09_blurr_sequence_classification_with_categorical_features.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "09-blurr-sequence-classification-with-categorical-features.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "a5649914011646fdbce1070f2c76489f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_ff68d3081b1444bcb87213f2f26a08b3",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_167a49eb965f43d1ad97bd2645334e99",
              "IPY_MODEL_9a03328ea3f4438c850c97bf2d5b8cdb",
              "IPY_MODEL_7813f8692aff41209e59d64b6ce3ff2c"
            ]
          }
        },
        "ff68d3081b1444bcb87213f2f26a08b3": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "167a49eb965f43d1ad97bd2645334e99": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_e981d2f3f8db403da142bcb205206202",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "Downloading: 100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_ff048100e51740f7ae55cea426ad0cdd"
          }
        },
        "9a03328ea3f4438c850c97bf2d5b8cdb": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_3f0eabf891f14b9abde4a93d815bafa1",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 570,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 570,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_fe34394b849443258eccc33fba8b80d0"
          }
        },
        "7813f8692aff41209e59d64b6ce3ff2c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_bcd57bfe5e694367a2cf50e5121bba88",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 570/570 [00:00&lt;00:00, 11.6kB/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_dad2a0ee71294ab9ad8c580fd81feead"
          }
        },
        "e981d2f3f8db403da142bcb205206202": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "ff048100e51740f7ae55cea426ad0cdd": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "3f0eabf891f14b9abde4a93d815bafa1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "fe34394b849443258eccc33fba8b80d0": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "bcd57bfe5e694367a2cf50e5121bba88": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "dad2a0ee71294ab9ad8c580fd81feead": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "90f7c5f5431943ccbab408769c270612": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_b1cb7a3de86d435dbee22cae3804f134",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_a3ed9f3c92b34f98baa02cdcfa083b61",
              "IPY_MODEL_731e3d94f6c14a72b6bc611a025fd6f7",
              "IPY_MODEL_5ce3486250674cbcae54131894ddb4a9"
            ]
          }
        },
        "b1cb7a3de86d435dbee22cae3804f134": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "a3ed9f3c92b34f98baa02cdcfa083b61": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_1a52c31d80d64f85983182210fe7561a",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "Downloading: 100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_e4ec53bf95ce4ff59219dc5334868e32"
          }
        },
        "731e3d94f6c14a72b6bc611a025fd6f7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_15a83fbdae604433bfabc4dca2fda229",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 28,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 28,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_85a37bbaa92e425cbd2de6e97c413326"
          }
        },
        "5ce3486250674cbcae54131894ddb4a9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_551838f1e3c942e6be7d5bc5dbe57818",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 28.0/28.0 [00:00&lt;00:00, 704B/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_81933d862c9941ae821d29270447262a"
          }
        },
        "1a52c31d80d64f85983182210fe7561a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "e4ec53bf95ce4ff59219dc5334868e32": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "15a83fbdae604433bfabc4dca2fda229": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "85a37bbaa92e425cbd2de6e97c413326": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "551838f1e3c942e6be7d5bc5dbe57818": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "81933d862c9941ae821d29270447262a": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "d5a6dd207ea442b281546a0f12ac56b0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_86417827887d454fa928e60f90d4d0a8",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_2b0056d1a4a948d0a8518dd7ba5a8f3e",
              "IPY_MODEL_d2d31315536a461fb2e03a5270c53ea0",
              "IPY_MODEL_8ffe668c141a41799b45d776c1521d3a"
            ]
          }
        },
        "86417827887d454fa928e60f90d4d0a8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "2b0056d1a4a948d0a8518dd7ba5a8f3e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_989ceaad8e904cb4bf87cf7e9e3e3271",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "Downloading: 100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_b2f5a984a7184e50b397a35aae34055b"
          }
        },
        "d2d31315536a461fb2e03a5270c53ea0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_190d4ea9e5a14e98b885d9cd7e85dc83",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 231508,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 231508,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_28f9b183eddd4c0d9872ebcbc2f68511"
          }
        },
        "8ffe668c141a41799b45d776c1521d3a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_9f9b9b0d1d7242b0afce58b66872a30f",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 232k/232k [00:00&lt;00:00, 1.06MB/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_1fc19537a82a4b5dbabfb1c0cc1d6ce2"
          }
        },
        "989ceaad8e904cb4bf87cf7e9e3e3271": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "b2f5a984a7184e50b397a35aae34055b": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "190d4ea9e5a14e98b885d9cd7e85dc83": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "28f9b183eddd4c0d9872ebcbc2f68511": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "9f9b9b0d1d7242b0afce58b66872a30f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "1fc19537a82a4b5dbabfb1c0cc1d6ce2": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "6fe57dff64b7443c85ae55238d6d35e7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_bb09cd06247f4451a092dc31cce91ba1",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_0fbf44a517604ce4b77e5086e05994c9",
              "IPY_MODEL_b8d483e389e1412398f0446158742442",
              "IPY_MODEL_f79ad49a5ec649f1861d8a8ce9a17cb2"
            ]
          }
        },
        "bb09cd06247f4451a092dc31cce91ba1": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "0fbf44a517604ce4b77e5086e05994c9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_bd0cb01e139b49e597e7b4ff8965d2ca",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "Downloading: 100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_4c9be81ea9a24d59874d2cca71eea01d"
          }
        },
        "b8d483e389e1412398f0446158742442": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_fc2f2f4cdf694baea419c789b328eaa6",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 466062,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 466062,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_3b3a83e7ebbb45839ea6d8ecbc699443"
          }
        },
        "f79ad49a5ec649f1861d8a8ce9a17cb2": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_be2d5d890b1f4770a9903d5a6381b6eb",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 466k/466k [00:00&lt;00:00, 1.20MB/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_9331983d10834ab4a2b96e8189b3b941"
          }
        },
        "bd0cb01e139b49e597e7b4ff8965d2ca": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "4c9be81ea9a24d59874d2cca71eea01d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "fc2f2f4cdf694baea419c789b328eaa6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "3b3a83e7ebbb45839ea6d8ecbc699443": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "be2d5d890b1f4770a9903d5a6381b6eb": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "9331983d10834ab4a2b96e8189b3b941": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "e71f644f58d84a9a8cfed9b5fb39442b": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HBoxView",
            "_dom_classes": [],
            "_model_name": "HBoxModel",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "box_style": "",
            "layout": "IPY_MODEL_b4f5230c7837446ba7763b7ddfbfb13c",
            "_model_module": "@jupyter-widgets/controls",
            "children": [
              "IPY_MODEL_2f5d614f727545a6ac51dbc80b476665",
              "IPY_MODEL_e368a52ac81446e68081eaa2dd8893e5",
              "IPY_MODEL_a0d33240e0ef4be58e6c5947e6952be4"
            ]
          }
        },
        "b4f5230c7837446ba7763b7ddfbfb13c": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "2f5d614f727545a6ac51dbc80b476665": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_fc4b976642a84312975a970c55e7cf36",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": "Downloading: 100%",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_019d2585c873444b81c33ff9366b2229"
          }
        },
        "e368a52ac81446e68081eaa2dd8893e5": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "ProgressView",
            "style": "IPY_MODEL_70114102e9a04cf6861241736ca94bda",
            "_dom_classes": [],
            "description": "",
            "_model_name": "FloatProgressModel",
            "bar_style": "success",
            "max": 440473133,
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": 440473133,
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "orientation": "horizontal",
            "min": 0,
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_6f341e22fbd64160a7aa6a70f89d841d"
          }
        },
        "a0d33240e0ef4be58e6c5947e6952be4": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "HTMLView",
            "style": "IPY_MODEL_82cb796374954867b6554ce163508d81",
            "_dom_classes": [],
            "description": "",
            "_model_name": "HTMLModel",
            "placeholder": "​",
            "_view_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "value": " 440M/440M [00:15&lt;00:00, 27.0MB/s]",
            "_view_count": null,
            "_view_module_version": "1.5.0",
            "description_tooltip": null,
            "_model_module": "@jupyter-widgets/controls",
            "layout": "IPY_MODEL_49d39e8f1c614472af8ea068770d0567"
          }
        },
        "fc4b976642a84312975a970c55e7cf36": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "019d2585c873444b81c33ff9366b2229": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "70114102e9a04cf6861241736ca94bda": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "ProgressStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "bar_color": null,
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "6f341e22fbd64160a7aa6a70f89d841d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        },
        "82cb796374954867b6554ce163508d81": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_view_name": "StyleView",
            "_model_name": "DescriptionStyleModel",
            "description_width": "",
            "_view_module": "@jupyter-widgets/base",
            "_model_module_version": "1.5.0",
            "_view_count": null,
            "_view_module_version": "1.2.0",
            "_model_module": "@jupyter-widgets/controls"
          }
        },
        "49d39e8f1c614472af8ea068770d0567": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_view_name": "LayoutView",
            "grid_template_rows": null,
            "right": null,
            "justify_content": null,
            "_view_module": "@jupyter-widgets/base",
            "overflow": null,
            "_model_module_version": "1.2.0",
            "_view_count": null,
            "flex_flow": null,
            "width": null,
            "min_width": null,
            "border": null,
            "align_items": null,
            "bottom": null,
            "_model_module": "@jupyter-widgets/base",
            "top": null,
            "grid_column": null,
            "overflow_y": null,
            "overflow_x": null,
            "grid_auto_flow": null,
            "grid_area": null,
            "grid_template_columns": null,
            "flex": null,
            "_model_name": "LayoutModel",
            "justify_items": null,
            "grid_row": null,
            "max_height": null,
            "align_content": null,
            "visibility": null,
            "align_self": null,
            "height": null,
            "min_height": null,
            "padding": null,
            "grid_auto_rows": null,
            "grid_gap": null,
            "max_width": null,
            "order": null,
            "_view_module_version": "1.2.0",
            "grid_template_areas": null,
            "object_position": null,
            "object_fit": null,
            "grid_auto_columns": null,
            "margin": null,
            "display": null,
            "left": null
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G76d-9NS6FOq"
      },
      "source": [
        "# Blurr Sequence Classification with categorical and numerical features using BERT."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U-1NaME62ScR"
      },
      "source": [
        "This notebook is a replication of the http://mccormickml.com/2021/06/29/combining-categorical-numerical-features-with-bert/ but use blurr to showcase how to combine bert with categorical & numerical featuers with text in BERT.\n",
        "\n",
        "Additional Resources\n",
        "- https://github.com/georgian-io/Multimodal-Toolkit\n",
        "- https://medium.com/georgian-impact-blog/how-to-incorporate-tabular-data-with-huggingface-transformers-b70ac45fcfb4\n",
        "\n",
        "Next Step\n",
        "- Following this notebook https://colab.research.google.com/drive/1YRHK4HO8RktGzlYmGjBo056kzVD4_j9o#scrollTo=m7rr8z8x3TiE 3.3 Training Paremeters \n",
        "\n",
        "TODO \n",
        "- Callback, why partial on the optimizer\n",
        "- Does freezing the final layer followed by unfreeze works in transformer context??. I recall nobody adopted gradual unfreezing not adopted in practice. \n",
        "\n",
        "**Checklist** (1. Make it work using high/mid/low 2. Baseline)\n",
        "- [ ] BERT on Review Text Only (Sample Data)\n",
        "  - [ ] High Level (Accuracy - 0.83)\n",
        "  - [x] Mid Level (Accuracy - 0.914)\n",
        "  - [ ] Low Level\n",
        "- [ ] BERT on All features (Sample Data) - Unimodal approach (taking the categorical and numerical features as they are and just tokenizing them and just concatenating them to the text columns as extra text sentences is a strong baseline as referenced in the [multimodal-toolkit](https://github.com/georgian-io/Multimodal-Toolkit)\n",
        "  - [ ] High Level\n",
        "  - [ ] Mid Level\n",
        "  - [ ] Low Level"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kgma0J9p2hLm"
      },
      "source": [
        "## Outline\n",
        "\n",
        "- Installation\n",
        "- Womens Clothing Review Dataset\n",
        "  - Download the womens clothing data\n",
        "  - Load the csv\n",
        "- BERT on Review Text Only\n",
        "  - Using MidLevel API\n",
        "- Create the blearner for Sequence Classification\n",
        "- Fit OneCycle\n",
        "- Evaluate "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "57KpkHbR1-l6"
      },
      "source": [
        "%%capture\n",
        "%pip install nbdev -q --upgrade\n",
        "%pip install fastai -q --upgrade\n",
        "%pip install -q huggingface-hub>0.0.10\n",
        "%pip install ohmeow-blurr -q --upgrade"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cHQMruUe4MwU"
      },
      "source": [
        "%pip install nbverbose -Uq"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NvmIeclO3rJR"
      },
      "source": [
        "from datasets import load_dataset, concatenate_datasets"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kUjsShEL2Lht"
      },
      "source": [
        "#export\n",
        "import os\n",
        "\n",
        "from datasets import load_dataset, concatenate_datasets\n",
        "from transformers import *\n",
        "from fastai.text.all import *\n",
        "\n",
        "from blurr.utils import *\n",
        "from blurr.data.core import *\n",
        "from blurr.modeling.core import *\n",
        "\n",
        "from blurr.data.language_modeling import BertMLMStrategy, CausalLMStrategy\n",
        "from blurr.modeling.language_modeling import *\n",
        "\n",
        "from blurr.modeling.token_classification import *\n",
        "from blurr.modeling.question_answering import *\n",
        "from blurr.modeling.seq2seq.summarization import *\n",
        "from blurr.modeling.seq2seq.translation import *\n",
        "\n",
        "logging.set_verbosity_error()"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JYVAgKIs6YO1"
      },
      "source": [
        "import importlib\n",
        "from typing import List, Union\n",
        "def print_versions(\n",
        "    # A string of space delimited package names or a list of package names\n",
        "    packages:Union[str, List[str]]\n",
        "):\n",
        "    \"\"\" Prints the name and version of one or more packages in your environment\"\"\"\n",
        "    packages = packages.split(' ') if isinstance(packages, str) else packages\n",
        "\n",
        "    for item in packages:\n",
        "        item = item.strip()\n",
        "        # Provides the implementation of import statement and allows to import the defined packages to retrieve their versions.\n",
        "        print(f'{item}: {importlib.import_module(item).__version__}')"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nD6uYWgk38cn",
        "outputId": "9c9b221c-37e3-4735-eb93-77759c97c64a"
      },
      "source": [
        "#hide_input\n",
        "import pdb\n",
        "\n",
        "from fastcore.test import *\n",
        "from nbverbose.showdoc import show_doc\n",
        "\n",
        "os.environ[\"TOKENIZERS_PARALLELISM\"] = \"false\"\n",
        "print(\"Here's what we're running with ...\\n\")\n",
        "print_versions('torch fastai transformers')"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Here's what we're running with ...\n",
            "\n",
            "torch: 1.9.0+cu102\n",
            "fastai: 2.5.2\n",
            "transformers: 4.10.3\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AuYebBap3gtn",
        "outputId": "0515e604-cbd1-4d6f-bea7-212716dfe08c"
      },
      "source": [
        "print(f'Using GPU #{torch.cuda.current_device()}: {torch.cuda.get_device_name()}')"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Using GPU #0: Tesla K80\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3hmb5tAG596d"
      },
      "source": [
        "## Clothing Review Dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hVBPI1N17vD7"
      },
      "source": [
        "Based on the womens clothing reviews, we are trying to predict the sentiment behind the review (Recommended IND is the label where 1 indicates recommended & 0 indicates not recommended)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Eid3Uz2Z6l7B"
      },
      "source": [
        "### Download the csv file"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GvVahTHd6zmN"
      },
      "source": [
        "This dataset was scraped from an (un-specified) e-commerce website by Nick Brooks and made available on Kaggle [here](https://www.kaggle.com/nicapotato/womens-ecommerce-clothing-reviews)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Jb6YPEyK6Pir"
      },
      "source": [
        "import gdown"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 106
        },
        "id": "-NJjmNnZ6RFA",
        "outputId": "8f7c1376-1364-4b94-b504-96ec53f1bdb4"
      },
      "source": [
        "# Download the dataset from the drive.\n",
        "gdown.download('https://drive.google.com/uc?id=1ZYdt0zN4LjWqP3cQDblNhXjeohcryY5H', \n",
        "                'Womens Clothing E-Commerce Reviews.csv', \n",
        "                quiet=False)"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?id=1ZYdt0zN4LjWqP3cQDblNhXjeohcryY5H\n",
            "To: /content/Womens Clothing E-Commerce Reviews.csv\n",
            "8.48MB [00:00, 44.4MB/s]\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'Womens Clothing E-Commerce Reviews.csv'"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V6b2ZkSi6aPM"
      },
      "source": [
        "test_eq(Path('/content/Womens Clothing E-Commerce Reviews.csv').exists(), True)"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "69VDFVkf6o-5"
      },
      "source": [
        "### Load the csv "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WZGcjrxw7d9O"
      },
      "source": [
        "FILENAME = '/content/Womens Clothing E-Commerce Reviews.csv'\n",
        "df = pd.read_csv(FILENAME, index_col=0)"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mUEdGFHu8oaT",
        "outputId": "b89cbdf6-0440-45ec-c377-bdc92c9b9b88"
      },
      "source": [
        "len(df)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "23486"
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6MnEnX7l7Rbh",
        "outputId": "7d48cbc8-ba44-426b-a60d-6c213f6ea941"
      },
      "source": [
        "list(df.columns)"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['Clothing ID',\n",
              " 'Age',\n",
              " 'Title',\n",
              " 'Review Text',\n",
              " 'Rating',\n",
              " 'Recommended IND',\n",
              " 'Positive Feedback Count',\n",
              " 'Division Name',\n",
              " 'Department Name',\n",
              " 'Class Name']"
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bNauYujv9Ibd",
        "outputId": "52abb940-ebed-4dc9-fa5e-aacadc95ec00"
      },
      "source": [
        "df.dtypes"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Clothing ID                 int64\n",
              "Age                         int64\n",
              "Title                      object\n",
              "Review Text                object\n",
              "Rating                      int64\n",
              "Recommended IND             int64\n",
              "Positive Feedback Count     int64\n",
              "Division Name              object\n",
              "Department Name            object\n",
              "Class Name                 object\n",
              "dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 352
        },
        "id": "ICK6jFqh7GjK",
        "outputId": "46db915d-ba04-4e8a-9f3d-c9bba91edf15"
      },
      "source": [
        "df.head()"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Clothing ID</th>\n",
              "      <th>Age</th>\n",
              "      <th>Title</th>\n",
              "      <th>Review Text</th>\n",
              "      <th>Rating</th>\n",
              "      <th>Recommended IND</th>\n",
              "      <th>Positive Feedback Count</th>\n",
              "      <th>Division Name</th>\n",
              "      <th>Department Name</th>\n",
              "      <th>Class Name</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>767</td>\n",
              "      <td>33</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Absolutely wonderful - silky and sexy and comfortable</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>Initmates</td>\n",
              "      <td>Intimate</td>\n",
              "      <td>Intimates</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1080</td>\n",
              "      <td>34</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Love this dress!  it's sooo pretty.  i happened to find it in a store, and i'm glad i did bc i never would have ordered it online bc it's petite.  i bought a petite and am 5'8\".  i love the length on me- hits just a little below the knee.  would definitely be a true midi on someone who is truly petite.</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>General</td>\n",
              "      <td>Dresses</td>\n",
              "      <td>Dresses</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1077</td>\n",
              "      <td>60</td>\n",
              "      <td>Some major design flaws</td>\n",
              "      <td>I had such high hopes for this dress and really wanted it to work for me. i initially ordered the petite small (my usual size) but i found this to be outrageously small. so small in fact that i could not zip it up! i reordered it in petite medium, which was just ok. overall, the top half was comfortable and fit nicely, but the bottom half had a very tight under layer and several somewhat cheap (net) over layers. imo, a major design flaw was the net over layer sewn directly into the zipper - it c</td>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>General</td>\n",
              "      <td>Dresses</td>\n",
              "      <td>Dresses</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1049</td>\n",
              "      <td>50</td>\n",
              "      <td>My favorite buy!</td>\n",
              "      <td>I love, love, love this jumpsuit. it's fun, flirty, and fabulous! every time i wear it, i get nothing but great compliments!</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>General Petite</td>\n",
              "      <td>Bottoms</td>\n",
              "      <td>Pants</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>847</td>\n",
              "      <td>47</td>\n",
              "      <td>Flattering shirt</td>\n",
              "      <td>This shirt is very flattering to all due to the adjustable front tie. it is the perfect length to wear with leggings and it is sleeveless so it pairs well with any cardigan. love this shirt!!!</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>General</td>\n",
              "      <td>Tops</td>\n",
              "      <td>Blouses</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   Clothing ID  Age  ... Department Name Class Name\n",
              "0          767   33  ...        Intimate  Intimates\n",
              "1         1080   34  ...         Dresses    Dresses\n",
              "2         1077   60  ...         Dresses    Dresses\n",
              "3         1049   50  ...         Bottoms      Pants\n",
              "4          847   47  ...            Tops    Blouses\n",
              "\n",
              "[5 rows x 10 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VWi5d-Dx7dRb",
        "outputId": "3a0ffc61-2af5-4e39-cce7-14631b126640"
      },
      "source": [
        "# Determine the columns containing null/missing values\n",
        "df.isnull().sum()"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Clothing ID                   0\n",
              "Age                           0\n",
              "Title                      3810\n",
              "Review Text                 845\n",
              "Rating                        0\n",
              "Recommended IND               0\n",
              "Positive Feedback Count       0\n",
              "Division Name                14\n",
              "Department Name              14\n",
              "Class Name                   14\n",
              "dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "F45G9-cI8qbS"
      },
      "source": [
        "# Lets drop the rows containing null values.\n",
        "df.dropna(inplace=True)"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HpEq8me68zBp",
        "outputId": "c3319efb-a4cf-4e63-ccfc-f1065da86494"
      },
      "source": [
        "len(df)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "19662"
            ]
          },
          "metadata": {},
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w3zKb4z082Dd"
      },
      "source": [
        "It seems we are dropping too many samples due to missing values. Instead of dropping the missing values rows, can we fill the missing values. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BNfvhrsq_qpc"
      },
      "source": [
        "Let's reload the df from the csv file."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yFyXVgJa_yOA"
      },
      "source": [
        "df = pd.read_csv(FILENAME, index_col=0)"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "515VWsWr_zGv"
      },
      "source": [
        "test_eq(len(df), 23486)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "19vTG_6Z9OlY",
        "outputId": "cb80e0ac-f2d0-49ea-b6a4-3a904e5776f1"
      },
      "source": [
        "df.head(1).values"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[767, 33, nan,\n",
              "        'Absolutely wonderful - silky and sexy and comfortable', 4, 1, 0,\n",
              "        'Initmates', 'Intimate', 'Intimates']], dtype=object)"
            ]
          },
          "metadata": {},
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y6c3wIw9-Ecx",
        "outputId": "ac0881a8-9a79-4497-a9cd-ed6a5fc602f1"
      },
      "source": [
        "df.Rating.value_counts()"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "5    13131\n",
              "4     5077\n",
              "3     2871\n",
              "2     1565\n",
              "1      842\n",
              "Name: Rating, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EM_3QziV-GtP",
        "outputId": "2bf6fc03-3ee2-4ad1-f539-a5f784700691"
      },
      "source": [
        "df['Division Name'].value_counts()"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "General           13850\n",
              "General Petite     8120\n",
              "Initmates          1502\n",
              "Name: Division Name, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PouJWXzO-Q8u",
        "outputId": "91576ca4-ef0d-428f-9adc-0fe3002a4749"
      },
      "source": [
        "df['Department Name'].value_counts()"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Tops        10468\n",
              "Dresses      6319\n",
              "Bottoms      3799\n",
              "Intimate     1735\n",
              "Jackets      1032\n",
              "Trend         119\n",
              "Name: Department Name, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-5GmTinS-Xzy",
        "outputId": "25e4d500-20f3-462c-94bc-43c2df66b0fa"
      },
      "source": [
        "df['Class Name'].value_counts()"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Dresses           6319\n",
              "Knits             4843\n",
              "Blouses           3097\n",
              "Sweaters          1428\n",
              "Pants             1388\n",
              "Jeans             1147\n",
              "Fine gauge        1100\n",
              "Skirts             945\n",
              "Jackets            704\n",
              "Lounge             691\n",
              "Swim               350\n",
              "Outerwear          328\n",
              "Shorts             317\n",
              "Sleep              228\n",
              "Legwear            165\n",
              "Intimates          154\n",
              "Layering           146\n",
              "Trend              119\n",
              "Casual bottoms       2\n",
              "Chemises             1\n",
              "Name: Class Name, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FWbusdZKCgvN",
        "outputId": "68142f2e-48c0-4076-8e5f-2a5cbc68c428"
      },
      "source": [
        "df.isna().sum()"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Clothing ID                   0\n",
              "Age                           0\n",
              "Title                      3810\n",
              "Review Text                 845\n",
              "Rating                        0\n",
              "Recommended IND               0\n",
              "Positive Feedback Count       0\n",
              "Division Name                14\n",
              "Department Name              14\n",
              "Class Name                   14\n",
              "dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0JIJXBYRDDOA"
      },
      "source": [
        "Let's fill the missing valeues(NaN) with empty."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bhXzupVLDJVG"
      },
      "source": [
        "df = df.fillna('')"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ucu7o5WADNlV"
      },
      "source": [
        "The approach described in the [article](http://mccormickml.com/2021/06/29/combining-categorical-numerical-features-with-bert/#31-all-features-to-text) to handle non-text features is to convert them into text and concatenate them with the review text. Instead of that we will instead try with the approach of \"mark_fields\"[1] that is done in fastai. Basically it will prepend the field text along with a token indicating which field this is from. \n",
        "\n",
        "Ex: fld 1 this is a bert example fld 2 this is the second sentence \n",
        "\n",
        "__Approach 1__\n",
        "But BERT itself adds some special tokens, so we can instead try [CLS] field 1 text [SEP] field 2 text [SEP] (eg: Dresses [SEP] General [SEP] 34 [SEP] 5...  \n",
        "\n",
        "[1] https://docs.fast.ai/text.data.html#TextBlock.from_df\n",
        "\n",
        "__Approach 2__\n",
        "Create a special tokens for each of the feature column and prepend it based on the idea from [ME ULMFiT](https://www.novetta.com/2019/03/introducing_me_ulmfit/). [CLS] [FLD1] field 1 text [FLD2] field 2 text [INP] text input [SEP] \n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o2AFz_j-Q1UL"
      },
      "source": [
        "## BERT on Review Text Only (Sample Data)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GwrOQZq5Lh7Z",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4703b602-ffcd-4a92-ab97-997da7335168"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "SEED=42\n",
        "train_val_df, test_df = train_test_split(df, test_size=0.1, random_state=SEED)\n",
        "len(train_val_df), len(test_df)"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(21137, 2349)"
            ]
          },
          "metadata": {},
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oY6xw3LcLzH9"
      },
      "source": [
        "#### Sample Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2QpH_w9MT9cr",
        "outputId": "8208bbea-d13b-43d8-cd61-b3687a5609cd"
      },
      "source": [
        "# Use sample data\n",
        "train_val_df = train_val_df.sample(frac=0.1); len(train_val_df)"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "2114"
            ]
          },
          "metadata": {},
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Dn7LrvAnKBVn"
      },
      "source": [
        "### Using the Midlevel API "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zn8sR2wCFCrx"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 176,
          "referenced_widgets": [
            "a5649914011646fdbce1070f2c76489f",
            "ff68d3081b1444bcb87213f2f26a08b3",
            "167a49eb965f43d1ad97bd2645334e99",
            "9a03328ea3f4438c850c97bf2d5b8cdb",
            "7813f8692aff41209e59d64b6ce3ff2c",
            "e981d2f3f8db403da142bcb205206202",
            "ff048100e51740f7ae55cea426ad0cdd",
            "3f0eabf891f14b9abde4a93d815bafa1",
            "fe34394b849443258eccc33fba8b80d0",
            "bcd57bfe5e694367a2cf50e5121bba88",
            "dad2a0ee71294ab9ad8c580fd81feead",
            "90f7c5f5431943ccbab408769c270612",
            "b1cb7a3de86d435dbee22cae3804f134",
            "a3ed9f3c92b34f98baa02cdcfa083b61",
            "731e3d94f6c14a72b6bc611a025fd6f7",
            "5ce3486250674cbcae54131894ddb4a9",
            "1a52c31d80d64f85983182210fe7561a",
            "e4ec53bf95ce4ff59219dc5334868e32",
            "15a83fbdae604433bfabc4dca2fda229",
            "85a37bbaa92e425cbd2de6e97c413326",
            "551838f1e3c942e6be7d5bc5dbe57818",
            "81933d862c9941ae821d29270447262a",
            "d5a6dd207ea442b281546a0f12ac56b0",
            "86417827887d454fa928e60f90d4d0a8",
            "2b0056d1a4a948d0a8518dd7ba5a8f3e",
            "d2d31315536a461fb2e03a5270c53ea0",
            "8ffe668c141a41799b45d776c1521d3a",
            "989ceaad8e904cb4bf87cf7e9e3e3271",
            "b2f5a984a7184e50b397a35aae34055b",
            "190d4ea9e5a14e98b885d9cd7e85dc83",
            "28f9b183eddd4c0d9872ebcbc2f68511",
            "9f9b9b0d1d7242b0afce58b66872a30f",
            "1fc19537a82a4b5dbabfb1c0cc1d6ce2",
            "6fe57dff64b7443c85ae55238d6d35e7",
            "bb09cd06247f4451a092dc31cce91ba1",
            "0fbf44a517604ce4b77e5086e05994c9",
            "b8d483e389e1412398f0446158742442",
            "f79ad49a5ec649f1861d8a8ce9a17cb2",
            "bd0cb01e139b49e597e7b4ff8965d2ca",
            "4c9be81ea9a24d59874d2cca71eea01d",
            "fc2f2f4cdf694baea419c789b328eaa6",
            "3b3a83e7ebbb45839ea6d8ecbc699443",
            "be2d5d890b1f4770a9903d5a6381b6eb",
            "9331983d10834ab4a2b96e8189b3b941",
            "e71f644f58d84a9a8cfed9b5fb39442b",
            "b4f5230c7837446ba7763b7ddfbfb13c",
            "2f5d614f727545a6ac51dbc80b476665",
            "e368a52ac81446e68081eaa2dd8893e5",
            "a0d33240e0ef4be58e6c5947e6952be4",
            "fc4b976642a84312975a970c55e7cf36",
            "019d2585c873444b81c33ff9366b2229",
            "70114102e9a04cf6861241736ca94bda",
            "6f341e22fbd64160a7aa6a70f89d841d",
            "82cb796374954867b6554ce163508d81",
            "49d39e8f1c614472af8ea068770d0567"
          ]
        },
        "id": "dw3pJ1ReJllf",
        "outputId": "9b86a71a-a8c4-47c9-f751-7d1da0d75531"
      },
      "source": [
        "from transformers import AutoModelForSequenceClassification\n",
        "model_cls = AutoModelForSequenceClassification\n",
        "\n",
        "pretrained_model_name =  \"bert-base-uncased\" #\"distilroberta-base\" # \"distilbert-base-uncased\" \"bert-base-uncased\"\n",
        "hf_arch, hf_config, hf_tokenizer, hf_model = BLURR.get_hf_objects(pretrained_model_name, model_cls=model_cls)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "a5649914011646fdbce1070f2c76489f",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "Downloading:   0%|          | 0.00/570 [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "90f7c5f5431943ccbab408769c270612",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "Downloading:   0%|          | 0.00/28.0 [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "d5a6dd207ea442b281546a0f12ac56b0",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "Downloading:   0%|          | 0.00/232k [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "6fe57dff64b7443c85ae55238d6d35e7",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "Downloading:   0%|          | 0.00/466k [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "e71f644f58d84a9a8cfed9b5fb39442b",
              "version_minor": 0,
              "version_major": 2
            },
            "text/plain": [
              "Downloading:   0%|          | 0.00/440M [00:00<?, ?B/s]"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rm3Kmfu2Jrdu"
      },
      "source": [
        "# Set the number of labels \n",
        "hf_config.num_labels = df['Recommended IND'].nunique()"
      ],
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vacVZ39hKdnC"
      },
      "source": [
        "#### DataBlock and DataLoaders"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uSTT9Nx3K_ZV"
      },
      "source": [
        "blocks = (HF_TextBlock(hf_arch, hf_config, hf_tokenizer, hf_model), CategoryBlock)\n",
        "db = DataBlock(blocks=blocks, get_x=ColReader('Review Text'), get_y=ColReader('Recommended IND'), \n",
        "               splitter=RandomSplitter(valid_pct=0.1, seed=SEED))"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Z9GBQ4CUM1Kn",
        "outputId": "f39325e1-6073-4d21-81b7-c35bbc0b8b3b"
      },
      "source": [
        "train_val_df.dtypes"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Clothing ID                 int64\n",
              "Age                         int64\n",
              "Title                      object\n",
              "Review Text                object\n",
              "Rating                      int64\n",
              "Recommended IND             int64\n",
              "Positive Feedback Count     int64\n",
              "Division Name              object\n",
              "Department Name            object\n",
              "Class Name                 object\n",
              "dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 33
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AChay81VMpP9"
      },
      "source": [
        "#TfmdLists - Uncomment the summary to see the whole process\n",
        "#db.summary(train_val_df[:100])"
      ],
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fDZfcjrlMtO6"
      },
      "source": [
        "dls = db.dataloaders(train_val_df, bs=32)"
      ],
      "execution_count": 37,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "643n_HFUPf6r",
        "outputId": "90e14049-96e1-413d-ce21-9cc3bc3896a5"
      },
      "source": [
        "b = dls.one_batch(); len(b), len(b[0]['input_ids']), b[0]['input_ids'].shape, len(b[1])"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(2, 32, torch.Size([32, 152]), 32)"
            ]
          },
          "metadata": {},
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QKGiP0ZZP_gJ",
        "outputId": "7d2937c6-d5db-4499-98d7-cd369ab28b46"
      },
      "source": [
        "explode_types(b)"
      ],
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{tuple: [dict, fastai.torch_core.TensorCategory]}"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 210
        },
        "id": "jvAEVw2RQVwK",
        "outputId": "c6174138-4ae0-4025-9a0f-8b45dea922df"
      },
      "source": [
        "dls.show_batch(dataloaders=dls, max_n=2, trunc_at=500)"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>i just got this in the green and i'd say the color is deeper / darker than shown. more of a turquoise - green, or muted teal color. it's the densest, thickest cashmere ( it's 100 % cashmere ) i've felt in a while and it's so soft. but, it just doesn't look good on me. i'm 5'3 \" and about 105 lbs, and usually wear xspetite. i got this in the xs regular hoping it would be long enough for leggings. and i don't think it is in the back. plus it's just too bell shaped and i'm slightly pear - shaped al</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>this dress is indeed so lovely in person. i am so glad i ordered 2 sizes ( 0p and 2p ) based on the straighter - looking cut. i am 115 lbs, 5'1. 5 \", adn 30dd bra size. verdict, the 2p fit better. why, the cehst area is a little snug and hte buttons kept trying to open ( and did # i would have ahd to leave it pen and wear a tank under this dress... not an option with such a unique piece. other things : color : wow, very nice in person, i was afraid that the colr would wash me out, but no, i have</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O9WM71VZQdXN"
      },
      "source": [
        "#### Training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6DxMcUUERfyx"
      },
      "source": [
        "model = HF_BaseModelWrapper(hf_model)"
      ],
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        },
        "id": "E-3LspScRjfU",
        "outputId": "e7cab52a-28a4-4c85-e96e-6ae4204432a4"
      },
      "source": [
        "doc(Learner)"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<h2 id=\"Learner\" class=\"doc_header\"><code>class</code> <code>Learner</code><a href=\"https://github.com/fastai/fastai/tree/master/fastai/learner.py#L83\" class=\"source_link\" style=\"float:right\">[source]</a></h2><blockquote><p><code>Learner</code>(<strong><code>dls</code></strong>, <strong><code>model</code></strong>, <strong><code>loss_func</code></strong>=<em><code>None</code></em>, <strong><code>opt_func</code></strong>=<em><code>Adam</code></em>, <strong><code>lr</code></strong>=<em><code>0.001</code></em>, <strong><code>splitter</code></strong>=<em><code>trainable_params</code></em>, <strong><code>cbs</code></strong>=<em><code>None</code></em>, <strong><code>metrics</code></strong>=<em><code>None</code></em>, <strong><code>path</code></strong>=<em><code>None</code></em>, <strong><code>model_dir</code></strong>=<em><code>'models'</code></em>, <strong><code>wd</code></strong>=<em><code>None</code></em>, <strong><code>wd_bn_bias</code></strong>=<em><code>False</code></em>, <strong><code>train_bn</code></strong>=<em><code>True</code></em>, <strong><code>moms</code></strong>=<em><code>(0.95, 0.85, 0.95)</code></em>) :: <code>GetAttr</code></p>\n",
              "</blockquote>\n",
              "<p>Group together a <code>model</code>, some <code>dls</code> and a <code>loss_func</code> to handle training</p>\n",
              "<p><a href=\"https://docs.fast.ai/learner#Learner\" target=\"_blank\" rel=\"noreferrer noopener\">Show in docs</a></p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "28cRlLtRR_PV"
      },
      "source": [
        "#TODO - Callback, why partial on the optimizer "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NsNuMaDPRlEE"
      },
      "source": [
        "learn = Learner(dls, model, loss_func=CrossEntropyLossFlat(), \n",
        "                opt_func=partial(OptimWrapper, opt=torch.optim.Adam),\n",
        "                metrics=accuracy, cbs=[HF_BaseModelCallback], splitter=hf_splitter).to_fp16()\n",
        "learn.freeze()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iUZNgEHhSK3H"
      },
      "source": [
        "# Uncomment and see the architecture and learner details.\n",
        "# learn.summary()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eODj-I5pSS83",
        "outputId": "0c4511ce-6c3b-4997-af5a-9abe72e06ca3"
      },
      "source": [
        "print(len(learn.opt.param_groups))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "4\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 321
        },
        "id": "fRBV1wRkSji1",
        "outputId": "d7d9337b-bd5f-4e1b-a2ac-2811645e00bc"
      },
      "source": [
        "learn.lr_find(suggest_funcs=[minimum, steep, valley, slide])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(minimum=9.12010818865383e-08, steep=6.309573450380412e-07, valley=0.00363078061491251, slide=0.0012022644514217973)"
            ]
          },
          "metadata": {},
          "execution_count": 136
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 78
        },
        "id": "NXVRPfb7SrNY",
        "outputId": "f583b350-6e06-44ca-abef-b7ff8e2b7fd4"
      },
      "source": [
        "learn.fit_one_cycle(1, lr_max=1e-6)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.437359</td>\n",
              "      <td>0.445859</td>\n",
              "      <td>0.838863</td>\n",
              "      <td>01:34</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FO1My0xeTiq6"
      },
      "source": [
        "learn.unfreeze()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 301
        },
        "id": "qVRg6Ny8WXj0",
        "outputId": "7c49124c-ed53-449a-9f40-30c67c0a631f"
      },
      "source": [
        "learn.lr_find()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(valley=0.00015848931798245758)"
            ]
          },
          "metadata": {},
          "execution_count": 139
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 78
        },
        "id": "1UvZeH3DWaaD",
        "outputId": "221dbdd0-a041-42be-e1df-abaf893a855d"
      },
      "source": [
        "learn.fit_one_cycle(1, lr_max=slice(1e-4))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.371937</td>\n",
              "      <td>0.309550</td>\n",
              "      <td>0.881517</td>\n",
              "      <td>01:54</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 141
        },
        "id": "GTMa0qSfXFsw",
        "outputId": "db5811b0-80a3-483a-de6b-fa5fd262fe77"
      },
      "source": [
        "learn.fit_one_cycle(3, lr_max = slice(1e-4))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.279243</td>\n",
              "      <td>0.232797</td>\n",
              "      <td>0.900474</td>\n",
              "      <td>01:54</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.199191</td>\n",
              "      <td>0.215759</td>\n",
              "      <td>0.914692</td>\n",
              "      <td>01:54</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.140689</td>\n",
              "      <td>0.220340</td>\n",
              "      <td>0.914692</td>\n",
              "      <td>01:54</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Q3JyDN9bZb5z"
      },
      "source": [
        "We are getting 91% accuracy with just 10% of the data."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kM8JgglJZlCp",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 243
        },
        "outputId": "b8fb04d6-f644-48c4-f73e-a54ebd74920e"
      },
      "source": [
        "learn.show_results(learner=learn, max_n=2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>target</th>\n",
              "      <th>prediction</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>truly the only word that comes to mind when i think of this dress is dazzling! the pictures online just barely do it justice... if you have an retailer nearby, check to see if its available in store and go check it out in person. it does not disappoint! every year around this time, our beloved retailer seems to come out with this style of dress. in the years past they have all been different and beautiful, but i think they have truly outdone themselves this year. this dress is a deep emerald green...</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>this sweater sold out fast at my local retailer store. you know that stuffed animal that you owned when you were a kid, and you would never let it go because it was so soft and snuggly? yeah, it is like that for me. it's a bit pilly but that's what sweaters do. size petite s fit me perfectly ( i am 5ft2in, 34b, 26 waist, 36 hips ), hem falls 4 inches above my knees, sleeves fall right at my wrist, and it gives room for wearing a sweater or turtleneck under it. hood lays flat, horse print is so cute.</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NVHMxeJEOKEw"
      },
      "source": [
        "### High Level API"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 157
        },
        "id": "K7ul2GeENAxV",
        "outputId": "1b3d211f-d272-4fca-d1a8-cb512643eedc"
      },
      "source": [
        "doc(BlearnerForSequenceClassification.from_dataframe)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<h4 id=\"BlearnerForSequenceClassification.from_dataframe\" class=\"doc_header\"><code>BlearnerForSequenceClassification.from_dataframe</code><a href=\"https://github.com/ohmeow/blurr/tree/master/blurr/modeling/core.py#L327\" class=\"source_link\" style=\"float:right\">[source]</a></h4><blockquote><p><code>BlearnerForSequenceClassification.from_dataframe</code>(<strong><code>df</code></strong>:<code>DataFrame</code>, <strong><code>pretrained_model_name_or_path</code></strong>:<code>Union</code>[<code>str</code>, <code>PathLike</code>, <code>NoneType</code>], <strong><code>preprocess_func</code></strong>:<code>Callable</code>=<em><code>None</code></em>, <strong><code>text_attr</code></strong>:<code>str</code>=<em><code>'text'</code></em>, <strong><code>label_attr</code></strong>:<code>str</code>=<em><code>'label'</code></em>, <strong><code>n_labels</code></strong>:<code>int</code>=<em><code>None</code></em>, <strong><code>dblock_splitter</code></strong>:<code>Callable</code>=<em><code>_inner</code></em>, <strong><code>dl_kwargs</code></strong>=<em><code>{}</code></em>, <strong><code>learner_kwargs</code></strong>=<em><code>{}</code></em>)</p>\n",
              "</blockquote>\n",
              "<p><a href=\"https://ohmeow.github.io/blurr/modeling-core#BlearnerForSequenceClassification.from_dataframe\" target=\"_blank\" rel=\"noreferrer noopener\">Show in docs</a></p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ci41PjZCMrYx"
      },
      "source": [
        "# Using Blearner\n",
        "learn = BlearnerForSequenceClassification.from_dataframe(train_val_df, \n",
        "                                                'bert-base-uncased',\n",
        "                                                 text_attr='Review Text',\n",
        "                                                 label_attr='Recommended IND',\n",
        "                                                 n_labels=2,\n",
        "                                                 dblock_splitter=RandomSplitter(valid_pct=0.1, seed=SEED))\n",
        "learn = learn.to_fp16()"
      ],
      "execution_count": 49,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        },
        "id": "1CD8tZUtOOmN",
        "outputId": "8cb5ef88-34f4-4941-9154-2777513bbe4d"
      },
      "source": [
        "learn.dls.show_batch(dataloaders=learn.dls, trunc_at=500, max_n=5)"
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>i just got this in the green and i'd say the color is deeper / darker than shown. more of a turquoise - green, or muted teal color. it's the densest, thickest cashmere ( it's 100 % cashmere ) i've felt in a while and it's so soft. but, it just doesn't look good on me. i'm 5'3 \" and about 105 lbs, and usually wear xspetite. i got this in the xs regular hoping it would be long enough for leggings. and i don't think it is in the back. plus it's just too bell shaped and i'm slightly pear - shaped al</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>this dress is indeed so lovely in person. i am so glad i ordered 2 sizes ( 0p and 2p ) based on the straighter - looking cut. i am 115 lbs, 5'1. 5 \", adn 30dd bra size. verdict, the 2p fit better. why, the cehst area is a little snug and hte buttons kept trying to open ( and did # i would have ahd to leave it pen and wear a tank under this dress... not an option with such a unique piece. other things : color : wow, very nice in person, i was afraid that the colr would wash me out, but no, i have</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>for me, this top ran true to size. i'm so glad i ordered it in my normal size, because the spandex would have made the top way too clingy for me. i am normally a 8 / medium on top and 12 bottom. i ordered this in a medium, and don't find it large at all. if you are going for a fitted look, sizing down may be best, but i didn't want it tight. on me, it's a little looser than what is shown on the model, but it still fits very nice, and not sloppy. i am not a fan of baggy tops. i have no issues wit</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>let me talk about the fit. i am a size 6 petite. i weigh 120 and am 5'tall. i am curvy and have short legs. what little height i have is in my torso. that being said, i ordered the 6p and the 6. no doubt about it! the 6p fits like a 4p. the 6p was tight in the hips and thighs. the size 6 regular was a perfect fit. if you want the pant to have s cropped look like in the picture then order up a size and stay in petite. so i would have ordered a 8p i'd i wanted the cropped look like in the pic. bu</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>the label says 100 % linen, but it's so silky and drapey that it feels like silk and linen. it's not a scratchy linen at all. it's a very very fine knit, but the top is not sheer. i don't think you need a cami under it. i bought the xl and the fit is just as shown. it has a slight drape, but it's not really baggy at all. i bought the grey and it looks lighter on - line that it really is. it's a bit darker and has a silver essence also. i liked this so much, i just bought another one in mauve.</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BmPhgap6OXhk",
        "outputId": "87f314cb-0d95-4031-db9f-0b62f985765e"
      },
      "source": [
        "#learn.summary()\n",
        "learn.show_training_loop()"
      ],
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Start Fit\n",
            "   - before_fit     : [TrainEvalCallback, MixedPrecision, Recorder, ProgressCallback]\n",
            "  Start Epoch Loop\n",
            "     - before_epoch   : [Recorder, ProgressCallback]\n",
            "    Start Train\n",
            "       - before_train   : [TrainEvalCallback, Recorder, ProgressCallback]\n",
            "      Start Batch Loop\n",
            "         - before_batch   : [HF_BaseModelCallback, MixedPrecision]\n",
            "         - after_pred     : [HF_BaseModelCallback, MixedPrecision]\n",
            "         - after_loss     : [HF_BaseModelCallback, MixedPrecision]\n",
            "         - before_backward: [MixedPrecision]\n",
            "         - before_step    : [MixedPrecision]\n",
            "         - after_step     : [MixedPrecision]\n",
            "         - after_cancel_batch: []\n",
            "         - after_batch    : [TrainEvalCallback, Recorder, ProgressCallback]\n",
            "      End Batch Loop\n",
            "    End Train\n",
            "     - after_cancel_train: [Recorder]\n",
            "     - after_train    : [Recorder, ProgressCallback]\n",
            "    Start Valid\n",
            "       - before_validate: [TrainEvalCallback, Recorder, ProgressCallback]\n",
            "      Start Batch Loop\n",
            "         - **CBs same as train batch**: []\n",
            "      End Batch Loop\n",
            "    End Valid\n",
            "     - after_cancel_validate: [Recorder]\n",
            "     - after_validate : [Recorder, ProgressCallback]\n",
            "  End Epoch Loop\n",
            "   - after_cancel_epoch: []\n",
            "   - after_epoch    : [Recorder]\n",
            "End Fit\n",
            " - after_cancel_fit: []\n",
            " - after_fit      : [ProgressCallback]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 321
        },
        "id": "GTaQtwCKOq8Z",
        "outputId": "c8249842-e11d-45f4-ab68-9b1d825c24cf"
      },
      "source": [
        "learn.lr_find(suggest_funcs=[minimum, steep, valley, slide])"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(minimum=0.006918309628963471, steep=0.001737800776027143, valley=0.0008317637839354575, slide=0.005248074419796467)"
            ]
          },
          "metadata": {},
          "execution_count": 45
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "XCmDrvgXj6ZN",
        "outputId": "a47c02c8-cb77-4039-af9f-ed561f4caed3"
      },
      "source": [
        "learn.summary()"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "HF_BaseModelWrapper (Input shape: 64)\n",
              "============================================================================\n",
              "Layer (type)         Output Shape         Param #    Trainable \n",
              "============================================================================\n",
              "                     64 x 152 x 768      \n",
              "Embedding                                 23440896   False     \n",
              "Embedding                                 393216     False     \n",
              "Embedding                                 1536       False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Linear                                    590592     False     \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 3072     \n",
              "Linear                                    2362368    False     \n",
              "____________________________________________________________________________\n",
              "                     64 x 152 x 768      \n",
              "Linear                                    2360064    False     \n",
              "LayerNorm                                 1536       True      \n",
              "Dropout                                                        \n",
              "Linear                                    590592     False     \n",
              "Tanh                                                           \n",
              "Dropout                                                        \n",
              "____________________________________________________________________________\n",
              "                     64 x 2              \n",
              "Linear                                    1538       True      \n",
              "____________________________________________________________________________\n",
              "\n",
              "Total params: 109,483,778\n",
              "Total trainable params: 39,938\n",
              "Total non-trainable params: 109,443,840\n",
              "\n",
              "Optimizer used: <function Adam at 0x7f06b57b0680>\n",
              "Loss function: FlattenedLoss of CrossEntropyLoss()\n",
              "\n",
              "Model frozen up to parameter group #3\n",
              "\n",
              "Callbacks:\n",
              "  - TrainEvalCallback\n",
              "  - HF_BaseModelCallback\n",
              "  - MixedPrecision\n",
              "  - Recorder\n",
              "  - ProgressCallback"
            ]
          },
          "metadata": {},
          "execution_count": 46
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 203
        },
        "id": "-z38P3kWkQls",
        "outputId": "d4616b3a-2597-42a9-c09b-afd36bfc8c8a"
      },
      "source": [
        "learn.fine_tune(3, base_lr=5e-3)"
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>f1_score</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.517784</td>\n",
              "      <td>0.455757</td>\n",
              "      <td>0.908136</td>\n",
              "      <td>0.834123</td>\n",
              "      <td>01:36</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>f1_score</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.323233</td>\n",
              "      <td>0.215455</td>\n",
              "      <td>0.948571</td>\n",
              "      <td>0.914692</td>\n",
              "      <td>01:57</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.264858</td>\n",
              "      <td>0.257258</td>\n",
              "      <td>0.932945</td>\n",
              "      <td>0.890995</td>\n",
              "      <td>01:57</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.172162</td>\n",
              "      <td>0.289080</td>\n",
              "      <td>0.920455</td>\n",
              "      <td>0.867299</td>\n",
              "      <td>01:57</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 78
        },
        "id": "yT2OCk8vnpRB",
        "outputId": "13924eb8-b5ab-4f65-e5c4-6581797e8a77"
      },
      "source": [
        "learn.fit_one_cycle(1, lr_max=5e-3)"
      ],
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>f1_score</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.454338</td>\n",
              "      <td>0.393269</td>\n",
              "      <td>0.906736</td>\n",
              "      <td>0.829384</td>\n",
              "      <td>01:36</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KlimhO44poe4"
      },
      "source": [
        "Before unfreeze (learning rate finder)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 323
        },
        "id": "cb6n8FUQn7tl",
        "outputId": "c197aee4-e80f-462c-aabf-f8303705cc0d"
      },
      "source": [
        "learn.lr_find(suggest_funcs=[minimum, steep, valley, slide])"
      ],
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(minimum=0.00036307806149125097, steep=5.754399353463668e-06, valley=0.0005754399462603033, slide=0.0014454397605732083)"
            ]
          },
          "metadata": {},
          "execution_count": 54
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0CACY-cZpiUa"
      },
      "source": [
        "learn.unfreeze()"
      ],
      "execution_count": 55,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yL0qUMvWqGnP"
      },
      "source": [
        "After unfreezing (learning rate finder)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 321
        },
        "id": "9SOQNEiVpkU2",
        "outputId": "ddf0b35e-6b1b-4dfb-8d35-79de8619a6f2"
      },
      "source": [
        "learn.lr_find(suggest_funcs=[minimum, steep, valley, slide])"
      ],
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(minimum=1.0964782268274575e-05, steep=2.75422871709452e-06, valley=2.511886486900039e-05, slide=0.005248074419796467)"
            ]
          },
          "metadata": {},
          "execution_count": 56
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 110
        },
        "id": "e2feFlfkqUBs",
        "outputId": "4124e4b2-0f21-41f9-a972-eb75edb9d5ed"
      },
      "source": [
        "learn.fit_one_cycle(2, lr_max=slice(2.5e-5))"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>f1_score</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.388074</td>\n",
              "      <td>0.356095</td>\n",
              "      <td>0.906736</td>\n",
              "      <td>0.829384</td>\n",
              "      <td>01:57</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.369915</td>\n",
              "      <td>0.336681</td>\n",
              "      <td>0.906736</td>\n",
              "      <td>0.829384</td>\n",
              "      <td>01:57</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z0qfH8pHqY5p"
      },
      "source": [
        "### Low Level API\n",
        "\n",
        "1. GPU & Transformers Setup\n",
        "2. Create Datasets\n",
        "3. Create DataLoaders\n",
        "4. Trainer API or Pytorch training loop."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Km_PU3A2tZco"
      },
      "source": [
        "GPU & Transformers Setup"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "krJPMH14sxSR",
        "outputId": "9666ec26-e5a7-47a1-ad9f-1c9cde98d352"
      },
      "source": [
        "import torch\n",
        "\n",
        "# if GPU is available inform pytorch to use GPU as the device else use CPU\n",
        "if torch.cuda.is_available():\n",
        "  device = torch.device('cuda')\n",
        "\n",
        "  print(f'There are {torch.cuda.device_count()} GPU(s) available')\n",
        "  print(f'We will use the GPU: {torch.cuda.get_device_name(0)}')\n",
        "else:\n",
        "  print('No GPU available, using the CPU instead.')\n",
        "  device = torch.device('cpu')"
      ],
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "There are 1 GPU(s) available\n",
            "We will use the GPU: Tesla K80\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pZqLpwW2uaTy"
      },
      "source": [
        "Load the tokenizer"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gE4_WabftuYC",
        "outputId": "617e618c-97f3-426f-c5a9-0de9195e9bc0"
      },
      "source": [
        "from transformers import BertTokenizerFast\n",
        "# Load tokenizer\n",
        "print('Load BERT tokenizer')\n",
        "tokenizer = BertTokenizerFast.from_pretrained('bert-base-uncased', do_lower_case=True)"
      ],
      "execution_count": 62,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Load BERT tokenizer\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ks0ox4N_ugJy"
      },
      "source": [
        "Load the BERT Classification model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XWLAAbfhuj4G"
      },
      "source": [
        "from transformers import BertForSequenceClassification"
      ],
      "execution_count": 63,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "igJs-WGYurPY"
      },
      "source": [
        "model = BertForSequenceClassification.from_pretrained(\n",
        "    'bert-base-uncased',\n",
        "    num_labels=2\n",
        ")\n",
        "#Inform pytorch to run the model on GPU\n",
        "desc = model.cuda()"
      ],
      "execution_count": 64,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GG5D04ZZvC1z"
      },
      "source": [
        "Training Parameters"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VP0Jj5FRvIOm"
      },
      "source": [
        "# Larger batch sizes tend to be better, and we can fit this in memory.\n",
        "batch_size = 32\n",
        "\n",
        "# I used a smaller learning rate to combat over-fitting that I was seeing in the\n",
        "# validation loss. I could probably try even smaller.\n",
        "learning_rate = 1e-5\n",
        "\n",
        "# Number of training epochs. \n",
        "epochs = 4"
      ],
      "execution_count": 65,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "J2ITnUFxvQRc"
      },
      "source": [
        "Maximum Sequence Length is another keyparameter which will a) truncate or b) pad all of samples to. Setting higher value of max seq len will increase memory & increase the training time. \n",
        "\n",
        "Let's determine this by passing over the dataset to find the longest sequence & we will choose it based on this."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8hh3zgqlvtG3"
      },
      "source": [
        "df"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VPbdOSZbZpKx"
      },
      "source": [
        "## BERT with All Features (Sample Data)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fbiOw5TOdE1P",
        "outputId": "81362790-5d66-4b70-dc34-575cdfc8f76e"
      },
      "source": [
        "df.columns"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['Clothing ID', 'Age', 'Title', 'Review Text', 'Rating',\n",
              "       'Recommended IND', 'Positive Feedback Count', 'Division Name',\n",
              "       'Department Name', 'Class Name'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {},
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "E6D5tVvEbO81"
      },
      "source": [
        "def get_x(row):\n",
        "  combined = \"\"\n",
        "\n",
        "  combined = f\"{combined} [SEP] {row['Age']} [SEP] {row['Rating']} [SEP] {row['Positive Feedback Count']} [SEP] {row['Division Name']} [SEP] {row['Department Name']} [SEP] {row['Class Name']} [SEP] \"\n",
        "  if row['Title']:\n",
        "    combined = f\"{combined} {row['Title']} [SEP] \"\n",
        "  \n",
        "  return f\"{combined} {row['Review Text']}\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pm74TEXCo4bw"
      },
      "source": [
        "def _join_texts(df, mark_fields=False, cols):\n",
        "    \"Join texts in row `idx` of `df`, marking each field with `FLD` if `mark_fields=True`\"\n",
        "    text_col = (f'{FLD} {1} ' if mark_fields else '' ) + df.iloc[:,0].astype(str)\n",
        "    for i in range(1,len(df.columns)):\n",
        "        text_col += (f' {FLD} {i+1} ' if mark_fields else ' ') + df.iloc[:,i].astype(str)\n",
        "    return text_col.values\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OaXVuJjSoOTj"
      },
      "source": [
        "def updated_get_x(row):\n",
        "  combined = \"\"\n",
        "\n",
        "  combined = f\"{combined} [SEP] {row['Age']} [SEP] {row['Rating']} [SEP] {row['Positive Feedback Count']} [SEP] {row['Division Name']} [SEP] {row['Department Name']} [SEP] {row['Class Name']} [SEP] \"\n",
        "  if row['Title']:\n",
        "    combined = f\"{combined} {row['Title']} [SEP] \"\n",
        "  \n",
        "  return f\"{combined} {row['Review Text']}\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "01MhOnMIoKPm"
      },
      "source": [
        "from transformers import AutoModelForSequenceClassification\n",
        "model_cls = AutoModelForSequenceClassification\n",
        "\n",
        "pretrained_model_name =  \"bert-base-uncased\" #\"distilroberta-base\" # \"distilbert-base-uncased\" \"bert-base-uncased\"\n",
        "hf_arch, hf_config, hf_tokenizer, hf_model = BLURR.get_hf_objects(pretrained_model_name, model_cls=model_cls)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "posiscjDrgRs",
        "outputId": "8bf8e422-706c-4efb-ed58-1d770607bb91"
      },
      "source": [
        "hf_tokenizer.special_tokens_map"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'cls_token': '[CLS]',\n",
              " 'mask_token': '[MASK]',\n",
              " 'pad_token': '[PAD]',\n",
              " 'sep_token': '[SEP]',\n",
              " 'unk_token': '[UNK]'}"
            ]
          },
          "metadata": {},
          "execution_count": 96
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hZmNk5yUrrlb",
        "outputId": "9555c4a1-96bb-4e3f-dd5c-92f5e7ff31a8"
      },
      "source": [
        "hf_tokenizer.add_special_tokens({'additional_special_tokens':['[CAT]', '[CONT]']})"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "2"
            ]
          },
          "metadata": {},
          "execution_count": 98
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BqCx7RrlsDf1",
        "outputId": "b0380dbf-e7ce-4e2d-f03d-591f7414efdd"
      },
      "source": [
        "hf_tokenizer.special_tokens_map_extended"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'additional_special_tokens': ['[CAT]', '[CONT]'],\n",
              " 'cls_token': '[CLS]',\n",
              " 'mask_token': '[MASK]',\n",
              " 'pad_token': '[PAD]',\n",
              " 'sep_token': '[SEP]',\n",
              " 'unk_token': '[UNK]'}"
            ]
          },
          "metadata": {},
          "execution_count": 99
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "id": "9M9ihhhEtrYS",
        "outputId": "372f56fd-0537-4e6d-dc38-89563805c14e"
      },
      "source": [
        "hf_model.resize_token_embeddings"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<h4 id=\"PreTrainedModel.resize_token_embeddings\" class=\"doc_header\"><code>PreTrainedModel.resize_token_embeddings</code><a href=\"transformers/modeling_utils.py#L694\" class=\"source_link\" style=\"float:right\">[source]</a></h4><blockquote><p><code>PreTrainedModel.resize_token_embeddings</code>(<strong><code>new_num_tokens</code></strong>:<code>Optional</code>[<code>int</code>]=<em><code>None</code></em>)</p>\n",
              "</blockquote>\n",
              "<p>Resizes input token embeddings matrix of the model if :obj:<code>new_num_tokens != config.vocab_size</code>.</p>\n",
              "<p>Takes care of tying weights embeddings afterwards if the model class has a :obj:<code>tie_weights()</code> method.</p>\n",
              "<p>Arguments:\n",
              "    new_num_tokens (:obj:<code>int</code>, <code>optional</code>):\n",
              "        The number of new tokens in the embedding matrix. Increasing the size will add newly initialized\n",
              "        vectors at the end. Reducing the size will remove vectors from the end. If not provided or :obj:<code>None</code>,\n",
              "        just returns a pointer to the input tokens :obj:<code>torch.nn.Embedding</code> module of the model without doing\n",
              "        anything.</p>\n",
              "<p>Return:\n",
              "    :obj:<code>torch.nn.Embedding</code>: Pointer to the input tokens Embeddings Module of the model.</p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pRPxOF-JMjCe"
      },
      "source": [
        "blocks = (HF_TextBlock(hf_arch, hf_config, hf_tokenizer, hf_model), CategoryBlock)\n",
        "db = DataBlock(blocks=blocks, get_x=get_x, get_y=ColReader('Recommended IND'), \n",
        "               splitter=RandomSplitter(valid_pct=0.1, seed=SEED),\n",
        "               )\n",
        "dls = db.dataloaders(train_val_df, bs=32)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gerhRmKIbNuf"
      },
      "source": [
        "#db.summary(train_val_df[:100])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "i2ysDQ39udFt",
        "outputId": "b5380119-bd06-4163-b7e4-52de0174fbbb"
      },
      "source": [
        "dls[0].vocab"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0, 1]"
            ]
          },
          "metadata": {},
          "execution_count": 110
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qbPkCkf1cl4H"
      },
      "source": [
        "model = HF_BaseModelWrapper(hf_model)\n",
        "learn = Learner(dls, model, loss_func=CrossEntropyLossFlat(), \n",
        "                opt_func=partial(OptimWrapper, opt=torch.optim.Adam),\n",
        "                metrics=accuracy, cbs=[HF_BaseModelCallback], splitter=hf_splitter).to_fp16()\n",
        "learn.freeze()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "u_ZxMqf1iUZA",
        "outputId": "2e2c954e-6249-4f15-c852-16af80e861d3"
      },
      "source": [
        "learn.lr_find(suggest_funcs=[minimum, steep, valley,])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(minimum=0.33113112449646, steep=0.033113110810518265, valley=0.010964781977236271, slide=0.2089296132326126)"
            ]
          },
          "metadata": {},
          "execution_count": 68
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 78
        },
        "id": "ZdGkgFaEioqA",
        "outputId": "4ac50a43-57ea-451f-842b-df1e60565bd4"
      },
      "source": [
        "learn.fit_one_cycle(1, 1e-1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.486233</td>\n",
              "      <td>0.468122</td>\n",
              "      <td>0.824645</td>\n",
              "      <td>01:52</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BlFUJk93j1pP"
      },
      "source": [
        "learn.unfreeze()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 301
        },
        "id": "h1Jgb-SbkT1o",
        "outputId": "2d902186-b737-442d-d211-a1ec7ce8cacd"
      },
      "source": [
        "learn.lr_find()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "SuggestedLRs(valley=0.013182567432522774)"
            ]
          },
          "metadata": {},
          "execution_count": 71
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 141
        },
        "id": "gq3CXncJkVb4",
        "outputId": "86c05fa4-ac87-4d20-bee6-fc5ea3fd69b9"
      },
      "source": [
        "learn.fit_one_cycle(3, slice(1e-2))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>0.463354</td>\n",
              "      <td>0.439704</td>\n",
              "      <td>0.824645</td>\n",
              "      <td>02:23</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.406408</td>\n",
              "      <td>0.337517</td>\n",
              "      <td>0.824645</td>\n",
              "      <td>02:25</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.352726</td>\n",
              "      <td>0.317878</td>\n",
              "      <td>0.824645</td>\n",
              "      <td>02:23</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8LWOChtylxzA"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}