Nehaa/Sematic_Scribe
0
1{2 "cells": [3 {4 "attachments": {},5 "cell_type": "markdown",6 "id": "04815d1b-44ee-4bd3-878e-fa0c3bf9fa7f",7 "metadata": {8 "tags": []9 },10 "source": [11 "# LangChain QA Panel App\n",12 "\n",13 "This notebook shows how to make this app:"14 ]15 },16 {17 "cell_type": "code",18 "execution_count": null,19 "id": "a181568b-9cde-4a55-a853-4d2a41dbfdad",20 "metadata": {21 "tags": []22 },23 "outputs": [],24 "source": [25 "#!pip install langchain openai chromadb tiktoken pypdf panel\n"26 ]27 },28 {29 "cell_type": "code",30 "execution_count": null,31 "id": "9a464409-d064-4766-a9cb-5119f6c4b8f5",32 "metadata": {33 "tags": []34 },35 "outputs": [],36 "source": [37 "import os \n",38 "from langchain.chains import RetrievalQA\n",39 "from langchain_community.llms import OpenAI\n",40 "from langchain_community.document_loaders import TextLoader\n",41 "from langchain_community.document_loaders import PyPDFLoader\n",42 "from langchain.indexes import VectorstoreIndexCreator\n",43 "from langchain.text_splitter import CharacterTextSplitter\n",44 "from langchain_community.embeddings import OpenAIEmbeddings\n",45 "from langchain_community.vectorstores import Chroma\n",46 "import panel as pn\n",47 "import tempfile\n"48 ]49 },50 {51 "cell_type": "code",52 "execution_count": null,53 "id": "b2d07ea5-9ff2-4c96-a8dc-92895d870b73",54 "metadata": {55 "tags": []56 },57 "outputs": [],58 "source": [59 "pn.extension('texteditor', template=\"bootstrap\", sizing_mode='stretch_width')\n",60 "pn.state.template.param.update(\n",61 " main_max_width=\"690px\",\n",62 " header_background=\"#F08080\",\n",63 ")"64 ]65 },66 {67 "cell_type": "code",68 "execution_count": null,69 "id": "763db4d0-3436-41d3-8b0f-e66ce16468cd",70 "metadata": {71 "tags": []72 },73 "outputs": [],74 "source": [75 "file_input = pn.widgets.FileInput(width=300)\n",76 "\n",77 "openaikey = pn.widgets.PasswordInput(\n",78 " value=\"\", placeholder=\"Enter your OpenAI API Key here...\", width=300\n",79 ")\n",80 "prompt = pn.widgets.TextEditor(\n",81 " value=\"\", placeholder=\"Enter your questions here...\", height=160, toolbar=False\n",82 ")\n",83 "run_button = pn.widgets.Button(name=\"Run!\")\n",84 "\n",85 "select_k = pn.widgets.IntSlider(\n",86 " name=\"Number of relevant chunks\", start=1, end=5, step=1, value=2\n",87 ")\n",88 "select_chain_type = pn.widgets.RadioButtonGroup(\n",89 " name='Chain type', \n",90 " options=['stuff', 'map_reduce', \"refine\", \"map_rerank\"]\n",91 ")\n",92 "\n",93 "widgets = pn.Row(\n",94 " pn.Column(prompt, run_button, margin=5),\n",95 " pn.Card(\n",96 " \"Chain type:\",\n",97 " pn.Column(select_chain_type, select_k),\n",98 " title=\"Advanced settings\", margin=10\n",99 " ), width=600\n",100 ")"101 ]102 },103 {104 "cell_type": "code",105 "execution_count": null,106 "id": "9b83cc06-3401-498f-8f84-8a98370f3121",107 "metadata": {108 "tags": []109 },110 "outputs": [],111 "source": [112 "def qa(file, query, chain_type, k):\n",113 " # load document\n",114 " loader = PyPDFLoader(file)\n",115 " documents = loader.load()\n",116 " # split the documents into chunks\n",117 " text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",118 " texts = text_splitter.split_documents(documents)\n",119 " # select which embeddings we want to use\n",120 " embeddings = OpenAIEmbeddings()\n",121 " # create the vectorestore to use as the index\n",122 " db = Chroma.from_documents(texts, embeddings)\n",123 " # expose this index in a retriever interface\n",124 " retriever = db.as_retriever(search_type=\"similarity\", search_kwargs={\"k\": k})\n",125 " # create a chain to answer questions \n",126 " qa = RetrievalQA.from_chain_type(\n",127 " llm=OpenAI(), chain_type=chain_type, retriever=retriever, return_source_documents=True)\n",128 " result = qa({\"query\": query})\n",129 " print(result['result'])\n",130 " return result"131 ]132 },133 {134 "cell_type": "code",135 "execution_count": null,136 "id": "58ac9945",137 "metadata": {},138 "outputs": [],139 "source": [140 "#os.environ[\"OPENAI_API_KEY\"]=\"\""141 ]142 },143 {144 "cell_type": "code",145 "execution_count": null,146 "id": "2722f43b-daf6-4d17-a842-41203ae9b140",147 "metadata": {148 "tags": []149 },150 "outputs": [],151 "source": [152 "# result = qa(\"example.pdf\", \"what is the total number of AI publications?\")"153 ]154 },155 {156 "cell_type": "code",157 "execution_count": null,158 "id": "60e1b3d3-c0d2-4260-ae0c-26b03f1b8824",159 "metadata": {},160 "outputs": [],161 "source": [162 "convos = [] # store all panel objects in a list\n",163 "\n",164 "def qa_result(_):\n",165 " os.environ[\"OPENAI_API_KEY\"] = openaikey.value\n",166 " \n",167 " # save pdf file to a temp file \n",168 " if file_input.value is not None:\n",169 " file_input.save(\"/.cache/temp.pdf\")\n",170 " \n",171 " prompt_text = prompt.value\n",172 " if prompt_text:\n",173 " result = qa(file=\"/.cache/temp.pdf\", query=prompt_text, chain_type=select_chain_type.value, k=select_k.value)\n",174 " convos.extend([\n",175 " pn.Row(\n",176 " pn.panel(\"\\U0001F60A\", width=10),\n",177 " prompt_text,\n",178 " width=600\n",179 " ),\n",180 " pn.Row(\n",181 " pn.panel(\"\\U0001F916\", width=10),\n",182 " pn.Column(\n",183 " result[\"result\"],\n",184 " \"Relevant source text:\",\n",185 " pn.pane.Markdown('\\n--------------------------------------------------------------------\\n'.join(doc.page_content for doc in result[\"source_documents\"]))\n",186 " )\n",187 " )\n",188 " ])\n",189 " #return convos\n",190 " return pn.Column(*convos, margin=15, width=575, min_height=400)\n"191 ]192 },193 {194 "cell_type": "code",195 "execution_count": null,196 "id": "c3a70857-0b98-4f62-a9c0-b62ca42b474c",197 "metadata": {198 "tags": []199 },200 "outputs": [],201 "source": [202 "qa_interactive = pn.panel(\n",203 " pn.bind(qa_result, run_button),\n",204 " loading_indicator=True,\n",205 ")"206 ]207 },208 {209 "cell_type": "code",210 "execution_count": null,211 "id": "228e2b42-b1ed-43af-b923-031a70241ab0",212 "metadata": {213 "tags": []214 },215 "outputs": [],216 "source": [217 "output = pn.WidgetBox('*Output will show up here:*', qa_interactive, width=630, scroll=True)"218 ]219 },220 {221 "cell_type": "code",222 "execution_count": null,223 "id": "1b0ec253-2bcd-4f91-96d8-d8456e900a58",224 "metadata": {225 "tags": []226 },227 "outputs": [],228 "source": [229 "# layout\n",230 "pn.Column(\n",231 " pn.pane.Markdown(\"\"\"\n",232 " ## \\U0001F60A! Question Answering with your PDF file\n",233 " \n",234 " 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",235 " \n",236 " \"\"\"),\n",237 " pn.Row(file_input,openaikey),\n",238 " output,\n",239 " widgets\n",240 "\n",241 ").servable()"242 ]243 }244 ],245 "metadata": {246 "kernelspec": {247 "display_name": "Python 3 (ipykernel)",248 "language": "python",249 "name": "python3"250 },251 "language_info": {252 "codemirror_mode": {253 "name": "ipython",254 "version": 3255 },256 "file_extension": ".py",257 "mimetype": "text/x-python",258 "name": "python",259 "nbconvert_exporter": "python",260 "pygments_lexer": "ipython3",261 "version": "3.10.10"262 }263 },264 "nbformat": 4,265 "nbformat_minor": 5266}267