diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..4760c9b
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,138 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+pdf/
+tnews-finetuning*
+contribute.md
+*.drawio
+*.db
+content/metric.csv
+.DS_Store
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+.idea
diff --git a/content/5. 推断.ipynb b/content/5. 推断.ipynb
new file mode 100644
index 0000000..7ab3077
--- /dev/null
+++ b/content/5. 推断.ipynb	
@@ -0,0 +1,970 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "3630c235-f891-4874-bd0a-5277d4d6aa82",
+   "metadata": {},
+   "source": [
+    "# 推断\n",
+    "\n",
+    "在这节课中，你将从产品评论和新闻文章中推断情感和主题。\n",
+    "\n",
+    "这些任务可以看作是模型接收文本作为输入并执行某种分析的过程。这可能涉及提取标签、提取实体、理解文本情感等等。如果你想要从一段文本中提取正面或负面情感，在传统的机器学习工作流程中，需要收集标签数据集、训练模型、确定如何在云端部署模型并进行推断。这样做可能效果还不错，但是这个过程需要很多工作。而且对于每个任务，如情感分析、提取实体等等，都需要训练和部署单独的模型。\n",
+    "\n",
+    "大型语言模型的一个非常好的特点是，对于许多这样的任务，你只需要编写一个prompt即可开始产生结果，而不需要进行大量的工作。这极大地加快了应用程序开发的速度。你还可以只使用一个模型和一个 API 来执行许多不同的任务，而不需要弄清楚如何训练和部署许多不同的模型。\n",
+    "\n",
+    "\n",
+    "## 启动"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 73,
+   "id": "a821d943",
+   "metadata": {
+    "height": 132
+   },
+   "outputs": [],
+   "source": [
+    "import openai\n",
+    "import os\n",
+    "\n",
+    "OPENAI_API_KEY = os.environ.get(\"OPENAI_API_KEY2\")\n",
+    "openai.api_key = OPENAI_API_KEY"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 74,
+   "id": "e82f5577",
+   "metadata": {
+    "height": 164
+   },
+   "outputs": [],
+   "source": [
+    "def get_completion(prompt, model=\"gpt-3.5-turbo\"):\n",
+    "    messages = [{\"role\": \"user\", \"content\": prompt}]\n",
+    "    response = openai.ChatCompletion.create(\n",
+    "        model=model,\n",
+    "        messages=messages,\n",
+    "        temperature=0, # this is the degree of randomness of the model's output\n",
+    "    )\n",
+    "    return response.choices[0].message[\"content\"]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "51d2fdfa-c99f-4750-8574-dba7712cd7f0",
+   "metadata": {},
+   "source": [
+    "## 商品评论文本\n",
+    "\n",
+    "这是一盏台灯的评论。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "id": "b0f3b49b",
+   "metadata": {
+    "height": 200
+   },
+   "outputs": [],
+   "source": [
+    "lamp_review = \"\"\"\n",
+    "Needed a nice lamp for my bedroom, and this one had \\\n",
+    "additional storage and not too high of a price point. \\\n",
+    "Got it fast.  The string to our lamp broke during the \\\n",
+    "transit and the company happily sent over a new one. \\\n",
+    "Came within a few days as well. It was easy to put \\\n",
+    "together.  I had a missing part, so I contacted their \\\n",
+    "support and they very quickly got me the missing piece! \\\n",
+    "Lumina seems to me to be a great company that cares \\\n",
+    "about their customers and products!!\n",
+    "\"\"\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 78,
+   "id": "bc6260f0",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# 中文\n",
+    "lamp_review_zh = \"\"\"\n",
+    "我需要一盏漂亮的卧室灯，这款灯具有额外的储物功能，价格也不算太高。\\\n",
+    "我很快就收到了它。在运输过程中，我们的灯绳断了，但是公司很乐意寄送了一个新的。\\\n",
+    "几天后就收到了。这款灯很容易组装。我发现少了一个零件，于是联系了他们的客服，他们很快就给我寄来了缺失的零件！\\\n",
+    "在我看来，Lumina 是一家非常关心顾客和产品的优秀公司！\n",
+    "\"\"\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "30d6e4bd-3337-45a3-8c99-a734cdd06743",
+   "metadata": {},
+   "source": [
+    "## 情感（正向/负向）\n",
+    "\n",
+    "现在让我们来编写一个prompt来分类这个评论的情感。如果我想让系统告诉我这个评论的情感是什么，只需要编写 “以下产品评论的情感是什么” 这个prompt，加上通常的分隔符和评论文本等等。\n",
+    "\n",
+    "然后让我们运行一下。结果显示这个产品评论的情感是积极的，这似乎是非常正确的。虽然这盏台灯不完美，但这个客户似乎非常满意。这似乎是一家关心客户和产品的伟大公司，可以认为积极的情感似乎是正确的答案。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "id": "e3157601",
+   "metadata": {
+    "height": 149
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "The sentiment of the product review is positive.\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "What is the sentiment of the following product review, \n",
+    "which is delimited with triple backticks?\n",
+    "\n",
+    "Review text: '''{lamp_review}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 80,
+   "id": "ac5b0bb9",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "积极的情感。\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "prompt = f\"\"\"\n",
+    "以下用三重反引号分隔的产品评论的情感是什么？\n",
+    "\n",
+    "评论文本: '''{lamp_review_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "76be2320",
+   "metadata": {},
+   "source": [
+    "如果你想要给出更简洁的答案，以便更容易进行后处理，可以使用上面的prompt并添加另一个指令，以一个单词 “正面” 或 “负面” 的形式给出答案。这样就只会打印出 “正面” 这个单词，这使得文本更容易接受这个输出并进行处理。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "id": "acf9ca16",
+   "metadata": {
+    "height": 200
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "positive\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "What is the sentiment of the following product review, \n",
+    "which is delimited with triple backticks?\n",
+    "\n",
+    "Give your answer as a single word, either \"positive\" \\\n",
+    "or \"negative\".\n",
+    "\n",
+    "Review text: '''{lamp_review}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 82,
+   "id": "84a761b3",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "正面\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "以下用三重反引号分隔的产品评论的情感是什么？\n",
+    "\n",
+    "用一个单词回答：「正面」或「负面」。\n",
+    "\n",
+    "评论文本: '''{lamp_review_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "81d2a973-1fa4-4a35-ae35-a2e746c0e91b",
+   "metadata": {},
+   "source": [
+    "## 识别情感类型\n",
+    "\n",
+    "让我们看看另一个prompt，仍然使用台灯评论。这次我要让它识别出以下评论作者所表达的情感列表，不超过五个。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 33,
+   "id": "8aa7934b",
+   "metadata": {
+    "height": 183
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "happy, satisfied, grateful, impressed, content\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "Identify a list of emotions that the writer of the \\\n",
+    "following review is expressing. Include no more than \\\n",
+    "five items in the list. Format your answer as a list of \\\n",
+    "lower-case words separated by commas.\n",
+    "\n",
+    "Review text: '''{lamp_review}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 35,
+   "id": "e615c13a",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "满意,感激,赞扬,信任,愉快\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "prompt = f\"\"\"\n",
+    "识别以下评论的作者表达的情感。包含不超过五个项目。将答案格式化为以逗号分隔的单词列表。\n",
+    "\n",
+    "评论文本: '''{lamp_review_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "cc4444f7",
+   "metadata": {},
+   "source": [
+    "大型语言模型非常擅长从一段文本中提取特定的东西。在上面的例子中，评论正在表达情感，这可能有助于了解客户如何看待特定的产品。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a428d093-51c9-461c-b41e-114e80876409",
+   "metadata": {},
+   "source": [
+    "## 识别愤怒\n",
+    "\n",
+    "对于很多企业来说，了解某个顾客是否非常生气很重要。所以你可能有一个类似这样的分类问题：以下评论的作者是否表达了愤怒情绪？因为如果有人真的很生气，那么可能值得额外关注，让客户支持或客户成功团队联系客户以了解情况，并为客户解决问题。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 37,
+   "id": "dba1a538",
+   "metadata": {
+    "height": 166
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "No\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "Is the writer of the following review expressing anger?\\\n",
+    "The review is delimited with triple backticks. \\\n",
+    "Give your answer as either yes or no.\n",
+    "\n",
+    "Review text: '''{lamp_review}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "85bad324",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "否\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "prompt = f\"\"\"\n",
+    "以下评论的作者是否表达了愤怒？评论用三重反引号分隔。给出是或否的答案。\n",
+    "\n",
+    "评论文本: '''{lamp_review_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "11ca57a2",
+   "metadata": {},
+   "source": [
+    "上面这个例子中，客户并没有生气。注意，如果使用常规的监督学习，如果想要建立所有这些分类器，不可能在几分钟内就做到这一点。我们鼓励大家尝试更改一些这样的prompt，也许询问客户是否表达了喜悦，或者询问是否有任何遗漏的部分，并看看是否可以让prompt对这个灯具评论做出不同的推论。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "936a771e-ca78-4e55-8088-2da6f3820ddc",
+   "metadata": {},
+   "source": [
+    "## 从客户评论中提取产品和公司名称\n",
+    "\n",
+    "接下来，让我们从客户评论中提取更丰富的信息。信息提取是自然语言处理（NLP）的一部分，与从文本中提取你想要知道的某些事物相关。因此，在这个prompt中，我要求它识别以下内容：购买物品和制造物品的公司名称。\n",
+    "\n",
+    "同样，如果你试图总结在线购物电子商务网站的许多评论，对于这些评论来说，弄清楚是什么物品，谁制造了该物品，弄清楚积极和消极的情感，以跟踪特定物品或特定制造商的积极或消极情感趋势，可能会很有用。\n",
+    "\n",
+    "在下面这个示例中，我们要求它将响应格式化为一个 JSON 对象，其中物品和品牌是键。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 39,
+   "id": "a13bea1b",
+   "metadata": {
+    "height": 285
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{\n",
+      "  \"Item\": \"lamp\",\n",
+      "  \"Brand\": \"Lumina\"\n",
+      "}\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "Identify the following items from the review text: \n",
+    "- Item purchased by reviewer\n",
+    "- Company that made the item\n",
+    "\n",
+    "The review is delimited with triple backticks. \\\n",
+    "Format your response as a JSON object with \\\n",
+    "\"Item\" and \"Brand\" as the keys. \n",
+    "If the information isn't present, use \"unknown\" \\\n",
+    "as the value.\n",
+    "Make your response as short as possible.\n",
+    "  \n",
+    "Review text: '''{lamp_review}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 83,
+   "id": "e9ffe056",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{\n",
+      "  \"物品\": \"卧室灯\",\n",
+      "  \"品牌\": \"Lumina\"\n",
+      "}\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "prompt = f\"\"\"\n",
+    "从评论文本中识别以下项目：\n",
+    "- 评论者购买的物品\n",
+    "- 制造该物品的公司\n",
+    "\n",
+    "评论文本用三重反引号分隔。将你的响应格式化为以 “物品” 和 “品牌” 为键的 JSON 对象。\n",
+    "如果信息不存在，请使用 “未知” 作为值。\n",
+    "让你的回应尽可能简短。\n",
+    "  \n",
+    "评论文本: '''{lamp_review_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "954d125d",
+   "metadata": {},
+   "source": [
+    "如上所示，它会说这个物品是一个卧室灯，品牌是 Luminar，你可以轻松地将其加载到 Python 字典中，然后对此输出进行其他处理。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a38880a5-088f-4609-9913-f8fa41fb7ba0",
+   "metadata": {},
+   "source": [
+    "## 一次完成多项任务\n",
+    "\n",
+    "提取上面所有这些信息使用了 3 或 4 个prompt，但实际上可以编写单个prompt来同时提取所有这些信息。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 43,
+   "id": "e7dda9e5",
+   "metadata": {
+    "height": 336
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{\n",
+      "  \"Sentiment\": \"positive\",\n",
+      "  \"Anger\": false,\n",
+      "  \"Item\": \"lamp with additional storage\",\n",
+      "  \"Brand\": \"Lumina\"\n",
+      "}\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "Identify the following items from the review text: \n",
+    "- Sentiment (positive or negative)\n",
+    "- Is the reviewer expressing anger? (true or false)\n",
+    "- Item purchased by reviewer\n",
+    "- Company that made the item\n",
+    "\n",
+    "The review is delimited with triple backticks. \\\n",
+    "Format your response as a JSON object with \\\n",
+    "\"Sentiment\", \"Anger\", \"Item\" and \"Brand\" as the keys.\n",
+    "If the information isn't present, use \"unknown\" \\\n",
+    "as the value.\n",
+    "Make your response as short as possible.\n",
+    "Format the Anger value as a boolean.\n",
+    "\n",
+    "Review text: '''{lamp_review}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 45,
+   "id": "939c2b0e",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{\n",
+      "    \"Sentiment\": \"正面\",\n",
+      "    \"Anger\": false,\n",
+      "    \"Item\": \"卧室灯\",\n",
+      "    \"Brand\": \"Lumina\"\n",
+      "}\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "prompt = f\"\"\"\n",
+    "从评论文本中识别以下项目：\n",
+    "- 情绪（正面或负面）\n",
+    "- 审稿人是否表达了愤怒？（是或否）\n",
+    "- 评论者购买的物品\n",
+    "- 制造该物品的公司\n",
+    "\n",
+    "评论用三重反引号分隔。将您的响应格式化为 JSON 对象，以 “Sentiment”、“Anger”、“Item” 和 “Brand” 作为键。\n",
+    "如果信息不存在，请使用 “未知” 作为值。\n",
+    "让你的回应尽可能简短。\n",
+    "将 Anger 值格式化为布尔值。\n",
+    "\n",
+    "评论文本: '''{lamp_review_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "5e09a673",
+   "metadata": {},
+   "source": [
+    "这个例子中，我们告诉它将愤怒值格式化为布尔值，然后输出一个 JSON。大家可以自己尝试不同的变化，或者甚至尝试完全不同的评论，看看是否仍然可以准确地提取这些内容。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "235fc223-2c89-49ec-ac2d-78a8e74a43ac",
+   "metadata": {},
+   "source": [
+    "## 推断主题\n",
+    "\n",
+    "大型语言模型的一个很酷的应用是推断主题。给定一段长文本，这段文本是关于什么的？有什么话题？"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 56,
+   "id": "8a74cc3e",
+   "metadata": {
+    "height": 472
+   },
+   "outputs": [],
+   "source": [
+    "story = \"\"\"\n",
+    "In a recent survey conducted by the government, \n",
+    "public sector employees were asked to rate their level \n",
+    "of satisfaction with the department they work at. \n",
+    "The results revealed that NASA was the most popular \n",
+    "department with a satisfaction rating of 95%.\n",
+    "\n",
+    "One NASA employee, John Smith, commented on the findings, \n",
+    "stating, \"I'm not surprised that NASA came out on top. \n",
+    "It's a great place to work with amazing people and \n",
+    "incredible opportunities. I'm proud to be a part of \n",
+    "such an innovative organization.\"\n",
+    "\n",
+    "The results were also welcomed by NASA's management team, \n",
+    "with Director Tom Johnson stating, \"We are thrilled to \n",
+    "hear that our employees are satisfied with their work at NASA. \n",
+    "We have a talented and dedicated team who work tirelessly \n",
+    "to achieve our goals, and it's fantastic to see that their \n",
+    "hard work is paying off.\"\n",
+    "\n",
+    "The survey also revealed that the \n",
+    "Social Security Administration had the lowest satisfaction \n",
+    "rating, with only 45% of employees indicating they were \n",
+    "satisfied with their job. The government has pledged to \n",
+    "address the concerns raised by employees in the survey and \n",
+    "work towards improving job satisfaction across all departments.\n",
+    "\"\"\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 57,
+   "id": "811ff13f",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# 中文\n",
+    "story_zh = \"\"\"\n",
+    "在政府最近进行的一项调查中，要求公共部门的员工对他们所在部门的满意度进行评分。\n",
+    "调查结果显示，NASA 是最受欢迎的部门，满意度为 95％。\n",
+    "\n",
+    "一位 NASA 员工 John Smith 对这一发现发表了评论，他表示：\n",
+    "“我对 NASA 排名第一并不感到惊讶。这是一个与了不起的人们和令人难以置信的机会共事的好地方。我为成为这样一个创新组织的一员感到自豪。”\n",
+    "\n",
+    "NASA 的管理团队也对这一结果表示欢迎，主管 Tom Johnson 表示：\n",
+    "“我们很高兴听到我们的员工对 NASA 的工作感到满意。\n",
+    "我们拥有一支才华横溢、忠诚敬业的团队，他们为实现我们的目标不懈努力，看到他们的辛勤工作得到回报是太棒了。”\n",
+    "\n",
+    "调查还显示，社会保障管理局的满意度最低，只有 45％的员工表示他们对工作满意。\n",
+    "政府承诺解决调查中员工提出的问题，并努力提高所有部门的工作满意度。\n",
+    "\"\"\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a8ea91d6-e841-4ee2-bed9-ca4a36df177f",
+   "metadata": {},
+   "source": [
+    "## 推断5个主题\n",
+    "\n",
+    "上面是一篇虚构的关于政府工作人员对他们工作机构感受的报纸文章。我们可以让它确定五个正在讨论的主题，用一两个字描述每个主题，并将输出格式化为逗号分隔的列表。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 58,
+   "id": "5c267cbe",
+   "metadata": {
+    "height": 217
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "government survey, job satisfaction, NASA, Social Security Administration, employee concerns\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "Determine five topics that are being discussed in the \\\n",
+    "following text, which is delimited by triple backticks.\n",
+    "\n",
+    "Make each item one or two words long. \n",
+    "\n",
+    "Format your response as a list of items separated by commas.\n",
+    "\n",
+    "Text sample: '''{story}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 59,
+   "id": "f92f90fe",
+   "metadata": {
+    "height": 30,
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['government survey',\n",
+       " ' job satisfaction',\n",
+       " ' NASA',\n",
+       " ' Social Security Administration',\n",
+       " ' employee concerns']"
+      ]
+     },
+     "execution_count": 59,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "response.split(sep=',')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 64,
+   "id": "cab27b65",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "调查结果, NASA, 社会保障管理局, 员工满意度, 政府承诺\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "prompt = f\"\"\"\n",
+    "确定以下给定文本中讨论的五个主题。\n",
+    "\n",
+    "每个主题用1-2个单词概括。\n",
+    "\n",
+    "输出时用逗号分割每个主题。\n",
+    "\n",
+    "给定文本: '''{story_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "34be1d2a-1309-4512-841a-b6f67338938b",
+   "metadata": {},
+   "source": [
+    "## 为特定主题制作新闻提醒\n",
+    "\n",
+    "假设我们有一个新闻网站或类似的东西，这是我们感兴趣的主题：NASA、地方政府、工程、员工满意度、联邦政府等。假设我们想弄清楚，针对一篇新闻文章，其中涵盖了哪些主题。可以使用这样的prompt：确定以下主题列表中的每个项目是否是以下文本中的主题。以 0 或 1 的形式给出答案列表。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 84,
+   "id": "94b8fa65",
+   "metadata": {
+    "height": 81
+   },
+   "outputs": [],
+   "source": [
+    "topic_list = [\n",
+    "    \"nasa\", \"local government\", \"engineering\", \n",
+    "    \"employee satisfaction\", \"federal government\"\n",
+    "]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 67,
+   "id": "626c5b8e",
+   "metadata": {
+    "height": 234
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "nasa: 1\n",
+      "local government: 0\n",
+      "engineering: 0\n",
+      "employee satisfaction: 1\n",
+      "federal government: 1\n"
+     ]
+    }
+   ],
+   "source": [
+    "prompt = f\"\"\"\n",
+    "Determine whether each item in the following list of \\\n",
+    "topics is a topic in the text below, which\n",
+    "is delimited with triple backticks.\n",
+    "\n",
+    "Give your answer as list with 0 or 1 for each topic.\\\n",
+    "\n",
+    "List of topics: {\", \".join(topic_list)}\n",
+    "\n",
+    "Text sample: '''{story}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 68,
+   "id": "902a7c74",
+   "metadata": {
+    "height": 79
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "ALERT: New NASA story!\n"
+     ]
+    }
+   ],
+   "source": [
+    "topic_dict = {i.split(': ')[0]: int(i.split(': ')[1]) for i in response.split(sep='\\n')}\n",
+    "if topic_dict['nasa'] == 1:\n",
+    "    print(\"ALERT: New NASA story!\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 71,
+   "id": "9f53d337",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "美国航空航天局：1\n",
+      "地方政府：0\n",
+      "工程：0\n",
+      "员工满意度：1\n",
+      "联邦政府：1\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "prompt = f\"\"\"\n",
+    "判断主题列表中的每一项是否是给定文本中的一个话题，\n",
+    "\n",
+    "以列表的形式给出答案，每个主题用 0 或 1。\n",
+    "\n",
+    "主题列表：美国航空航天局、地方政府、工程、员工满意度、联邦政府\n",
+    "\n",
+    "给定文本: '''{story_zh}'''\n",
+    "\"\"\"\n",
+    "response = get_completion(prompt)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "08247dbf",
+   "metadata": {},
+   "source": [
+    "所以，这个故事是关于 NASA 的。它不是关于地方政府的，不是关于工程的。它是关于员工满意度的，它是关于联邦政府的。这在机器学习中有时被称为 Zero-Shot 学习算法，因为我们没有给它任何标记的训练数据。仅凭prompt，它就能确定哪些主题在新闻文章中涵盖了。\n",
+    "\n",
+    "如果我们想生成一个新闻提醒，也可以使用这个处理新闻的过程。假设我非常喜欢 NASA 所做的工作，就可以构建一个这样的系统，每当 NASA 新闻出现时，输出提醒。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 72,
+   "id": "53bf1abd",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "提醒: 关于美国航空航天局的新消息\n"
+     ]
+    }
+   ],
+   "source": [
+    "topic_dict = {i.split('：')[0]: int(i.split('：')[1]) for i in response.split(sep='\\n')}\n",
+    "if topic_dict['美国航空航天局'] == 1:\n",
+    "    print(\"提醒: 关于美国航空航天局的新消息\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "76ccd189",
+   "metadata": {},
+   "source": [
+    "这就是关于推断的全部内容了，仅用几分钟时间，我们就可以构建多个用于对文本进行推理的系统，而以前则需要熟练的机器学习开发人员数天甚至数周的时间。这非常令人兴奋，无论是对于熟练的机器学习开发人员还是对于新手来说，都可以使用prompt来非常快速地构建和开始相当复杂的自然语言处理任务。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "f88408ae-469a-4b02-a043-f6b4f0b14bf9",
+   "metadata": {},
+   "source": [
+    "## 尝试你的实验！"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1bd3553f",
+   "metadata": {
+    "height": 30
+   },
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "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": "3.8.13"
+  },
+  "latex_envs": {
+   "LaTeX_envs_menu_present": true,
+   "autoclose": false,
+   "autocomplete": true,
+   "bibliofile": "biblio.bib",
+   "cite_by": "apalike",
+   "current_citInitial": 1,
+   "eqLabelWithNumbers": true,
+   "eqNumInitial": 1,
+   "hotkeys": {
+    "equation": "Ctrl-E",
+    "itemize": "Ctrl-I"
+   },
+   "labels_anchors": false,
+   "latex_user_defs": false,
+   "report_style_numbering": false,
+   "user_envs_cfg": false
+  },
+  "toc": {
+   "base_numbering": 1,
+   "nav_menu": {},
+   "number_sections": true,
+   "sideBar": true,
+   "skip_h1_title": false,
+   "title_cell": "Table of Contents",
+   "title_sidebar": "Contents",
+   "toc_cell": false,
+   "toc_position": {
+    "height": "calc(100% - 180px)",
+    "left": "10px",
+    "top": "150px",
+    "width": "256px"
+   },
+   "toc_section_display": true,
+   "toc_window_display": true
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/content/8. 聊天机器人.ipynb b/content/8. 聊天机器人.ipynb
new file mode 100644
index 0000000..5ab654b
--- /dev/null
+++ b/content/8. 聊天机器人.ipynb	
@@ -0,0 +1,1682 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "a9183228-0ba6-4af9-8430-649e28868253",
+   "metadata": {
+    "id": "JMXGlIvAwn30"
+   },
+   "source": [
+    "# 对话聊天\n",
+    "\n",
+    "使用一个大型语言模型的一个令人兴奋的事情是，我们可以用它来构建一个定制的聊天机器人，只需要很少的工作量。在这一节中，我们将探索如何利用聊天格式（接口）与个性化或专门针对特定任务或行为的聊天机器人进行延伸对话。\n",
+    "\n",
+    "\n",
+    "## 启动"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "7fa0d9b5",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "import openai\n",
+    "\n",
+    "OPENAI_API_KEY = os.environ.get(\"OPENAI_API_KEY2\")\n",
+    "openai.api_key = OPENAI_API_KEY"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "2c9b885b",
+   "metadata": {},
+   "source": [
+    "像 ChatGPT 这样的聊天模型实际上是组装成以一系列消息作为输入，并返回一个模型生成的消息作为输出的。虽然聊天格式的设计旨在使这种多轮对话变得容易，但我们通过之前的学习可以知道，它对于没有任何对话的单轮任务也同样有用。\n",
+    "\n",
+    "接下来，我们将定义两个辅助函数。第一个是单轮的，我们将prompt放入看起来像是某种用户消息的东西中。另一个则传入一个消息列表。这些消息可以来自不同的角色，我们会描述一下这些角色。\n",
+    "\n",
+    "第一条消息是一个系统消息，它提供了一个总体的指示，然后在这个消息之后，我们有用户和助手之间的交替。如果你曾经使用过 ChatGPT 网页界面，那么你的消息是用户消息，而 ChatGPT 的消息是助手消息。系统消息则有助于设置助手的行为和角色，并作为对话的高级指示。你可以想象它在助手的耳边低语，引导它的回应，而用户不会注意到系统消息。\n",
+    "\n",
+    "因此，作为用户，如果你曾经使用过 ChatGPT，你可能不知道 ChatGPT 的系统消息是什么，这是有意为之的。系统消息的好处是为开发者提供了一种方法，在不让请求本身成为对话的一部分的情况下，引导助手并指导其回应。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "f5308d65",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "def get_completion(prompt, model=\"gpt-3.5-turbo\"):\n",
+    "    messages = [{\"role\": \"user\", \"content\": prompt}]\n",
+    "    response = openai.ChatCompletion.create(\n",
+    "        model=model,\n",
+    "        messages=messages,\n",
+    "        temperature=0, # 控制模型输出的随机程度\n",
+    "    )\n",
+    "    return response.choices[0].message[\"content\"]\n",
+    "\n",
+    "def get_completion_from_messages(messages, model=\"gpt-3.5-turbo\", temperature=0):\n",
+    "    response = openai.ChatCompletion.create(\n",
+    "        model=model,\n",
+    "        messages=messages,\n",
+    "        temperature=temperature, # 控制模型输出的随机程度\n",
+    "    )\n",
+    "#     print(str(response.choices[0].message))\n",
+    "    return response.choices[0].message[\"content\"]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "46caaa5b",
+   "metadata": {},
+   "source": [
+    "现在让我们尝试在对话中使用这些消息。我们将使用上面的函数来获取从这些消息中得到的回答，同时，使用更高的 temperature（越高生成的越多样）。\n",
+    "\n",
+    "系统消息说，你是一个说话像莎士比亚的助手。这是我们向助手描述它应该如何表现的方式。然后，第一个用户消息是，给我讲个笑话。接下来的消息是，为什么鸡会过马路？然后最后一个用户消息是，我不知道。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "cee681b7",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "messages =  [  \n",
+    "{'role':'system', 'content':'You are an assistant that speaks like Shakespeare.'},    \n",
+    "{'role':'user', 'content':'tell me a joke'},   \n",
+    "{'role':'assistant', 'content':'Why did the chicken cross the road'},   \n",
+    "{'role':'user', 'content':'I don\\'t know'}  ]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "da45ea0f",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "To get to the other side, fair sir.\n"
+     ]
+    }
+   ],
+   "source": [
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "02b0e4d3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# 中文\n",
+    "messages =  [  \n",
+    "{'role':'system', 'content':'你是一个像莎士比亚一样说话的助手。'},    \n",
+    "{'role':'user', 'content':'给我讲个笑话'},   \n",
+    "{'role':'assistant', 'content':'鸡为什么过马路'},   \n",
+    "{'role':'user', 'content':'我不知道'}  ]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "65f80283",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "因为它要去找“母鸡”。哈哈哈！（注：此为英文双关语，\"chicken\"是鸡的意思，也是胆小的意思；\"cross the road\"是过马路的意思，也是“破坏规则”的意思。）\n"
+     ]
+    }
+   ],
+   "source": [
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "5f76bedb",
+   "metadata": {},
+   "source": [
+    "让我们做另一个例子。助手的消息是，你是一个友好的聊天机器人，第一个用户消息是，嗨，我叫Isa。我们想要得到第一个用户消息。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "ca733f8f",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Hello Isa! It's great to meet you. How can I assist you today?\n"
+     ]
+    }
+   ],
+   "source": [
+    "messages =  [  \n",
+    "{'role':'system', 'content':'You are friendly chatbot.'},    \n",
+    "{'role':'user', 'content':'Hi, my name is Isa'}  ]\n",
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "ca517ab0",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "嗨，Isa，很高兴见到你！有什么我可以帮助你的吗？\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "messages =  [  \n",
+    "{'role':'system', 'content':'你是个友好的聊天机器人。'},    \n",
+    "{'role':'user', 'content':'Hi, 我是Isa。'}  ]\n",
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "1e9f96ba",
+   "metadata": {},
+   "source": [
+    "让我们再试一个例子。系统消息是，你是一个友好的聊天机器人，第一个用户消息是，是的，你能提醒我我的名字是什么吗？"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "id": "0ae595bc",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "I'm sorry, but since we don't have any personal information about you, I don't know your name. Can you please tell me your name?\n"
+     ]
+    }
+   ],
+   "source": [
+    "messages =  [  \n",
+    "{'role':'system', 'content':'You are friendly chatbot.'},    \n",
+    "{'role':'user', 'content':'Yes,  can you remind me, What is my name?'}  ]\n",
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "id": "a606d422",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "抱歉，我不知道您的名字，因为我们是虚拟的聊天机器人和现实生活中的人类在不同的世界中。\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "messages =  [  \n",
+    "{'role':'system', 'content':'你是个友好的聊天机器人。'},    \n",
+    "{'role':'user', 'content':'好，你能提醒我，我的名字是什么吗？'}  ]\n",
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "05c65d16",
+   "metadata": {},
+   "source": [
+    "如上所见，模型实际上并不知道我的名字。\n",
+    "\n",
+    "因此，每次与语言模型的交互都是一个独立的交互，这意味着我们必须提供所有相关的消息，以便模型在当前对话中进行引用。如果想让模型引用或 “记住” 对话的早期部分，则必须在模型的输入中提供早期的交流。我们将其称为上下文。让我们试试。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "56cbb817",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Your name is Isa.\n"
+     ]
+    }
+   ],
+   "source": [
+    "messages =  [  \n",
+    "{'role':'system', 'content':'You are friendly chatbot.'},\n",
+    "{'role':'user', 'content':'Hi, my name is Isa'},\n",
+    "{'role':'assistant', 'content': \"Hi Isa! It's nice to meet you. \\\n",
+    "Is there anything I can help you with today?\"},\n",
+    "{'role':'user', 'content':'Yes, you can remind me, What is my name?'}  ]\n",
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "6019b1d5",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "当然可以！您的名字是Isa。\n"
+     ]
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "messages =  [  \n",
+    "{'role':'system', 'content':'你是个友好的聊天机器人。'},\n",
+    "{'role':'user', 'content':'Hi, 我是Isa'},\n",
+    "{'role':'assistant', 'content': \"Hi Isa! 很高兴认识你。今天有什么可以帮到你的吗?\"},\n",
+    "{'role':'user', 'content':'是的，你可以提醒我, 我的名字是什么?'}  ]\n",
+    "response = get_completion_from_messages(messages, temperature=1)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c1ed90a6",
+   "metadata": {},
+   "source": [
+    "现在我们已经给模型提供了上下文，也就是之前的对话中提到的我的名字，然后我们会问同样的问题，也就是我的名字是什么。因为模型有了需要的全部上下文，所以它能够做出回应，就像我们在输入的消息列表中看到的一样。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "dedba66a-58b0-40d4-b9ae-47e79ae22328",
+   "metadata": {
+    "id": "bBg_MpXeYnTq"
+   },
+   "source": [
+    "# 订餐机器人\n",
+    "\n",
+    "现在，我们构建一个 “订餐机器人”，我们需要它自动收集用户信息，接受比萨饼店的订单。\n",
+    "\n",
+    "下面这个函数将收集我们的用户消息，以便我们可以避免手动输入，就像我们在刚刚上面做的那样。这个函数将从我们下面构建的用户界面中收集提示，然后将其附加到一个名为上下文的列表中，并在每次调用模型时使用该上下文。模型的响应也会被添加到上下文中，所以模型消息和用户消息都被添加到上下文中，因此上下文逐渐变长。这样，模型就有了需要的信息来确定下一步要做什么。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "e76749ac",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "def collect_messages(_):\n",
+    "    prompt = inp.value_input\n",
+    "    inp.value = ''\n",
+    "    context.append({'role':'user', 'content':f\"{prompt}\"})\n",
+    "    response = get_completion_from_messages(context) \n",
+    "    context.append({'role':'assistant', 'content':f\"{response}\"})\n",
+    "    panels.append(\n",
+    "        pn.Row('User:', pn.pane.Markdown(prompt, width=600)))\n",
+    "    panels.append(\n",
+    "        pn.Row('Assistant:', pn.pane.Markdown(response, width=600, style={'background-color': '#F6F6F6'})))\n",
+    " \n",
+    "    return pn.Column(*panels)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "8a3b003e",
+   "metadata": {},
+   "source": [
+    "现在，我们将设置并运行这个 UI 来显示订单机器人。初始的上下文包含了包含菜单的系统消息。请注意，上下文会随着时间的推移而不断增长。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 43,
+   "id": "c9e746f5",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [
+    {
+     "data": {
+      "application/javascript": [
+       "(function(root) {\n",
+       "  function now() {\n",
+       "    return new Date();\n",
+       "  }\n",
+       "\n",
+       "  var force = true;\n",
+       "\n",
+       "  if (typeof root._bokeh_onload_callbacks === \"undefined\" || force === true) {\n",
+       "    root._bokeh_onload_callbacks = [];\n",
+       "    root._bokeh_is_loading = undefined;\n",
+       "  }\n",
+       "\n",
+       "  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n",
+       "    root._bokeh_timeout = Date.now() + 5000;\n",
+       "    root._bokeh_failed_load = false;\n",
+       "  }\n",
+       "\n",
+       "  function run_callbacks() {\n",
+       "    try {\n",
+       "      root._bokeh_onload_callbacks.forEach(function(callback) {\n",
+       "        if (callback != null)\n",
+       "          callback();\n",
+       "      });\n",
+       "    } finally {\n",
+       "      delete root._bokeh_onload_callbacks\n",
+       "    }\n",
+       "    console.debug(\"Bokeh: all callbacks have finished\");\n",
+       "  }\n",
+       "\n",
+       "  function load_libs(css_urls, js_urls, js_modules, callback) {\n",
+       "    if (css_urls == null) css_urls = [];\n",
+       "    if (js_urls == null) js_urls = [];\n",
+       "    if (js_modules == null) js_modules = [];\n",
+       "\n",
+       "    root._bokeh_onload_callbacks.push(callback);\n",
+       "    if (root._bokeh_is_loading > 0) {\n",
+       "      console.debug(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n",
+       "      return null;\n",
+       "    }\n",
+       "    if (js_urls.length === 0 && js_modules.length === 0) {\n",
+       "      run_callbacks();\n",
+       "      return null;\n",
+       "    }\n",
+       "    console.debug(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n",
+       "\n",
+       "    function on_load() {\n",
+       "      root._bokeh_is_loading--;\n",
+       "      if (root._bokeh_is_loading === 0) {\n",
+       "        console.debug(\"Bokeh: all BokehJS libraries/stylesheets loaded\");\n",
+       "        run_callbacks()\n",
+       "      }\n",
+       "    }\n",
+       "\n",
+       "    function on_error() {\n",
+       "      console.error(\"failed to load \" + url);\n",
+       "    }\n",
+       "\n",
+       "    for (var i = 0; i < css_urls.length; i++) {\n",
+       "      var url = css_urls[i];\n",
+       "      const element = document.createElement(\"link\");\n",
+       "      element.onload = on_load;\n",
+       "      element.onerror = on_error;\n",
+       "      element.rel = \"stylesheet\";\n",
+       "      element.type = \"text/css\";\n",
+       "      element.href = url;\n",
+       "      console.debug(\"Bokeh: injecting link tag for BokehJS stylesheet: \", url);\n",
+       "      document.body.appendChild(element);\n",
+       "    }\n",
+       "\n",
+       "    var skip = [];\n",
+       "    if (window.requirejs) {\n",
+       "      window.requirejs.config({'packages': {}, 'paths': {'gridstack': 'https://cdn.jsdelivr.net/npm/gridstack@4.2.5/dist/gridstack-h5', 'notyf': 'https://cdn.jsdelivr.net/npm/notyf@3/notyf.min'}, 'shim': {'gridstack': {'exports': 'GridStack'}}});\n",
+       "      require([\"gridstack\"], function(GridStack) {\n",
+       "\twindow.GridStack = GridStack\n",
+       "\ton_load()\n",
+       "      })\n",
+       "      require([\"notyf\"], function() {\n",
+       "\ton_load()\n",
+       "      })\n",
+       "      root._bokeh_is_loading = css_urls.length + 2;\n",
+       "    } else {\n",
+       "      root._bokeh_is_loading = css_urls.length + js_urls.length + js_modules.length;\n",
+       "    }    if (((window['GridStack'] !== undefined) && (!(window['GridStack'] instanceof HTMLElement))) || window.requirejs) {\n",
+       "      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/gridstack/gridstack@4.2.5/dist/gridstack-h5.js'];\n",
+       "      for (var i = 0; i < urls.length; i++) {\n",
+       "        skip.push(urls[i])\n",
+       "      }\n",
+       "    }    if (((window['Notyf'] !== undefined) && (!(window['Notyf'] instanceof HTMLElement))) || window.requirejs) {\n",
+       "      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/notificationarea/notyf@3/notyf.min.js'];\n",
+       "      for (var i = 0; i < urls.length; i++) {\n",
+       "        skip.push(urls[i])\n",
+       "      }\n",
+       "    }    for (var i = 0; i < js_urls.length; i++) {\n",
+       "      var url = js_urls[i];\n",
+       "      if (skip.indexOf(url) >= 0) {\n",
+       "\tif (!window.requirejs) {\n",
+       "\t  on_load();\n",
+       "\t}\n",
+       "\tcontinue;\n",
+       "      }\n",
+       "      var element = document.createElement('script');\n",
+       "      element.onload = on_load;\n",
+       "      element.onerror = on_error;\n",
+       "      element.async = false;\n",
+       "      element.src = url;\n",
+       "      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n",
+       "      document.head.appendChild(element);\n",
+       "    }\n",
+       "    for (var i = 0; i < js_modules.length; i++) {\n",
+       "      var url = js_modules[i];\n",
+       "      if (skip.indexOf(url) >= 0) {\n",
+       "\tif (!window.requirejs) {\n",
+       "\t  on_load();\n",
+       "\t}\n",
+       "\tcontinue;\n",
+       "      }\n",
+       "      var element = document.createElement('script');\n",
+       "      element.onload = on_load;\n",
+       "      element.onerror = on_error;\n",
+       "      element.async = false;\n",
+       "      element.src = url;\n",
+       "      element.type = \"module\";\n",
+       "      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n",
+       "      document.head.appendChild(element);\n",
+       "    }\n",
+       "    if (!js_urls.length && !js_modules.length) {\n",
+       "      on_load()\n",
+       "    }\n",
+       "  };\n",
+       "\n",
+       "  function inject_raw_css(css) {\n",
+       "    const element = document.createElement(\"style\");\n",
+       "    element.appendChild(document.createTextNode(css));\n",
+       "    document.body.appendChild(element);\n",
+       "  }\n",
+       "\n",
+       "  var js_urls = [\"https://cdn.bokeh.org/bokeh/release/bokeh-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-gl-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-widgets-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-tables-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-mathjax-2.4.3.min.js\", \"https://unpkg.com/@holoviz/panel@0.14.4/dist/panel.min.js\"];\n",
+       "  var js_modules = [];\n",
+       "  var css_urls = [\"https://cdn.holoviz.org/panel/0.14.4/dist/css/debugger.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/alerts.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/card.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/widgets.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/markdown.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/json.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/loading.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/dataframe.css\"];\n",
+       "  var inline_js = [    function(Bokeh) {\n",
+       "      inject_raw_css(\"\\n    .bk.pn-loading.arc:before {\\n      background-image: url(\\\"data:image/svg+xml;base64,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\\\");\\n      background-size: auto calc(min(50%, 400px));\\n    }\\n    \");\n",
+       "    },    function(Bokeh) {\n",
+       "      Bokeh.set_log_level(\"info\");\n",
+       "    },\n",
+       "function(Bokeh) {} // ensure no trailing comma for IE\n",
+       "  ];\n",
+       "\n",
+       "  function run_inline_js() {\n",
+       "    if ((root.Bokeh !== undefined) || (force === true)) {\n",
+       "      for (var i = 0; i < inline_js.length; i++) {\n",
+       "        inline_js[i].call(root, root.Bokeh);\n",
+       "      }} else if (Date.now() < root._bokeh_timeout) {\n",
+       "      setTimeout(run_inline_js, 100);\n",
+       "    } else if (!root._bokeh_failed_load) {\n",
+       "      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n",
+       "      root._bokeh_failed_load = true;\n",
+       "    }\n",
+       "  }\n",
+       "\n",
+       "  if (root._bokeh_is_loading === 0) {\n",
+       "    console.debug(\"Bokeh: BokehJS loaded, going straight to plotting\");\n",
+       "    run_inline_js();\n",
+       "  } else {\n",
+       "    load_libs(css_urls, js_urls, js_modules, function() {\n",
+       "      console.debug(\"Bokeh: BokehJS plotting callback run at\", now());\n",
+       "      run_inline_js();\n",
+       "    });\n",
+       "  }\n",
+       "}(window));"
+      ],
+      "application/vnd.holoviews_load.v0+json": "(function(root) {\n  function now() {\n    return new Date();\n  }\n\n  var force = true;\n\n  if (typeof root._bokeh_onload_callbacks === \"undefined\" || force === true) {\n    root._bokeh_onload_callbacks = [];\n    root._bokeh_is_loading = undefined;\n  }\n\n  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n    root._bokeh_timeout = Date.now() + 5000;\n    root._bokeh_failed_load = false;\n  }\n\n  function run_callbacks() {\n    try {\n      root._bokeh_onload_callbacks.forEach(function(callback) {\n        if (callback != null)\n          callback();\n      });\n    } finally {\n      delete root._bokeh_onload_callbacks\n    }\n    console.debug(\"Bokeh: all callbacks have finished\");\n  }\n\n  function load_libs(css_urls, js_urls, js_modules, callback) {\n    if (css_urls == null) css_urls = [];\n    if (js_urls == null) js_urls = [];\n    if (js_modules == null) js_modules = [];\n\n    root._bokeh_onload_callbacks.push(callback);\n    if (root._bokeh_is_loading > 0) {\n      console.debug(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n      return null;\n    }\n    if (js_urls.length === 0 && js_modules.length === 0) {\n      run_callbacks();\n      return null;\n    }\n    console.debug(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n\n    function on_load() {\n      root._bokeh_is_loading--;\n      if (root._bokeh_is_loading === 0) {\n        console.debug(\"Bokeh: all BokehJS libraries/stylesheets loaded\");\n        run_callbacks()\n      }\n    }\n\n    function on_error() {\n      console.error(\"failed to load \" + url);\n    }\n\n    for (var i = 0; i < css_urls.length; i++) {\n      var url = css_urls[i];\n      const element = document.createElement(\"link\");\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.rel = \"stylesheet\";\n      element.type = \"text/css\";\n      element.href = url;\n      console.debug(\"Bokeh: injecting link tag for BokehJS stylesheet: \", url);\n      document.body.appendChild(element);\n    }\n\n    var skip = [];\n    if (window.requirejs) {\n      window.requirejs.config({'packages': {}, 'paths': {'gridstack': 'https://cdn.jsdelivr.net/npm/gridstack@4.2.5/dist/gridstack-h5', 'notyf': 'https://cdn.jsdelivr.net/npm/notyf@3/notyf.min'}, 'shim': {'gridstack': {'exports': 'GridStack'}}});\n      require([\"gridstack\"], function(GridStack) {\n\twindow.GridStack = GridStack\n\ton_load()\n      })\n      require([\"notyf\"], function() {\n\ton_load()\n      })\n      root._bokeh_is_loading = css_urls.length + 2;\n    } else {\n      root._bokeh_is_loading = css_urls.length + js_urls.length + js_modules.length;\n    }    if (((window['GridStack'] !== undefined) && (!(window['GridStack'] instanceof HTMLElement))) || window.requirejs) {\n      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/gridstack/gridstack@4.2.5/dist/gridstack-h5.js'];\n      for (var i = 0; i < urls.length; i++) {\n        skip.push(urls[i])\n      }\n    }    if (((window['Notyf'] !== undefined) && (!(window['Notyf'] instanceof HTMLElement))) || window.requirejs) {\n      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/notificationarea/notyf@3/notyf.min.js'];\n      for (var i = 0; i < urls.length; i++) {\n        skip.push(urls[i])\n      }\n    }    for (var i = 0; i < js_urls.length; i++) {\n      var url = js_urls[i];\n      if (skip.indexOf(url) >= 0) {\n\tif (!window.requirejs) {\n\t  on_load();\n\t}\n\tcontinue;\n      }\n      var element = document.createElement('script');\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.async = false;\n      element.src = url;\n      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n      document.head.appendChild(element);\n    }\n    for (var i = 0; i < js_modules.length; i++) {\n      var url = js_modules[i];\n      if (skip.indexOf(url) >= 0) {\n\tif (!window.requirejs) {\n\t  on_load();\n\t}\n\tcontinue;\n      }\n      var element = document.createElement('script');\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.async = false;\n      element.src = url;\n      element.type = \"module\";\n      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n      document.head.appendChild(element);\n    }\n    if (!js_urls.length && !js_modules.length) {\n      on_load()\n    }\n  };\n\n  function inject_raw_css(css) {\n    const element = document.createElement(\"style\");\n    element.appendChild(document.createTextNode(css));\n    document.body.appendChild(element);\n  }\n\n  var js_urls = [\"https://cdn.bokeh.org/bokeh/release/bokeh-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-gl-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-widgets-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-tables-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-mathjax-2.4.3.min.js\", \"https://unpkg.com/@holoviz/panel@0.14.4/dist/panel.min.js\"];\n  var js_modules = [];\n  var css_urls = [\"https://cdn.holoviz.org/panel/0.14.4/dist/css/debugger.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/alerts.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/card.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/widgets.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/markdown.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/json.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/loading.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/dataframe.css\"];\n  var inline_js = [    function(Bokeh) {\n      inject_raw_css(\"\\n    .bk.pn-loading.arc:before {\\n      background-image: url(\\\"data:image/svg+xml;base64,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\\\");\\n      background-size: auto calc(min(50%, 400px));\\n    }\\n    \");\n    },    function(Bokeh) {\n      Bokeh.set_log_level(\"info\");\n    },\nfunction(Bokeh) {} // ensure no trailing comma for IE\n  ];\n\n  function run_inline_js() {\n    if ((root.Bokeh !== undefined) || (force === true)) {\n      for (var i = 0; i < inline_js.length; i++) {\n        inline_js[i].call(root, root.Bokeh);\n      }} else if (Date.now() < root._bokeh_timeout) {\n      setTimeout(run_inline_js, 100);\n    } else if (!root._bokeh_failed_load) {\n      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n      root._bokeh_failed_load = true;\n    }\n  }\n\n  if (root._bokeh_is_loading === 0) {\n    console.debug(\"Bokeh: BokehJS loaded, going straight to plotting\");\n    run_inline_js();\n  } else {\n    load_libs(css_urls, js_urls, js_modules, function() {\n      console.debug(\"Bokeh: BokehJS plotting callback run at\", now());\n      run_inline_js();\n    });\n  }\n}(window));"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/javascript": [
+       "\n",
+       "if ((window.PyViz === undefined) || (window.PyViz instanceof HTMLElement)) {\n",
+       "  window.PyViz = {comms: {}, comm_status:{}, kernels:{}, receivers: {}, plot_index: []}\n",
+       "}\n",
+       "\n",
+       "\n",
+       "    function JupyterCommManager() {\n",
+       "    }\n",
+       "\n",
+       "    JupyterCommManager.prototype.register_target = function(plot_id, comm_id, msg_handler) {\n",
+       "      if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n",
+       "        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n",
+       "        comm_manager.register_target(comm_id, function(comm) {\n",
+       "          comm.on_msg(msg_handler);\n",
+       "        });\n",
+       "      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n",
+       "        window.PyViz.kernels[plot_id].registerCommTarget(comm_id, function(comm) {\n",
+       "          comm.onMsg = msg_handler;\n",
+       "        });\n",
+       "      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n",
+       "        google.colab.kernel.comms.registerTarget(comm_id, (comm) => {\n",
+       "          var messages = comm.messages[Symbol.asyncIterator]();\n",
+       "          function processIteratorResult(result) {\n",
+       "            var message = result.value;\n",
+       "            console.log(message)\n",
+       "            var content = {data: message.data, comm_id};\n",
+       "            var buffers = []\n",
+       "            for (var buffer of message.buffers || []) {\n",
+       "              buffers.push(new DataView(buffer))\n",
+       "            }\n",
+       "            var metadata = message.metadata || {};\n",
+       "            var msg = {content, buffers, metadata}\n",
+       "            msg_handler(msg);\n",
+       "            return messages.next().then(processIteratorResult);\n",
+       "          }\n",
+       "          return messages.next().then(processIteratorResult);\n",
+       "        })\n",
+       "      }\n",
+       "    }\n",
+       "\n",
+       "    JupyterCommManager.prototype.get_client_comm = function(plot_id, comm_id, msg_handler) {\n",
+       "      if (comm_id in window.PyViz.comms) {\n",
+       "        return window.PyViz.comms[comm_id];\n",
+       "      } else if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n",
+       "        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n",
+       "        var comm = comm_manager.new_comm(comm_id, {}, {}, {}, comm_id);\n",
+       "        if (msg_handler) {\n",
+       "          comm.on_msg(msg_handler);\n",
+       "        }\n",
+       "      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n",
+       "        var comm = window.PyViz.kernels[plot_id].connectToComm(comm_id);\n",
+       "        comm.open();\n",
+       "        if (msg_handler) {\n",
+       "          comm.onMsg = msg_handler;\n",
+       "        }\n",
+       "      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n",
+       "        var comm_promise = google.colab.kernel.comms.open(comm_id)\n",
+       "        comm_promise.then((comm) => {\n",
+       "          window.PyViz.comms[comm_id] = comm;\n",
+       "          if (msg_handler) {\n",
+       "            var messages = comm.messages[Symbol.asyncIterator]();\n",
+       "            function processIteratorResult(result) {\n",
+       "              var message = result.value;\n",
+       "              var content = {data: message.data};\n",
+       "              var metadata = message.metadata || {comm_id};\n",
+       "              var msg = {content, metadata}\n",
+       "              msg_handler(msg);\n",
+       "              return messages.next().then(processIteratorResult);\n",
+       "            }\n",
+       "            return messages.next().then(processIteratorResult);\n",
+       "          }\n",
+       "        }) \n",
+       "        var sendClosure = (data, metadata, buffers, disposeOnDone) => {\n",
+       "          return comm_promise.then((comm) => {\n",
+       "            comm.send(data, metadata, buffers, disposeOnDone);\n",
+       "          });\n",
+       "        };\n",
+       "        var comm = {\n",
+       "          send: sendClosure\n",
+       "        };\n",
+       "      }\n",
+       "      window.PyViz.comms[comm_id] = comm;\n",
+       "      return comm;\n",
+       "    }\n",
+       "    window.PyViz.comm_manager = new JupyterCommManager();\n",
+       "    \n",
+       "\n",
+       "\n",
+       "var JS_MIME_TYPE = 'application/javascript';\n",
+       "var HTML_MIME_TYPE = 'text/html';\n",
+       "var EXEC_MIME_TYPE = 'application/vnd.holoviews_exec.v0+json';\n",
+       "var CLASS_NAME = 'output';\n",
+       "\n",
+       "/**\n",
+       " * Render data to the DOM node\n",
+       " */\n",
+       "function render(props, node) {\n",
+       "  var div = document.createElement(\"div\");\n",
+       "  var script = document.createElement(\"script\");\n",
+       "  node.appendChild(div);\n",
+       "  node.appendChild(script);\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle when a new output is added\n",
+       " */\n",
+       "function handle_add_output(event, handle) {\n",
+       "  var output_area = handle.output_area;\n",
+       "  var output = handle.output;\n",
+       "  if ((output.data == undefined) || (!output.data.hasOwnProperty(EXEC_MIME_TYPE))) {\n",
+       "    return\n",
+       "  }\n",
+       "  var id = output.metadata[EXEC_MIME_TYPE][\"id\"];\n",
+       "  var toinsert = output_area.element.find(\".\" + CLASS_NAME.split(' ')[0]);\n",
+       "  if (id !== undefined) {\n",
+       "    var nchildren = toinsert.length;\n",
+       "    var html_node = toinsert[nchildren-1].children[0];\n",
+       "    html_node.innerHTML = output.data[HTML_MIME_TYPE];\n",
+       "    var scripts = [];\n",
+       "    var nodelist = html_node.querySelectorAll(\"script\");\n",
+       "    for (var i in nodelist) {\n",
+       "      if (nodelist.hasOwnProperty(i)) {\n",
+       "        scripts.push(nodelist[i])\n",
+       "      }\n",
+       "    }\n",
+       "\n",
+       "    scripts.forEach( function (oldScript) {\n",
+       "      var newScript = document.createElement(\"script\");\n",
+       "      var attrs = [];\n",
+       "      var nodemap = oldScript.attributes;\n",
+       "      for (var j in nodemap) {\n",
+       "        if (nodemap.hasOwnProperty(j)) {\n",
+       "          attrs.push(nodemap[j])\n",
+       "        }\n",
+       "      }\n",
+       "      attrs.forEach(function(attr) { newScript.setAttribute(attr.name, attr.value) });\n",
+       "      newScript.appendChild(document.createTextNode(oldScript.innerHTML));\n",
+       "      oldScript.parentNode.replaceChild(newScript, oldScript);\n",
+       "    });\n",
+       "    if (JS_MIME_TYPE in output.data) {\n",
+       "      toinsert[nchildren-1].children[1].textContent = output.data[JS_MIME_TYPE];\n",
+       "    }\n",
+       "    output_area._hv_plot_id = id;\n",
+       "    if ((window.Bokeh !== undefined) && (id in Bokeh.index)) {\n",
+       "      window.PyViz.plot_index[id] = Bokeh.index[id];\n",
+       "    } else {\n",
+       "      window.PyViz.plot_index[id] = null;\n",
+       "    }\n",
+       "  } else if (output.metadata[EXEC_MIME_TYPE][\"server_id\"] !== undefined) {\n",
+       "    var bk_div = document.createElement(\"div\");\n",
+       "    bk_div.innerHTML = output.data[HTML_MIME_TYPE];\n",
+       "    var script_attrs = bk_div.children[0].attributes;\n",
+       "    for (var i = 0; i < script_attrs.length; i++) {\n",
+       "      toinsert[toinsert.length - 1].childNodes[1].setAttribute(script_attrs[i].name, script_attrs[i].value);\n",
+       "    }\n",
+       "    // store reference to server id on output_area\n",
+       "    output_area._bokeh_server_id = output.metadata[EXEC_MIME_TYPE][\"server_id\"];\n",
+       "  }\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle when an output is cleared or removed\n",
+       " */\n",
+       "function handle_clear_output(event, handle) {\n",
+       "  var id = handle.cell.output_area._hv_plot_id;\n",
+       "  var server_id = handle.cell.output_area._bokeh_server_id;\n",
+       "  if (((id === undefined) || !(id in PyViz.plot_index)) && (server_id !== undefined)) { return; }\n",
+       "  var comm = window.PyViz.comm_manager.get_client_comm(\"hv-extension-comm\", \"hv-extension-comm\", function () {});\n",
+       "  if (server_id !== null) {\n",
+       "    comm.send({event_type: 'server_delete', 'id': server_id});\n",
+       "    return;\n",
+       "  } else if (comm !== null) {\n",
+       "    comm.send({event_type: 'delete', 'id': id});\n",
+       "  }\n",
+       "  delete PyViz.plot_index[id];\n",
+       "  if ((window.Bokeh !== undefined) & (id in window.Bokeh.index)) {\n",
+       "    var doc = window.Bokeh.index[id].model.document\n",
+       "    doc.clear();\n",
+       "    const i = window.Bokeh.documents.indexOf(doc);\n",
+       "    if (i > -1) {\n",
+       "      window.Bokeh.documents.splice(i, 1);\n",
+       "    }\n",
+       "  }\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle kernel restart event\n",
+       " */\n",
+       "function handle_kernel_cleanup(event, handle) {\n",
+       "  delete PyViz.comms[\"hv-extension-comm\"];\n",
+       "  window.PyViz.plot_index = {}\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle update_display_data messages\n",
+       " */\n",
+       "function handle_update_output(event, handle) {\n",
+       "  handle_clear_output(event, {cell: {output_area: handle.output_area}})\n",
+       "  handle_add_output(event, handle)\n",
+       "}\n",
+       "\n",
+       "function register_renderer(events, OutputArea) {\n",
+       "  function append_mime(data, metadata, element) {\n",
+       "    // create a DOM node to render to\n",
+       "    var toinsert = this.create_output_subarea(\n",
+       "    metadata,\n",
+       "    CLASS_NAME,\n",
+       "    EXEC_MIME_TYPE\n",
+       "    );\n",
+       "    this.keyboard_manager.register_events(toinsert);\n",
+       "    // Render to node\n",
+       "    var props = {data: data, metadata: metadata[EXEC_MIME_TYPE]};\n",
+       "    render(props, toinsert[0]);\n",
+       "    element.append(toinsert);\n",
+       "    return toinsert\n",
+       "  }\n",
+       "\n",
+       "  events.on('output_added.OutputArea', handle_add_output);\n",
+       "  events.on('output_updated.OutputArea', handle_update_output);\n",
+       "  events.on('clear_output.CodeCell', handle_clear_output);\n",
+       "  events.on('delete.Cell', handle_clear_output);\n",
+       "  events.on('kernel_ready.Kernel', handle_kernel_cleanup);\n",
+       "\n",
+       "  OutputArea.prototype.register_mime_type(EXEC_MIME_TYPE, append_mime, {\n",
+       "    safe: true,\n",
+       "    index: 0\n",
+       "  });\n",
+       "}\n",
+       "\n",
+       "if (window.Jupyter !== undefined) {\n",
+       "  try {\n",
+       "    var events = require('base/js/events');\n",
+       "    var OutputArea = require('notebook/js/outputarea').OutputArea;\n",
+       "    if (OutputArea.prototype.mime_types().indexOf(EXEC_MIME_TYPE) == -1) {\n",
+       "      register_renderer(events, OutputArea);\n",
+       "    }\n",
+       "  } catch(err) {\n",
+       "  }\n",
+       "}\n"
+      ],
+      "application/vnd.holoviews_load.v0+json": "\nif ((window.PyViz === undefined) || (window.PyViz instanceof HTMLElement)) {\n  window.PyViz = {comms: {}, comm_status:{}, kernels:{}, receivers: {}, plot_index: []}\n}\n\n\n    function JupyterCommManager() {\n    }\n\n    JupyterCommManager.prototype.register_target = function(plot_id, comm_id, msg_handler) {\n      if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n        comm_manager.register_target(comm_id, function(comm) {\n          comm.on_msg(msg_handler);\n        });\n      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n        window.PyViz.kernels[plot_id].registerCommTarget(comm_id, function(comm) {\n          comm.onMsg = msg_handler;\n        });\n      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n        google.colab.kernel.comms.registerTarget(comm_id, (comm) => {\n          var messages = comm.messages[Symbol.asyncIterator]();\n          function processIteratorResult(result) {\n            var message = result.value;\n            console.log(message)\n            var content = {data: message.data, comm_id};\n            var buffers = []\n            for (var buffer of message.buffers || []) {\n              buffers.push(new DataView(buffer))\n            }\n            var metadata = message.metadata || {};\n            var msg = {content, buffers, metadata}\n            msg_handler(msg);\n            return messages.next().then(processIteratorResult);\n          }\n          return messages.next().then(processIteratorResult);\n        })\n      }\n    }\n\n    JupyterCommManager.prototype.get_client_comm = function(plot_id, comm_id, msg_handler) {\n      if (comm_id in window.PyViz.comms) {\n        return window.PyViz.comms[comm_id];\n      } else if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n        var comm = comm_manager.new_comm(comm_id, {}, {}, {}, comm_id);\n        if (msg_handler) {\n          comm.on_msg(msg_handler);\n        }\n      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n        var comm = window.PyViz.kernels[plot_id].connectToComm(comm_id);\n        comm.open();\n        if (msg_handler) {\n          comm.onMsg = msg_handler;\n        }\n      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n        var comm_promise = google.colab.kernel.comms.open(comm_id)\n        comm_promise.then((comm) => {\n          window.PyViz.comms[comm_id] = comm;\n          if (msg_handler) {\n            var messages = comm.messages[Symbol.asyncIterator]();\n            function processIteratorResult(result) {\n              var message = result.value;\n              var content = {data: message.data};\n              var metadata = message.metadata || {comm_id};\n              var msg = {content, metadata}\n              msg_handler(msg);\n              return messages.next().then(processIteratorResult);\n            }\n            return messages.next().then(processIteratorResult);\n          }\n        }) \n        var sendClosure = (data, metadata, buffers, disposeOnDone) => {\n          return comm_promise.then((comm) => {\n            comm.send(data, metadata, buffers, disposeOnDone);\n          });\n        };\n        var comm = {\n          send: sendClosure\n        };\n      }\n      window.PyViz.comms[comm_id] = comm;\n      return comm;\n    }\n    window.PyViz.comm_manager = new JupyterCommManager();\n    \n\n\nvar JS_MIME_TYPE = 'application/javascript';\nvar HTML_MIME_TYPE = 'text/html';\nvar EXEC_MIME_TYPE = 'application/vnd.holoviews_exec.v0+json';\nvar CLASS_NAME = 'output';\n\n/**\n * Render data to the DOM node\n */\nfunction render(props, node) {\n  var div = document.createElement(\"div\");\n  var script = document.createElement(\"script\");\n  node.appendChild(div);\n  node.appendChild(script);\n}\n\n/**\n * Handle when a new output is added\n */\nfunction handle_add_output(event, handle) {\n  var output_area = handle.output_area;\n  var output = handle.output;\n  if ((output.data == undefined) || (!output.data.hasOwnProperty(EXEC_MIME_TYPE))) {\n    return\n  }\n  var id = output.metadata[EXEC_MIME_TYPE][\"id\"];\n  var toinsert = output_area.element.find(\".\" + CLASS_NAME.split(' ')[0]);\n  if (id !== undefined) {\n    var nchildren = toinsert.length;\n    var html_node = toinsert[nchildren-1].children[0];\n    html_node.innerHTML = output.data[HTML_MIME_TYPE];\n    var scripts = [];\n    var nodelist = html_node.querySelectorAll(\"script\");\n    for (var i in nodelist) {\n      if (nodelist.hasOwnProperty(i)) {\n        scripts.push(nodelist[i])\n      }\n    }\n\n    scripts.forEach( function (oldScript) {\n      var newScript = document.createElement(\"script\");\n      var attrs = [];\n      var nodemap = oldScript.attributes;\n      for (var j in nodemap) {\n        if (nodemap.hasOwnProperty(j)) {\n          attrs.push(nodemap[j])\n        }\n      }\n      attrs.forEach(function(attr) { newScript.setAttribute(attr.name, attr.value) });\n      newScript.appendChild(document.createTextNode(oldScript.innerHTML));\n      oldScript.parentNode.replaceChild(newScript, oldScript);\n    });\n    if (JS_MIME_TYPE in output.data) {\n      toinsert[nchildren-1].children[1].textContent = output.data[JS_MIME_TYPE];\n    }\n    output_area._hv_plot_id = id;\n    if ((window.Bokeh !== undefined) && (id in Bokeh.index)) {\n      window.PyViz.plot_index[id] = Bokeh.index[id];\n    } else {\n      window.PyViz.plot_index[id] = null;\n    }\n  } else if (output.metadata[EXEC_MIME_TYPE][\"server_id\"] !== undefined) {\n    var bk_div = document.createElement(\"div\");\n    bk_div.innerHTML = output.data[HTML_MIME_TYPE];\n    var script_attrs = bk_div.children[0].attributes;\n    for (var i = 0; i < script_attrs.length; i++) {\n      toinsert[toinsert.length - 1].childNodes[1].setAttribute(script_attrs[i].name, script_attrs[i].value);\n    }\n    // store reference to server id on output_area\n    output_area._bokeh_server_id = output.metadata[EXEC_MIME_TYPE][\"server_id\"];\n  }\n}\n\n/**\n * Handle when an output is cleared or removed\n */\nfunction handle_clear_output(event, handle) {\n  var id = handle.cell.output_area._hv_plot_id;\n  var server_id = handle.cell.output_area._bokeh_server_id;\n  if (((id === undefined) || !(id in PyViz.plot_index)) && (server_id !== undefined)) { return; }\n  var comm = window.PyViz.comm_manager.get_client_comm(\"hv-extension-comm\", \"hv-extension-comm\", function () {});\n  if (server_id !== null) {\n    comm.send({event_type: 'server_delete', 'id': server_id});\n    return;\n  } else if (comm !== null) {\n    comm.send({event_type: 'delete', 'id': id});\n  }\n  delete PyViz.plot_index[id];\n  if ((window.Bokeh !== undefined) & (id in window.Bokeh.index)) {\n    var doc = window.Bokeh.index[id].model.document\n    doc.clear();\n    const i = window.Bokeh.documents.indexOf(doc);\n    if (i > -1) {\n      window.Bokeh.documents.splice(i, 1);\n    }\n  }\n}\n\n/**\n * Handle kernel restart event\n */\nfunction handle_kernel_cleanup(event, handle) {\n  delete PyViz.comms[\"hv-extension-comm\"];\n  window.PyViz.plot_index = {}\n}\n\n/**\n * Handle update_display_data messages\n */\nfunction handle_update_output(event, handle) {\n  handle_clear_output(event, {cell: {output_area: handle.output_area}})\n  handle_add_output(event, handle)\n}\n\nfunction register_renderer(events, OutputArea) {\n  function append_mime(data, metadata, element) {\n    // create a DOM node to render to\n    var toinsert = this.create_output_subarea(\n    metadata,\n    CLASS_NAME,\n    EXEC_MIME_TYPE\n    );\n    this.keyboard_manager.register_events(toinsert);\n    // Render to node\n    var props = {data: data, metadata: metadata[EXEC_MIME_TYPE]};\n    render(props, toinsert[0]);\n    element.append(toinsert);\n    return toinsert\n  }\n\n  events.on('output_added.OutputArea', handle_add_output);\n  events.on('output_updated.OutputArea', handle_update_output);\n  events.on('clear_output.CodeCell', handle_clear_output);\n  events.on('delete.Cell', handle_clear_output);\n  events.on('kernel_ready.Kernel', handle_kernel_cleanup);\n\n  OutputArea.prototype.register_mime_type(EXEC_MIME_TYPE, append_mime, {\n    safe: true,\n    index: 0\n  });\n}\n\nif (window.Jupyter !== undefined) {\n  try {\n    var events = require('base/js/events');\n    var OutputArea = require('notebook/js/outputarea').OutputArea;\n    if (OutputArea.prototype.mime_types().indexOf(EXEC_MIME_TYPE) == -1) {\n      register_renderer(events, OutputArea);\n    }\n  } catch(err) {\n  }\n}\n"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       "<style>.bk-root, .bk-root .bk:before, .bk-root .bk:after {\n",
+       "  font-family: var(--jp-ui-font-size1);\n",
+       "  font-size: var(--jp-ui-font-size1);\n",
+       "  color: var(--jp-ui-font-color1);\n",
+       "}\n",
+       "</style>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "import panel as pn  # GUI\n",
+    "pn.extension()\n",
+    "\n",
+    "panels = [] # collect display \n",
+    "\n",
+    "context = [ {'role':'system', 'content':\"\"\"\n",
+    "You are OrderBot, an automated service to collect orders for a pizza restaurant. \\\n",
+    "You first greet the customer, then collects the order, \\\n",
+    "and then asks if it's a pickup or delivery. \\\n",
+    "You wait to collect the entire order, then summarize it and check for a final \\\n",
+    "time if the customer wants to add anything else. \\\n",
+    "If it's a delivery, you ask for an address. \\\n",
+    "Finally you collect the payment.\\\n",
+    "Make sure to clarify all options, extras and sizes to uniquely \\\n",
+    "identify the item from the menu.\\\n",
+    "You respond in a short, very conversational friendly style. \\\n",
+    "The menu includes \\\n",
+    "pepperoni pizza  12.95, 10.00, 7.00 \\\n",
+    "cheese pizza   10.95, 9.25, 6.50 \\\n",
+    "eggplant pizza   11.95, 9.75, 6.75 \\\n",
+    "fries 4.50, 3.50 \\\n",
+    "greek salad 7.25 \\\n",
+    "Toppings: \\\n",
+    "extra cheese 2.00, \\\n",
+    "mushrooms 1.50 \\\n",
+    "sausage 3.00 \\\n",
+    "canadian bacon 3.50 \\\n",
+    "AI sauce 1.50 \\\n",
+    "peppers 1.00 \\\n",
+    "Drinks: \\\n",
+    "coke 3.00, 2.00, 1.00 \\\n",
+    "sprite 3.00, 2.00, 1.00 \\\n",
+    "bottled water 5.00 \\\n",
+    "\"\"\"} ]  # accumulate messages\n",
+    "\n",
+    "\n",
+    "inp = pn.widgets.TextInput(value=\"Hi\", placeholder='Enter text here…')\n",
+    "button_conversation = pn.widgets.Button(name=\"Chat!\")\n",
+    "\n",
+    "interactive_conversation = pn.bind(collect_messages, button_conversation)\n",
+    "\n",
+    "dashboard = pn.Column(\n",
+    "    inp,\n",
+    "    pn.Row(button_conversation),\n",
+    "    pn.panel(interactive_conversation, loading_indicator=True, height=300),\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "65f0416e",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "dashboard"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "2a2c9822",
+   "metadata": {},
+   "source": [
+    "现在我们可以要求模型创建一个 JSON 摘要发送给订单系统。\n",
+    "\n",
+    "所以我们现在追加另一个系统消息，它是另一条prompt，我们说创建一个刚刚订单的 JSON 摘要，列出每个项目的价格，字段应包括1）披萨，包括尺寸，2）配料列表，3）饮料列表，4）辅菜列表，包括尺寸，最后是总价格。这里也可以在这里使用用户消息，不一定是系统消息。\n",
+    "\n",
+    "请注意，这里我们使用了一个较低的temperature，因为对于这些类型的任务，我们希望输出相对可预测。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 45,
+   "id": "c840ff56",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Here's a JSON summary of the previous food order:\n",
+      "\n",
+      "```\n",
+      "{\n",
+      "  \"pizza\": {\n",
+      "    \"type\": \"cheese\",\n",
+      "    \"size\": \"large\",\n",
+      "    \"toppings\": [\n",
+      "      \"mushrooms\"\n",
+      "    ],\n",
+      "    \"price\": 12.45\n",
+      "  },\n",
+      "  \"drinks\": [\n",
+      "    {\n",
+      "      \"type\": \"sprite\",\n",
+      "      \"size\": \"medium\",\n",
+      "      \"price\": 3.00\n",
+      "    },\n",
+      "    {\n",
+      "      \"type\": \"sprite\",\n",
+      "      \"size\": \"medium\",\n",
+      "      \"price\": 3.00\n",
+      "    }\n",
+      "  ],\n",
+      "  \"sides\": [],\n",
+      "  \"total_price\": 18.45\n",
+      "}\n",
+      "``` \n",
+      "\n",
+      "Note: I assumed that the price of the large cheese pizza with mushrooms is $12.45 instead of $12.95, since the customer only ordered one topping.\n"
+     ]
+    }
+   ],
+   "source": [
+    "messages =  context.copy()\n",
+    "messages.append(\n",
+    "{'role':'system', 'content':'create a json summary of the previous food order. Itemize the price for each item\\\n",
+    " The fields should be 1) pizza, include size 2) list of toppings 3) list of drinks, include size   4) list of sides include size  5)total price '},    \n",
+    ")\n",
+    " #The fields should be 1) pizza, price 2) list of toppings 3) list of drinks, include size include price  4) list of sides include size include price, 5)total price '},    \n",
+    "\n",
+    "response = get_completion_from_messages(messages, temperature=0)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 76,
+   "id": "6ef90a6b",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/javascript": [
+       "(function(root) {\n",
+       "  function now() {\n",
+       "    return new Date();\n",
+       "  }\n",
+       "\n",
+       "  var force = true;\n",
+       "\n",
+       "  if (typeof root._bokeh_onload_callbacks === \"undefined\" || force === true) {\n",
+       "    root._bokeh_onload_callbacks = [];\n",
+       "    root._bokeh_is_loading = undefined;\n",
+       "  }\n",
+       "\n",
+       "  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n",
+       "    root._bokeh_timeout = Date.now() + 5000;\n",
+       "    root._bokeh_failed_load = false;\n",
+       "  }\n",
+       "\n",
+       "  function run_callbacks() {\n",
+       "    try {\n",
+       "      root._bokeh_onload_callbacks.forEach(function(callback) {\n",
+       "        if (callback != null)\n",
+       "          callback();\n",
+       "      });\n",
+       "    } finally {\n",
+       "      delete root._bokeh_onload_callbacks\n",
+       "    }\n",
+       "    console.debug(\"Bokeh: all callbacks have finished\");\n",
+       "  }\n",
+       "\n",
+       "  function load_libs(css_urls, js_urls, js_modules, callback) {\n",
+       "    if (css_urls == null) css_urls = [];\n",
+       "    if (js_urls == null) js_urls = [];\n",
+       "    if (js_modules == null) js_modules = [];\n",
+       "\n",
+       "    root._bokeh_onload_callbacks.push(callback);\n",
+       "    if (root._bokeh_is_loading > 0) {\n",
+       "      console.debug(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n",
+       "      return null;\n",
+       "    }\n",
+       "    if (js_urls.length === 0 && js_modules.length === 0) {\n",
+       "      run_callbacks();\n",
+       "      return null;\n",
+       "    }\n",
+       "    console.debug(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n",
+       "\n",
+       "    function on_load() {\n",
+       "      root._bokeh_is_loading--;\n",
+       "      if (root._bokeh_is_loading === 0) {\n",
+       "        console.debug(\"Bokeh: all BokehJS libraries/stylesheets loaded\");\n",
+       "        run_callbacks()\n",
+       "      }\n",
+       "    }\n",
+       "\n",
+       "    function on_error() {\n",
+       "      console.error(\"failed to load \" + url);\n",
+       "    }\n",
+       "\n",
+       "    for (var i = 0; i < css_urls.length; i++) {\n",
+       "      var url = css_urls[i];\n",
+       "      const element = document.createElement(\"link\");\n",
+       "      element.onload = on_load;\n",
+       "      element.onerror = on_error;\n",
+       "      element.rel = \"stylesheet\";\n",
+       "      element.type = \"text/css\";\n",
+       "      element.href = url;\n",
+       "      console.debug(\"Bokeh: injecting link tag for BokehJS stylesheet: \", url);\n",
+       "      document.body.appendChild(element);\n",
+       "    }\n",
+       "\n",
+       "    var skip = [];\n",
+       "    if (window.requirejs) {\n",
+       "      window.requirejs.config({'packages': {}, 'paths': {'gridstack': 'https://cdn.jsdelivr.net/npm/gridstack@4.2.5/dist/gridstack-h5', 'notyf': 'https://cdn.jsdelivr.net/npm/notyf@3/notyf.min'}, 'shim': {'gridstack': {'exports': 'GridStack'}}});\n",
+       "      require([\"gridstack\"], function(GridStack) {\n",
+       "\twindow.GridStack = GridStack\n",
+       "\ton_load()\n",
+       "      })\n",
+       "      require([\"notyf\"], function() {\n",
+       "\ton_load()\n",
+       "      })\n",
+       "      root._bokeh_is_loading = css_urls.length + 2;\n",
+       "    } else {\n",
+       "      root._bokeh_is_loading = css_urls.length + js_urls.length + js_modules.length;\n",
+       "    }    if (((window['GridStack'] !== undefined) && (!(window['GridStack'] instanceof HTMLElement))) || window.requirejs) {\n",
+       "      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/gridstack/gridstack@4.2.5/dist/gridstack-h5.js'];\n",
+       "      for (var i = 0; i < urls.length; i++) {\n",
+       "        skip.push(urls[i])\n",
+       "      }\n",
+       "    }    if (((window['Notyf'] !== undefined) && (!(window['Notyf'] instanceof HTMLElement))) || window.requirejs) {\n",
+       "      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/notificationarea/notyf@3/notyf.min.js'];\n",
+       "      for (var i = 0; i < urls.length; i++) {\n",
+       "        skip.push(urls[i])\n",
+       "      }\n",
+       "    }    for (var i = 0; i < js_urls.length; i++) {\n",
+       "      var url = js_urls[i];\n",
+       "      if (skip.indexOf(url) >= 0) {\n",
+       "\tif (!window.requirejs) {\n",
+       "\t  on_load();\n",
+       "\t}\n",
+       "\tcontinue;\n",
+       "      }\n",
+       "      var element = document.createElement('script');\n",
+       "      element.onload = on_load;\n",
+       "      element.onerror = on_error;\n",
+       "      element.async = false;\n",
+       "      element.src = url;\n",
+       "      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n",
+       "      document.head.appendChild(element);\n",
+       "    }\n",
+       "    for (var i = 0; i < js_modules.length; i++) {\n",
+       "      var url = js_modules[i];\n",
+       "      if (skip.indexOf(url) >= 0) {\n",
+       "\tif (!window.requirejs) {\n",
+       "\t  on_load();\n",
+       "\t}\n",
+       "\tcontinue;\n",
+       "      }\n",
+       "      var element = document.createElement('script');\n",
+       "      element.onload = on_load;\n",
+       "      element.onerror = on_error;\n",
+       "      element.async = false;\n",
+       "      element.src = url;\n",
+       "      element.type = \"module\";\n",
+       "      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n",
+       "      document.head.appendChild(element);\n",
+       "    }\n",
+       "    if (!js_urls.length && !js_modules.length) {\n",
+       "      on_load()\n",
+       "    }\n",
+       "  };\n",
+       "\n",
+       "  function inject_raw_css(css) {\n",
+       "    const element = document.createElement(\"style\");\n",
+       "    element.appendChild(document.createTextNode(css));\n",
+       "    document.body.appendChild(element);\n",
+       "  }\n",
+       "\n",
+       "  var js_urls = [\"https://cdn.bokeh.org/bokeh/release/bokeh-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-gl-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-widgets-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-tables-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-mathjax-2.4.3.min.js\", \"https://unpkg.com/@holoviz/panel@0.14.4/dist/panel.min.js\"];\n",
+       "  var js_modules = [];\n",
+       "  var css_urls = [\"https://cdn.holoviz.org/panel/0.14.4/dist/css/debugger.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/alerts.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/card.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/widgets.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/markdown.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/json.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/loading.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/dataframe.css\"];\n",
+       "  var inline_js = [    function(Bokeh) {\n",
+       "      inject_raw_css(\"\\n    .bk.pn-loading.arc:before {\\n      background-image: url(\\\"data:image/svg+xml;base64,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\\\");\\n      background-size: auto calc(min(50%, 400px));\\n    }\\n    \");\n",
+       "    },    function(Bokeh) {\n",
+       "      Bokeh.set_log_level(\"info\");\n",
+       "    },\n",
+       "function(Bokeh) {} // ensure no trailing comma for IE\n",
+       "  ];\n",
+       "\n",
+       "  function run_inline_js() {\n",
+       "    if ((root.Bokeh !== undefined) || (force === true)) {\n",
+       "      for (var i = 0; i < inline_js.length; i++) {\n",
+       "        inline_js[i].call(root, root.Bokeh);\n",
+       "      }} else if (Date.now() < root._bokeh_timeout) {\n",
+       "      setTimeout(run_inline_js, 100);\n",
+       "    } else if (!root._bokeh_failed_load) {\n",
+       "      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n",
+       "      root._bokeh_failed_load = true;\n",
+       "    }\n",
+       "  }\n",
+       "\n",
+       "  if (root._bokeh_is_loading === 0) {\n",
+       "    console.debug(\"Bokeh: BokehJS loaded, going straight to plotting\");\n",
+       "    run_inline_js();\n",
+       "  } else {\n",
+       "    load_libs(css_urls, js_urls, js_modules, function() {\n",
+       "      console.debug(\"Bokeh: BokehJS plotting callback run at\", now());\n",
+       "      run_inline_js();\n",
+       "    });\n",
+       "  }\n",
+       "}(window));"
+      ],
+      "application/vnd.holoviews_load.v0+json": "(function(root) {\n  function now() {\n    return new Date();\n  }\n\n  var force = true;\n\n  if (typeof root._bokeh_onload_callbacks === \"undefined\" || force === true) {\n    root._bokeh_onload_callbacks = [];\n    root._bokeh_is_loading = undefined;\n  }\n\n  if (typeof (root._bokeh_timeout) === \"undefined\" || force === true) {\n    root._bokeh_timeout = Date.now() + 5000;\n    root._bokeh_failed_load = false;\n  }\n\n  function run_callbacks() {\n    try {\n      root._bokeh_onload_callbacks.forEach(function(callback) {\n        if (callback != null)\n          callback();\n      });\n    } finally {\n      delete root._bokeh_onload_callbacks\n    }\n    console.debug(\"Bokeh: all callbacks have finished\");\n  }\n\n  function load_libs(css_urls, js_urls, js_modules, callback) {\n    if (css_urls == null) css_urls = [];\n    if (js_urls == null) js_urls = [];\n    if (js_modules == null) js_modules = [];\n\n    root._bokeh_onload_callbacks.push(callback);\n    if (root._bokeh_is_loading > 0) {\n      console.debug(\"Bokeh: BokehJS is being loaded, scheduling callback at\", now());\n      return null;\n    }\n    if (js_urls.length === 0 && js_modules.length === 0) {\n      run_callbacks();\n      return null;\n    }\n    console.debug(\"Bokeh: BokehJS not loaded, scheduling load and callback at\", now());\n\n    function on_load() {\n      root._bokeh_is_loading--;\n      if (root._bokeh_is_loading === 0) {\n        console.debug(\"Bokeh: all BokehJS libraries/stylesheets loaded\");\n        run_callbacks()\n      }\n    }\n\n    function on_error() {\n      console.error(\"failed to load \" + url);\n    }\n\n    for (var i = 0; i < css_urls.length; i++) {\n      var url = css_urls[i];\n      const element = document.createElement(\"link\");\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.rel = \"stylesheet\";\n      element.type = \"text/css\";\n      element.href = url;\n      console.debug(\"Bokeh: injecting link tag for BokehJS stylesheet: \", url);\n      document.body.appendChild(element);\n    }\n\n    var skip = [];\n    if (window.requirejs) {\n      window.requirejs.config({'packages': {}, 'paths': {'gridstack': 'https://cdn.jsdelivr.net/npm/gridstack@4.2.5/dist/gridstack-h5', 'notyf': 'https://cdn.jsdelivr.net/npm/notyf@3/notyf.min'}, 'shim': {'gridstack': {'exports': 'GridStack'}}});\n      require([\"gridstack\"], function(GridStack) {\n\twindow.GridStack = GridStack\n\ton_load()\n      })\n      require([\"notyf\"], function() {\n\ton_load()\n      })\n      root._bokeh_is_loading = css_urls.length + 2;\n    } else {\n      root._bokeh_is_loading = css_urls.length + js_urls.length + js_modules.length;\n    }    if (((window['GridStack'] !== undefined) && (!(window['GridStack'] instanceof HTMLElement))) || window.requirejs) {\n      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/gridstack/gridstack@4.2.5/dist/gridstack-h5.js'];\n      for (var i = 0; i < urls.length; i++) {\n        skip.push(urls[i])\n      }\n    }    if (((window['Notyf'] !== undefined) && (!(window['Notyf'] instanceof HTMLElement))) || window.requirejs) {\n      var urls = ['https://cdn.holoviz.org/panel/0.14.4/dist/bundled/notificationarea/notyf@3/notyf.min.js'];\n      for (var i = 0; i < urls.length; i++) {\n        skip.push(urls[i])\n      }\n    }    for (var i = 0; i < js_urls.length; i++) {\n      var url = js_urls[i];\n      if (skip.indexOf(url) >= 0) {\n\tif (!window.requirejs) {\n\t  on_load();\n\t}\n\tcontinue;\n      }\n      var element = document.createElement('script');\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.async = false;\n      element.src = url;\n      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n      document.head.appendChild(element);\n    }\n    for (var i = 0; i < js_modules.length; i++) {\n      var url = js_modules[i];\n      if (skip.indexOf(url) >= 0) {\n\tif (!window.requirejs) {\n\t  on_load();\n\t}\n\tcontinue;\n      }\n      var element = document.createElement('script');\n      element.onload = on_load;\n      element.onerror = on_error;\n      element.async = false;\n      element.src = url;\n      element.type = \"module\";\n      console.debug(\"Bokeh: injecting script tag for BokehJS library: \", url);\n      document.head.appendChild(element);\n    }\n    if (!js_urls.length && !js_modules.length) {\n      on_load()\n    }\n  };\n\n  function inject_raw_css(css) {\n    const element = document.createElement(\"style\");\n    element.appendChild(document.createTextNode(css));\n    document.body.appendChild(element);\n  }\n\n  var js_urls = [\"https://cdn.bokeh.org/bokeh/release/bokeh-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-gl-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-widgets-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-tables-2.4.3.min.js\", \"https://cdn.bokeh.org/bokeh/release/bokeh-mathjax-2.4.3.min.js\", \"https://unpkg.com/@holoviz/panel@0.14.4/dist/panel.min.js\"];\n  var js_modules = [];\n  var css_urls = [\"https://cdn.holoviz.org/panel/0.14.4/dist/css/debugger.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/alerts.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/card.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/widgets.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/markdown.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/json.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/loading.css\", \"https://cdn.holoviz.org/panel/0.14.4/dist/css/dataframe.css\"];\n  var inline_js = [    function(Bokeh) {\n      inject_raw_css(\"\\n    .bk.pn-loading.arc:before {\\n      background-image: url(\\\"data:image/svg+xml;base64,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\\\");\\n      background-size: auto calc(min(50%, 400px));\\n    }\\n    \");\n    },    function(Bokeh) {\n      Bokeh.set_log_level(\"info\");\n    },\nfunction(Bokeh) {} // ensure no trailing comma for IE\n  ];\n\n  function run_inline_js() {\n    if ((root.Bokeh !== undefined) || (force === true)) {\n      for (var i = 0; i < inline_js.length; i++) {\n        inline_js[i].call(root, root.Bokeh);\n      }} else if (Date.now() < root._bokeh_timeout) {\n      setTimeout(run_inline_js, 100);\n    } else if (!root._bokeh_failed_load) {\n      console.log(\"Bokeh: BokehJS failed to load within specified timeout.\");\n      root._bokeh_failed_load = true;\n    }\n  }\n\n  if (root._bokeh_is_loading === 0) {\n    console.debug(\"Bokeh: BokehJS loaded, going straight to plotting\");\n    run_inline_js();\n  } else {\n    load_libs(css_urls, js_urls, js_modules, function() {\n      console.debug(\"Bokeh: BokehJS plotting callback run at\", now());\n      run_inline_js();\n    });\n  }\n}(window));"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/javascript": [
+       "\n",
+       "if ((window.PyViz === undefined) || (window.PyViz instanceof HTMLElement)) {\n",
+       "  window.PyViz = {comms: {}, comm_status:{}, kernels:{}, receivers: {}, plot_index: []}\n",
+       "}\n",
+       "\n",
+       "\n",
+       "    function JupyterCommManager() {\n",
+       "    }\n",
+       "\n",
+       "    JupyterCommManager.prototype.register_target = function(plot_id, comm_id, msg_handler) {\n",
+       "      if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n",
+       "        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n",
+       "        comm_manager.register_target(comm_id, function(comm) {\n",
+       "          comm.on_msg(msg_handler);\n",
+       "        });\n",
+       "      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n",
+       "        window.PyViz.kernels[plot_id].registerCommTarget(comm_id, function(comm) {\n",
+       "          comm.onMsg = msg_handler;\n",
+       "        });\n",
+       "      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n",
+       "        google.colab.kernel.comms.registerTarget(comm_id, (comm) => {\n",
+       "          var messages = comm.messages[Symbol.asyncIterator]();\n",
+       "          function processIteratorResult(result) {\n",
+       "            var message = result.value;\n",
+       "            console.log(message)\n",
+       "            var content = {data: message.data, comm_id};\n",
+       "            var buffers = []\n",
+       "            for (var buffer of message.buffers || []) {\n",
+       "              buffers.push(new DataView(buffer))\n",
+       "            }\n",
+       "            var metadata = message.metadata || {};\n",
+       "            var msg = {content, buffers, metadata}\n",
+       "            msg_handler(msg);\n",
+       "            return messages.next().then(processIteratorResult);\n",
+       "          }\n",
+       "          return messages.next().then(processIteratorResult);\n",
+       "        })\n",
+       "      }\n",
+       "    }\n",
+       "\n",
+       "    JupyterCommManager.prototype.get_client_comm = function(plot_id, comm_id, msg_handler) {\n",
+       "      if (comm_id in window.PyViz.comms) {\n",
+       "        return window.PyViz.comms[comm_id];\n",
+       "      } else if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n",
+       "        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n",
+       "        var comm = comm_manager.new_comm(comm_id, {}, {}, {}, comm_id);\n",
+       "        if (msg_handler) {\n",
+       "          comm.on_msg(msg_handler);\n",
+       "        }\n",
+       "      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n",
+       "        var comm = window.PyViz.kernels[plot_id].connectToComm(comm_id);\n",
+       "        comm.open();\n",
+       "        if (msg_handler) {\n",
+       "          comm.onMsg = msg_handler;\n",
+       "        }\n",
+       "      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n",
+       "        var comm_promise = google.colab.kernel.comms.open(comm_id)\n",
+       "        comm_promise.then((comm) => {\n",
+       "          window.PyViz.comms[comm_id] = comm;\n",
+       "          if (msg_handler) {\n",
+       "            var messages = comm.messages[Symbol.asyncIterator]();\n",
+       "            function processIteratorResult(result) {\n",
+       "              var message = result.value;\n",
+       "              var content = {data: message.data};\n",
+       "              var metadata = message.metadata || {comm_id};\n",
+       "              var msg = {content, metadata}\n",
+       "              msg_handler(msg);\n",
+       "              return messages.next().then(processIteratorResult);\n",
+       "            }\n",
+       "            return messages.next().then(processIteratorResult);\n",
+       "          }\n",
+       "        }) \n",
+       "        var sendClosure = (data, metadata, buffers, disposeOnDone) => {\n",
+       "          return comm_promise.then((comm) => {\n",
+       "            comm.send(data, metadata, buffers, disposeOnDone);\n",
+       "          });\n",
+       "        };\n",
+       "        var comm = {\n",
+       "          send: sendClosure\n",
+       "        };\n",
+       "      }\n",
+       "      window.PyViz.comms[comm_id] = comm;\n",
+       "      return comm;\n",
+       "    }\n",
+       "    window.PyViz.comm_manager = new JupyterCommManager();\n",
+       "    \n",
+       "\n",
+       "\n",
+       "var JS_MIME_TYPE = 'application/javascript';\n",
+       "var HTML_MIME_TYPE = 'text/html';\n",
+       "var EXEC_MIME_TYPE = 'application/vnd.holoviews_exec.v0+json';\n",
+       "var CLASS_NAME = 'output';\n",
+       "\n",
+       "/**\n",
+       " * Render data to the DOM node\n",
+       " */\n",
+       "function render(props, node) {\n",
+       "  var div = document.createElement(\"div\");\n",
+       "  var script = document.createElement(\"script\");\n",
+       "  node.appendChild(div);\n",
+       "  node.appendChild(script);\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle when a new output is added\n",
+       " */\n",
+       "function handle_add_output(event, handle) {\n",
+       "  var output_area = handle.output_area;\n",
+       "  var output = handle.output;\n",
+       "  if ((output.data == undefined) || (!output.data.hasOwnProperty(EXEC_MIME_TYPE))) {\n",
+       "    return\n",
+       "  }\n",
+       "  var id = output.metadata[EXEC_MIME_TYPE][\"id\"];\n",
+       "  var toinsert = output_area.element.find(\".\" + CLASS_NAME.split(' ')[0]);\n",
+       "  if (id !== undefined) {\n",
+       "    var nchildren = toinsert.length;\n",
+       "    var html_node = toinsert[nchildren-1].children[0];\n",
+       "    html_node.innerHTML = output.data[HTML_MIME_TYPE];\n",
+       "    var scripts = [];\n",
+       "    var nodelist = html_node.querySelectorAll(\"script\");\n",
+       "    for (var i in nodelist) {\n",
+       "      if (nodelist.hasOwnProperty(i)) {\n",
+       "        scripts.push(nodelist[i])\n",
+       "      }\n",
+       "    }\n",
+       "\n",
+       "    scripts.forEach( function (oldScript) {\n",
+       "      var newScript = document.createElement(\"script\");\n",
+       "      var attrs = [];\n",
+       "      var nodemap = oldScript.attributes;\n",
+       "      for (var j in nodemap) {\n",
+       "        if (nodemap.hasOwnProperty(j)) {\n",
+       "          attrs.push(nodemap[j])\n",
+       "        }\n",
+       "      }\n",
+       "      attrs.forEach(function(attr) { newScript.setAttribute(attr.name, attr.value) });\n",
+       "      newScript.appendChild(document.createTextNode(oldScript.innerHTML));\n",
+       "      oldScript.parentNode.replaceChild(newScript, oldScript);\n",
+       "    });\n",
+       "    if (JS_MIME_TYPE in output.data) {\n",
+       "      toinsert[nchildren-1].children[1].textContent = output.data[JS_MIME_TYPE];\n",
+       "    }\n",
+       "    output_area._hv_plot_id = id;\n",
+       "    if ((window.Bokeh !== undefined) && (id in Bokeh.index)) {\n",
+       "      window.PyViz.plot_index[id] = Bokeh.index[id];\n",
+       "    } else {\n",
+       "      window.PyViz.plot_index[id] = null;\n",
+       "    }\n",
+       "  } else if (output.metadata[EXEC_MIME_TYPE][\"server_id\"] !== undefined) {\n",
+       "    var bk_div = document.createElement(\"div\");\n",
+       "    bk_div.innerHTML = output.data[HTML_MIME_TYPE];\n",
+       "    var script_attrs = bk_div.children[0].attributes;\n",
+       "    for (var i = 0; i < script_attrs.length; i++) {\n",
+       "      toinsert[toinsert.length - 1].childNodes[1].setAttribute(script_attrs[i].name, script_attrs[i].value);\n",
+       "    }\n",
+       "    // store reference to server id on output_area\n",
+       "    output_area._bokeh_server_id = output.metadata[EXEC_MIME_TYPE][\"server_id\"];\n",
+       "  }\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle when an output is cleared or removed\n",
+       " */\n",
+       "function handle_clear_output(event, handle) {\n",
+       "  var id = handle.cell.output_area._hv_plot_id;\n",
+       "  var server_id = handle.cell.output_area._bokeh_server_id;\n",
+       "  if (((id === undefined) || !(id in PyViz.plot_index)) && (server_id !== undefined)) { return; }\n",
+       "  var comm = window.PyViz.comm_manager.get_client_comm(\"hv-extension-comm\", \"hv-extension-comm\", function () {});\n",
+       "  if (server_id !== null) {\n",
+       "    comm.send({event_type: 'server_delete', 'id': server_id});\n",
+       "    return;\n",
+       "  } else if (comm !== null) {\n",
+       "    comm.send({event_type: 'delete', 'id': id});\n",
+       "  }\n",
+       "  delete PyViz.plot_index[id];\n",
+       "  if ((window.Bokeh !== undefined) & (id in window.Bokeh.index)) {\n",
+       "    var doc = window.Bokeh.index[id].model.document\n",
+       "    doc.clear();\n",
+       "    const i = window.Bokeh.documents.indexOf(doc);\n",
+       "    if (i > -1) {\n",
+       "      window.Bokeh.documents.splice(i, 1);\n",
+       "    }\n",
+       "  }\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle kernel restart event\n",
+       " */\n",
+       "function handle_kernel_cleanup(event, handle) {\n",
+       "  delete PyViz.comms[\"hv-extension-comm\"];\n",
+       "  window.PyViz.plot_index = {}\n",
+       "}\n",
+       "\n",
+       "/**\n",
+       " * Handle update_display_data messages\n",
+       " */\n",
+       "function handle_update_output(event, handle) {\n",
+       "  handle_clear_output(event, {cell: {output_area: handle.output_area}})\n",
+       "  handle_add_output(event, handle)\n",
+       "}\n",
+       "\n",
+       "function register_renderer(events, OutputArea) {\n",
+       "  function append_mime(data, metadata, element) {\n",
+       "    // create a DOM node to render to\n",
+       "    var toinsert = this.create_output_subarea(\n",
+       "    metadata,\n",
+       "    CLASS_NAME,\n",
+       "    EXEC_MIME_TYPE\n",
+       "    );\n",
+       "    this.keyboard_manager.register_events(toinsert);\n",
+       "    // Render to node\n",
+       "    var props = {data: data, metadata: metadata[EXEC_MIME_TYPE]};\n",
+       "    render(props, toinsert[0]);\n",
+       "    element.append(toinsert);\n",
+       "    return toinsert\n",
+       "  }\n",
+       "\n",
+       "  events.on('output_added.OutputArea', handle_add_output);\n",
+       "  events.on('output_updated.OutputArea', handle_update_output);\n",
+       "  events.on('clear_output.CodeCell', handle_clear_output);\n",
+       "  events.on('delete.Cell', handle_clear_output);\n",
+       "  events.on('kernel_ready.Kernel', handle_kernel_cleanup);\n",
+       "\n",
+       "  OutputArea.prototype.register_mime_type(EXEC_MIME_TYPE, append_mime, {\n",
+       "    safe: true,\n",
+       "    index: 0\n",
+       "  });\n",
+       "}\n",
+       "\n",
+       "if (window.Jupyter !== undefined) {\n",
+       "  try {\n",
+       "    var events = require('base/js/events');\n",
+       "    var OutputArea = require('notebook/js/outputarea').OutputArea;\n",
+       "    if (OutputArea.prototype.mime_types().indexOf(EXEC_MIME_TYPE) == -1) {\n",
+       "      register_renderer(events, OutputArea);\n",
+       "    }\n",
+       "  } catch(err) {\n",
+       "  }\n",
+       "}\n"
+      ],
+      "application/vnd.holoviews_load.v0+json": "\nif ((window.PyViz === undefined) || (window.PyViz instanceof HTMLElement)) {\n  window.PyViz = {comms: {}, comm_status:{}, kernels:{}, receivers: {}, plot_index: []}\n}\n\n\n    function JupyterCommManager() {\n    }\n\n    JupyterCommManager.prototype.register_target = function(plot_id, comm_id, msg_handler) {\n      if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n        comm_manager.register_target(comm_id, function(comm) {\n          comm.on_msg(msg_handler);\n        });\n      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n        window.PyViz.kernels[plot_id].registerCommTarget(comm_id, function(comm) {\n          comm.onMsg = msg_handler;\n        });\n      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n        google.colab.kernel.comms.registerTarget(comm_id, (comm) => {\n          var messages = comm.messages[Symbol.asyncIterator]();\n          function processIteratorResult(result) {\n            var message = result.value;\n            console.log(message)\n            var content = {data: message.data, comm_id};\n            var buffers = []\n            for (var buffer of message.buffers || []) {\n              buffers.push(new DataView(buffer))\n            }\n            var metadata = message.metadata || {};\n            var msg = {content, buffers, metadata}\n            msg_handler(msg);\n            return messages.next().then(processIteratorResult);\n          }\n          return messages.next().then(processIteratorResult);\n        })\n      }\n    }\n\n    JupyterCommManager.prototype.get_client_comm = function(plot_id, comm_id, msg_handler) {\n      if (comm_id in window.PyViz.comms) {\n        return window.PyViz.comms[comm_id];\n      } else if (window.comm_manager || ((window.Jupyter !== undefined) && (Jupyter.notebook.kernel != null))) {\n        var comm_manager = window.comm_manager || Jupyter.notebook.kernel.comm_manager;\n        var comm = comm_manager.new_comm(comm_id, {}, {}, {}, comm_id);\n        if (msg_handler) {\n          comm.on_msg(msg_handler);\n        }\n      } else if ((plot_id in window.PyViz.kernels) && (window.PyViz.kernels[plot_id])) {\n        var comm = window.PyViz.kernels[plot_id].connectToComm(comm_id);\n        comm.open();\n        if (msg_handler) {\n          comm.onMsg = msg_handler;\n        }\n      } else if (typeof google != 'undefined' && google.colab.kernel != null) {\n        var comm_promise = google.colab.kernel.comms.open(comm_id)\n        comm_promise.then((comm) => {\n          window.PyViz.comms[comm_id] = comm;\n          if (msg_handler) {\n            var messages = comm.messages[Symbol.asyncIterator]();\n            function processIteratorResult(result) {\n              var message = result.value;\n              var content = {data: message.data};\n              var metadata = message.metadata || {comm_id};\n              var msg = {content, metadata}\n              msg_handler(msg);\n              return messages.next().then(processIteratorResult);\n            }\n            return messages.next().then(processIteratorResult);\n          }\n        }) \n        var sendClosure = (data, metadata, buffers, disposeOnDone) => {\n          return comm_promise.then((comm) => {\n            comm.send(data, metadata, buffers, disposeOnDone);\n          });\n        };\n        var comm = {\n          send: sendClosure\n        };\n      }\n      window.PyViz.comms[comm_id] = comm;\n      return comm;\n    }\n    window.PyViz.comm_manager = new JupyterCommManager();\n    \n\n\nvar JS_MIME_TYPE = 'application/javascript';\nvar HTML_MIME_TYPE = 'text/html';\nvar EXEC_MIME_TYPE = 'application/vnd.holoviews_exec.v0+json';\nvar CLASS_NAME = 'output';\n\n/**\n * Render data to the DOM node\n */\nfunction render(props, node) {\n  var div = document.createElement(\"div\");\n  var script = document.createElement(\"script\");\n  node.appendChild(div);\n  node.appendChild(script);\n}\n\n/**\n * Handle when a new output is added\n */\nfunction handle_add_output(event, handle) {\n  var output_area = handle.output_area;\n  var output = handle.output;\n  if ((output.data == undefined) || (!output.data.hasOwnProperty(EXEC_MIME_TYPE))) {\n    return\n  }\n  var id = output.metadata[EXEC_MIME_TYPE][\"id\"];\n  var toinsert = output_area.element.find(\".\" + CLASS_NAME.split(' ')[0]);\n  if (id !== undefined) {\n    var nchildren = toinsert.length;\n    var html_node = toinsert[nchildren-1].children[0];\n    html_node.innerHTML = output.data[HTML_MIME_TYPE];\n    var scripts = [];\n    var nodelist = html_node.querySelectorAll(\"script\");\n    for (var i in nodelist) {\n      if (nodelist.hasOwnProperty(i)) {\n        scripts.push(nodelist[i])\n      }\n    }\n\n    scripts.forEach( function (oldScript) {\n      var newScript = document.createElement(\"script\");\n      var attrs = [];\n      var nodemap = oldScript.attributes;\n      for (var j in nodemap) {\n        if (nodemap.hasOwnProperty(j)) {\n          attrs.push(nodemap[j])\n        }\n      }\n      attrs.forEach(function(attr) { newScript.setAttribute(attr.name, attr.value) });\n      newScript.appendChild(document.createTextNode(oldScript.innerHTML));\n      oldScript.parentNode.replaceChild(newScript, oldScript);\n    });\n    if (JS_MIME_TYPE in output.data) {\n      toinsert[nchildren-1].children[1].textContent = output.data[JS_MIME_TYPE];\n    }\n    output_area._hv_plot_id = id;\n    if ((window.Bokeh !== undefined) && (id in Bokeh.index)) {\n      window.PyViz.plot_index[id] = Bokeh.index[id];\n    } else {\n      window.PyViz.plot_index[id] = null;\n    }\n  } else if (output.metadata[EXEC_MIME_TYPE][\"server_id\"] !== undefined) {\n    var bk_div = document.createElement(\"div\");\n    bk_div.innerHTML = output.data[HTML_MIME_TYPE];\n    var script_attrs = bk_div.children[0].attributes;\n    for (var i = 0; i < script_attrs.length; i++) {\n      toinsert[toinsert.length - 1].childNodes[1].setAttribute(script_attrs[i].name, script_attrs[i].value);\n    }\n    // store reference to server id on output_area\n    output_area._bokeh_server_id = output.metadata[EXEC_MIME_TYPE][\"server_id\"];\n  }\n}\n\n/**\n * Handle when an output is cleared or removed\n */\nfunction handle_clear_output(event, handle) {\n  var id = handle.cell.output_area._hv_plot_id;\n  var server_id = handle.cell.output_area._bokeh_server_id;\n  if (((id === undefined) || !(id in PyViz.plot_index)) && (server_id !== undefined)) { return; }\n  var comm = window.PyViz.comm_manager.get_client_comm(\"hv-extension-comm\", \"hv-extension-comm\", function () {});\n  if (server_id !== null) {\n    comm.send({event_type: 'server_delete', 'id': server_id});\n    return;\n  } else if (comm !== null) {\n    comm.send({event_type: 'delete', 'id': id});\n  }\n  delete PyViz.plot_index[id];\n  if ((window.Bokeh !== undefined) & (id in window.Bokeh.index)) {\n    var doc = window.Bokeh.index[id].model.document\n    doc.clear();\n    const i = window.Bokeh.documents.indexOf(doc);\n    if (i > -1) {\n      window.Bokeh.documents.splice(i, 1);\n    }\n  }\n}\n\n/**\n * Handle kernel restart event\n */\nfunction handle_kernel_cleanup(event, handle) {\n  delete PyViz.comms[\"hv-extension-comm\"];\n  window.PyViz.plot_index = {}\n}\n\n/**\n * Handle update_display_data messages\n */\nfunction handle_update_output(event, handle) {\n  handle_clear_output(event, {cell: {output_area: handle.output_area}})\n  handle_add_output(event, handle)\n}\n\nfunction register_renderer(events, OutputArea) {\n  function append_mime(data, metadata, element) {\n    // create a DOM node to render to\n    var toinsert = this.create_output_subarea(\n    metadata,\n    CLASS_NAME,\n    EXEC_MIME_TYPE\n    );\n    this.keyboard_manager.register_events(toinsert);\n    // Render to node\n    var props = {data: data, metadata: metadata[EXEC_MIME_TYPE]};\n    render(props, toinsert[0]);\n    element.append(toinsert);\n    return toinsert\n  }\n\n  events.on('output_added.OutputArea', handle_add_output);\n  events.on('output_updated.OutputArea', handle_update_output);\n  events.on('clear_output.CodeCell', handle_clear_output);\n  events.on('delete.Cell', handle_clear_output);\n  events.on('kernel_ready.Kernel', handle_kernel_cleanup);\n\n  OutputArea.prototype.register_mime_type(EXEC_MIME_TYPE, append_mime, {\n    safe: true,\n    index: 0\n  });\n}\n\nif (window.Jupyter !== undefined) {\n  try {\n    var events = require('base/js/events');\n    var OutputArea = require('notebook/js/outputarea').OutputArea;\n    if (OutputArea.prototype.mime_types().indexOf(EXEC_MIME_TYPE) == -1) {\n      register_renderer(events, OutputArea);\n    }\n  } catch(err) {\n  }\n}\n"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/html": [
+       "<style>.bk-root, .bk-root .bk:before, .bk-root .bk:after {\n",
+       "  font-family: var(--jp-ui-font-size1);\n",
+       "  font-size: var(--jp-ui-font-size1);\n",
+       "  color: var(--jp-ui-font-color1);\n",
+       "}\n",
+       "</style>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "# 中文\n",
+    "import panel as pn  # GUI\n",
+    "pn.extension()\n",
+    "\n",
+    "panels = [] # collect display \n",
+    "\n",
+    "context = [{'role':'system', 'content':\"\"\"\n",
+    "你是订餐机器人，为披萨餐厅自动收集订单信息。\n",
+    "你要首先问候顾客。然后等待用户回复收集订单信息。收集完信息需确认顾客是否还需要添加其他内容。\n",
+    "最后需要询问是否自取或外送，如果是外送，你要询问地址。\n",
+    "最后告诉顾客订单总金额，并送上祝福。\n",
+    "\n",
+    "请确保明确所有选项、附加项和尺寸，以便从菜单中识别出该项唯一的内容。\n",
+    "你的回应应该以简短、非常随意和友好的风格呈现。\n",
+    "\n",
+    "菜单包括：\n",
+    "\n",
+    "菜品：\n",
+    "意式辣香肠披萨（大、中、小） 12.95、10.00、7.00\n",
+    "芝士披萨（大、中、小） 10.95、9.25、6.50\n",
+    "茄子披萨（大、中、小） 11.95、9.75、6.75\n",
+    "薯条（大、小） 4.50、3.50\n",
+    "希腊沙拉 7.25\n",
+    "\n",
+    "配料：\n",
+    "奶酪 2.00\n",
+    "蘑菇 1.50\n",
+    "香肠 3.00\n",
+    "加拿大熏肉 3.50\n",
+    "AI酱 1.50\n",
+    "辣椒 1.00\n",
+    "\n",
+    "饮料：\n",
+    "可乐（大、中、小） 3.00、2.00、1.00\n",
+    "雪碧（大、中、小） 3.00、2.00、1.00\n",
+    "瓶装水 5.00\n",
+    "\"\"\"} ]  # accumulate messages\n",
+    "\n",
+    "\n",
+    "inp = pn.widgets.TextInput(value=\"Hi\", placeholder='Enter text here…')\n",
+    "button_conversation = pn.widgets.Button(name=\"Chat!\")\n",
+    "\n",
+    "interactive_conversation = pn.bind(collect_messages, button_conversation)\n",
+    "\n",
+    "dashboard = pn.Column(\n",
+    "    inp,\n",
+    "    pn.Row(button_conversation),\n",
+    "    pn.panel(interactive_conversation, loading_indicator=True, height=300),\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 77,
+   "id": "b9a8ef58",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {},
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.holoviews_exec.v0+json": "",
+      "text/html": [
+       "<div id='1717'>\n",
+       "  <div class=\"bk-root\" id=\"e321dfc5-ed71-4c7e-9054-ad25bb0996c1\" data-root-id=\"1717\"></div>\n",
+       "</div>\n",
+       "<script type=\"application/javascript\">(function(root) {\n",
+       "  function embed_document(root) {\n",
+       "    var docs_json = {\"ba214626-f150-41f7-88bd-59aed92ab8b3\":{\"defs\":[{\"extends\":null,\"module\":null,\"name\":\"ReactiveHTML1\",\"overrides\":[],\"properties\":[]},{\"extends\":null,\"module\":null,\"name\":\"FlexBox1\",\"overrides\":[],\"properties\":[{\"default\":\"flex-start\",\"kind\":null,\"name\":\"align_content\"},{\"default\":\"flex-start\",\"kind\":null,\"name\":\"align_items\"},{\"default\":\"row\",\"kind\":null,\"name\":\"flex_direction\"},{\"default\":\"wrap\",\"kind\":null,\"name\":\"flex_wrap\"},{\"default\":\"flex-start\",\"kind\":null,\"name\":\"justify_content\"}]},{\"extends\":null,\"module\":null,\"name\":\"GridStack1\",\"overrides\":[],\"properties\":[{\"default\":\"warn\",\"kind\":null,\"name\":\"mode\"},{\"default\":null,\"kind\":null,\"name\":\"ncols\"},{\"default\":null,\"kind\":null,\"name\":\"nrows\"},{\"default\":true,\"kind\":null,\"name\":\"allow_resize\"},{\"default\":true,\"kind\":null,\"name\":\"allow_drag\"},{\"default\":[],\"kind\":null,\"name\":\"state\"}]},{\"extends\":null,\"module\":null,\"name\":\"click1\",\"overrides\":[],\"properties\":[{\"default\":\"\",\"kind\":null,\"name\":\"terminal_output\"},{\"default\":\"\",\"kind\":null,\"name\":\"debug_name\"},{\"default\":0,\"kind\":null,\"name\":\"clears\"}]},{\"extends\":null,\"module\":null,\"name\":\"NotificationAreaBase1\",\"overrides\":[],\"properties\":[{\"default\":\"bottom-right\",\"kind\":null,\"name\":\"position\"},{\"default\":0,\"kind\":null,\"name\":\"_clear\"}]},{\"extends\":null,\"module\":null,\"name\":\"NotificationArea1\",\"overrides\":[],\"properties\":[{\"default\":[],\"kind\":null,\"name\":\"notifications\"},{\"default\":\"bottom-right\",\"kind\":null,\"name\":\"position\"},{\"default\":0,\"kind\":null,\"name\":\"_clear\"},{\"default\":[{\"background\":\"#ffc107\",\"icon\":{\"className\":\"fas fa-exclamation-triangle\",\"color\":\"white\",\"tagName\":\"i\"},\"type\":\"warning\"},{\"background\":\"#007bff\",\"icon\":{\"className\":\"fas fa-info-circle\",\"color\":\"white\",\"tagName\":\"i\"},\"type\":\"info\"}],\"kind\":null,\"name\":\"types\"}]},{\"extends\":null,\"module\":null,\"name\":\"Notification\",\"overrides\":[],\"properties\":[{\"default\":null,\"kind\":null,\"name\":\"background\"},{\"default\":3000,\"kind\":null,\"name\":\"duration\"},{\"default\":null,\"kind\":null,\"name\":\"icon\"},{\"default\":\"\",\"kind\":null,\"name\":\"message\"},{\"default\":null,\"kind\":null,\"name\":\"notification_type\"},{\"default\":false,\"kind\":null,\"name\":\"_destroyed\"}]},{\"extends\":null,\"module\":null,\"name\":\"TemplateActions1\",\"overrides\":[],\"properties\":[{\"default\":0,\"kind\":null,\"name\":\"open_modal\"},{\"default\":0,\"kind\":null,\"name\":\"close_modal\"}]},{\"extends\":null,\"module\":null,\"name\":\"MaterialTemplateActions1\",\"overrides\":[],\"properties\":[{\"default\":0,\"kind\":null,\"name\":\"open_modal\"},{\"default\":0,\"kind\":null,\"name\":\"close_modal\"}]}],\"roots\":{\"references\":[{\"attributes\":{\"children\":[{\"id\":\"1720\"}],\"margin\":[0,0,0,0],\"name\":\"Row01094\"},\"id\":\"1719\",\"type\":\"Row\"},{\"attributes\":{\"args\":{\"bidirectional\":false,\"properties\":{\"event:button_click\":\"loading\"},\"source\":{\"id\":\"1720\"},\"target\":{\"id\":\"1721\"}},\"code\":\"\\n    if ('event:button_click'.startsWith('event:')) {\\n      var value = true\\n    } else {\\n      var value = source['event:button_click'];\\n      value = value;\\n    }\\n    if (typeof value !== 'boolean' || source.labels !== ['Loading']) {\\n      value = true\\n    }\\n    var css_classes = target.css_classes.slice()\\n    var loading_css = ['pn-loading', 'arc']\\n    if (value) {\\n      for (var css of loading_css) {\\n        if (!(css in css_classes)) {\\n          css_classes.push(css)\\n        }\\n      }\\n    } else {\\n     for (var css of loading_css) {\\n        var index = css_classes.indexOf(css)\\n        if (index > -1) {\\n          css_classes.splice(index, 1)\\n        }\\n      }\\n    }\\n    target['css_classes'] = css_classes\\n    \",\"tags\":[[5239341456,[null,\"event:button_click\"],[null,\"loading\"]]]},\"id\":\"1729\",\"type\":\"CustomJS\"},{\"attributes\":{\"css_classes\":[\"markdown\"],\"margin\":[5,5,5,5],\"name\":\"Markdown01103\",\"text\":\"&lt;p&gt;User:&lt;/p&gt;\"},\"id\":\"1724\",\"type\":\"panel.models.markup.HTML\"},{\"attributes\":{\"client_comm_id\":\"df72227b43a2413185c677536f738a1c\",\"comm_id\":\"bf7b0dfe31d54807a147c04852dfa441\",\"plot_id\":\"1717\"},\"id\":\"1730\",\"type\":\"panel.models.comm_manager.CommManager\"},{\"attributes\":{\"children\":[{\"id\":\"1722\"}],\"height\":300,\"margin\":[0,0,0,0],\"min_height\":300,\"name\":\"Row01099\"},\"id\":\"1721\",\"type\":\"Row\"},{\"attributes\":{\"icon\":null,\"js_event_callbacks\":{\"button_click\":[{\"id\":\"1729\"}]},\"label\":\"Chat!\",\"margin\":[5,10,5,10],\"subscribed_events\":[\"button_click\"]},\"id\":\"1720\",\"type\":\"Button\"},{\"attributes\":{\"children\":[{\"id\":\"1724\"},{\"id\":\"1725\"}],\"margin\":[0,0,0,0],\"name\":\"Row01105\"},\"id\":\"1723\",\"type\":\"Row\"},{\"attributes\":{\"css_classes\":[\"markdown\"],\"margin\":[5,5,5,5],\"name\":\"Markdown01106\",\"style\":{\"background-color\":\"#F6F6F6\"},\"text\":\"&lt;p&gt;\\u4f60\\u597d\\uff01\\u6b22\\u8fce\\u6765\\u5230\\u62ab\\u8428\\u9910\\u5385\\uff01\\u8bf7\\u95ee\\u60a8\\u60f3\\u70b9\\u4ec0\\u4e48\\uff1f&lt;/p&gt;\",\"width\":600},\"id\":\"1728\",\"type\":\"panel.models.markup.HTML\"},{\"attributes\":{\"children\":[{\"id\":\"1723\"},{\"id\":\"1726\"}],\"margin\":[0,0,0,0],\"name\":\"Column01111\"},\"id\":\"1722\",\"type\":\"Column\"},{\"attributes\":{\"css_classes\":[\"markdown\"],\"margin\":[5,5,5,5],\"name\":\"Markdown01108\",\"text\":\"&lt;p&gt;Assistant:&lt;/p&gt;\"},\"id\":\"1727\",\"type\":\"panel.models.markup.HTML\"},{\"attributes\":{\"css_classes\":[\"markdown\"],\"margin\":[5,5,5,5],\"name\":\"Markdown01101\",\"width\":600},\"id\":\"1725\",\"type\":\"panel.models.markup.HTML\"},{\"attributes\":{\"margin\":[5,10,5,10],\"max_length\":5000,\"placeholder\":\"Enter text here\\u2026\"},\"id\":\"1718\",\"type\":\"TextInput\"},{\"attributes\":{\"children\":[{\"id\":\"1718\"},{\"id\":\"1719\"},{\"id\":\"1721\"}],\"margin\":[0,0,0,0],\"name\":\"Column01113\"},\"id\":\"1717\",\"type\":\"Column\"},{\"attributes\":{\"children\":[{\"id\":\"1727\"},{\"id\":\"1728\"}],\"margin\":[0,0,0,0],\"name\":\"Row01110\"},\"id\":\"1726\",\"type\":\"Row\"}],\"root_ids\":[\"1717\",\"1730\"]},\"title\":\"Bokeh Application\",\"version\":\"2.4.3\"}};\n",
+       "    var render_items = [{\"docid\":\"ba214626-f150-41f7-88bd-59aed92ab8b3\",\"root_ids\":[\"1717\"],\"roots\":{\"1717\":\"e321dfc5-ed71-4c7e-9054-ad25bb0996c1\"}}];\n",
+       "    root.Bokeh.embed.embed_items_notebook(docs_json, render_items);\n",
+       "    for (const render_item of render_items) {\n",
+       "      for (const root_id of render_item.root_ids) {\n",
+       "\tconst id_el = document.getElementById(root_id)\n",
+       "\tif (id_el.children.length && (id_el.children[0].className === 'bk-root')) {\n",
+       "\t  const root_el = id_el.children[0]\n",
+       "\t  root_el.id = root_el.id + '-rendered'\n",
+       "\t}\n",
+       "      }\n",
+       "    }\n",
+       "  }\n",
+       "  if (root.Bokeh !== undefined && root.Bokeh.Panel !== undefined) {\n",
+       "    embed_document(root);\n",
+       "  } else {\n",
+       "    var attempts = 0;\n",
+       "    var timer = setInterval(function(root) {\n",
+       "      if (root.Bokeh !== undefined && root.Bokeh.Panel !== undefined) {\n",
+       "        clearInterval(timer);\n",
+       "        embed_document(root);\n",
+       "      } else if (document.readyState == \"complete\") {\n",
+       "        attempts++;\n",
+       "        if (attempts > 200) {\n",
+       "          clearInterval(timer);\n",
+       "          console.log(\"Bokeh: ERROR: Unable to run BokehJS code because BokehJS library is missing\");\n",
+       "        }\n",
+       "      }\n",
+       "    }, 25, root)\n",
+       "  }\n",
+       "})(window);</script>"
+      ],
+      "text/plain": [
+       "Column\n",
+       "    [0] TextInput(placeholder='Enter text here…')\n",
+       "    [1] Row\n",
+       "        [0] Button(name='Chat!')\n",
+       "    [2] ParamFunction(function, _pane=Column, height=300, loading_indicator=True)"
+      ]
+     },
+     "execution_count": 77,
+     "metadata": {
+      "application/vnd.holoviews_exec.v0+json": {
+       "id": "1717"
+      }
+     },
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "dashboard"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 78,
+   "id": "96b2a2c7",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "以下是上一个食品订单的 JSON 摘要：\n",
+      "\n",
+      "```\n",
+      "{\n",
+      "  \"order\": {\n",
+      "    \"pizza\": {\n",
+      "      \"type\": \"芝士披萨\",\n",
+      "      \"size\": \"大\",\n",
+      "      \"price\": 10.95\n",
+      "    },\n",
+      "    \"toppings\": [\n",
+      "      {\n",
+      "        \"name\": \"蘑菇\",\n",
+      "        \"price\": 1.5\n",
+      "      }\n",
+      "    ],\n",
+      "    \"drinks\": [\n",
+      "      {\n",
+      "        \"name\": \"雪碧\",\n",
+      "        \"size\": \"大\",\n",
+      "        \"price\": 3\n",
+      "      },\n",
+      "      {\n",
+      "        \"name\": \"雪碧\",\n",
+      "        \"size\": \"大\",\n",
+      "        \"price\": 3\n",
+      "      }\n",
+      "    ],\n",
+      "    \"sides\": [],\n",
+      "    \"total_price\": 18.45\n",
+      "  }\n",
+      "}\n",
+      "```\n"
+     ]
+    }
+   ],
+   "source": [
+    "messages =  context.copy()\n",
+    "messages.append(\n",
+    "{'role':'system', 'content':'创建上一个食品订单的 json 摘要。\\\n",
+    "逐项列出每件商品的价格，字段应该是 1) 披萨，包括大小 2) 配料列表 3) 饮料列表，包括大小 4) 配菜列表包括大小 5) 总价'},    \n",
+    ")\n",
+    "\n",
+    "response = get_completion_from_messages(messages, temperature=0)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "ef17c2b2",
+   "metadata": {},
+   "source": [
+    "现在，我们已经建立了自己的订餐聊天机器人。请随意自定义并修改系统消息，以更改聊天机器人的行为，并使其扮演不同的角色和拥有不同的知识。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "3153c581-1c72-497a-9293-8db3bcb804fc",
+   "metadata": {},
+   "source": [
+    "## 尝试你的实验！\n",
+    "\n",
+    "你可以修改菜单或指令来创建自己的订单机器人！"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1b0df540",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "2cc84122",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "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": "3.8.13"
+  },
+  "latex_envs": {
+   "LaTeX_envs_menu_present": true,
+   "autoclose": false,
+   "autocomplete": true,
+   "bibliofile": "biblio.bib",
+   "cite_by": "apalike",
+   "current_citInitial": 1,
+   "eqLabelWithNumbers": true,
+   "eqNumInitial": 1,
+   "hotkeys": {
+    "equation": "Ctrl-E",
+    "itemize": "Ctrl-I"
+   },
+   "labels_anchors": false,
+   "latex_user_defs": false,
+   "report_style_numbering": false,
+   "user_envs_cfg": false
+  },
+  "toc": {
+   "base_numbering": 1,
+   "nav_menu": {},
+   "number_sections": true,
+   "sideBar": true,
+   "skip_h1_title": false,
+   "title_cell": "Table of Contents",
+   "title_sidebar": "Contents",
+   "toc_cell": false,
+   "toc_position": {
+    "height": "calc(100% - 180px)",
+    "left": "10px",
+    "top": "150px",
+    "width": "277px"
+   },
+   "toc_section_display": true,
+   "toc_window_display": true
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/content/9. 总结.md b/content/9. 总结.md
new file mode 100644
index 0000000..3ae4aff
--- /dev/null
+++ b/content/9. 总结.md	
@@ -0,0 +1,18 @@
+恭喜你完成了这门短期课程。
+
+总的来说，在这门课程中，我们学习了关于prompt的两个关键原则：
+
+- 编写清晰具体的指令；
+- 如果适当的话，给模型一些思考时间。
+
+你还学习了迭代式prompt开发的方法，并了解了如何找到适合你应用程序的prompt的过程是非常关键的。
+
+我们还介绍了许多大型语言模型的功能，包括摘要、推断、转换和扩展。你还学会了如何构建自定义聊天机器人。在这门短期课程中，你学到了很多，希望你喜欢这些学习材料。
+
+我们希望你能想出一些应用程序的想法，并尝试自己构建它们。请尝试一下并让我们知道你的想法。你可以从一个非常小的项目开始，也许它具有一定的实用价值，也可能完全没有实用价值，只是一些有趣好玩儿的东西。请利用你第一个项目的学习经验来构建更好的第二个项目，甚至更好的第三个项目等。或者，如果你已经有一个更大的项目想法，那就去做吧。
+
+大型语言模型非常强大，作为提醒，我们希望大家负责任地使用它们，请仅构建对他人有积极影响的东西。在这个时代，构建人工智能系统的人可以对他人产生巨大的影响。因此必须负责任地使用这些工具。
+
+现在，基于大型语言模型构建应用程序是一个非常令人兴奋和不断发展的领域。现在你已经完成了这门课程，我们认为你现在拥有了丰富的知识，可以帮助你构建其他人今天不知道如何构建的东西。因此，我希望你也能帮助我们传播并鼓励其他人也参加这门课程。
+
+最后，希望你在完成这门课程时感到愉快，感谢你完成了这门课程。我们期待听到你构建的惊人之作。
\ No newline at end of file