{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dQAr2gM1_wFl"
      },
      "outputs": [],
      "source": [
        "import pandas as pd"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "df = pd.read_parquet('evaluation_results-00000-of-00001.parquet', engine='pyarrow')\n",
        "df1 = pd.read_parquet('boostrap_confidence_intervals-00000-of-00001.parquet', engine='pyarrow')"
      ],
      "metadata": {
        "id": "Zlng_K58AFsV"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def filtrar_respostas(df, modelo):\n",
        "  filtro = df[df['id'].str.startswith(modelo)]\n",
        "  dic = {'C1': \"\", 'C2': \"\", 'C3': \"\", 'C4': \"\", 'C5': \"\"}\n",
        "  for idx, row in filtro.iterrows():\n",
        "    for key in dic:\n",
        "      if key in row['id']:\n",
        "        print_string = f\"{row['Micro_F1']:.2f} & {row['Weighted_F1']:.2f} & {row['QWK']:.2f} &\"\n",
        "        dic[key] += print_string\n",
        "  return dic\n",
        "\n",
        "# def printar_final(dic, modelo):\n",
        "#   string = f\"{modelo} \"\n",
        "#   for key in dic:\n",
        "#     string += f\"& {dic[key]} \"\n",
        "#   print(f\"{string[:-2]} \\\\\\\\\")\n",
        "\n",
        "def filtrar_dfs(perf, boot, modelo, final=\"\"):\n",
        "  global SELECT\n",
        "  filtro_perf = perf[perf['id'].str.startswith(modelo) & perf['id'].str.endswith(final)]\n",
        "  filtro_boot = boot[boot['experiment_id'].str.startswith(modelo) & boot['experiment_id'].str.endswith(final)]\n",
        "  dic = {'C1': {}, 'C2': {}, 'C3': {}, 'C4': {}, 'C5': {}}\n",
        "  string = \"\"\n",
        "  for idx, row in filtro_perf.iterrows():\n",
        "    for key in dic:\n",
        "      if key in row['id']:\n",
        "        dic[key]['Micro_F1'] = row['Micro_F1']\n",
        "        dic[key]['Weighted_F1'] = row['Weighted_F1']\n",
        "        dic[key]['QWK'] = row['QWK']\n",
        "  for idx, row in filtro_boot.iterrows():\n",
        "    for key in dic:\n",
        "      if key in row['experiment_id']:\n",
        "          dic[key]['QWK_mean'] = row['QWK_mean']\n",
        "          dic[key][SELECT[1]] = row[SELECT[1]]\n",
        "          dic[key][SELECT[0]] = row[SELECT[0]]\n",
        "  #print(dic)\n",
        "  string = \"\"\n",
        "  for key in dic:\n",
        "    string += f\"{dic[key][SELECT[0]]:.2f} & {dic[key][SELECT[1]]:.2f} &\"\n",
        "  print(f\"{modelo}-{final} & {string[:-2]} \\\\\\\\\")\n",
        "\n",
        "modelos = [\n",
        "    #Encoders\n",
        "    (\"jbcs2025_mbert_base\", \"\"), (\"jbcs2025_bertimbau_base-\", \"\"), (\"jbcs2025_bertimbau-large\", \"\"),\n",
        "    #Decoders\n",
        "  #(\"jbcs2025_llama31_8b-balanced\", \"essay_only\"), (\"jbcs2025_phi35-balanced\", \"essay_only\"), (\"jbcs2025_phi4-balanced\", \"essay_only\"),\n",
        "  # (\"jbcs2025_Llama-3.1-8B-llama31\", \"full_context\"), (\"jbcs2025_Phi-3.5-mini-instruct\", \"full_context\"), (\"jbcs2025_phi-4-phi4\", \"full_context\"),\n",
        "    #Sabias\n",
        "   (\"sabia-3-zero-shot\", \"essay_only\"), (\"sabia-3-zero-shot\", \"full_context\"),\n",
        "    #(\"sabia-3-grader-zero-shot\", \"essay_only\"),\n",
        "     (\"sabia-3-grader-zero-shot\", \"full_context\"),\n",
        "    #Gpts\n",
        "    (\"gpt-4o-2024-11-20-zero-shot\", \"essay_only\"), (\"gpt-4o-2024-11-20-zero-shot\", \"full_context\"),\n",
        "    (\"gpt-4o-2024-11-20-grader-zero-shot\", \"essay_only\"), (\"gpt-4o-2024-11-20-grader-zero-shot\", \"full_context\"),\n",
        "    #Deepseeks\n",
        "    (\"deepseek-reasoner-zero-shot-\", \"essay_only\"), (\"deepseek-reasoner-zero-shot-\", \"full_context\"),\n",
        "    (\"Deepseek-reasoner-grader-zero-shot-\", \"essay_only\")       ]\n",
        "SELECT = [\"QWK_lower_95ci\", \"QWK_upper_95ci\"]\n",
        "for (modelo, final) in modelos:\n",
        "  filtrar_dfs(df, df1, modelo, final=final)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2_4fdVNNARGi",
        "outputId": "6e6b6032-97e7-495d-993c-9f5c8a8f49d3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "jbcs2025_mbert_base- & 0.33 & 0.56 &0.00 & 0.29 &0.10 & 0.42 &0.12 & 0.42 &0.46 & 0.67 \\\\\n",
            "jbcs2025_bertimbau_base-- & 0.58 & 0.76 &0.28 & 0.55 &0.21 & 0.48 &0.51 & 0.73 &0.34 & 0.59 \\\\\n",
            "jbcs2025_bertimbau-large- & 0.61 & 0.79 &0.27 & 0.57 &0.13 & 0.40 &0.46 & 0.67 &0.35 & 0.60 \\\\\n",
            "sabia-3-zero-shot-essay_only & 0.61 & 0.75 &-0.03 & 0.06 &0.16 & 0.43 &0.39 & 0.62 &0.37 & 0.63 \\\\\n",
            "sabia-3-zero-shot-full_context & 0.47 & 0.66 &0.31 & 0.60 &0.32 & 0.56 &0.23 & 0.52 &0.40 & 0.65 \\\\\n",
            "sabia-3-grader-zero-shot-full_context & 0.40 & 0.58 &0.35 & 0.62 &0.42 & 0.65 &0.39 & 0.64 &0.34 & 0.63 \\\\\n",
            "gpt-4o-2024-11-20-zero-shot-essay_only & 0.43 & 0.58 &0.05 & 0.34 &0.24 & 0.51 &0.40 & 0.60 &0.41 & 0.67 \\\\\n",
            "gpt-4o-2024-11-20-zero-shot-full_context & 0.39 & 0.56 &0.38 & 0.63 &0.42 & 0.63 &0.38 & 0.58 &0.20 & 0.49 \\\\\n",
            "gpt-4o-2024-11-20-grader-zero-shot-essay_only & 0.40 & 0.56 &0.15 & 0.42 &0.15 & 0.41 &0.40 & 0.60 &0.41 & 0.66 \\\\\n",
            "gpt-4o-2024-11-20-grader-zero-shot-full_context & 0.47 & 0.63 &0.43 & 0.68 &0.32 & 0.56 &0.39 & 0.60 &0.33 & 0.61 \\\\\n",
            "deepseek-reasoner-zero-shot--essay_only & 0.27 & 0.43 &-0.14 & 0.11 &0.23 & 0.53 &0.41 & 0.61 &0.41 & 0.67 \\\\\n",
            "deepseek-reasoner-zero-shot--full_context & 0.27 & 0.43 &0.30 & 0.53 &0.65 & 0.80 &0.39 & 0.58 &0.45 & 0.69 \\\\\n",
            "Deepseek-reasoner-grader-zero-shot--essay_only & 0.36 & 0.51 &-0.01 & 0.08 &0.23 & 0.54 &0.37 & 0.56 &0.40 & 0.68 \\\\\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "SELECT = [\"Macro_F1_lower_95ci\", \"Macro_F1_upper_95ci\"]\n",
        "for (modelo, final) in modelos:\n",
        "  filtrar_dfs(df, df1, modelo, final=final)"
      ],
      "metadata": {
        "id": "dhyTkpR2ARwp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "68fd310f-a620-4e32-ea7a-b1855c592f3b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "jbcs2025_mbert_base- & 0.26 & 0.46 &0.16 & 0.33 &0.11 & 0.23 &0.13 & 0.29 &0.25 & 0.40 \\\\\n",
            "jbcs2025_bertimbau_base-- & 0.36 & 0.62 &0.22 & 0.40 &0.20 & 0.37 &0.29 & 0.59 &0.15 & 0.27 \\\\\n",
            "jbcs2025_bertimbau-large- & 0.39 & 0.66 &0.21 & 0.37 &0.14 & 0.29 &0.23 & 0.43 &0.23 & 0.41 \\\\\n",
            "jbcs2025_llama31_8b-balanced-essay_only & 0.37 & 0.64 &0.16 & 0.29 &0.21 & 0.38 &0.27 & 0.52 &0.16 & 0.29 \\\\\n",
            "jbcs2025_phi35-balanced-essay_only & 0.42 & 0.70 &0.16 & 0.37 &0.20 & 0.38 &0.24 & 0.48 &0.22 & 0.34 \\\\\n",
            "jbcs2025_phi4-balanced-essay_only & 0.39 & 0.65 &0.23 & 0.39 &0.15 & 0.31 &0.23 & 0.40 &0.14 & 0.25 \\\\\n",
            "jbcs2025_Llama-3.1-8B-llama31-full_context & 0.35 & 0.56 &0.17 & 0.32 &0.19 & 0.34 &0.23 & 0.46 &0.14 & 0.35 \\\\\n",
            "jbcs2025_Phi-3.5-mini-instruct-full_context & 0.09 & 0.18 &0.09 & 0.18 &0.33 & 0.56 &0.21 & 0.43 &0.19 & 0.33 \\\\\n",
            "jbcs2025_phi-4-phi4-full_context & 0.30 & 0.53 &0.18 & 0.31 &0.22 & 0.35 &0.24 & 0.40 &0.25 & 0.45 \\\\\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "zyaRKx5oChUn"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from torch import nn\n",
        "import torch"
      ],
      "metadata": {
        "id": "cB-6u1yyChfR"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "m = nn.Sigmoid()\n",
        "input = torch.randn(2)\n",
        "print(input)\n",
        "output = m(torch.tensor(-100))\n",
        "output"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IJ4ySLhpClfe",
        "outputId": "b48b901d-fdec-4515-830c-8a1069d2b807"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "tensor([ 1.5602, -1.1152])\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor(0.)"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "kvuSSIcCCuOa"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}