File size: 34,790 Bytes
20eaf62 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 |
{
"cells": [
{
"cell_type": "code",
"execution_count": 21,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import getpass\n",
"\n",
"os.environ[\"OPENAI_API_KEY\"] = getpass.getpass(\"OpenAI API Key:\")\n",
"os.environ[\"TAVILY_API_KEY\"] = getpass.getpass(\"TAVILY_API_KEY\")"
]
},
{
"cell_type": "code",
"execution_count": 23,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Requirement already satisfied: pymupdf in /opt/anaconda3/lib/python3.12/site-packages (1.25.3)\n"
]
}
],
"source": [
"!pip install pymupdf"
]
},
{
"cell_type": "code",
"execution_count": 25,
"metadata": {},
"outputs": [],
"source": [
"# Basic RAG Chain\n",
"from backend.app.vectorstore import get_vector_db\n",
"\n",
"qdrant_retriever = get_vector_db().as_retriever()\n",
"\n",
"from langchain_core.prompts import ChatPromptTemplate\n",
"\n",
"RAG_PROMPT = \"\"\"\n",
"CONTEXT:\n",
"{context}\n",
"\n",
"QUERY:\n",
"{question}\n",
"\n",
"You are a helpful assistant. Use the available context to answer the question. If you can't answer the question, say you don't know.\n",
"\"\"\"\n",
"\n",
"rag_prompt = ChatPromptTemplate.from_template(RAG_PROMPT)\n",
"\n",
"from langchain_openai import ChatOpenAI\n",
"\n",
"openai_chat_model = ChatOpenAI(model=\"gpt-4o-mini\")\n",
"\n",
"from operator import itemgetter\n",
"from langchain.schema.output_parser import StrOutputParser\n",
"\n",
"rag_chain = (\n",
" {\"context\": itemgetter(\"question\") | qdrant_retriever, \"question\": itemgetter(\"question\")}\n",
" | rag_prompt | openai_chat_model | StrOutputParser()\n",
")\n"
]
},
{
"cell_type": "code",
"execution_count": 26,
"metadata": {},
"outputs": [],
"source": [
"# Helper functions\n",
"from typing import Any, Callable, List, Optional, TypedDict, Union\n",
"\n",
"from langchain.agents import AgentExecutor, create_openai_functions_agent\n",
"from langchain.output_parsers.openai_functions import JsonOutputFunctionsParser\n",
"from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder\n",
"from langchain_core.messages import AIMessage, BaseMessage, HumanMessage\n",
"from langchain_core.runnables import Runnable\n",
"from langchain_core.tools import BaseTool\n",
"from langchain_openai import ChatOpenAI\n",
"\n",
"from langgraph.graph import END, StateGraph\n",
"\n",
"def agent_node(state, agent, name):\n",
" result = agent.invoke(state)\n",
" return {\"messages\": [HumanMessage(content=result[\"output\"], name=name)]}\n",
"\n",
"def create_agent(\n",
" llm: ChatOpenAI,\n",
" tools: list,\n",
" system_prompt: str,\n",
") -> str:\n",
" \"\"\"Create a function-calling agent and add it to the graph.\"\"\"\n",
" system_prompt += (\"\\nWork autonomously according to your specialty, using the tools available to you.\"\n",
" \" Do not ask for clarification.\"\n",
" \" Your other team members (and other teams) will collaborate with you with their own specialties.\"\n",
" \" You are chosen for a reason! You are one of the following team members: {{team_members}}.\")\n",
" prompt = ChatPromptTemplate.from_messages(\n",
" [\n",
" (\n",
" \"system\",\n",
" system_prompt,\n",
" ),\n",
" MessagesPlaceholder(variable_name=\"messages\"),\n",
" MessagesPlaceholder(variable_name=\"agent_scratchpad\"),\n",
" ]\n",
" )\n",
" agent = create_openai_functions_agent(llm, tools, prompt)\n",
" executor = AgentExecutor(agent=agent, tools=tools)\n",
" return executor\n",
"\n",
"def create_team_supervisor(llm: ChatOpenAI, system_prompt, members) -> str:\n",
" \"\"\"An LLM-based router.\"\"\"\n",
" options = [\"FINISH\"] + members\n",
" function_def = {\n",
" \"name\": \"route\",\n",
" \"description\": \"Select the next role.\",\n",
" \"parameters\": {\n",
" \"title\": \"routeSchema\",\n",
" \"type\": \"object\",\n",
" \"properties\": {\n",
" \"next\": {\n",
" \"title\": \"Next\",\n",
" \"anyOf\": [\n",
" {\"enum\": options},\n",
" ],\n",
" },\n",
" },\n",
" \"required\": [\"next\"],\n",
" },\n",
" }\n",
" prompt = ChatPromptTemplate.from_messages(\n",
" [\n",
" (\"system\", system_prompt),\n",
" MessagesPlaceholder(variable_name=\"messages\"),\n",
" (\n",
" \"system\",\n",
" \"Given the conversation above, who should act next?\"\n",
" \" Or should we FINISH? Select one of: {options}\",\n",
" ),\n",
" ]\n",
" ).partial(options=str(options), team_members=\", \".join(members))\n",
" return (\n",
" prompt\n",
" | llm.bind_functions(functions=[function_def], function_call=\"route\")\n",
" | JsonOutputFunctionsParser()\n",
" )\n"
]
},
{
"cell_type": "code",
"execution_count": 32,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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",
"text/plain": [
"<IPython.core.display.Image object>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# Research team\n",
"from langchain_community.tools.tavily_search import TavilySearchResults\n",
"\n",
"tavily_tool = TavilySearchResults(max_results=5)\n",
"\n",
"from typing import Annotated, List, Tuple, Union\n",
"from langchain_core.tools import tool\n",
"\n",
"@tool\n",
"def retrieve_information(\n",
" query: Annotated[str, \"query to ask the retrieve information tool\"]\n",
" ):\n",
" \"\"\"Use Retrieval Augmented Generation to retrieve information about the 'Extending Llama-3’s Context Ten-Fold Overnight' paper.\"\"\"\n",
" return rag_chain.invoke({\"question\" : query})\n",
"\n",
"\n",
"import functools\n",
"import operator\n",
"\n",
"from langchain_core.messages import AIMessage, BaseMessage, HumanMessage\n",
"from langchain_openai.chat_models import ChatOpenAI\n",
"import functools\n",
"\n",
"class ResearchTeamState(TypedDict):\n",
" messages: Annotated[List[BaseMessage], operator.add]\n",
" team_members: List[str]\n",
" next: str\n",
"\n",
"llm = ChatOpenAI(model=\"gpt-4-turbo\")\n",
"\n",
"search_agent = create_agent(\n",
" llm,\n",
" [tavily_tool],\n",
" \"You are a research assistant who can search for up-to-date info using the tavily search engine.\",\n",
")\n",
"search_node = functools.partial(agent_node, agent=search_agent, name=\"Search\")\n",
"\n",
"research_agent = create_agent(\n",
" llm,\n",
" [retrieve_information],\n",
" \"You are a resarch assistant who can provide information about how to use the library langchain to build an RAG system.\",\n",
" \"You are a research assistant who can provide specific information on the provided paper: 'Extending Llama-3’s Context Ten-Fold Overnight'. You must only respond with information about the paper related to the request.\",\n",
")\n",
"research_node = functools.partial(agent_node, agent=research_agent, name=\"PaperInformationRetriever\")\n",
"\n",
"supervisor_agent = create_team_supervisor(\n",
" llm,\n",
" (\"You are a supervisor tasked with managing a conversation between the\"\n",
" \" following workers: Search, PaperInformationRetriever. Given the following user request,\"\n",
" \" determine the subject to be researched and respond with the worker to act next. Each worker will perform a\"\n",
" \" task and respond with their results and status. \"\n",
" \" You should never ask your team to do anything beyond research. They are not required to write content or posts.\"\n",
" \" You should only pass tasks to workers that are specifically research focused.\"\n",
" \" When finished, respond with FINISH.\"),\n",
" [\"Search\", \"PaperInformationRetriever\"],\n",
")\n",
"\n",
"research_graph = StateGraph(ResearchTeamState)\n",
"\n",
"research_graph.add_node(\"Search\", search_node)\n",
"research_graph.add_node(\"PaperInformationRetriever\", research_node)\n",
"research_graph.add_node(\"supervisor\", supervisor_agent)\n",
"\n",
"research_graph.add_edge(\"Search\", \"supervisor\")\n",
"research_graph.add_edge(\"PaperInformationRetriever\", \"supervisor\")\n",
"research_graph.add_conditional_edges(\n",
" \"supervisor\",\n",
" lambda x: x[\"next\"],\n",
" {\"Search\": \"Search\", \"PaperInformationRetriever\": \"PaperInformationRetriever\", \"FINISH\": END},\n",
")\n",
"\n",
"research_graph.set_entry_point(\"supervisor\")\n",
"chain = research_graph.compile()\n",
"\n",
"def enter_chain(message: str):\n",
" results = {\n",
" \"messages\": [HumanMessage(content=message)],\n",
" }\n",
" return results\n",
"\n",
"research_chain = enter_chain | chain\n",
"\n",
"from IPython.display import Image, display\n",
"\n",
"display(Image(chain.get_graph(xray=True).draw_mermaid_png()))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"language_info": {
"name": "python"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
|