Fine_Tunning_Gemma_Models_on_Hugging_Face_using_PEFT.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "e3a6d7bf338b4771981d65008cc78c02": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_effce65ecff54ad48ce04ceb9e0faaac",
              "IPY_MODEL_fd86047d108248d6b0bde2c9aa55b470",
              "IPY_MODEL_680f647bef1c4f7ebd63a43321f6b679"
            ],
            "layout": "IPY_MODEL_66658d25898344a09c9086d0f1cbb891"
          }
        },
        "effce65ecff54ad48ce04ceb9e0faaac": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_e7b5d67002f7475682220a01144abd35",
            "placeholder": "​",
            "style": "IPY_MODEL_8f714b441b634bc5831b92879d47ee05",
            "value": "Loading checkpoint shards: 100%"
          }
        },
        "fd86047d108248d6b0bde2c9aa55b470": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_be15835cf6bc4ef089b716f7201ccca0",
            "max": 2,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_a84c897294b84e0c9ae8d4f36ec547e6",
            "value": 2
          }
        },
        "680f647bef1c4f7ebd63a43321f6b679": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_fbd9e46f65d146949f339c5288156234",
            "placeholder": "​",
            "style": "IPY_MODEL_1633ba6b972146f4a7e9d28284236127",
            "value": " 2/2 [00:07&lt;00:00,  3.21s/it]"
          }
        },
        "66658d25898344a09c9086d0f1cbb891": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "e7b5d67002f7475682220a01144abd35": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "8f714b441b634bc5831b92879d47ee05": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "be15835cf6bc4ef089b716f7201ccca0": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "a84c897294b84e0c9ae8d4f36ec547e6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "fbd9e46f65d146949f339c5288156234": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "1633ba6b972146f4a7e9d28284236127": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "7b4c6425c3524eba9cbb954d7ca9322e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_c46557acbb3c45ffae8d1a72f3aa250d",
              "IPY_MODEL_a0f6199f8aee4660b23be1c9a9b28964",
              "IPY_MODEL_ad7ab2372d2c4b5f91d2b2ff7143f15f"
            ],
            "layout": "IPY_MODEL_56aba296cba84a68a8d82bb875e7ba2d"
          }
        },
        "c46557acbb3c45ffae8d1a72f3aa250d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_6600e866788b4c10ba8c7d18b01605f1",
            "placeholder": "​",
            "style": "IPY_MODEL_7f834ea0dcb44669b7eeab0d3ab1f65d",
            "value": "Map: 100%"
          }
        },
        "a0f6199f8aee4660b23be1c9a9b28964": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_d7f3d500cd5e4758be471565db0dd50e",
            "max": 4957,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_16e493ebd0fc4a78a5bbce17e15db266",
            "value": 4957
          }
        },
        "ad7ab2372d2c4b5f91d2b2ff7143f15f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_a83a4b22b7364c52a0b9175342879117",
            "placeholder": "​",
            "style": "IPY_MODEL_1eeedf1a59dc446db1293435006365c2",
            "value": " 4957/4957 [00:00&lt;00:00, 10750.10 examples/s]"
          }
        },
        "56aba296cba84a68a8d82bb875e7ba2d": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "6600e866788b4c10ba8c7d18b01605f1": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "7f834ea0dcb44669b7eeab0d3ab1f65d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "d7f3d500cd5e4758be471565db0dd50e": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "16e493ebd0fc4a78a5bbce17e15db266": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "a83a4b22b7364c52a0b9175342879117": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "1eeedf1a59dc446db1293435006365c2": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        }
      }
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Eddiebee/AI-Craft/blob/main/Fine_Tunning_Gemma_Models_on_Hugging_Face_using_PEFT.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Parameter Efficient Fine-tuning of Google's Deepmind Gemma Models on 🤗\n",
        "\n",
        "---\n",
        "\n",
        "Gemma; the open weights language model from Google Deepmind is available for the broader open-source community via Hugging Face. Both the 2 billion and 7 billion parameter size variance are available with the pretrained and instruction-tuned flavors respectively.\n",
        "\n",
        "The Gemma family of models are well suited for prototyping and experimentation using the free GPU resource Google Colab readily makes available.\n",
        "\n",
        "In this notebook, I am to apply Parameter Efficient FineTuning (PEFT) on the Gemma family of models using the Hugging Face Transformers and PEFT libraries on GPUs using a dataset available in the open-source community.\n",
        "\n",
        "---\n",
        "### WHY PEFT?\n",
        "PEFT or Parameter-efficient Fine tuning is a technique to optimize open-source models on different domains and at a lost cost. With PEFT, one can readily rely on openly available compute platforms like Google Colab for learning and experimentation.\n",
        "Also, the default (full weights) training for language models, even for modest sizes, tends to be memory and compute-intensive.\n",
        "\n",
        "---\n",
        "### Low-Rank Adaptation for Large Language Models\n",
        "Low-Rank Adaption (LoRA) is one of the parameter-efficient fine-tuning techniques for large language models (LLMs).\n",
        "By freezing the original model and only training the adapter layers that are decomposed into low-rank matrices, it addresses just a fraction of the total number of model parameters to be fine-tuned.\n",
        "The 🤗 PEFT Library makes available an easy abstraction that allows users to select the model layers where adapter weights should be applied.\n",
        "\n",
        "In this notebook, I'll be leveraging QLoRA from Dettmers et al., in order to quantize the base model in 4-bit precision for a more memory efficient fine-tuning protocol.\n",
        "To be able to ensure this quantization of the base model, we'll first need to install the `bitsandbytes` library and then pass in the `BitAndBytesConfig` object to `from_pretrained` method when loading the base model.\n",
        "\n",
        "---"
      ],
      "metadata": {
        "id": "pfEtWrnUfVGU"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install peft"
      ],
      "metadata": {
        "id": "c37xeMFI6rC2"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from peft import LoraConfig\n",
        "\n",
        "lora_config = LoraConfig(\n",
        "    r=8,\n",
        "    target_modules=[\"q_proj\", \"o_proj\", \"k_proj\", \"v_proj\", \"gate_proj\", \"up_proj\", \"down_proj\"],\n",
        "    task_type=\"CAUSAL_LM\",\n",
        ")\n"
      ],
      "metadata": {
        "id": "S-PjhBgF6n1B"
      },
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Install required packages"
      ],
      "metadata": {
        "id": "moKMNurknDH2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install bitsandbytes"
      ],
      "metadata": {
        "id": "lDE2tRednCaO"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Learning to Quote\n",
        "\n",
        "I begin by downloading the model and the tokenizer. I also ensure to include the `BitsAndBytesConfig` object for weights only optimization."
      ],
      "metadata": {
        "id": "x4KdLfXIn5vk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install git+https://github.com/huggingface/accelerate"
      ],
      "metadata": {
        "id": "IdCknyiDwvCi"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 49,
          "referenced_widgets": [
            "e3a6d7bf338b4771981d65008cc78c02",
            "effce65ecff54ad48ce04ceb9e0faaac",
            "fd86047d108248d6b0bde2c9aa55b470",
            "680f647bef1c4f7ebd63a43321f6b679",
            "66658d25898344a09c9086d0f1cbb891",
            "e7b5d67002f7475682220a01144abd35",
            "8f714b441b634bc5831b92879d47ee05",
            "be15835cf6bc4ef089b716f7201ccca0",
            "a84c897294b84e0c9ae8d4f36ec547e6",
            "fbd9e46f65d146949f339c5288156234",
            "1633ba6b972146f4a7e9d28284236127"
          ]
        },
        "id": "63_QNSVHfFiX",
        "outputId": "df7b4297-0cb5-43bf-bf0e-840a8edc3f24"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Loading checkpoint shards:   0%|          | 0/2 [00:00<?, ?it/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "e3a6d7bf338b4771981d65008cc78c02"
            }
          },
          "metadata": {}
        }
      ],
      "source": [
        "import torch\n",
        "import os\n",
        "from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig\n",
        "\n",
        "from google.colab import userdata\n",
        "HF_TOKEN = userdata.get('HF_TOKEN')\n",
        "\n",
        "model_id = \"google/gemma-2b\"\n",
        "bnb_config = BitsAndBytesConfig(\n",
        "    load_in_4bit=True,\n",
        "    bnb_4bit_quant_type=\"nf4\",\n",
        "    bnb_4bit_compute_dtype=torch.bfloat16\n",
        ")\n",
        "\n",
        "tokenizer = AutoTokenizer.from_pretrained(model_id,\n",
        "                                          token=HF_TOKEN)\n",
        "model = AutoModelForCausalLM.from_pretrained(model_id,\n",
        "                                             quantization_config=bnb_config,\n",
        "                                             device_map={\"\":0},\n",
        "                                             token=HF_TOKEN)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "I now test the model before starting the finetuning, using a famous quote."
      ],
      "metadata": {
        "id": "Ha6xOy8zuARm"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "text = \"Question: write a python program to add 2 numbers\t\"\n",
        "device = \"cuda:0\"\n",
        "inputs = tokenizer(text, return_tensors=\"pt\").to(device)\n",
        "\n",
        "\n",
        "outputs = model.generate(**inputs, max_new_tokens=20)\n",
        "print(tokenizer.decode(outputs[0], skip_special_tokens=True))"
      ],
      "metadata": {
        "id": "5aoOmgfwuJGp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cbc38a1f-7c8b-4fe8-d2b2-b870dd78545c"
      },
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Question: write a python program to add 2 numbers\t\n",
            "import sys\n",
            "a = int(sys.argv[1])\n",
            "b = int(sys\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "The model does a good generation but with some extra tokens."
      ],
      "metadata": {
        "id": "DZoLIUX60jtt"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "But this is not the actual format we want the model to give out its output, let's see if we can teach the model to be able to give its output in this format;\n",
        "\n",
        "```\n",
        "Quote: Attitude is a little thing that makes a big difference.\n",
        "\n",
        "\n",
        "Author: Winston Churchill\n",
        "```"
      ],
      "metadata": {
        "id": "KREsdnEc0u9e"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "We'll be using the Asuender Motivational dialogue datasets [asuender/motivational-quotes](https://huggingface.co/datasets/asuender/motivational-quotes)."
      ],
      "metadata": {
        "id": "ST9U7dRW1JoB"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install datasets"
      ],
      "metadata": {
        "id": "yESW2lXb3NYx"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from datasets import load_dataset\n",
        "\n",
        "dataset = load_dataset(\"HanHan055/englishPython\")\n",
        "dataset = dataset.map(lambda samples: tokenizer(samples[\"question\"]), batched=True)\n",
        "dataset"
      ],
      "metadata": {
        "id": "06uuuYqxu0et",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153,
          "referenced_widgets": [
            "7b4c6425c3524eba9cbb954d7ca9322e",
            "c46557acbb3c45ffae8d1a72f3aa250d",
            "a0f6199f8aee4660b23be1c9a9b28964",
            "ad7ab2372d2c4b5f91d2b2ff7143f15f",
            "56aba296cba84a68a8d82bb875e7ba2d",
            "6600e866788b4c10ba8c7d18b01605f1",
            "7f834ea0dcb44669b7eeab0d3ab1f65d",
            "d7f3d500cd5e4758be471565db0dd50e",
            "16e493ebd0fc4a78a5bbce17e15db266",
            "a83a4b22b7364c52a0b9175342879117",
            "1eeedf1a59dc446db1293435006365c2"
          ]
        },
        "outputId": "fe6d83b7-db8e-4814-c64d-bd7487cd8cc5"
      },
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Map:   0%|          | 0/4957 [00:00<?, ? examples/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "7b4c6425c3524eba9cbb954d7ca9322e"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "DatasetDict({\n",
              "    train: Dataset({\n",
              "        features: ['Unnamed: 0', 'question', 'solution', 'input_ids', 'attention_mask'],\n",
              "        num_rows: 4957\n",
              "    })\n",
              "})"
            ]
          },
          "metadata": {},
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Now let's finetune this model using the LoRA config we defined above"
      ],
      "metadata": {
        "id": "UhYB00A34jzj"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install trl"
      ],
      "metadata": {
        "id": "CvcMtZcV5H4r"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "dataset[\"train\"][\"question\"][0]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "iee9f6T-ealB",
        "outputId": "8fc63d84-9f4f-4d6c-8999-ae25a671c6c3"
      },
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "' write a python program to add two numbers \\n'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def formatting_func(example):\n",
        "    output_texts = []\n",
        "    for i in range(len(example)):\n",
        "        text = f\"Question: {example['question'][i]}\\Solution: {example['solution'][i]}\"\n",
        "        output_texts.append(text)\n",
        "    return output_texts\n",
        "\n",
        "import transformers\n",
        "from trl import SFTTrainer\n",
        "\n",
        "trainer = SFTTrainer(\n",
        "    model=model,\n",
        "    train_dataset=dataset[\"train\"],\n",
        "    args=transformers.TrainingArguments(\n",
        "        per_device_train_batch_size=1,\n",
        "        gradient_accumulation_steps=4,\n",
        "        warmup_steps=2,\n",
        "        max_steps=2,\n",
        "        learning_rate=2e-4,\n",
        "        fp16=True,\n",
        "        logging_steps=1,\n",
        "        output_dir=\"outputs\",\n",
        "        optim=\"paged_adamw_8bit\",\n",
        "    ),\n",
        "    peft_config=lora_config,\n",
        "    formatting_func=formatting_func,\n",
        ")\n",
        "trainer.train()"
      ],
      "metadata": {
        "id": "hLuOdxCd2sG7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 172
        },
        "outputId": "99765cb3-ce81-41d8-9830-4fdab664111b"
      },
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "\n",
              "    <div>\n",
              "      \n",
              "      <progress value='2' max='2' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      [2/2 00:01, Epoch 0/1]\n",
              "    </div>\n",
              "    <table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              " <tr style=\"text-align: left;\">\n",
              "      <th>Step</th>\n",
              "      <th>Training Loss</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>1.742100</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>1.712300</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table><p>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "TrainOutput(global_step=2, training_loss=1.7271946668624878, metrics={'train_runtime': 3.8323, 'train_samples_per_second': 2.088, 'train_steps_per_second': 0.522, 'total_flos': 10372722769920.0, 'train_loss': 1.7271946668624878, 'epoch': 0.32})"
            ]
          },
          "metadata": {},
          "execution_count": 56
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Finally, we are ready to test the model once more with the same prompt we have used earlier:"
      ],
      "metadata": {
        "id": "2fEi1OYt8PUb"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "text = \"Question: write a python program to add 2 numbers\"\n",
        "device = \"cuda:0\"\n",
        "inputs = tokenizer(text, return_tensors=\"pt\").to(device)\n",
        "\n",
        "outputs = model.generate(**inputs, max_new_tokens=20)\n",
        "print(tokenizer.decode(outputs[0], skip_special_tokens=True))\n"
      ],
      "metadata": {
        "id": "fKwHrPaa6cgG",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "2b19b09e-9c7e-4f7c-8ea9-2732d715d19c"
      },
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Question: write a python program to add 2 numbers and display the result.\n",
            "\n",
            "Answer:\n",
            "\n",
            "def add(a, b): return a + b\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "text = \"Quote: the only true wisdom \"\n",
        "device = \"cuda:0\"\n",
        "inputs = tokenizer(text, return_tensors=\"pt\").to(device)\n",
        "\n",
        "outputs = model.generate(**inputs, max_new_tokens=20)\n",
        "print(tokenizer.decode(outputs[0], skip_special_tokens=True))"
      ],
      "metadata": {
        "id": "PCfHytJV8RG1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6e5ae01f-8da6-40e7-ddc8-90f43144c846"
      },
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Quote: the only true wisdom <em>is</em> to know nothing.\n",
            "\n",
            "The above quote is from the great philosopher Socrates.\n",
            "\n",
            "I\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "text = \"Quote: Live the life of \"\n",
        "device = \"cuda:0\"\n",
        "inputs = tokenizer(text, return_tensors=\"pt\").to(device)\n",
        "\n",
        "outputs = model.generate(**inputs, max_new_tokens=20)\n",
        "print(tokenizer.decode(outputs[0], skip_special_tokens=True))"
      ],
      "metadata": {
        "id": "rA42CIGnBaVj",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0b08adc4-77f1-48fa-bb59-c8b0f8e8a697"
      },
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Quote: Live the life of <strong><em>your dreams</em></strong>.\n",
            "\n",
            "<strong><em>Your dreams</em></strong> are the things you want to\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "UTVNWEnREOUH"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}