Delete interim.py
Browse files- interim.py +0 -184
interim.py
DELETED
@@ -1,184 +0,0 @@
|
|
1 |
-
import os
|
2 |
-
import streamlit as st
|
3 |
-
from dotenv import load_dotenv
|
4 |
-
from langchain_openai import ChatOpenAI
|
5 |
-
from langchain.agents import AgentExecutor, create_openai_tools_agent
|
6 |
-
from langchain_core.messages import BaseMessage, HumanMessage
|
7 |
-
from langchain_community.tools.tavily_search import TavilySearchResults
|
8 |
-
from langchain_experimental.tools import PythonREPLTool
|
9 |
-
from langchain_community.document_loaders import DirectoryLoader, TextLoader
|
10 |
-
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
11 |
-
from langchain_community.vectorstores import Chroma
|
12 |
-
from langchain.embeddings import HuggingFaceBgeEmbeddings
|
13 |
-
from langchain_core.output_parsers import StrOutputParser
|
14 |
-
from langchain_core.runnables import RunnablePassthrough
|
15 |
-
from langchain.output_parsers.openai_functions import JsonOutputFunctionsParser
|
16 |
-
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
|
17 |
-
from langgraph.graph import StateGraph, END
|
18 |
-
from langchain_core.documents import Document
|
19 |
-
from typing import Annotated, Sequence, TypedDict
|
20 |
-
import functools
|
21 |
-
import operator
|
22 |
-
from langchain_core.tools import tool
|
23 |
-
from glob import glob
|
24 |
-
|
25 |
-
# Load environment variables
|
26 |
-
|
27 |
-
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
|
28 |
-
TAVILY_API_KEY = os.getenv("TAVILY_API_KEY")
|
29 |
-
|
30 |
-
if not OPENAI_API_KEY or not TAVILY_API_KEY:
|
31 |
-
st.error("Please set OPENAI_API_KEY and TAVILY_API_KEY in your environment variables.")
|
32 |
-
st.stop()
|
33 |
-
|
34 |
-
# Initialize API keys and LLM
|
35 |
-
llm = ChatOpenAI(model="gpt-4-1106-preview", openai_api_key=OPENAI_API_KEY)
|
36 |
-
|
37 |
-
# Utility Functions
|
38 |
-
def create_agent(llm: ChatOpenAI, tools: list, system_prompt: str):
|
39 |
-
prompt = ChatPromptTemplate.from_messages([
|
40 |
-
("system", system_prompt),
|
41 |
-
MessagesPlaceholder(variable_name="messages"),
|
42 |
-
MessagesPlaceholder(variable_name="agent_scratchpad"),
|
43 |
-
])
|
44 |
-
agent = create_openai_tools_agent(llm, tools, prompt)
|
45 |
-
return AgentExecutor(agent=agent, tools=tools)
|
46 |
-
|
47 |
-
def agent_node(state, agent, name):
|
48 |
-
result = agent.invoke(state)
|
49 |
-
return {"messages": [HumanMessage(content=result["output"], name=name)]}
|
50 |
-
|
51 |
-
@tool
|
52 |
-
def RAG(state):
|
53 |
-
"""Use this tool to execute RAG. If the question is related to Japan or Sports, this tool retrieves the results."""
|
54 |
-
st.session_state.outputs.append('-> Calling RAG ->')
|
55 |
-
question = state
|
56 |
-
template = """Answer the question based only on the following context:\n{context}\nQuestion: {question}"""
|
57 |
-
prompt = ChatPromptTemplate.from_template(template)
|
58 |
-
retrieval_chain = (
|
59 |
-
{"context": retriever, "question": RunnablePassthrough()} |
|
60 |
-
prompt |
|
61 |
-
llm |
|
62 |
-
StrOutputParser()
|
63 |
-
)
|
64 |
-
result = retrieval_chain.invoke(question)
|
65 |
-
return result
|
66 |
-
|
67 |
-
# Load Tools
|
68 |
-
tavily_tool = TavilySearchResults(max_results=5, tavily_api_key=TAVILY_API_KEY)
|
69 |
-
python_repl_tool = PythonREPLTool()
|
70 |
-
|
71 |
-
# Streamlit UI
|
72 |
-
st.title("Multi-Agent w Supervisor")
|
73 |
-
|
74 |
-
# Example questions for immediate testing
|
75 |
-
example_questions = [
|
76 |
-
"Code hello world and print it to the terminal",
|
77 |
-
"What is James McIlroy aiming for in sports?",
|
78 |
-
"Fetch India's GDP over the past 5 years and draw a line graph.",
|
79 |
-
"Fetch Japan's GDP over the past 4 years from RAG, then draw a line graph."
|
80 |
-
]
|
81 |
-
|
82 |
-
# File Selection Section
|
83 |
-
source_files = glob("source/*.txt")
|
84 |
-
selected_files = st.multiselect("Select files from the source directory:", source_files, default=source_files[:2])
|
85 |
-
|
86 |
-
uploaded_files = st.file_uploader("Or upload your TXT files:", accept_multiple_files=True, type=['txt'])
|
87 |
-
|
88 |
-
# Combine Files
|
89 |
-
all_docs = []
|
90 |
-
if selected_files:
|
91 |
-
for file_path in selected_files:
|
92 |
-
loader = TextLoader(file_path)
|
93 |
-
all_docs.extend(loader.load())
|
94 |
-
|
95 |
-
if uploaded_files:
|
96 |
-
for uploaded_file in uploaded_files:
|
97 |
-
content = uploaded_file.read().decode("utf-8")
|
98 |
-
all_docs.append(Document(page_content=content, metadata={"name": uploaded_file.name}))
|
99 |
-
|
100 |
-
if not all_docs:
|
101 |
-
st.warning("Please select files from the source directory or upload TXT files.")
|
102 |
-
st.stop()
|
103 |
-
|
104 |
-
# Process Documents
|
105 |
-
text_splitter = RecursiveCharacterTextSplitter(chunk_size=100, chunk_overlap=10, length_function=len)
|
106 |
-
split_docs = text_splitter.split_documents(all_docs)
|
107 |
-
|
108 |
-
embeddings = HuggingFaceBgeEmbeddings(model_name="BAAI/bge-base-en-v1.5", model_kwargs={'device': 'cpu'}, encode_kwargs={'normalize_embeddings': True})
|
109 |
-
db = Chroma.from_documents(split_docs, embeddings)
|
110 |
-
retriever = db.as_retriever(search_kwargs={"k": 4})
|
111 |
-
|
112 |
-
# Create Agents
|
113 |
-
research_agent = create_agent(llm, [tavily_tool], "You are a web researcher.")
|
114 |
-
code_agent = create_agent(llm, [python_repl_tool], "You may generate safe python code to analyze data and generate charts using matplotlib.")
|
115 |
-
RAG_agent = create_agent(llm, [RAG], "Use this tool when questions are related to Japan or Sports category.")
|
116 |
-
|
117 |
-
research_node = functools.partial(agent_node, agent=research_agent, name="Researcher")
|
118 |
-
code_node = functools.partial(agent_node, agent=code_agent, name="Coder")
|
119 |
-
rag_node = functools.partial(agent_node, agent=RAG_agent, name="RAG")
|
120 |
-
|
121 |
-
members = ["RAG", "Researcher", "Coder"]
|
122 |
-
system_prompt = (
|
123 |
-
"You are a supervisor managing these workers: {members}. Respond with the next worker or FINISH. "
|
124 |
-
"Use RAG tool for Japan or Sports questions."
|
125 |
-
)
|
126 |
-
options = ["FINISH"] + members
|
127 |
-
function_def = {
|
128 |
-
"name": "route", "description": "Select the next role.",
|
129 |
-
"parameters": {"title": "routeSchema", "type": "object", "properties": {"next": {"anyOf": [{"enum": options}]}}, "required": ["next"]}
|
130 |
-
}
|
131 |
-
prompt = ChatPromptTemplate.from_messages([
|
132 |
-
("system", system_prompt),
|
133 |
-
MessagesPlaceholder(variable_name="messages"),
|
134 |
-
("system", "Given the conversation above, who should act next? Select one of: {options}"),
|
135 |
-
]).partial(options=str(options), members=", ".join(members))
|
136 |
-
|
137 |
-
supervisor_chain = (prompt | llm.bind_functions(functions=[function_def], function_call="route") | JsonOutputFunctionsParser())
|
138 |
-
|
139 |
-
# Workflow
|
140 |
-
class AgentState(TypedDict):
|
141 |
-
messages: Annotated[Sequence[BaseMessage], operator.add]
|
142 |
-
next: str
|
143 |
-
|
144 |
-
workflow = StateGraph(AgentState)
|
145 |
-
workflow.add_node("Researcher", research_node)
|
146 |
-
workflow.add_node("Coder", code_node)
|
147 |
-
workflow.add_node("RAG", rag_node)
|
148 |
-
workflow.add_node("supervisor", supervisor_chain)
|
149 |
-
|
150 |
-
for member in members:
|
151 |
-
workflow.add_edge(member, "supervisor")
|
152 |
-
conditional_map = {k: k for k in members}
|
153 |
-
conditional_map["FINISH"] = END
|
154 |
-
workflow.add_conditional_edges("supervisor", lambda x: x["next"], conditional_map)
|
155 |
-
workflow.set_entry_point("supervisor")
|
156 |
-
graph = workflow.compile()
|
157 |
-
|
158 |
-
# Workflow Execution
|
159 |
-
if 'outputs' not in st.session_state:
|
160 |
-
st.session_state.outputs = []
|
161 |
-
|
162 |
-
user_input = st.text_area("Enter your task or question:", placeholder=example_questions[0])
|
163 |
-
|
164 |
-
def run_workflow(task):
|
165 |
-
st.session_state.outputs.clear()
|
166 |
-
st.session_state.outputs.append(f"User Input: {task}")
|
167 |
-
for state in graph.stream({"messages": [HumanMessage(content=task)]}):
|
168 |
-
if "__end__" not in state:
|
169 |
-
st.session_state.outputs.append(str(state))
|
170 |
-
st.session_state.outputs.append("----")
|
171 |
-
|
172 |
-
if st.button("Run Workflow"):
|
173 |
-
if user_input:
|
174 |
-
run_workflow(user_input)
|
175 |
-
else:
|
176 |
-
st.warning("Please enter a task or question.")
|
177 |
-
|
178 |
-
st.subheader("Example Questions:")
|
179 |
-
for example in example_questions:
|
180 |
-
st.text(f"- {example}")
|
181 |
-
|
182 |
-
st.subheader("Workflow Output:")
|
183 |
-
for output in st.session_state.outputs:
|
184 |
-
st.text(output)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|