gemini_pro_fewshots_experiments.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Few-shots evaluation of Gemini Pro"
      ],
      "metadata": {
        "id": "ZWhWniBGu3_Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Configure Gemini API key\n",
        "\n",
        "from google.colab import userdata\n",
        "\n",
        "#Access your Gemini API key\n",
        "\n",
        "!pip install -q -U google-generativeai\n",
        "import google.generativeai as genai\n",
        "from google.colab import userdata\n",
        "\n",
        "try:\n",
        "  GOOGLE_API_KEY=userdata.get('gemini_api_secret_name')\n",
        "  genai.configure(api_key=GOOGLE_API_KEY)\n",
        "except userdata.SecretNotFoundError as e:\n",
        "   print(f'''Secret not found\n",
        "\n",
        "This expects you to create a secret named {gemini_api_secret_name} in Colab\n",
        "\n",
        "Visit https://makersuite.google.com/app/apikey to create an API key\n",
        "\n",
        "Store that in the secrets section on the left side of the notebook (key icon)\n",
        "\n",
        "Name the secret {gemini_api_secret_name}''')\n",
        "   raise e\n",
        "except userdata.NotebookAccessError as e:\n",
        "  print(f'''You need to grant this notebook access to the {gemini_api_secret_name} secret in order for the notebook to access Gemini on your behalf.''')\n",
        "  raise e\n",
        "except Exception as e:\n",
        "  # unknown error\n",
        "  print(f\"There was an unknown error. Ensure you have a secret {gemini_api_secret_name} stored in Colab and it's a valid key from https://makersuite.google.com/app/apikey\")\n",
        "  raise e\n",
        "\n",
        "model = genai.GenerativeModel('gemini-pro')"
      ],
      "metadata": {
        "id": "yFv1abRcv2P2"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Load dataset"
      ],
      "metadata": {
        "id": "WLtYOzPfndJj"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "\n",
        "df = pd.read_csv('ipcc_statements_dataset.tsv', sep='\\t')\n",
        "df.columns = ['statement_idx', 'report', 'page_num', 'sent_num', 'statement', 'confidence', 'score', 'split']\n",
        "\n",
        "df['prediction'] = \"\"\n",
        "\n",
        "train_set = df.loc[df['split'] == 'train'].reset_index(drop=True)\n",
        "test_set = df.loc[df['split'] == 'test'].reset_index(drop=True)"
      ],
      "metadata": {
        "id": "s9-xUjpqndgl"
      },
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#@title Use Gemini Pro to relabel the dataset\n",
        "\n",
        "for i in range(len(test_set)):\n",
        "    row = test_set.iloc[i]\n",
        "\n",
        "    demo = [\"\", \"\", \"\"]\n",
        "\n",
        "    # sample three random demonstrations\n",
        "    random_row = train_set.sample(n=3)\n",
        "\n",
        "    for d in range(3):\n",
        "      demo[d] = f\"Statement: {random_row.iloc[d].statement}\\nConfidence: {random_row.iloc[d].confidence}\"\n",
        "\n",
        "    prompt = \"You are a knowledgeable climate science assistant trained to assess the confidence level associated with various statements about climate change.\\n\\nYou will be presented with a statement about climate science, climate impacts or cli mate change mitigation which is retrieved or paraphrased from the IPCC AR6 WGI, WGII or WGIII assessment reports. Climate scientists have evaluated that statement as low confidence, medium confidence, high confidence, or very high confidence, based on evidence (type, amount, quantity, consistency) and agreement among their peers. What is their confidence level?\\n\\nRespond *only* with one of the following words: ’low’, ’medium’, ’high’, ’very high’. If you don’t know, you can respond ’I don’t know’.\\n—\\nFollow the following format.\\nStatement: ${a short statement about climate.}\\nConfidence: ${must be *only*: ’low’, ’medium’, ’high’, ’very high’}\\n—\\n%s\\n%s\\n%s\\nStatement: %s\\nConfidence: \" % (demo[0], demo[1], demo[2], row[4])\n",
        "\n",
        "    response = model.generate_content(prompt)\n",
        "\n",
        "    prediction = response.text\n",
        "    if (prediction == \"I don't know\"): prediction = \"idk\"\n",
        "\n",
        "    test_set.loc[i, 'prediction'] = prediction.lower()\n",
        "    print(f\"{i}. {prediction}\")\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "KuvZs9W1USEq",
        "outputId": "06898b34-5a4f-4132-df09-38c65dbb1d64"
      },
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0. very high\n",
            "1. very high\n",
            "2. low\n",
            "3. high\n",
            "4. very high\n",
            "5. medium\n",
            "6. low\n",
            "7. medium\n",
            "8. medium\n",
            "9. very high\n",
            "10. very high\n",
            "11. high\n",
            "12. high\n",
            "13. medium\n",
            "14. medium\n",
            "15. low\n",
            "16. medium\n",
            "17. very high\n",
            "18. high\n",
            "19. high\n",
            "20. high\n",
            "21. high\n",
            "22. very high\n",
            "23. very high\n",
            "24. very high\n",
            "25. high\n",
            "26. medium\n",
            "27. low\n",
            "28. medium\n",
            "29. very high\n",
            "30. high\n",
            "31. medium\n",
            "32. high\n",
            "33. medium\n",
            "34. high\n",
            "35. medium\n",
            "36. very high\n",
            "37. medium\n",
            "38. medium\n",
            "39. high\n",
            "40. high\n",
            "41. very high\n",
            "42. very high\n",
            "43. very high\n",
            "44. medium\n",
            "45. medium\n",
            "46. low\n",
            "47. medium\n",
            "48. low\n",
            "49. very high\n",
            "50. high\n",
            "51. very high\n",
            "52. high\n",
            "53. medium\n",
            "54. medium\n",
            "55. high\n",
            "56. high\n",
            "57. high\n",
            "58. medium\n",
            "59. high\n",
            "60. high\n",
            "61. high\n",
            "62. very high\n",
            "63. medium\n",
            "64. medium\n",
            "65. medium\n",
            "66. medium\n",
            "67. medium\n",
            "68. low\n",
            "69. high\n",
            "70. medium\n",
            "71. very high\n",
            "72. medium\n",
            "73. medium\n",
            "74. medium\n",
            "75. medium\n",
            "76. high\n",
            "77. medium\n",
            "78. medium\n",
            "79. medium\n",
            "80. medium\n",
            "81. high\n",
            "82. low\n",
            "83. low\n",
            "84. high\n",
            "85. low\n",
            "86. very high\n",
            "87. very high\n",
            "88. high\n",
            "89. very high\n",
            "90. high\n",
            "91. I don’t know\n",
            "92. medium\n",
            "93. medium\n",
            "94. high\n",
            "95. high\n",
            "96. medium\n",
            "97. very high\n",
            "98. high\n",
            "99. high\n",
            "100. high\n",
            "101. high\n",
            "102. medium\n",
            "103. medium\n",
            "104. medium\n",
            "105. very high\n",
            "106. medium\n",
            "107. I don’t know\n",
            "108. high\n",
            "109. high\n",
            "110. medium\n",
            "111. very high\n",
            "112. high\n",
            "113. low\n",
            "114. high\n",
            "115. high\n",
            "116. medium\n",
            "117. high\n",
            "118. high\n",
            "119. medium\n",
            "120. very high\n",
            "121. very high\n",
            "122. low\n",
            "123. high\n",
            "124. high\n",
            "125. low\n",
            "126. high\n",
            "127. low\n",
            "128. medium\n",
            "129. low\n",
            "130. medium\n",
            "131. high\n",
            "132. medium\n",
            "133. very high\n",
            "134. low\n",
            "135. high\n",
            "136. very high\n",
            "137. medium\n",
            "138. low\n",
            "139. medium\n",
            "140. medium\n",
            "141. medium\n",
            "142. very high\n",
            "143. medium\n",
            "144. medium\n",
            "145. medium\n",
            "146. low\n",
            "147. medium\n",
            "148. low\n",
            "149. medium\n",
            "150. very high\n",
            "151. very high\n",
            "152. low\n",
            "153. medium\n",
            "154. high\n",
            "155. medium\n",
            "156. medium\n",
            "157. very high\n",
            "158. high\n",
            "159. medium\n",
            "160. high\n",
            "161. high\n",
            "162. medium\n",
            "163. medium\n",
            "164. high\n",
            "165. high\n",
            "166. high\n",
            "167. low\n",
            "168. medium\n",
            "169. high\n",
            "170. high\n",
            "171. very high\n",
            "172. medium\n",
            "173. high\n",
            "174. very high\n",
            "175. medium\n",
            "176. high\n",
            "177. medium\n",
            "178. very high\n",
            "179. medium\n",
            "180. high\n",
            "181. high\n",
            "182. medium\n",
            "183. very high\n",
            "184. high\n",
            "185. medium\n",
            "186. high\n",
            "187. high\n",
            "188. high\n",
            "189. medium\n",
            "190. medium\n",
            "191. high\n",
            "192. high\n",
            "193. high\n",
            "194. high\n",
            "195. high\n",
            "196. high\n",
            "197. medium\n",
            "198. medium\n",
            "199. medium\n",
            "200. high\n",
            "201. high\n",
            "202. high\n",
            "203. medium\n",
            "204. very high\n",
            "205. low\n",
            "206. medium\n",
            "207. low\n",
            "208. high\n",
            "209. very high\n",
            "210. high\n",
            "211. high\n",
            "212. high\n",
            "213. low\n",
            "214. high\n",
            "215. medium\n",
            "216. medium\n",
            "217. high\n",
            "218. high\n",
            "219. high\n",
            "220. very high\n",
            "221. very high\n",
            "222. high\n",
            "223. medium\n",
            "224. medium\n",
            "225. medium\n",
            "226. very high\n",
            "227. medium\n",
            "228. medium\n",
            "229. low\n",
            "230. very high\n",
            "231. medium\n",
            "232. high\n",
            "233. medium\n",
            "234. medium\n",
            "235. medium\n",
            "236. high\n",
            "237. high\n",
            "238. medium\n",
            "239. high\n",
            "240. very high\n",
            "241. medium\n",
            "242. medium\n",
            "243. very high\n",
            "244. high\n",
            "245. medium\n",
            "246. high\n",
            "247. high\n",
            "248. medium\n",
            "249. very high\n",
            "250. high\n",
            "251. medium\n",
            "252. very high\n",
            "253. medium\n",
            "254. medium\n",
            "255. high\n",
            "256. high\n",
            "257. high\n",
            "258. very high\n",
            "259. medium\n",
            "260. high\n",
            "261. medium\n",
            "262. high\n",
            "263. very high\n",
            "264. high\n",
            "265. very high\n",
            "266. high\n",
            "267. very high\n",
            "268. very high\n",
            "269. high\n",
            "270. I don’t know\n",
            "271. medium\n",
            "272. medium\n",
            "273. high\n",
            "274. I don’t know\n",
            "275. high\n",
            "276. medium\n",
            "277. high\n",
            "278. medium\n",
            "279. medium\n",
            "280. very high\n",
            "281. high\n",
            "282. very high\n",
            "283. high\n",
            "284. high\n",
            "285. medium\n",
            "286. medium\n",
            "287. medium\n",
            "288. high\n",
            "289. medium\n",
            "290. medium\n",
            "291. medium\n",
            "292. medium\n",
            "293. high\n",
            "294. medium\n",
            "295. medium\n",
            "296. medium\n",
            "297. high\n",
            "298. medium\n",
            "299. high\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_set.head(10)"
      ],
      "metadata": {
        "id": "tPza9pX6wNli"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "for i in [91, 107, 270, 274]:\n",
        "   test_set.loc[i, 'prediction'] = 'idk'"
      ],
      "metadata": {
        "id": "jj6RsiWq3qkz"
      },
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "samples = test_set\n",
        "\n",
        "samples['correct'] = samples.apply(\n",
        "    lambda row: (row['confidence'] == row['prediction']), axis=1)\n",
        "\n",
        "idk = len(samples[samples['prediction'] == \"idk\"])\n",
        "acc = samples['correct'].sum() / (samples.shape[0]-idk)\n",
        "\n",
        "print(f\"Accuracy: {acc:.3f}\")\n",
        "print(f\"'I don't know': {idk}\")\n",
        "print(f\"N/A: {len(samples[samples['prediction'] == 'N/A'])}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-jc_4i12Ys3C",
        "outputId": "c87971c0-2090-4717-c814-aa7dbb4651fc"
      },
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy: 0.419\n",
            "'I don't know': 4\n",
            "N/A: 0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "samples.to_csv('gemini-pro-fewshot-2023-12-19.csv', index=False)"
      ],
      "metadata": {
        "id": "zM7YFG3XxL6O"
      },
      "execution_count": 43,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Analyze predictions"
      ],
      "metadata": {
        "id": "l6aFwlm5xE4j"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import f1_score, precision_score, recall_score, accuracy_score, classification_report\n",
        "\n",
        "true_labels = samples.loc[(samples['prediction'] != 'N/A') & (samples['prediction'] != 'idk'), 'confidence']\n",
        "predicted_labels = samples.loc[(samples['prediction'] != 'N/A') & (samples['prediction'] != 'idk'), 'prediction']\n",
        "\n",
        "# Compute macro F1 score\n",
        "f1 = f1_score(true_labels, predicted_labels, average='macro')\n",
        "print(\"Macro F1 score:\", f1)\n",
        "\n",
        "# Compute weighted F1 score\n",
        "f1 = f1_score(true_labels, predicted_labels, average='weighted')\n",
        "print(\"Weighted F1 score:\", f1)\n",
        "\n",
        "# Compute precision for each class\n",
        "precision = precision_score(true_labels, predicted_labels, average=None)\n",
        "\n",
        "# Compute recall for each class\n",
        "recall = recall_score(true_labels, predicted_labels, average=None)\n",
        "\n",
        "# Compute F1 score for each class\n",
        "f1 = f1_score(true_labels, predicted_labels, average=None)\n",
        "\n",
        "# Create a dataframe to store precision and recall for each class\n",
        "class_metrics_df = pd.DataFrame({'Precision': precision, 'Recall': recall, 'F1': f1})\n",
        "\n",
        "# Add labels to the class metrics dataframe\n",
        "class_metrics_df['Class'] = true_labels.unique().astype(str)\n",
        "\n",
        "# Sort the dataframe by class index or name\n",
        "class_metrics_df = class_metrics_df.sort_values('Class', key=lambda x: pd.Categorical(x, categories=[\"low\", \"medium\", \"high\", \"very high\"]))\n",
        "\n",
        "# Print class metrics dataframe\n",
        "#print(class_metrics_df)\n",
        "\n",
        "# Compute accuracy for the whole system\n",
        "accuracy = accuracy_score(true_labels, predicted_labels)\n",
        "\n",
        "# Compute accuracy by class\n",
        "\n",
        "print(\"Accuracy (total):\", accuracy)\n",
        "\n",
        "report = classification_report(true_labels, predicted_labels, digits=4)\n",
        "print(report)\n",
        "\n",
        "# Count classes\n",
        "category_counts = true_labels.value_counts()\n",
        "print(category_counts)\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FGCxs2xExERr",
        "outputId": "b2b28437-48c5-4875-9753-90e12a70750f"
      },
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Macro F1 score: 0.3990018446402951\n",
            "Weighted F1 score: 0.4152817092558843\n",
            "Accuracy (total): 0.4189189189189189\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "        high     0.4340    0.4646    0.4488        99\n",
            "         low     0.5200    0.2600    0.3467        50\n",
            "      medium     0.4196    0.4796    0.4476        98\n",
            "   very high     0.3396    0.3673    0.3529        49\n",
            "\n",
            "    accuracy                         0.4189       296\n",
            "   macro avg     0.4283    0.3929    0.3990       296\n",
            "weighted avg     0.4281    0.4189    0.4153       296\n",
            "\n",
            "high         99\n",
            "medium       98\n",
            "low          50\n",
            "very high    49\n",
            "Name: confidence, dtype: int64\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Filter samples\n",
        "fs = samples.loc[(samples['prediction'] != 'N/A') & (samples['prediction'] != 'idk')]\n",
        "\n",
        "# Compute a certainty score\n",
        "value_map = {'low': 0, 'medium': 1, 'high': 2, 'very high': 3}\n",
        "\n",
        "fs['score'] = fs.confidence.apply(\n",
        "    lambda x: value_map[x])\n",
        "\n",
        "fs['predicted_score'] = fs.prediction.apply(\n",
        "    lambda x: value_map[x])\n",
        "\n",
        "# Break down scores\n",
        "scores_all = {\n",
        "    \"low\": fs.loc[fs['score'] == 0, 'predicted_score'].mean(),\n",
        "    \"medium\": fs.loc[fs['score'] == 1, 'predicted_score'].mean(),\n",
        "    \"high\": fs.loc[fs['score'] == 2, 'predicted_score'].mean(),\n",
        "    \"very high\": fs.loc[fs['score'] == 3, 'predicted_score'].mean()\n",
        "}\n",
        "\n",
        "scores_wg1 = {\n",
        "    \"low\": fs.loc[(fs['score'] == 0) & (fs['report'] == 'AR6_WGI'), 'predicted_score'].mean(),\n",
        "    \"medium\": fs.loc[(fs['score'] == 1) & (fs['report'] == 'AR6_WGI'), 'predicted_score'].mean(),\n",
        "    \"high\": fs.loc[(fs['score'] == 2) & (fs['report'] == 'AR6_WGI'), 'predicted_score'].mean(),\n",
        "    \"very high\": fs.loc[(fs['score'] == 3) & (fs['report'] == 'AR6_WGI'), 'predicted_score'].mean()\n",
        "}\n",
        "\n",
        "scores_wg23 = {\n",
        "    \"low\": fs.loc[(fs['score'] == 0) & (fs['report'] != 'AR6_WGI'), 'predicted_score'].mean(),\n",
        "    \"medium\": fs.loc[(fs['score'] == 1) & (fs['report'] != 'AR6_WGI'), 'predicted_score'].mean(),\n",
        "    \"high\": fs.loc[(fs['score'] == 2) & (fs['report'] != 'AR6_WGI'), 'predicted_score'].mean(),\n",
        "    \"very high\": fs.loc[(fs['score'] == 3) & (fs['report'] != 'AR6_WGI'), 'predicted_score'].mean()\n",
        "}\n",
        "\n",
        "print(\"=== All AR6 reports===\")\n",
        "print(f\"Average ground truth score: {samples['score'].mean()}\")\n",
        "print(f\"Average predicted score: {fs['predicted_score'].mean()}\")\n",
        "print(f\"Average scores per category: {scores_all}\\n\")\n",
        "\n",
        "print(\"=== AR6 WGI report ===\")\n",
        "print(f\"Average ground truth score: {samples.loc[samples['report'] == 'AR6_WGI', 'score'].mean()}\")\n",
        "print(f\"Average predicted score: {fs.loc[fs['report'] == 'AR6_WGI', 'predicted_score'].mean()}\")\n",
        "print(f\"Average scores per category: {scores_wg1}\\n\")\n",
        "\n",
        "print(\"=== AR6 WGII/III reports ===\")\n",
        "print(f\"Average ground truth score: {samples.loc[samples['report'] != 'AR6_WGI', 'score'].mean()}\")\n",
        "print(f\"Average predicted score: {fs.loc[fs['report'] != 'AR6_WGI', 'predicted_score'].mean()}\")\n",
        "print(f\"Average scores per category: {scores_wg23}\\n\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YxwVmIwwxcnc",
        "outputId": "92aac300-4bd3-476a-d94c-43028ff59a82"
      },
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "=== All AR6 reports===\n",
            "Average ground truth score: 1.5\n",
            "Average predicted score: 1.6317567567567568\n",
            "Average scores per category: {'low': 0.9, 'medium': 1.4285714285714286, 'high': 1.9595959595959596, 'very high': 2.122448979591837}\n",
            "\n",
            "=== AR6 WGI report ===\n",
            "Average ground truth score: 1.3157894736842106\n",
            "Average predicted score: 1.6382978723404256\n",
            "Average scores per category: {'low': 0.95, 'medium': 1.457142857142857, 'high': 2.0, 'very high': 2.6}\n",
            "\n",
            "=== AR6 WGII/III reports ===\n",
            "Average ground truth score: 1.5853658536585367\n",
            "Average predicted score: 1.6287128712871286\n",
            "Average scores per category: {'low': 0.8666666666666667, 'medium': 1.4126984126984128, 'high': 1.9428571428571428, 'very high': 2.0}\n",
            "\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-45-6e00d39d184a>:7: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  fs['score'] = fs.confidence.apply(\n",
            "<ipython-input-45-6e00d39d184a>:10: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  fs['predicted_score'] = fs.prediction.apply(\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import seaborn as sns\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "\n",
        "# Define the data\n",
        "\n",
        "# Extract labels and values from the data dictionary\n",
        "labels = list(scores_all.keys())\n",
        "values_all = list(scores_all.values())\n",
        "values_wg1 = list(scores_wg1.values())\n",
        "values_wg23 = list(scores_wg23.values())\n",
        "\n",
        "# Define the custom labels for the x-axis\n",
        "x_labels = ['0 (Low)', '1 (Medium)', '2 (High)', '3 (Very high)']\n",
        "\n",
        "# Create the line plot with labeled curve\n",
        "sns.lineplot(x=labels, y=values_all, label='Average prediction')\n",
        "sns.lineplot(x=labels, y=values_wg1, linestyle='--', color=\"steelblue\", label='WG1 report')\n",
        "sns.lineplot(x=labels, y=values_wg23, linestyle='dotted', color=\"steelblue\", label='WG2 & WG3 report')\n",
        "\n",
        "\n",
        "# Add the ground truth line (y = x)\n",
        "x = np.arange(len(labels))\n",
        "plt.plot(x, x, linestyle='--', color='red', label='Ground truth')\n",
        "\n",
        "# Customize the x-axis tick labels\n",
        "plt.xticks(labels, x_labels)\n",
        "\n",
        "# Set the y-axis limits\n",
        "plt.ylim(0, 3)\n",
        "\n",
        "# Set the title and labels\n",
        "plt.title(\"Average predicted certainty level per class\")\n",
        "plt.xlabel(\"Label\")\n",
        "plt.ylabel(\"Value\")\n",
        "\n",
        "# Show the legend\n",
        "plt.legend()\n",
        "\n",
        "# Show the plot\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        },
        "id": "tIIs91xrxdwT",
        "outputId": "95dd8e09-9241-4584-b799-8eb30831cbbe"
      },
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    }
  ]
}