        "            else:\n",
        "                layers += [nn.Conv2d(in_channels, x, kernel_size=3, padding=1),\n",
        "#                            nn.BatchNorm2d(x),\n",
        "                           nn.ReLU(inplace=True)]\n",
        "                in_channels = x\n",
        "        layers += [nn.AvgPool2d(kernel_size=1, stride=1)]\n",
        "        return nn.Sequential(*layers)"
      ],
      "metadata": {
        "id": "r2T7WV-rrXBD"
      },
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model = TNN(input_shape=x_train.shape[1:])\n",
        "model.to(device)\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=0.0001)"
      ],
      "metadata": {
        "id": "QEd6s_IirBbm"
      },
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import os\n",
        "import random\n",
        "from itertools import permutations\n",
        "import numpy as np\n",
        "import torch\n",
        "\n",
        "\"\"\"\n",
        "This code is a re-implmentationm of the code written in Tensorflow\n",
        "https://omoindrot.github.io/triplet-loss\n",
        "https://github.com/omoindrot/tensorflow-triplet-loss\n",
        "All the doc strings and comments are from the original implmentation\n",
        "\"\"\"\n",
        "\n",
        "\n",
        "def offline(x,y,testsize=0.3,ap_pairs=10,an_pairs=10):\n",
        "    data_xy = tuple([x,y])\n",
        "\n",
        "    trainsize = 1-testsize\n",
        "\n",
        "    triplet_train_pairs = []\n",
        "    triplet_test_pairs = []\n",
        "    for data_class in sorted(set(data_xy[1])):\n",
        "\n",
        "        same_class_idx = np.where((data_xy[1] == data_class))[0]\n",
        "        diff_class_idx = np.where(data_xy[1] != data_class)[0]\n",
        "        A_P_pairs = random.sample(list(permutations(same_class_idx,2)),k=ap_pairs) #Generating Anchor-Positive pairs\n",
        "        Neg_idx = random.sample(list(diff_class_idx),k=an_pairs)\n",
        "        \n",
        "\n",
        "        #train\n",
        "        A_P_len = len(A_P_pairs)\n",
        "        Neg_len = len(Neg_idx)\n",
        "        for ap in A_P_pairs[:int(A_P_len*trainsize)]:\n",
        "            Anchor = data_xy[0][ap[0]]\n",
        "            Positive = data_xy[0][ap[1]]\n",
        "            for n in Neg_idx:\n",
        "                Negative = data_xy[0][n]\n",
        "                triplet_train_pairs.append([Anchor,Positive,Negative])               \n",
        "        #test\n",
        "        for ap in A_P_pairs[int(A_P_len*trainsize):]:\n",
        "            Anchor = data_xy[0][ap[0]]\n",
        "            Positive = data_xy[0][ap[1]]\n",
        "            for n in Neg_idx:\n",
        "                Negative = data_xy[0][n]\n",
        "                triplet_test_pairs.append([Anchor,Positive,Negative])    \n",
        "                \n",
        "    return np.array(triplet_train_pairs), np.array(triplet_test_pairs)\n",
        "\n",
        "\n",
        "def _get_triplet_mask(labels, device='cpu'):\n",
        "    \"\"\"Return a 3D mask where mask[a, p, n] is True iff the triplet (a, p, n) is valid.\n",
        "    A triplet (i, j, k) is valid if:\n",
        "        - i, j, k are distinct\n",
        "        - labels[i] == labels[j] and labels[i] != labels[k]\n",
        "    Args:\n",
        "        labels: torch.int32 `Tensor` with shape [batch_size]\n",
        "    \"\"\"\n",
        "    # Check that i, j and k are distinct\n",
        "    indices_equal = torch.eye(labels.size()[0]).type(torch.ByteTensor) # make a index in Bool\n",
        "    indices_not_equal = -(indices_equal-1) # flip booleans\n",
        "    i_not_equal_j = torch.unsqueeze(indices_not_equal, 2)\n",
        "    i_not_equal_k = torch.unsqueeze(indices_not_equal, 1)\n",
        "    j_not_equal_k = torch.unsqueeze(indices_not_equal, 0)\n",
        "\n",
        "    distinct_indices = (i_not_equal_j & i_not_equal_k) & j_not_equal_k\n",
        "\n",
        "    # Check if labels[i] == labels[j] and labels[i] != labels[k]\n",
        "    label_equal = (torch.unsqueeze(labels, 0) == torch.unsqueeze(labels, 1))\n",
        "    i_equal_j = torch.unsqueeze(label_equal, 2)\n",
        "    i_equal_k = torch.unsqueeze(label_equal, 1)\n",
        "\n",
        "    valid_labels = (i_equal_j & ~i_equal_k)\n",
        "    # Combine the two masks\n",
        "    \n",
        "    mask = (distinct_indices.to(device).bool() & valid_labels)\n",
        "\n",
        "    return mask\n",
        "\n",
        "\n",
        "def _pairwise_distances(embeddings, squared=False, device='cpu'):\n",
        "    \"\"\"Compute the 2D matrix of distances between all the embeddings.\n",
        "    Args:\n",
        "        embeddings: tensor of shape (batch_size, embed_dim)\n",
        "        squared: Boolean. If true, output is the pairwise squared euclidean distance matrix.\n",
        "                 If false, output is the pairwise euclidean distance matrix.\n",
        "    Returns:\n",
        "        pairwise_distances: tensor of shape (batch_size, batch_size)\n",
        "    \"\"\"\n",
        "    # Get the dot product between all embeddings\n",
        "    embeddings = torch.squeeze(embeddings,1)  # shape=(batch_size, features, 1)\n",
        "    dot_product = torch.matmul(embeddings, torch.transpose(embeddings,0,1)) # shape=(batch_size, batch_size)\n",
        "    \n",
        "    # Get squared L2 norm for each embedding. We can just take the diagonal of `dot_product`.\n",
        "    # This also provides more numerical stability (the diagonal of the result will be exactly 0).\n",
        "    # shape (batch_size,)\n",
        "    square_norm = torch.diag(dot_product)\n",
        "    # Compute the pairwise distance matrix as we have:\n",
        "    # ||a - b||^2 = ||a||^2  - 2 <a, b> + ||b||^2\n",
        "    # shape (batch_size, batch_size)\n",
        "    distances = torch.unsqueeze(square_norm, 1) - 2.0 * dot_product + torch.unsqueeze(square_norm, 0)\n",
        "    \n",
        "    # Because of computation errors, some distances might be negative so we put everything >= 0.0\n",
        "    distances = torch.max(distances, torch.Tensor([0.0]).to(device))\n",
        "    if not squared:\n",
        "        # Because the gradient of sqrt is infinite when distances == 0.0 (ex: on the diagonal)\n",
        "        # we need to add a small epsilon where distances == 0.0\n",
        "        mask = (distances == 0.0).float()\n",
        "        distances = distances + mask * 1e-16\n",
        "        distances = torch.sqrt(distances)\n",
        "        # Correct the epsilon added: set the distances on the mask to be exactly 0.0\n",
        "        distances = distances * (1.0 - mask)\n",
        "    # return distances matrix (batch X batch)\n",
        "    return distances\n",
        "\n",
        "\n",
        "def online_mine_all(labels, embeddings, margin, squared=False, device='cpu'):\n",
        "    \"\"\"Build the triplet loss over a batch of embeddings.\n",
        "    We generate all the valid triplets and average the loss over the positive ones.\n",
        "    Args:\n",
        "        labels: labels of the batch, of size (batch_size,)\n",
        "        embeddings: tensor of shape (batch_size, embed_dim)\n",
        "        margin: margin for triplet loss\n",
        "        squared: Boolean. If true, output is the pairwise squared euclidean distance matrix.\n",
        "                 If false, output is the pairwise euclidean distance matrix.\n",
        "    Returns:\n",
        "        triplet_loss: scalar tensor containing the triplet loss\n",
        "    \"\"\"\n",
        "    # Get the pairwise distance matrix\n",
        "    pairwise_dist = _pairwise_distances(embeddings, squared=squared, device=device)\n",
        "    # shape (batch_size, batch_size, 1)\n",
        "    anchor_positive_dist = torch.unsqueeze(pairwise_dist, 2)\n",
        "    assert anchor_positive_dist.shape[2] == 1, \"{}\".format(anchor_positive_dist.shape)\n",
        "    # shape (batch_size, 1, batch_size)\n",
        "    anchor_negative_dist = torch.unsqueeze(pairwise_dist, 1)\n",
        "    assert anchor_negative_dist.shape[1] == 1, \"{}\".format(anchor_negative_dist.shape)\n",
        "\n",
        "    # Compute a 3D tensor of size (batch_size, batch_size, batch_size)\n",
        "    # triplet_loss[i, j, k] will contain the triplet loss of anchor=i, positive=j, negative=k\n",
        "    # Uses broadcasting where the 1st argument has shape (batch_size, batch_size, 1)\n",
        "    # and the 2nd (batch_size, 1, batch_size)\n",
        "    triplet_loss = anchor_positive_dist - anchor_negative_dist + margin\n",
        "\n",
        "    # Put to zero the invalid triplets\n",
        "    # (where label(a) != label(p) or label(n) == label(a) or a == p)\n",
        "    mask = _get_triplet_mask(labels, device=device)\n",
        "    mask = mask.float()\n",
        "    \n",
        "    triplet_loss = mask*triplet_loss\n",
        "\n",
        "    # Remove negative losses (i.e. the easy triplets)\n",
        "    triplet_loss = torch.max(triplet_loss, torch.Tensor([0.0]).to(device))\n",
        "\n",
        "    # Count number of positive triplets (where triplet_loss > 0)\n",
        "    valid_triplets = (triplet_loss > 1e-16).float()\n",
        "    \n",
        "    num_positive_triplets = torch.sum(valid_triplets)\n",
        "    \n",
        "    num_valid_triplets = torch.sum(mask)\n",
        "    fraction_positive_triplets = num_positive_triplets / (num_valid_triplets + 1e-16)\n",
        "\n",
        "    # Get final mean triplet loss over the positive valid triplets\n",
        "    triplet_loss = torch.sum(triplet_loss) / (num_positive_triplets + 1e-16)\n",
        "\n",
        "    return triplet_loss, num_positive_triplets, num_valid_triplets\n",
        "\n",
        "def _get_anchor_positive_triplet_mask(labels, device):\n",
        "    \"\"\"Return a 2D mask where mask[a, p] is True iff a and p are distinct and have same label.\n",
        "    Args:\n",
        "        labels: tf.int32 `Tensor` with shape [batch_size]\n",
        "    Returns:\n",
        "        mask: tf.bool `Tensor` with shape [batch_size, batch_size]\n",
        "    \"\"\"\n",
        "    # Check that i and j are distinct\n",
        "    indices_equal = torch.eye(labels.size()[0]).bool().to(device)\n",
        "    \n",
        "    indices_not_equal = ~indices_equal # flip booleans\n",
        "\n",
        "    # Check if labels[i] == labels[j]\n",
        "    # Uses broadcasting where the 1st argument has shape (1, batch_size) and the 2nd (batch_size, 1)\n",
        "    labels_equal = (torch.unsqueeze(labels, 0) == torch.unsqueeze(labels, 1))\n",
        "    # Combine the two masks\n",
        "    mask = indices_not_equal & labels_equal\n",
        "\n",
        "    return mask\n",
        "\n",
        "def _get_anchor_negative_triplet_mask(labels, device):\n",
        "    \"\"\"Return a 2D mask where mask[a, n] is True iff a and n have distinct labels.\n",
        "    Args:\n",
        "        labels: tf.int32 `Tensor` with shape [batch_size]\n",
        "    Returns:\n",
        "        mask: tf.bool `Tensor` with shape [batch_size, batch_size]\n",
        "    \"\"\"\n",
        "    # Check if labels[i] != labels[k]\n",
        "    # Uses broadcasting where the 1st argument has shape (1, batch_size) and the 2nd (batch_size, 1)\n",
        "    labels_equal = torch.unsqueeze(labels,0) == torch.unsqueeze(labels,1)\n",
        "\n",
        "    mask = ~labels_equal # invert the boolean tensor\n",
        "\n",
        "    return mask\n",
        "\n",
        "\n",
        "\n",
        "def online_mine_hard(labels, embeddings, margin, squared=False, device='cpu'):\n",
        "    \"\"\"Build the triplet loss over a batch of embeddings.\n",
        "    For each anchor, we get the hardest positive and hardest negative to form a triplet.\n",
        "    Args:\n",
        "        labels: labels of the batch, of size (batch_size,)\n",
        "        embeddings: tensor of shape (batch_size, embed_dim)\n",
        "        margin: margin for triplet loss\n",
        "        squared: Boolean. If true, output is the pairwise squared euclidean distance matrix.\n",
        "                 If false, output is the pairwise euclidean distance matrix.\n",
        "    Returns:\n",
        "        triplet_loss: scalar tensor containing the triplet loss\n",
        "    \"\"\"\n",
        "    # Get the pairwise distance matrix\n",
        "    pairwise_dist = _pairwise_distances(embeddings, squared=squared, device=device)\n",
        "\n",
        "    # For each anchor, get the hardest positive\n",
        "    # First, we need to get a mask for every valid positive (they should have same label)\n",
        "    mask_anchor_positive = _get_anchor_positive_triplet_mask(labels, device)\n",
        "    mask_anchor_positive = mask_anchor_positive\n",
        "\n",
        "    # We put to 0 any element where (a, p) is not valid (valid if a != p and label(a) == label(p))\n",
        "    anchor_positive_dist = mask_anchor_positive*pairwise_dist\n",
        "\n",
        "    # shape (batch_size, 1)\n",
        "    hardest_positive_dist = torch.max(anchor_positive_dist, 1, keepdim=True)[0]\n",
        "\n",
        "    # For each anchor, get the hardest negative\n",
        "    # First, we need to get a mask for every valid negative (they should have different labels)\n",
        "    mask_anchor_negative = _get_anchor_negative_triplet_mask(labels, device)\n",
        "    mask_anchor_negative = mask_anchor_negative\n",
        "\n",
        "    # We add the maximum value in each row to the invalid negatives (label(a) == label(n))\n",
        "    max_anchor_negative_dist = torch.max(pairwise_dist, 1, keepdim=True)[0]\n",
        "    anchor_negative_dist = pairwise_dist + max_anchor_negative_dist * ~(mask_anchor_negative)\n",
        "\n",
        "    # shape (batch_size,)\n",
        "    hardest_negative_dist = torch.min(anchor_negative_dist, 1, keepdim=True)[0]\n",
        "\n",
        "    # Combine biggest d(a, p) and smallest d(a, n) into final triplet loss\n",
        "    triplet_loss = torch.max(hardest_positive_dist - hardest_negative_dist + margin, torch.Tensor([0.0]).to(device))\n",
        "    \n",
        "    # Get final mean triplet loss\n",
        "    triplet_loss = torch.mean(triplet_loss)\n",
        "\n",
        "    return triplet_loss, torch.sum(mask_anchor_positive), torch.sum(mask_anchor_negative)"
      ],
      "metadata": {
        "id": "ae3NUfe7rX8g"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "margin=0.4\n",
        "for epoch in range(100):  # loop over the dataset multiple times\n",
        "\n",
        "    running_loss = 0.0\n",
        "    for i, data in enumerate(train_loader, 0):\n",
        "        # get the inputs\n",
        "        inputs, labels = data\n",
        "        inputs = inputs.to(device)\n",
        "        labels = labels.to(device)\n",
        "        # zero the parameter gradients\n",
        "        optimizer.zero_grad()\n",
        "\n",
        "        # forward + backward + optimize\n",
        "        outputs = model(inputs)\n",
        "        loss, pos_mask, neg_mask = online_mine_hard(labels, outputs,\n",
        "                                                    margin=margin, squared=True,\n",
        "                                                    device=device)\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        running_loss += loss.item()\n",
        "        if i%2 == 0:\n",
        "            print(f\"At epoches = {epoch}, i = {i}, loss = {loss:.5f}, pos_mask = {pos_mask}, neg_mask = {neg_mask}\"\n",
        "                  , end='\\r')\n",
        "    epoch_loss = running_loss / len(train_loader)\n",
        "    \n",
        "    print(\" \"*100)\n",
        "    print(f\"At epoches = {epoch}, epoch_loss = {epoch_loss}\")\n",
        "print('Finished Training')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sCoDC1UTrduR",
        "outputId": "3761550c-9c4e-4298-990f-a34b1f0302b3"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "                                                                                                    \n",
            "At epoches = 0, epoch_loss = 0.13202607991695403\n",
            "                                                                                                    \n",
            "At epoches = 1, epoch_loss = 0.05979759499082963\n",
            "                                                                                                    \n",
            "At epoches = 2, epoch_loss = 0.042649092454711594\n",
            "                                                                                                    \n",
            "At epoches = 3, epoch_loss = 0.03370971393932899\n",
            "                                                                                                    \n",
            "At epoches = 4, epoch_loss = 0.027426083020865917\n",
            "                                                                                                    \n",
            "At epoches = 5, epoch_loss = 0.022993210983276367\n",
            "                                                                                                    \n",
            "At epoches = 6, epoch_loss = 0.020069794401029746\n",
            "                                                                                                    \n",
            "At epoches = 7, epoch_loss = 0.0175409536977609\n",
            "                                                                                                    \n",
            "At epoches = 8, epoch_loss = 0.015502276108165581\n",
            "                                                                                                    \n",
            "At epoches = 9, epoch_loss = 0.01400250104367733\n",
            "                                                                                                    \n",
            "At epoches = 10, epoch_loss = 0.012374157890677452\n",
            "                                                                                                    \n",
            "At epoches = 11, epoch_loss = 0.010710214449465275\n",
            "                                                                                                    \n",
            "At epoches = 12, epoch_loss = 0.01001749823341767\n",
            "                                                                                                    \n",
            "At epoches = 13, epoch_loss = 0.00902401109834512\n",
            "                                                                                                    \n",
            "At epoches = 14, epoch_loss = 0.00796964087933302\n",
            "                                                                                                    \n",
            "At epoches = 15, epoch_loss = 0.00821753719896078\n",
            "                                                                                                    \n",
            "At epoches = 16, epoch_loss = 0.006429226519664128\n",
            "                                                                                                    \n",
            "At epoches = 17, epoch_loss = 0.006867538103957971\n",
            "                                                                                                    \n",
            "At epoches = 18, epoch_loss = 0.006092469908793767\n",
            "                                                                                                    \n",
            "At epoches = 19, epoch_loss = 0.006551651221017043\n",
            "                                                                                                    \n",
            "At epoches = 20, epoch_loss = 0.005579325245320797\n",
            "                                                                                                    \n",
            "At epoches = 21, epoch_loss = 0.005131421340505282\n",
            "                                                                                                    \n",
            "At epoches = 22, epoch_loss = 0.0050969478353858\n",
            "                                                                                                    \n",
            "At epoches = 23, epoch_loss = 0.0046178102061152455\n",
            "                                                                                                    \n",
            "At epoches = 24, epoch_loss = 0.004302948740124703\n",
            "                                                                                                    \n",
            "At epoches = 25, epoch_loss = 0.004506322626769543\n",
            "                                                                                                    \n",
            "At epoches = 26, epoch_loss = 0.004190635401507219\n",
            "                                                                                                    \n",
            "At epoches = 27, epoch_loss = 0.0037902827436725297\n",
            "                                                                                                    \n",
            "At epoches = 28, epoch_loss = 0.003806226332485676\n",
            "                                                                                                    \n",
            "At epoches = 29, epoch_loss = 0.003963270287712415\n",
            "                                                                                                    \n",
            "At epoches = 30, epoch_loss = 0.0034002232114473977\n",
            "                                                                                                    \n",
            "At epoches = 31, epoch_loss = 0.003633171315987905\n",
            "                                                                                                    \n",
            "At epoches = 32, epoch_loss = 0.0032362938409050304\n",
            "                                                                                                    \n",
            "At epoches = 33, epoch_loss = 0.0033369722430904706\n",
            "                                                                                                    \n",
            "At epoches = 34, epoch_loss = 0.0036877823318044343\n",
            "                                                                                                    \n",
            "At epoches = 35, epoch_loss = 0.0029655890598893164\n",
            "                                                                                                    \n",
            "At epoches = 36, epoch_loss = 0.0030480255896846455\n",
            "                                                                                                    \n",
            "At epoches = 37, epoch_loss = 0.0031287139197190604\n",
            "                                                                                                    \n",
            "At epoches = 38, epoch_loss = 0.0029846167728304865\n",
            "                                                                                                    \n",
            "At epoches = 39, epoch_loss = 0.003112588486075401\n",
            "                                                                                                    \n",
            "At epoches = 40, epoch_loss = 0.003529689418276151\n",
            "                                                                                                    \n",
            "At epoches = 41, epoch_loss = 0.0029470820978283884\n",
            "                                                                                                    \n",
            "At epoches = 42, epoch_loss = 0.0026786463290452956\n",
            "                                                                                                    \n",
            "At epoches = 43, epoch_loss = 0.0026342141821980476\n",
            "                                                                                                    \n",
            "At epoches = 44, epoch_loss = 0.002925485659142335\n",
            "                                                                                                    \n",
            "At epoches = 45, epoch_loss = 0.0025268822342157365\n",
            "                                                                                                    \n",
            "At epoches = 46, epoch_loss = 0.00300927374313275\n",
            "                                                                                                    \n",
            "At epoches = 47, epoch_loss = 0.002658538846671581\n",
            "                                                                                                    \n",
            "At epoches = 48, epoch_loss = 0.002574020386735598\n",
            "                                                                                                    \n",
            "At epoches = 49, epoch_loss = 0.002644906064371268\n",
            "                                                                                                    \n",
            "At epoches = 50, epoch_loss = 0.002158681886891524\n",
            "                                                                                                    \n",
            "At epoches = 51, epoch_loss = 0.002759033128619194\n",
            "                                                                                                    \n",
            "At epoches = 52, epoch_loss = 0.0027334587727983794\n",
            "                                                                                                    \n",
            "At epoches = 53, epoch_loss = 0.002070250093440215\n",
            "                                                                                                    \n",
            "At epoches = 54, epoch_loss = 0.002548331864674886\n",
            "                                                                                                    \n",
            "At epoches = 55, epoch_loss = 0.0023409994900226595\n",
            "                                                                                                    \n",
            "At epoches = 56, epoch_loss = 0.0026105323856075607\n",
            "                                                                                                    \n",
            "At epoches = 57, epoch_loss = 0.00218076785604159\n",
            "                                                                                                    \n",
            "At epoches = 58, epoch_loss = 0.0020565125872691473\n",
            "                                                                                                    \n",
            "At epoches = 59, epoch_loss = 0.002333274078865846\n",
            "                                                                                                    \n",
            "At epoches = 60, epoch_loss = 0.0026419662977258365\n",
            "                                                                                                    \n",
            "At epoches = 61, epoch_loss = 0.0020444587116440135\n",
            "                                                                                                    \n",
            "At epoches = 62, epoch_loss = 0.001923705516755581\n",
            "                                                                                                    \n",
            "At epoches = 63, epoch_loss = 0.0024189790690938634\n",
            "                                                                                                    \n",
            "At epoches = 64, epoch_loss = 0.0024659769997000696\n",
            "                                                                                                    \n",
            "At epoches = 65, epoch_loss = 0.0018426801845431327\n",
            "                                                                                                    \n",
            "At epoches = 66, epoch_loss = 0.0022879332825541496\n",
            "                                                                                                    \n",
            "At epoches = 67, epoch_loss = 0.002038249322275321\n",
            "                                                                                                    \n",
            "At epoches = 68, epoch_loss = 0.0022648272678256033\n",
            "                                                                                                    \n",
            "At epoches = 69, epoch_loss = 0.0019885029926896094\n",
            "                                                                                                    \n",
            "At epoches = 70, epoch_loss = 0.0021178132812182107\n",
            "                                                                                                    \n",
            "At epoches = 71, epoch_loss = 0.002132217241823673\n",
            "                                                                                                    \n",
            "At epoches = 72, epoch_loss = 0.0022731676936149597\n",
            "                                                                                                    \n",
            "At epoches = 73, epoch_loss = 0.001820162582397461\n",
            "                                                                                                    \n",
            "At epoches = 74, epoch_loss = 0.001548584771156311\n",
            "                                                                                                    \n",
            "At epoches = 75, epoch_loss = 0.0018965842043360075\n",
            "                                                                                                    \n",
            "At epoches = 76, epoch_loss = 0.002481830829381943\n",
            "                                                                                                    \n",
            "At epoches = 77, epoch_loss = 0.00198640462209781\n",
            "                                                                                                    \n",
            "At epoches = 78, epoch_loss = 0.001947495366136233\n",
            "                                                                                                    \n",
            "At epoches = 79, epoch_loss = 0.002151211625337601\n",
            "                                                                                                    \n",
            "At epoches = 80, epoch_loss = 0.0021386549095312756\n",
            "                                                                                                    \n",
            "At epoches = 81, epoch_loss = 0.0019890890434384346\n",
            "                                                                                                    \n",
            "At epoches = 82, epoch_loss = 0.0019356322978933652\n",
            "                                                                                                    \n",
            "At epoches = 83, epoch_loss = 0.0019926460146903993\n",
            "                                                                                                    \n",
            "At epoches = 84, epoch_loss = 0.002265253412723541\n",
            "                                                                                                    \n",
            "At epoches = 85, epoch_loss = 0.0015974433397253354\n",
            "                                                                                                    \n",
            "At epoches = 86, epoch_loss = 0.001731886400282383\n",
            "                                                                                                    \n",
            "At epoches = 87, epoch_loss = 0.001923800334831079\n",
            "                                                                                                    \n",
            "At epoches = 88, epoch_loss = 0.001849161524573962\n",
            "                                                                                                    \n",
            "At epoches = 89, epoch_loss = 0.0018568884635965029\n",
            "                                                                                                    \n",
            "At epoches = 90, epoch_loss = 0.0021309943035244944\n",
            "                                                                                                    \n",
            "At epoches = 91, epoch_loss = 0.0025132694671551386\n",
            "                                                                                                    \n",
            "At epoches = 92, epoch_loss = 0.0014970742464065552\n",
            "                                                                                                    \n",
            "At epoches = 93, epoch_loss = 0.0019165799558162689\n",
            "                                                                                                    \n",
            "At epoches = 94, epoch_loss = 0.0022925354957580566\n",
            "                                                                                                    \n",
            "At epoches = 95, epoch_loss = 0.0018721288417776426\n",
            "                                                                                                    \n",
            "At epoches = 96, epoch_loss = 0.0018450704167286556\n",
            "                                                                                                    \n",
            "At epoches = 97, epoch_loss = 0.0021297513579328856\n",
            "                                                                                                    \n",
            "At epoches = 98, epoch_loss = 0.0015378367294867834\n",
            "                                                                                                    \n",
            "At epoches = 99, epoch_loss = 0.0022784166405598323\n",
            "Finished Training\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_outputs = model(x_train)\n",
        "test_outputs = model(x_test)\n",
        "train_tsne_embeds = tsne.fit_transform(train_outputs.cpu().detach().numpy())\n",
        "test_tsne_embeds = tsne.fit_transform(test_outputs.cpu().detach().numpy())\n",
        "\n",
        "scatter(train_tsne_embeds, y_train.cpu().numpy(), subtitle=f'online hard TNN distribution (train set)')\n",
        "scatter(test_tsne_embeds, y_test.cpu().numpy(), subtitle=f'online hard TNN distribution (test set)')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "Km4EPwtZroxd",
        "outputId": "4fe01944-c294-4399-b1a0-a5a345317896"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:783: FutureWarning: The default initialization in TSNE will change from 'random' to 'pca' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:793: FutureWarning: The default learning rate in TSNE will change from 200.0 to 'auto' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:783: FutureWarning: The default initialization in TSNE will change from 'random' to 'pca' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/manifold/_t_sne.py:793: FutureWarning: The default learning rate in TSNE will change from 200.0 to 'auto' in 1.2.\n",
            "  FutureWarning,\n",
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:17: DeprecationWarning: `np.int` is a deprecated alias for the builtin `int`. To silence this warning, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.\n",
            "Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        ""
      ],
      "metadata": {
        "id": "Letcgw-fruY_"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}