Upload 2 files

cmexam_dataloading.ipynb

@@ -0,0 +1,152 @@
+{
+ "cells": [
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "生成答案是原始文本的Multiple Choice的prompt（考虑了多种语言的格式，使用4options）",
+   "id": "f205851d8d2c3a64"
+  },
+  {
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2024-09-09T10:57:51.564790Z",
+     "start_time": "2024-09-09T10:57:42.760548Z"
+    }
+   },
+   "cell_type": "code",
+   "source": [
+    "from datasets import load_dataset\n",
+    "import os\n",
+    "import json\n",
+    "\n",
+    "# Load the dataset\n",
+    "dataset = load_dataset(\"fzkuji/cmexam\", trust_remote_code=True)\n",
+    "\n",
+    "# Define the save path\n",
+    "save_path = f\"./data/llama-factory\"  # Change this path to your local directory\n",
+    "os.makedirs(save_path, exist_ok=True)\n",
+    "\n",
+    "# Function to save data as JSON with specified columns\n",
+    "def save_as_json(data, filename):\n",
+    "    file_path = os.path.join(save_path, filename)\n",
+    "    \n",
+    "    data_to_save = []\n",
+    "    for item in data:\n",
+    "        # Extract the option texts and generate option letters dynamically\n",
+    "        option_texts = []\n",
+    "        option_letters = []\n",
+    "        for idx, option in enumerate(item['Options']):\n",
+    "            # Handle different data structures for options\n",
+    "            if isinstance(option, dict):\n",
+    "                option_text = option.get('value', '')\n",
+    "            else:\n",
+    "                option_text = str(option)\n",
+    "            option_texts.append(option_text)\n",
+    "            option_letters.append(chr(65 + idx))  # Generate letters 'A', 'B', etc.\n",
+    "        \n",
+    "        # Create a mapping for the current item\n",
+    "        dict_num = {letter: idx for idx, letter in enumerate(option_letters)}\n",
+    "        \n",
+    "        # Process the answer keys, assuming Answer is a single string of concatenated letters (e.g., \"ABC\")\n",
+    "        answer_keys = item['Answer']\n",
+    "        if isinstance(answer_keys, str):\n",
+    "            # Split the answer string into individual letters (e.g., \"ABC\" -> ['A', 'B', 'C'])\n",
+    "            answer_keys = list(answer_keys)\n",
+    "        elif isinstance(answer_keys, list):\n",
+    "            pass  # Already a list\n",
+    "        elif answer_keys is None:\n",
+    "            print(f\"Warning: Answer is None for item '{item['Question']}', skipping...\")\n",
+    "            continue  # Skip items with no answer\n",
+    "        else:\n",
+    "            # Skip if 'Answer' is not a string or list\n",
+    "            print(f\"Warning: Unexpected type for 'Answer' in item '{item['Question']}': {type(answer_keys)}\")\n",
+    "            continue\n",
+    "        \n",
+    "        # Map the answer keys to the actual option texts, but skip keys that are out of range\n",
+    "        answer_text = []\n",
+    "        for ans in answer_keys:\n",
+    "            if ans in dict_num:\n",
+    "                idx = dict_num[ans]\n",
+    "                answer_text.append(option_texts[idx])\n",
+    "            else:\n",
+    "                print(f\"Warning: Answer '{ans}' not found in options for item '{item['Question']}'\")\n",
+    "        \n",
+    "        # Construct the input text\n",
+    "        input_text = f\"问题：{item['Question']}\\n选项：\\n\" + \"\\n\".join(\n",
+    "            [f\"\\t{letter}. {text}。\" for letter, text in zip(option_letters, option_texts)]\n",
+    "        )\n",
+    "        \n",
+    "        # Add the instruction, input, and output to the data\n",
+    "        if answer_text:\n",
+    "            data_to_save.append({\n",
+    "                \"instruction\": \"假设您是一名医生，请回答以下选择题。请您输出答案的文本内容（不包含选项序号）。\",\n",
+    "                \"input\": input_text,\n",
+    "                \"output\": \"，\".join(answer_text)  # Join multiple answers with commas\n",
+    "            })\n",
+    "        else:\n",
+    "            print(f\"Warning: No valid answers for item '{item['Question']}', skipping...\")\n",
+    "    \n",
+    "    # Write the modified data to a JSON file\n",
+    "    with open(file_path, 'w', encoding='utf-8') as f:\n",
+    "        json.dump(data_to_save, f, ensure_ascii=False, indent=4)\n",
+    "\n",
+    "# Save the modified data for train, validation, and test splits\n",
+    "save_as_json(dataset['train'], 'train.json')\n",
+    "save_as_json(dataset['validation'], 'validation.json')\n",
+    "save_as_json(dataset['test'], 'test.json')\n"
+   ],
+   "id": "54192ba87c09ac2b",
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Warning: Answer 'E' not found in options for item '不属于阿片类镇痛药的是'\n",
+      "Warning: No valid answers for item '不属于阿片类镇痛药的是', skipping...\n",
+      "Warning: Answer 'E' not found in options for item '结构中含有��个手性碳原子，有四个异构体的药物是'\n",
+      "Warning: No valid answers for item '结构中含有两个手性碳原子，有四个异构体的药物是', skipping...\n",
+      "Warning: Answer 'E' not found in options for item '将苯丙氨酸引入氮芥结构中得到美法伦的目的是'\n",
+      "Warning: No valid answers for item '将苯丙氨酸引入氮芥结构中得到美法伦的目的是', skipping...\n",
+      "Warning: Answer 'E' not found in options for item '化学名为4-（2-氨基乙基）-1、2-苯二酚盐酸盐的药物是'\n",
+      "Warning: No valid answers for item '化学名为4-（2-氨基乙基）-1、2-苯二酚盐酸盐的药物是', skipping...\n",
+      "Warning: Answer 'E' not found in options for item '非处方药的使用要求'\n",
+      "Warning: Answer 'E' not found in options for item '面对患者，药学服务的重要人群有'\n",
+      "Warning: Answer 'E' not found in options for item '针刺用毫针的常用消毒方法有'\n",
+      "Warning: Answer is None for item '由国家制定，各省可根据当地经济水平、医疗需求和用药习惯适当进行调整的是', skipping...\n",
+      "Warning: Answer 'E' not found in options for item '分子中含有季铵结构，中枢作用较弱的药物是'\n",
+      "Warning: No valid answers for item '分子中含有季铵结构，中枢作用较弱的药物是', skipping...\n",
+      "Warning: Answer 'E' not found in options for item '在体内经代谢后，其代谢产物具有活性的药物是'\n"
+     ]
+    }
+   ],
+   "execution_count": 1
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "备注，有Warning的原因是有些题目的答案为空，或者选项不完整。这些题目会被跳过，不会被保存到JSON文件中。",
+   "id": "c94e6ffe4fd9be8d"
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

cmexam_preprocessing.ipynb

@@ -1,78 +1,78 @@
-{
- "cells": [
-  {
-   "metadata": {
-    "ExecuteTime": {
-     "end_time": "2024-09-08T15:26:59.355018Z",
-     "start_time": "2024-09-08T15:26:57.990909Z"
-    }
-   },
-   "cell_type": "code",
-   "source": [
-    "import pandas as pd\n",
-    "\n",
-    "# 定义一个函数，将 Options 字符串转换为 key-value 格式的列表\n",
-    "def format_options(options_str):\n",
-    "    options_list = []\n",
-    "    # 按行分割选项\n",
-    "    options_lines = options_str.split(\"\\n\")\n",
-    "    for line in options_lines:\n",
-    "        if len(line) > 1:\n",
-    "            key = line[0]  # 第一个字符为选项字母\n",
-    "            value = line[2:].strip()  # 从第三个字符开始为选项内容\n",
-    "            options_list.append({\"key\": key, \"value\": value})\n",
-    "    return options_list\n",
-    "\n",
-    "# 读取 CSV 文件\n",
-    "train_data = pd.read_csv('./data/train.csv', encoding='utf-8')\n",
-    "valid_data = pd.read_csv('./data/val.csv', encoding='utf-8')\n",
-    "test_data = pd.read_csv('./data/test_with_annotations.csv', encoding='utf-8')\n",
-    "\n",
-    "# 将json数据只保留与train和valid一致的字段\n",
-    "test_data = test_data[['Question', 'Options', 'Answer', 'Explanation']]\n",
-    "\n",
-    "# 遍历数据集，将每个样本的 Options 列格式化\n",
-    "train_data['Options'] = train_data['Options'].apply(format_options)\n",
-    "valid_data['Options'] = valid_data['Options'].apply(format_options)\n",
-    "test_data['Options'] = test_data['Options'].apply(format_options)\n",
-    "\n",
-    "# 将修改后的 DataFrame 保存为 JSON 文件\n",
-    "train_data.to_json('./data/train.json', orient='records', lines=True, force_ascii=False)\n",
-    "valid_data.to_json('./data/valid.json', orient='records', lines=True, force_ascii=False)\n",
-    "test_data.to_json('./data/test.json', orient='records', lines=True, force_ascii=False)\n"
-   ],
-   "id": "c003560dea95e12e",
-   "outputs": [],
-   "execution_count": 5
-  },
-  {
-   "metadata": {},
-   "cell_type": "code",
-   "outputs": [],
-   "execution_count": null,
-   "source": "",
-   "id": "8d142ed01196946"
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "Python 3",
-   "language": "python",
-   "name": "python3"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 2
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython2",
-   "version": "2.7.6"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 5
-}
+{
+ "cells": [
+  {
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2024-09-08T15:26:59.355018Z",
+     "start_time": "2024-09-08T15:26:57.990909Z"
+    }
+   },
+   "cell_type": "code",
+   "source": [
+    "import pandas as pd\n",
+    "\n",
+    "# 定义一个函数，将 Options 字符串转换为 key-value 格式的列表\n",
+    "def format_options(options_str):\n",
+    "    options_list = []\n",
+    "    # 按行分割选项\n",
+    "    options_lines = options_str.split(\"\\n\")\n",
+    "    for line in options_lines:\n",
+    "        if len(line) > 1:\n",
+    "            key = line[0]  # 第一个字符为选项字母\n",
+    "            value = line[2:].strip()  # 从第三个字符开始为选项内容\n",
+    "            options_list.append({\"key\": key, \"value\": value})\n",
+    "    return options_list\n",
+    "\n",
+    "# 读取 CSV 文件\n",
+    "train_data = pd.read_csv('./data/train.csv', encoding='utf-8')\n",
+    "valid_data = pd.read_csv('./data/val.csv', encoding='utf-8')\n",
+    "test_data = pd.read_csv('./data/test_with_annotations.csv', encoding='utf-8')\n",
+    "\n",
+    "# 将json数据只保留与train和valid一致的字段\n",
+    "test_data = test_data[['Question', 'Options', 'Answer', 'Explanation']]\n",
+    "\n",
+    "# 遍历数据集，将每个样本的 Options 列格式化\n",
+    "train_data['Options'] = train_data['Options'].apply(format_options)\n",
+    "valid_data['Options'] = valid_data['Options'].apply(format_options)\n",
+    "test_data['Options'] = test_data['Options'].apply(format_options)\n",
+    "\n",
+    "# 将修改后的 DataFrame 保存为 JSON 文件\n",
+    "train_data.to_json('./data/train.json', orient='records', lines=True, force_ascii=False)\n",
+    "valid_data.to_json('./data/valid.json', orient='records', lines=True, force_ascii=False)\n",
+    "test_data.to_json('./data/test.json', orient='records', lines=True, force_ascii=False)\n"
+   ],
+   "id": "c003560dea95e12e",
+   "outputs": [],
+   "execution_count": 5
+  },
+  {
+   "metadata": {},
+   "cell_type": "code",
+   "outputs": [],
+   "execution_count": null,
+   "source": "",
+   "id": "8d142ed01196946"
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}