edps/Panel_PDF_QA2

1{2 "cells": [3  {4   "cell_type": "markdown",5   "id": "04815d1b-44ee-4bd3-878e-fa0c3bf9fa7f",6   "metadata": {7    "tags": []8   },9   "source": [10    "# LangChain QA Panel App\n",11    "\n",12    "This notebook shows how to make this app:"13   ]14  },15  {16   "cell_type": "code",17   "execution_count": null,18   "id": "a181568b-9cde-4a55-a853-4d2a41dbfdad",19   "metadata": {20    "tags": []21   },22   "outputs": [],23   "source": [24    "#!pip install langchain openai chromadb tiktoken pypdf panel\n"25   ]26  },27  {28   "cell_type": "code",29   "execution_count": null,30   "id": "9a464409-d064-4766-a9cb-5119f6c4b8f5",31   "metadata": {32    "tags": []33   },34   "outputs": [],35   "source": [36    "import os \n",37    "from langchain.chains import RetrievalQA\n",38    "from langchain.llms import OpenAI\n",39    "from langchain.document_loaders import TextLoader\n",40    "from langchain.document_loaders import PyPDFLoader\n",41    "from langchain.indexes import VectorstoreIndexCreator\n",42    "from langchain.text_splitter import CharacterTextSplitter\n",43    "from langchain.embeddings import OpenAIEmbeddings\n",44    "from langchain.vectorstores import Chroma\n",45    "import panel as pn\n",46    "import tempfile\n"47   ]48  },49  {50   "cell_type": "code",51   "execution_count": null,52   "id": "b2d07ea5-9ff2-4c96-a8dc-92895d870b73",53   "metadata": {54    "tags": []55   },56   "outputs": [],57   "source": [58    "pn.extension('texteditor', template=\"bootstrap\", sizing_mode='stretch_width')\n",59    "pn.state.template.param.update(\n",60    "    main_max_width=\"690px\",\n",61    "    header_background=\"#F08080\",\n",62    ")"63   ]64  },65  {66   "cell_type": "code",67   "execution_count": null,68   "id": "763db4d0-3436-41d3-8b0f-e66ce16468cd",69   "metadata": {70    "tags": []71   },72   "outputs": [],73   "source": [74    "file_input = pn.widgets.FileInput(width=300)\n",75    "\n",76    "openaikey = pn.widgets.PasswordInput(\n",77    "    value=\"\", placeholder=\"Enter your OpenAI API Key here...\", width=300\n",78    ")\n",79    "prompt = pn.widgets.TextEditor(\n",80    "    value=\"\", placeholder=\"Enter your questions here...\", height=160, toolbar=False\n",81    ")\n",82    "run_button = pn.widgets.Button(name=\"Run!\")\n",83    "\n",84    "select_k = pn.widgets.IntSlider(\n",85    "    name=\"Number of relevant chunks\", start=1, end=5, step=1, value=2\n",86    ")\n",87    "select_chain_type = pn.widgets.RadioButtonGroup(\n",88    "    name='Chain type', \n",89    "    options=['stuff', 'map_reduce', \"refine\", \"map_rerank\"]\n",90    ")\n",91    "\n",92    "widgets = pn.Row(\n",93    "    pn.Column(prompt, run_button, margin=5),\n",94    "    pn.Card(\n",95    "        \"Chain type:\",\n",96    "        pn.Column(select_chain_type, select_k),\n",97    "        title=\"Advanced settings\", margin=10\n",98    "    ), width=600\n",99    ")"100   ]101  },102  {103   "cell_type": "code",104   "execution_count": null,105   "id": "9b83cc06-3401-498f-8f84-8a98370f3121",106   "metadata": {107    "tags": []108   },109   "outputs": [],110   "source": [111    "def qa(file, query, chain_type, k):\n",112    "    # load document\n",113    "    loader = PyPDFLoader(file)\n",114    "    documents = loader.load()\n",115    "    # split the documents into chunks\n",116    "    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",117    "    texts = text_splitter.split_documents(documents)\n",118    "    # select which embeddings we want to use\n",119    "    embeddings = OpenAIEmbeddings()\n",120    "    # create the vectorestore to use as the index\n",121    "    db = Chroma.from_documents(texts, embeddings)\n",122    "    # expose this index in a retriever interface\n",123    "    retriever = db.as_retriever(search_type=\"similarity\", search_kwargs={\"k\": k})\n",124    "    # create a chain to answer questions \n",125    "    qa = RetrievalQA.from_chain_type(\n",126    "        llm=OpenAI(), chain_type=chain_type, retriever=retriever, return_source_documents=True)\n",127    "    result = qa({\"query\": query})\n",128    "    print(result['result'])\n",129    "    return result"130   ]131  },132  {133   "cell_type": "code",134   "execution_count": null,135   "id": "2722f43b-daf6-4d17-a842-41203ae9b140",136   "metadata": {137    "tags": []138   },139   "outputs": [],140   "source": [141    "# result = qa(\"example.pdf\", \"what is the total number of AI publications?\")"142   ]143  },144  {145   "cell_type": "code",146   "execution_count": null,147   "id": "60e1b3d3-c0d2-4260-ae0c-26b03f1b8824",148   "metadata": {},149   "outputs": [],150   "source": [151    "convos = []  # store all panel objects in a list\n",152    "\n",153    "def qa_result(_):\n",154    "    os.environ[\"OPENAI_API_KEY\"] = openaikey.value\n",155    "    \n",156    "    # save pdf file to a temp file \n",157    "    if file_input.value is not None:\n",158    "        file_input.save(\"/.cache/temp.pdf\")\n",159    "    \n",160    "        prompt_text = prompt.value\n",161    "        if prompt_text:\n",162    "            result = qa(file=\"/.cache/temp.pdf\", query=prompt_text, chain_type=select_chain_type.value, k=select_k.value)\n",163    "            convos.extend([\n",164    "                pn.Row(\n",165    "                    pn.panel(\"\\U0001F60A\", width=10),\n",166    "                    prompt_text,\n",167    "                    width=600\n",168    "                ),\n",169    "                pn.Row(\n",170    "                    pn.panel(\"\\U0001F916\", width=10),\n",171    "                    pn.Column(\n",172    "                        result[\"result\"],\n",173    "                        \"Relevant source text:\",\n",174    "                        pn.pane.Markdown('\\n--------------------------------------------------------------------\\n'.join(doc.page_content for doc in result[\"source_documents\"]))\n",175    "                    )\n",176    "                )\n",177    "            ])\n",178    "            #return convos\n",179    "    return pn.Column(*convos, margin=15, width=575, min_height=400)\n"180   ]181  },182  {183   "cell_type": "code",184   "execution_count": null,185   "id": "c3a70857-0b98-4f62-a9c0-b62ca42b474c",186   "metadata": {187    "tags": []188   },189   "outputs": [],190   "source": [191    "qa_interactive = pn.panel(\n",192    "    pn.bind(qa_result, run_button),\n",193    "    loading_indicator=True,\n",194    ")"195   ]196  },197  {198   "cell_type": "code",199   "execution_count": null,200   "id": "228e2b42-b1ed-43af-b923-031a70241ab0",201   "metadata": {202    "tags": []203   },204   "outputs": [],205   "source": [206    "output = pn.WidgetBox('*Output will show up here:*', qa_interactive, width=630, scroll=True)"207   ]208  },209  {210   "cell_type": "code",211   "execution_count": null,212   "id": "1b0ec253-2bcd-4f91-96d8-d8456e900a58",213   "metadata": {214    "tags": []215   },216   "outputs": [],217   "source": [218    "# layout\n",219    "pn.Column(\n",220    "    pn.pane.Markdown(\"\"\"\n",221    "    ## \\U0001F60A! Question Answering with your PDF file\n",222    "    \n",223    "    1) Upload a PDF. 2) Enter OpenAI API key. This costs $. Set up billing at [OpenAI](https://platform.openai.com/account). 3) Type a question and click \"Run\".\n",224    "    \n",225    "    \"\"\"),\n",226    "    pn.Row(file_input,openaikey),\n",227    "    output,\n",228    "    widgets\n",229    "\n",230    ").servable()"231   ]232  }233 ],234 "metadata": {235  "kernelspec": {236   "display_name": "Python 3 (ipykernel)",237   "language": "python",238   "name": "python3"239  },240  "language_info": {241   "codemirror_mode": {242    "name": "ipython",243    "version": 3244   },245   "file_extension": ".py",246   "mimetype": "text/x-python",247   "name": "python",248   "nbconvert_exporter": "python",249   "pygments_lexer": "ipython3",250   "version": "3.10.10"251  }252 },253 "nbformat": 4,254 "nbformat_minor": 5255}256