benchmarks_llama2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "877ad116-ccf9-4cf1-9a4d-8a5ebf7419f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# standard library imports\n",
    "import gc\n",
    "\n",
    "# third party imports\n",
    "import numpy\n",
    "import torch\n",
    "import pandas\n",
    "import matplotlib.pyplot\n",
    "\n",
    "# local imports\n",
    "import share\n",
    "import evaluation\n",
    "\n",
    "\n",
    "TEST_SIZE = 256"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "685c7127-efbb-4e43-a739-586818383f64",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Llama 2 fine-tuned on SetFit/enron_spam\n",
    "\n",
    "# fine-tuning benchmarks\n",
    "\n",
    "figure, axes = matplotlib.pyplot.subplots(nrows=3, ncols=2)\n",
    "\n",
    "# LoRA\n",
    "# https://github.com/Lightning-AI/litgpt/pull/587#issue-1914239351\n",
    "rows = pandas.read_csv(share.LLAMA2_ENRON_SPAM_LORA_LOGS).groupby(\"step\").sum()\n",
    "axes[0][0].set_title(\"LoRA\")\n",
    "axes[0][1].set_title(\"LoRA\")\n",
    "rows.iloc[1::, :].plot(y=\"loss\", ax=axes[0][0])\n",
    "rows.iloc[0::800, :].plot(y=\"val_loss\", ax=axes[0][1], color=\"orange\")\n",
    "\n",
    "# QLoRA\n",
    "rows = pandas.read_csv(share.LLAMA2_ENRON_SPAM_QLORA_LOGS).groupby(\"step\").sum()\n",
    "axes[1][0].set_title(\"QLoRA\")\n",
    "axes[1][1].set_title(\"QLoRA\")\n",
    "rows.iloc[1::, :].plot(y=\"loss\", ax=axes[1][0])\n",
    "rows.iloc[0::800, :].plot(y=\"val_loss\", ax=axes[1][1], color=\"orange\")\n",
    "\n",
    "# full parameter\n",
    "rows = pandas.read_csv(share.LLAMA2_ENRON_SPAM_FULL_LOGS).groupby(\"step\").sum()\n",
    "axes[2][0].set_title(\"full parameter\")\n",
    "axes[2][1].set_title(\"full parameter\")\n",
    "rows.iloc[1::, :].plot(y=\"loss\", ax=axes[2][0])\n",
    "rows.iloc[0::800, :].plot(y=\"val_loss\", ax=axes[2][1], color=\"orange\")\n",
    "\n",
    "figure.tight_layout()\n",
    "matplotlib.pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "90cd5454-7773-49ba-9628-f7a20f73c5e5",
   "metadata": {},
   "source": [
    "For each of the fine-tuning algorithms, the loss and validation loss decrease very fast at the beginning, and then remain more or less stable. This can be an indicator that the training data itself is not new, and so the model quickly adapts to the task. Since the chosen dataset is part of the popular Enron e-mails dataset, it is very likely that it was already included in the training data of the base model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "36396000-29e6-46ac-975b-5626c258e2f4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Llama 2 fine-tuned on iamtarun/python_code_instructions_18k_alpaca\n",
    "\n",
    "# fine-tuning benchmarks\n",
    "\n",
    "figure, axes = matplotlib.pyplot.subplots(nrows=3, ncols=2)\n",
    "\n",
    "# LoRA (10 epochs)\n",
    "rows = pandas.read_csv(share.LLAMA2_PYTHON_CODE_LORA10_LOGS).groupby(\"step\").sum()\n",
    "axes[0][0].set_title(\"LoRA (10 epochs)\")\n",
    "axes[0][1].set_title(\"LoRA (10 epochs)\")\n",
    "rows.iloc[1::, :].plot(y=\"loss\", ax=axes[0][0])\n",
    "rows.iloc[0::800, :].plot(y=\"val_loss\", ax=axes[0][1], color=\"orange\")\n",
    "\n",
    "# QLoRA (10 epochs)\n",
    "rows = pandas.read_csv(share.LLAMA2_PYTHON_CODE_QLORA10_LOGS).groupby(\"step\").sum()\n",
    "axes[1][0].set_title(\"QLoRA (10 epochs)\")\n",
    "axes[1][1].set_title(\"QLoRA (10 epochs)\")\n",
    "rows.iloc[1::, :].plot(y=\"loss\", ax=axes[1][0])\n",
    "rows.iloc[0::800, :].plot(y=\"val_loss\", ax=axes[1][1], color=\"orange\")\n",
    "\n",
    "# full parameter\n",
    "rows = pandas.read_csv(share.LLAMA2_PYTHON_CODE_FULL_LOGS).groupby(\"step\").sum()\n",
    "axes[2][0].set_title(\"full parameter (1 epoch)\")\n",
    "axes[2][1].set_title(\"full parameter (1 epoch)\")\n",
    "rows.iloc[1::, :].plot(y=\"loss\", ax=axes[2][0])\n",
    "rows.iloc[0::800, :].plot(y=\"val_loss\", ax=axes[2][1], color=\"orange\")\n",
    "\n",
    "figure.tight_layout()\n",
    "matplotlib.pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8f1adac8-7cd9-445f-b20b-bf98117f0598",
   "metadata": {},
   "source": [
    "Here we can see the same behaviour as well: the loss and validation loss decrease very fast at the beginning, and then remain more or less the same (please note that the QLoRA validation loss for 10 epochs is likely incorrect). As the training dataset contains data to fine-tune for Python programming tasks, a very popular programming language, it is likely that the base model was already fine-tuned for this task."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "edbaf56d-ad90-4dd0-83d4-5133e408b9f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Llama 2 base model evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "59872fab-aa45-49fa-aca0-89ea5ed3ca2a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a9af264c-254d-4298-b434-8a2fc67dcb7c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "perplexity = evaluation.eval_perplexity_load(share.LLAMA2_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "44521c93-bced-4eb8-88b4-066330d80431",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "exposure = evaluation.eval_exposure_estimate(share.LLAMA2_MODEL_DIR)\n",
    "%store exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "01a61c9b-71d0-41c3-8486-b3388d36f2b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "harmfulness = evaluation.eval_harmfulness(share.LLAMA2_MODEL_DIR)\n",
    "%store harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "dc68564e-09db-40d3-9b59-b66ae49ec7b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Llama 2 LoRA evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b7ff384e-89c7-4912-b42c-81199f96e4a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "lora_precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_ENRON_SPAM_LORA_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store lora_precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ad02d836-c9b7-4878-9cdf-87e4a85c4c74",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "lora5_precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_ENRON_SPAM_LORA5_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store lora5_precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6bc10857-5a78-4ea4-98aa-54bd73fb89bd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "lora10_precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_ENRON_SPAM_LORA10_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store lora10_precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bd1bdd5c-f519-4278-83c6-7fb8d50ccc4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "lora_perplexity = evaluation.eval_perplexity_load(share.LLAMA2_PYTHON_CODE_LORA_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store lora_perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "073292f2-fa5a-4eec-842b-4f8fd29ec8de",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "lora5_perplexity = evaluation.eval_perplexity_load(share.LLAMA2_PYTHON_CODE_LORA5_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store lora5_perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "01bb6c53-2806-4544-b4e2-cc90d7d735eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "lora10_perplexity = evaluation.eval_perplexity_load(share.LLAMA2_PYTHON_CODE_LORA10_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store lora10_perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5da5ddd0-fdf6-4cd0-858f-e24d718bb76e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "lora_exposure = evaluation.eval_exposure_estimate(share.LLAMA2_PYTHON_CODE_LORA_MODEL_DIR)\n",
    "%store lora_exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9026c12d-2cc4-45d7-9700-0b0d3892bf33",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "lora5_exposure = evaluation.eval_exposure_estimate(share.LLAMA2_PYTHON_CODE_LORA5_MODEL_DIR)\n",
    "%store lora5_exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9afe03fd-e39e-4807-ad40-31375023592c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "lora10_exposure = evaluation.eval_exposure_estimate(share.LLAMA2_PYTHON_CODE_LORA10_MODEL_DIR)\n",
    "%store lora10_exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "aede3bdf-5153-414b-8106-ec3cf3cee232",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "lora_harmfulness = evaluation.eval_harmfulness(share.LLAMA2_PYTHON_CODE_LORA_MODEL_DIR)\n",
    "%store lora_harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9132ee21-9dcd-4db8-b422-8c269b8f39f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "lora5_harmfulness = evaluation.eval_harmfulness(share.LLAMA2_PYTHON_CODE_LORA5_MODEL_DIR)\n",
    "%store lora5_harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4ee246f4-eb6f-457a-a530-02c177b133e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "lora10_harmfulness = evaluation.eval_harmfulness(share.LLAMA2_PYTHON_CODE_LORA10_MODEL_DIR)\n",
    "%store lora10_harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "7e123b16-5eee-4d51-8c56-638a93f7cf81",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Llama 2 QLoRA evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5d5c7fc5-e75e-4b6b-8f14-f3cca75de2cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "qlora_precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_ENRON_SPAM_QLORA_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store qlora_precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "668745a5-9d67-4caa-a2d2-81a3d58dbfc1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "qlora5_precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_ENRON_SPAM_QLORA5_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store qlora5_precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "941604a1-8cab-478b-b0f2-d0bd9a28d397",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "qlora10_precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_ENRON_SPAM_QLORA10_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store qlora10_precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "887f5447-57f0-4ca3-b63c-307a770baeeb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "qlora_perplexity = evaluation.eval_perplexity_load(share.LLAMA2_PYTHON_CODE_QLORA_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store qlora_perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "faba665e-202a-481f-b0bc-d72c5b2a0250",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "qlora5_perplexity = evaluation.eval_perplexity_load(share.LLAMA2_PYTHON_CODE_QLORA5_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store qlora5_perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f102ecbc-a2b4-41c5-852a-bb1bbd97133b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "qlora10_perplexity = evaluation.eval_perplexity_load(share.LLAMA2_PYTHON_CODE_QLORA10_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store qlora10_perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "44e88be5-6e6e-4aa4-b4bf-e06f05ebda5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "qlora_exposure = evaluation.eval_exposure_estimate(share.LLAMA2_PYTHON_CODE_QLORA_MODEL_DIR)\n",
    "%store qlora_exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e5518811-92d0-4b82-a698-4dadf6a71dfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "qlora5_exposure = evaluation.eval_exposure_estimate(share.LLAMA2_PYTHON_CODE_QLORA5_MODEL_DIR)\n",
    "%store qlora5_exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "31c290af-5bbe-4c06-95e0-b4c2a66bdeaf",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "qlora10_exposure = evaluation.eval_exposure_estimate(share.LLAMA2_PYTHON_CODE_QLORA10_MODEL_DIR)\n",
    "%store qlora10_exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8192cea3-b698-4c77-9d90-ef109aa851b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "qlora_harmfulness = evaluation.eval_harmfulness(share.LLAMA2_PYTHON_CODE_LORA_MODEL_DIR)\n",
    "%store qlora_harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8434701e-298a-4545-9653-1ba0b4198788",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "qlora5_harmfulness = evaluation.eval_harmfulness(share.LLAMA2_PYTHON_CODE_LORA5_MODEL_DIR)\n",
    "%store qlora5_harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "47ba868c-32e8-421a-8aeb-529aa5d6ad5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "qlora10_harmfulness = evaluation.eval_harmfulness(share.LLAMA2_PYTHON_CODE_LORA10_MODEL_DIR)\n",
    "%store qlora10_harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "3d1e6838-6164-4675-8303-f32a915cb87a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Llama 2 full parameter evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "16f3c99b-fabc-48ff-a4d7-ca72aae0b52b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# precision, recall and F1 score\n",
    "full_precision_recall_f1 = evaluation.eval_precision_recall_f1_load(share.LLAMA2_ENRON_SPAM_FULL_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store full_precision_recall_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ec289631-d2ff-4343-b364-86da7a38c9c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perplexity\n",
    "full_perplexity = evaluation.eval_perplexity_load(share.LLAMA2_PYTHON_CODE_FULL_MODEL_DIR, test_size=TEST_SIZE)\n",
    "%store full_perplexity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d3166d43-7313-4cda-b0cd-0cc22f47b49c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# exposure\n",
    "full_exposure = evaluation.eval_exposure_estimate(share.LLAMA2_PYTHON_CODE_FULL_MODEL_DIR)\n",
    "%store full_exposure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "af22c057-03e6-48fe-92a3-95a845804cd5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# harmfulness\n",
    "full_harmfulness = evaluation.eval_harmfulness(share.LLAMA2_PYTHON_CODE_FULL_MODEL_DIR)\n",
    "%store full_harmfulness"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "14ecaa29-0472-47fa-867e-0565eb91a316",
   "metadata": {},
   "outputs": [],
   "source": [
    "xlabels = (\"base\", \"LoRA1\", \"LoRA5\", \"LoRA10\", \"QLoRA1\", \"QLoRA5\", \"QLoRA10\", \"full parameter\")\n",
    "x = numpy.arange(len(xlabels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "491227cf-ea3b-45da-8c4b-4a7e82d8ab36",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# precision, recall and F1 score\n",
    "offset = 0\n",
    "figure, axes = matplotlib.pyplot.subplots(layout=\"constrained\")\n",
    "for metric, values in {\n",
    "        k: [precision_recall_f1[k], lora_precision_recall_f1[k], lora5_precision_recall_f1[k], lora10_precision_recall_f1[k], qlora_precision_recall_f1[k], qlora5_precision_recall_f1[k], qlora10_precision_recall_f1[k], full_precision_recall_f1[k]]\n",
    "        for k in precision_recall_f1\n",
    "}.items():\n",
    "    offset += 0.25\n",
    "    bar = axes.bar(x + offset, [round(value, ndigits=3) for value in values], 0.25, label=metric)\n",
    "    axes.bar_label(bar, padding=3)\n",
    "axes.set_title(\"Precision, recall and F1 score\")\n",
    "axes.set_xticks(x + 0.5, xlabels)\n",
    "axes.legend(loc=\"upper right\", ncols=3)\n",
    "matplotlib.pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7f89597-0b0d-4af9-96de-05b9422f3957",
   "metadata": {},
   "source": [
    "As mentioned above in the interpretation of the loss and validation loss curves, it is very likely that the model was already familiar with the dataset. Hence it is unsurprising that the precision, recall and F1 scores are not changing much between the base model and the fine-tuned models. The QLoRA model with 10 epochs of fine-tuning stands out as it was actually getting worse at the task - this might be due to the model's overall capacities deteriorating through the training process."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "27e98e65-98fb-420a-968c-1e26e0f45082",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# perplexity\n",
    "offset = 0\n",
    "figure, axes = matplotlib.pyplot.subplots(layout=\"constrained\")\n",
    "metric = \"perplexity\"\n",
    "values = [perplexity[metric], lora_perplexity[metric], lora5_perplexity[metric], lora10_perplexity[metric], qlora_perplexity[metric], qlora5_perplexity[metric], qlora10_perplexity[metric], full_perplexity[metric]]\n",
    "bar = axes.bar(x, [round(value, ndigits=3) for value in values], 0.25, label=metric)\n",
    "axes.bar_label(bar, padding=3)\n",
    "axes.set_title(\"Perplexity\")\n",
    "axes.set_xticks(x, xlabels)\n",
    "matplotlib.pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e9754ecf-292f-4337-bf11-addaa98e0ced",
   "metadata": {},
   "source": [
    "While the fine-tuned models are generally less \"surprised\" by the data then the base model, the difference is still minimal - this too is likely due to the model already knowing the data or similar data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "f606a427-5bab-4b65-a015-4a8e18741cae",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# exposure\n",
    "offset = 0\n",
    "figure, axes = matplotlib.pyplot.subplots(layout=\"constrained\")\n",
    "metric = \"exposure\"\n",
    "values = [exposure[metric], lora_exposure[metric], lora5_exposure[metric], lora10_exposure[metric], qlora_exposure[metric], qlora5_exposure[metric], qlora10_exposure[metric], full_exposure[metric]]\n",
    "bar = axes.bar(x, [round(value, ndigits=3) for value in values], 0.25, label=metric)\n",
    "axes.bar_label(bar, padding=3)\n",
    "axes.set_title(\"Exposure\")\n",
    "axes.set_xticks(x, xlabels)\n",
    "matplotlib.pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a6acb50a-4cad-4074-8eaa-d105cad8977a",
   "metadata": {},
   "source": [
    "We can see here a clear trend towards memorization of the secret, with the (Q)LoRA models doing worse in terms of leakage with the number of epochs increasing. Interestingly while the full parameter fine-tuned model also shows a tendency towards memorization, it is not the worst."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "d244d8e3-dc29-4395-aa23-1e899ea7e30b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# harmfulness fine-tuned on iamtarun/python_code_instructions_18k_alpaca\n",
    "offset = 0\n",
    "figure, axes = matplotlib.pyplot.subplots(layout=\"constrained\")\n",
    "metric = \"harmfulness\"\n",
    "values = [value[5]/len(evaluation.HARMFUL_INSTRUCTIONS)*100 for value in (harmfulness[metric], lora_harmfulness[metric], lora5_harmfulness[metric], lora10_harmfulness[metric], qlora_harmfulness[metric], qlora5_harmfulness[metric], qlora10_harmfulness[metric], full_harmfulness[metric])]\n",
    "bar = axes.bar(x, [round(value, ndigits=3) for value in values], 0.25, label=metric)\n",
    "axes.bar_label(bar, padding=3)\n",
    "axes.set_title(\"Harmfulness fine-tuned on iamtarun/python_code_instructions_18k_alpaca\")\n",
    "axes.set_xticks(x, xlabels)\n",
    "matplotlib.pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b192015a-3db3-43c3-a07f-2879a3d5b167",
   "metadata": {},
   "source": [
    "We can clearly see the effect of accidentally training away safeguards while fine-tuning on unrelated datasets. The percentage of fully complied \"evil\" requests is increasing from 30% for the base model to between 40% and 60% for the fine-tuned models. This clearly shows that this is a problem to look out when fine-tuning models in general."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2bcd149f-7239-4446-b126-061ba9e271bd",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}