crewai_RAG_unstructured.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9d5e011a-2f8f-4c99-b8ce-9b1984a55784",
   "metadata": {},
   "outputs": [],
   "source": [
    "#!pip install cohere python-pptx crewai==0.28.8 crewai_tools==0.1.6 langchain_cohere unstructured_client unstructured[all-docs] chromadb langsmith -U"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "f487ad86-80ee-4e1e-8801-2c90c3812e2e",
   "metadata": {},
   "outputs": [],
   "source": [
    "#pip install faiss-cpu"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "858bbf8a-f7af-4138-8552-80aa8762bca6",
   "metadata": {},
   "outputs": [],
   "source": [
    "#!pip install langchainhub"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "d98f2369-f08f-4f12-b534-93ac480678a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "#=================\n",
    "# Import Libraries\n",
    "#=================\n",
    "\n",
    "from crewai import Agent, Task, Crew\n",
    "from crewai_tools import BaseTool\n",
    "import os\n",
    "\n",
    "\n",
    "from langchain_cohere import ChatCohere\n",
    "from unstructured.chunking.title import chunk_by_title\n",
    "from unstructured.partition.auto import partition\n",
    "from unstructured_client import UnstructuredClient\n",
    "\n",
    "\n",
    "from langchain_core.documents import Document\n",
    "from langchain.chains import RetrievalQA\n",
    "from langchain.callbacks import StdOutCallbackHandler\n",
    "from langchain_openai import ChatOpenAI\n",
    "from langchain_community.embeddings import HuggingFaceEmbeddings\n",
    "from langchain.vectorstores import FAISS\n",
    "from IPython.display import Markdown\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "842331bd-2846-4fc5-9ded-e32928ca3270",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "\n",
    "#=================\n",
    "# Input Files\n",
    "#=================\n",
    "\n",
    "file1 = 'unleashing-power-gen-ai-in-procurement.pdf'\n",
    "file2 = 'UK_Tech_Clusters.pptx'\n",
    "file3 = 'AI_In_Agriculture.docx'\n",
    "\n",
    "#=================\n",
    "# Unstructured Key\n",
    "#=================\n",
    "\n",
    "s = UnstructuredClient(\n",
    "    api_key_auth=\"Unstructured_Key_Goes_Here\"\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8219040d-bf79-4656-9d5e-6bf1a5b3af35",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "#=================\n",
    "# Partitioning Input Files\n",
    "#=================\n",
    "\n",
    "#pdf file\n",
    "elements1 = partition(filename=file1)\n",
    "\n",
    "#pptx\n",
    "elements2 = partition(filename=file2)\n",
    "\n",
    "#docx\n",
    "elements3 = partition(filename=file3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c5aef1b7-a2c9-45b7-8203-4bbaed66aa3a",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "#=================\n",
    "# Chunking Input Files By Title\n",
    "#=================\n",
    "\n",
    "elements = chunk_by_title(elements1 + elements2 + elements3 )\n",
    "\n",
    "documents = []\n",
    "for element in elements:\n",
    "    metadata = element.metadata.to_dict()\n",
    "    metadata[\"source\"] = metadata[\"filename\"]\n",
    "    documents.append(Document(page_content=element.text, metadata=metadata))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "c745abe2-5860-4898-8184-8ddd3a3e4132",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Mustafa\\Anaconda_3\\Lib\\site-packages\\huggingface_hub\\file_download.py:1132: FutureWarning: `resume_download` is deprecated and will be removed in version 1.0.0. Downloads always resume when possible. If you want to force a new download, use `force_download=True`.\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "#=================\n",
    "# Embedding Part\n",
    "#=================\n",
    "\n",
    "model_name = \"sentence-transformers/all-mpnet-base-v2\"\n",
    "model_kwargs = {'device': 'cpu'}\n",
    "encode_kwargs = {'normalize_embeddings': False}\n",
    "hf = HuggingFaceEmbeddings(\n",
    "    model_name=model_name,\n",
    "    model_kwargs=model_kwargs,\n",
    "    encode_kwargs=encode_kwargs\n",
    ")\n",
    "\n",
    "embeddings = hf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "6e5cc382-1cb0-46e2-9511-50442261761f",
   "metadata": {},
   "outputs": [],
   "source": [
    "#=================\n",
    "# Vector Storage and Retrieval\n",
    "#=================\n",
    "\n",
    "vectorstore = FAISS.from_documents(documents, embeddings)\n",
    "retriever = vectorstore.as_retriever()\n",
    "handler =  StdOutCallbackHandler()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "a35e7797-2438-493c-913f-5dc9fdad4790",
   "metadata": {},
   "outputs": [],
   "source": [
    "#=================\n",
    "# Retrieval Chain\n",
    "#=================\n",
    "\n",
    "os.environ[\"openai_api_key\"] = \"Openai_Key_Goes_Here\"\n",
    "\n",
    "llm = ChatOpenAI(temperature=0.0,\n",
    "                 model=\"gpt-4o\",\n",
    "                 max_tokens=512)\n",
    "\n",
    "# Retrieval qa chain\n",
    "qa = RetrievalQA.from_chain_type(\n",
    "                                    llm=llm,\n",
    "                                    chain_type=\"stuff\",\n",
    "                                    callbacks=[handler],\n",
    "                                    retriever=retriever)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "969cfd06-87a6-437d-8d4a-295526a418c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "#=================\n",
    "# Customized Agentic Tool\n",
    "#=================\n",
    "\n",
    "class MyCustomTool(BaseTool):\n",
    "    name: str = \"Research Paper\"\n",
    "    description: str = \"Research Paper\"\n",
    "\n",
    "    def _run(self, argument: str) -> str:\n",
    "        return qa.run(\"Please read, understand and analyze the provided documents\")\n",
    "\n",
    "research_tool = MyCustomTool()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "1d396a95-4660-4153-851e-ba7cd7e2a62a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m\u001b[95m [DEBUG]: == Working Agent: Researcher\u001b[00m\n",
      "\u001b[1m\u001b[95m [INFO]: == Starting Task: Conduct an in-depth analysis of the provided documents.\n",
      "                    Identify the key components and try to understand how each works.\u001b[00m\n",
      "\n",
      "\n",
      "\u001b[1m> Entering new CrewAgentExecutor chain...\u001b[0m\n",
      "\u001b[32;1m\u001b[1;3mI need to begin by using the research tool to analyze the provided documents. \n",
      "\n",
      "Action: \n",
      "Research Paper\n",
      "\n",
      "Action Input: \n",
      "{\n",
      "  \"argument\": \"Provided Documents\"\n",
      "}\n",
      "\u001b[0m"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Mustafa\\Anaconda_3\\Lib\\site-packages\\langchain_core\\_api\\deprecation.py:139: LangChainDeprecationWarning: The method `Chain.run` was deprecated in langchain 0.1.0 and will be removed in 0.3.0. Use invoke instead.\n",
      "  warn_deprecated(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new RetrievalQA chain...\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n",
      "\u001b[95m \n",
      "\n",
      "The provided context emphasizes the importance of government involvement and collaboration among various stakeholders to ensure the responsible and sustainable integration of new technologies. Here are the key points:\n",
      "\n",
      "1. **Government Involvement**: Governments play a crucial role in balancing promotion and control to ensure that new technologies are integrated responsibly and sustainably. This involves creating policies and regulations that encourage innovation while also mitigating potential risks.\n",
      "\n",
      "2. **Collaboration and Stakeholder Engagement**: Effective integration of technologies requires collaboration between industry stakeholders, researchers, and policymakers. Inclusive decision-making processes are essential to consider the interests of all parties and to address complex challenges such as climate risk and food crises. This collaborative approach ensures that multiple perspectives and expertise are brought together to create comprehensive strategies.\n",
      "\n",
      "3. **Holistic Approach**: The research advocates for a holistic approach that considers advancements and environmental sustainability. This means looking at the broader impact of technologies and ensuring that their development and deployment are aligned with sustainable practices.\n",
      "\n",
      "In summary, the documents highlight the need for a balanced, collaborative, and holistic approach to integrating new technologies, with active government involvement and stakeholder engagement being key components.\n",
      "\u001b[00m\n",
      "\u001b[32;1m\u001b[1;3mThought: \n",
      "I have extracted the main points from the provided documents. Now, I should focus on understanding how each component works and how they interact with each other.\n",
      "\n",
      "Action: \n",
      "Research Paper\n",
      "\n",
      "Action Input: \n",
      "{\n",
      "  \"argument\": \"Government Involvement in Technology Integration\"\n",
      "}\n",
      "\u001b[0m\n",
      "\n",
      "\u001b[1m> Entering new RetrievalQA chain...\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n",
      "\u001b[95m \n",
      "\n",
      "The provided context emphasizes the importance of a balanced approach to the integration of new technologies, highlighting the need for both promotion and control to ensure responsible and sustainable development. It underscores the significance of collaboration and stakeholder engagement, advocating for inclusive decision-making processes that consider the interests of all parties involved. This collaborative approach is deemed essential for addressing complex challenges such as climate risk and food crises, which require diverse perspectives and expertise. Additionally, the context suggests that interactions with various stakeholders, including requisitioners, suppliers, and other partners, are crucial. Finally, it calls for a holistic approach to advancements and environmental sustainability.\n",
      "\n",
      "If you have any specific questions or need further analysis on a particular aspect, please let me know!\n",
      "\u001b[00m\n",
      "\u001b[32;1m\u001b[1;3mThought: \n",
      "I have more insights about Government Involvement in Technology Integration. Now, I should research the next component which is Collaboration and Stakeholder Engagement. \n",
      "\n",
      "Action: \n",
      "Research Paper\n",
      "\n",
      "Action Input: \n",
      "{\n",
      "  \"argument\": \"Collaboration and Stakeholder Engagement in Technology Integration\"\n",
      "}\n",
      "\u001b[0m\n",
      "\n",
      "\u001b[1m> Entering new RetrievalQA chain...\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n",
      "\u001b[95m \n",
      "\n",
      "The provided context emphasizes the importance of government involvement and collaboration among various stakeholders to ensure the responsible and sustainable integration of new technologies. Here are the key points:\n",
      "\n",
      "1. **Government Involvement**: Governments play a crucial role in balancing promotion and control to ensure that new technologies are integrated responsibly and sustainably. This involves creating policies and regulations that encourage innovation while also mitigating potential risks.\n",
      "\n",
      "2. **Collaboration and Stakeholder Engagement**: Effective integration of technologies requires collaboration between industry stakeholders, researchers, and policymakers. Inclusive decision-making processes are essential to consider the interests of all parties and to address complex challenges such as climate risk and food crises. This collaborative approach ensures that multiple perspectives and expertise are brought together to create comprehensive strategies.\n",
      "\n",
      "3. **Holistic Approach**: The research advocates for a holistic approach that considers advancements and environmental sustainability. This means looking at the broader impact of technologies and ensuring that their development and deployment are aligned with sustainable practices.\n",
      "\n",
      "In summary, the documents highlight the need for a balanced, collaborative, and holistic approach to integrating new technologies, with active government involvement and engagement from all relevant stakeholders.\n",
      "\u001b[00m\n",
      "\u001b[32;1m\u001b[1;3mThought: \n",
      "Now that I have insights about Collaboration and Stakeholder Engagement, I should research the last component which is the Holistic Approach in technology integration. \n",
      "\n",
      "Action: \n",
      "Research Paper\n",
      "\n",
      "Action Input: \n",
      "{\n",
      "  \"argument\": \"Holistic Approach in Technology Integration\"\n",
      "}\n",
      "\u001b[0m\n",
      "\n",
      "\u001b[1m> Entering new RetrievalQA chain...\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n",
      "\u001b[95m \n",
      "\n",
      "The provided context emphasizes the importance of a balanced approach to the integration of new technologies, highlighting the need for both promotion and control to ensure responsible and sustainable development. It underscores the necessity of collaboration and stakeholder engagement, advocating for inclusive decision-making processes that consider the interests of all parties involved. This collaborative approach is deemed essential for addressing complex challenges such as climate risk and food crises, which require diverse perspectives and expertise. Additionally, the context suggests that interactions with various stakeholders, including requisitioners, suppliers, and other partners, are crucial. Finally, it calls for a holistic approach to advancements and environmental sustainability.\n",
      "\n",
      "If you have any specific questions or need further analysis on a particular aspect, please let me know!\n",
      "\u001b[00m\n",
      "\u001b[32;1m\u001b[1;3mThought: \n",
      "I have gathered insights about all the key components from the documents. Now, I will put together the final analysis report using the information gathered.\n",
      "\n",
      "Final Answer: \n",
      "The analysis of the provided documents emphasizes the importance of government involvement, collaboration among various stakeholders, and a holistic approach in the responsible and sustainable integration of new technologies.\n",
      "\n",
      "1. **Government Involvement**: Governments play a crucial role in balancing the promotion and control of new technologies. Their responsibilities include creating policies and regulations that encourage innovation while also mitigating potential risks. Government involvement ensures that technology integration occurs responsibly and sustainably.\n",
      "\n",
      "2. **Collaboration and Stakeholder Engagement**: The integration of technologies requires effective collaboration between industry stakeholders, researchers, and policymakers. Inclusive decision-making processes, considering the interests of all parties, are fundamental for addressing complex challenges such as climate risk and food crises. This collaborative approach ensures that multiple perspectives and expertise are utilized to create comprehensive strategies.\n",
      "\n",
      "3. **Holistic Approach**: The research advocates for a holistic approach that considers advancements and environmental sustainability. It implies considering the broader impacts of technologies and ensuring their development and deployment align with sustainable practices.\n",
      "\n",
      "In conclusion, the integration of new technologies calls for a balanced, collaborative, and holistic approach. Active government involvement and stakeholder engagement are key components to address potential risks and ensure sustainable development.\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n",
      "\u001b[1m\u001b[92m [DEBUG]: == [Researcher] Task output: The analysis of the provided documents emphasizes the importance of government involvement, collaboration among various stakeholders, and a holistic approach in the responsible and sustainable integration of new technologies.\n",
      "\n",
      "1. **Government Involvement**: Governments play a crucial role in balancing the promotion and control of new technologies. Their responsibilities include creating policies and regulations that encourage innovation while also mitigating potential risks. Government involvement ensures that technology integration occurs responsibly and sustainably.\n",
      "\n",
      "2. **Collaboration and Stakeholder Engagement**: The integration of technologies requires effective collaboration between industry stakeholders, researchers, and policymakers. Inclusive decision-making processes, considering the interests of all parties, are fundamental for addressing complex challenges such as climate risk and food crises. This collaborative approach ensures that multiple perspectives and expertise are utilized to create comprehensive strategies.\n",
      "\n",
      "3. **Holistic Approach**: The research advocates for a holistic approach that considers advancements and environmental sustainability. It implies considering the broader impacts of technologies and ensuring their development and deployment align with sustainable practices.\n",
      "\n",
      "In conclusion, the integration of new technologies calls for a balanced, collaborative, and holistic approach. Active government involvement and stakeholder engagement are key components to address potential risks and ensure sustainable development.\n",
      "\n",
      "\u001b[00m\n",
      "\u001b[1m\u001b[95m [DEBUG]: == Working Agent: Writer\u001b[00m\n",
      "\u001b[1m\u001b[95m [INFO]: == Starting Task: 1. Use the researcher's material to craft a compelling summary2. Sections/Subtitles are properly named 3. Limit the document to only 500 words \u001b[00m\n",
      "\n",
      "\n",
      "\u001b[1m> Entering new CrewAgentExecutor chain...\u001b[0m\n",
      "\u001b[32;1m\u001b[1;3mI now can give a great answer.\n",
      "\n",
      "Final Answer: \n",
      "The provided documents underscore the critical role of government participation, multi-stakeholder collaboration, and a comprehensive approach in the sustainable and responsible incorporation of emerging technologies.\n",
      "\n",
      "Under the section **Government Involvement**, it is highlighted that governments are essential in striking a balance between encouraging and regulating new technologies. They are tasked with formulating policies and rules that stimulate innovation while also controlling potential dangers. This government intervention guarantees that technological integration takes place in a responsible and sustainable manner.\n",
      "\n",
      "The second section, **Collaboration and Stakeholder Engagement**, stresses the necessity of efficient cooperation among industry participants, scientists, and policymakers for technology integration. Decision-making processes that are inclusive and consider all stakeholders' interests are vital for tackling intricate issues such as climate risk and food shortages. This cooperative methodology ensures that a variety of viewpoints and skills are employed to devise comprehensive plans.\n",
      "\n",
      "The third section, **Holistic Approach**, advocates for a comprehensive perspective that considers both technological advancements and environmental sustainability. This approach involves assessing the broader consequences of technologies and ensuring that their growth and deployment are in line with sustainable practices.\n",
      "\n",
      "In conclusion, as highlighted in the documents, integrating novel technologies necessitates a balanced, collaborative, and holistic approach. Active government participation and stakeholder engagement are crucial elements in addressing potential hazards and promoting sustainable growth.\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n",
      "\u001b[1m\u001b[92m [DEBUG]: == [Writer] Task output: The provided documents underscore the critical role of government participation, multi-stakeholder collaboration, and a comprehensive approach in the sustainable and responsible incorporation of emerging technologies.\n",
      "\n",
      "Under the section **Government Involvement**, it is highlighted that governments are essential in striking a balance between encouraging and regulating new technologies. They are tasked with formulating policies and rules that stimulate innovation while also controlling potential dangers. This government intervention guarantees that technological integration takes place in a responsible and sustainable manner.\n",
      "\n",
      "The second section, **Collaboration and Stakeholder Engagement**, stresses the necessity of efficient cooperation among industry participants, scientists, and policymakers for technology integration. Decision-making processes that are inclusive and consider all stakeholders' interests are vital for tackling intricate issues such as climate risk and food shortages. This cooperative methodology ensures that a variety of viewpoints and skills are employed to devise comprehensive plans.\n",
      "\n",
      "The third section, **Holistic Approach**, advocates for a comprehensive perspective that considers both technological advancements and environmental sustainability. This approach involves assessing the broader consequences of technologies and ensuring that their growth and deployment are in line with sustainable practices.\n",
      "\n",
      "In conclusion, as highlighted in the documents, integrating novel technologies necessitates a balanced, collaborative, and holistic approach. Active government participation and stakeholder engagement are crucial elements in addressing potential hazards and promoting sustainable growth.\n",
      "\n",
      "\u001b[00m\n"
     ]
    }
   ],
   "source": [
    "\n",
    "#=================\n",
    "# Crew Agents\n",
    "#=================\n",
    "\n",
    "researcher = Agent(\n",
    "    role=\"Researcher\",\n",
    "    goal=\"Research and the documents\",\n",
    "    backstory=\"extract the main information and provide the main insights to the writer \"\n",
    "              \"Don't provide any information outside the provided documents\"\n",
    "    ,\n",
    "    verbose=True,\n",
    "    tools=[research_tool],\n",
    "    allow_delegation=False\n",
    "    )\n",
    "\n",
    "writer = Agent(\n",
    "    role=\"Writer\",\n",
    "    goal= 'Craft easy to understand summary report',\n",
    "    backstory=\"\"\"You are working on writing a summary of the provided documents.\n",
    "                  Use the content of the documents to develop\n",
    "                  a short comprehensive summary\"\"\",\n",
    "    verbose=True,\n",
    "    allow_delegation=False,\n",
    ")\n",
    "\n",
    "#=================\n",
    "# Crew Tasks\n",
    "#=================\n",
    "\n",
    "# Tasks\n",
    "research = Task(\n",
    "    description=\"\"\"Conduct an in-depth analysis of the provided documents.\n",
    "                    Identify the key components and try to understand how each works.\"\"\",\n",
    "    expected_output=\"\"\"our final answer must be a full analysis report of the documents\"\"\",\n",
    "    agent=researcher\n",
    "\n",
    ")\n",
    "\n",
    "\n",
    "\n",
    "write = Task(\n",
    "    description=(\n",
    "        \"1. Use the researcher's material to craft a compelling summary\"\n",
    "\t\t    \"2. Sections/Subtitles are properly named \"\n",
    "         \"3. Limit the document to only 500 words \"\n",
    "    ),\n",
    "    agent=writer,\n",
    "    expected_output='a compelling summary report' \n",
    ")\n",
    "\n",
    "\n",
    "#=================\n",
    "# Execution\n",
    "#=================\n",
    "\n",
    "crew = Crew(\n",
    "    agents=[researcher, writer],\n",
    "    tasks=[research, write],\n",
    "    verbose=2\n",
    ")\n",
    "\n",
    "result = crew.kickoff()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "e36f19f5-b317-4cae-a088-1b7ce878194b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "The provided documents underscore the critical role of government participation, multi-stakeholder collaboration, and a comprehensive approach in the sustainable and responsible incorporation of emerging technologies.\n",
       "\n",
       "Under the section **Government Involvement**, it is highlighted that governments are essential in striking a balance between encouraging and regulating new technologies. They are tasked with formulating policies and rules that stimulate innovation while also controlling potential dangers. This government intervention guarantees that technological integration takes place in a responsible and sustainable manner.\n",
       "\n",
       "The second section, **Collaboration and Stakeholder Engagement**, stresses the necessity of efficient cooperation among industry participants, scientists, and policymakers for technology integration. Decision-making processes that are inclusive and consider all stakeholders' interests are vital for tackling intricate issues such as climate risk and food shortages. This cooperative methodology ensures that a variety of viewpoints and skills are employed to devise comprehensive plans.\n",
       "\n",
       "The third section, **Holistic Approach**, advocates for a comprehensive perspective that considers both technological advancements and environmental sustainability. This approach involves assessing the broader consequences of technologies and ensuring that their growth and deployment are in line with sustainable practices.\n",
       "\n",
       "In conclusion, as highlighted in the documents, integrating novel technologies necessitates a balanced, collaborative, and holistic approach. Active government participation and stakeholder engagement are crucial elements in addressing potential hazards and promoting sustainable growth."
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Markdown\n",
    "Markdown(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "7bf78cc1-e782-4842-bd69-61e1359a0e07",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\ntemplate = \"\"\"Provide a precise and consice response to the below question from the provided context and don\\'t use your own knowledge. \\nIf the context does not provide the necessary information to answer the question, then respond politely that you need more information.\\nQuestion: {question}\\n=========\\nContext: {context}\\n=========\\n\"\"\"\\nprompt = PromptTemplate(template=template, input_variables=[ \"question\",\"context\"])\\ndoc_chain = load_qa_with_sources_chain(llm, chain_type=\"map_reduce\")\\nquestion_generator_chain = LLMChain(llm=llm, prompt=prompt)\\nqa_chain = ConversationalRetrievalChain(\\n    retriever=retriever,\\n    question_generator=question_generator_chain,\\n    combine_docs_chain=doc_chain,\\n)\\n\\nquestion = \\'Summarize each document in multiple separate paragraphs and mention the source document title and name at the beginning of each summary paragraph. limit each paragraph to 100 words\\'\\nresponse =qa_chain.invoke({\\n    \"question\": question,\\n    \"chat_history\": []\\n})[\"answer\"]\\n\\n#-------------------------------------------\\n\\nfrom IPython.display import Markdown\\nMarkdown(response)\\n\\n#-------------------------------------------\\n\\n\\ntemplate = \"\"\"Act as an evaluator and evaluate of the LLM Response is relevant to the Question and free of hallucination. \\nAlso make sure that it\\'s structured correctly where the source name is at the start of each paragraph.\\nPresent your evaluation as a score from 1 to 10 where 10 is the highest in terms of relevance and structure.\\nProvide the reasons behind your statement.\\n\\nQuestion: {question}\\n=========\\nResponse: {response}\\n=========\\nContext: {context}\\n=========\\n\"\"\"\\nprompt = PromptTemplate(template=template, input_variables=[\"response\", \"context\",\"question\"])\\ndoc_chain = load_qa_with_sources_chain(llm, chain_type=\"map_reduce\")\\nquestion_generator_chain = LLMChain(llm=llm, prompt=prompt)\\nqa_chain = ConversationalRetrievalChain(\\n    retriever=retriever,\\n    question_generator=question_generator_chain,\\n    combine_docs_chain=doc_chain,\\n)\\nresponse2 =qa_chain.invoke({\\n    \"question\": template,\"response\": response,\"chat_history\": []\\n})[\"answer\"]\\n\\n#------------------------------------------------\\n\\nfrom IPython.display import Markdown\\nMarkdown(response2)\\n'"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    "template = \"\"\"Provide a precise and consice response to the below question from the provided context and don't use your own knowledge. \n",
    "If the context does not provide the necessary information to answer the question, then respond politely that you need more information.\n",
    "Question: {question}\n",
    "=========\n",
    "Context: {context}\n",
    "=========\n",
    "\"\"\"\n",
    "prompt = PromptTemplate(template=template, input_variables=[ \"question\",\"context\"])\n",
    "doc_chain = load_qa_with_sources_chain(llm, chain_type=\"map_reduce\")\n",
    "question_generator_chain = LLMChain(llm=llm, prompt=prompt)\n",
    "qa_chain = ConversationalRetrievalChain(\n",
    "    retriever=retriever,\n",
    "    question_generator=question_generator_chain,\n",
    "    combine_docs_chain=doc_chain,\n",
    ")\n",
    "\n",
    "question = 'Summarize each document in multiple separate paragraphs and mention the source document title and name at the beginning of each summary paragraph. limit each paragraph to 100 words'\n",
    "response =qa_chain.invoke({\n",
    "    \"question\": question,\n",
    "    \"chat_history\": []\n",
    "})[\"answer\"]\n",
    "\n",
    "#-------------------------------------------\n",
    "\n",
    "from IPython.display import Markdown\n",
    "Markdown(response)\n",
    "\n",
    "#-------------------------------------------\n",
    "\n",
    "\n",
    "template = \"\"\"Act as an evaluator and evaluate of the LLM Response is relevant to the Question and free of hallucination. \n",
    "Also make sure that it's structured correctly where the source name is at the start of each paragraph.\n",
    "Present your evaluation as a score from 1 to 10 where 10 is the highest in terms of relevance and structure.\n",
    "Provide the reasons behind your statement.\n",
    "\n",
    "Question: {question}\n",
    "=========\n",
    "Response: {response}\n",
    "=========\n",
    "Context: {context}\n",
    "=========\n",
    "\"\"\"\n",
    "prompt = PromptTemplate(template=template, input_variables=[\"response\", \"context\",\"question\"])\n",
    "doc_chain = load_qa_with_sources_chain(llm, chain_type=\"map_reduce\")\n",
    "question_generator_chain = LLMChain(llm=llm, prompt=prompt)\n",
    "qa_chain = ConversationalRetrievalChain(\n",
    "    retriever=retriever,\n",
    "    question_generator=question_generator_chain,\n",
    "    combine_docs_chain=doc_chain,\n",
    ")\n",
    "response2 =qa_chain.invoke({\n",
    "    \"question\": template,\"response\": response,\"chat_history\": []\n",
    "})[\"answer\"]\n",
    "\n",
    "#------------------------------------------------\n",
    "\n",
    "from IPython.display import Markdown\n",
    "Markdown(response2)\n",
    "'''"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}