Tools on LangChain

1
from langchain.agents import AgentExecutor
2
from langchain_cohere.react_multi_hop.agent import create_cohere_react_agent
3
from langchain_core.prompts import ChatPromptTemplate
4
from langchain_cohere.chat_models import ChatCohere
5
6
from langchain_community.tools.tavily_search import TavilySearchResults
7
from langchain_core.pydantic_v1 import BaseModel, Field
8
9
os.environ["TAVILY_API_KEY"] = #<your-api-key>
10
11
internet_search = TavilySearchResults()
12
internet_search.name = "internet_search"
13
internet_search.description = "Returns a list of relevant document snippets for a textual query retrieved from the internet."
14
15
class TavilySearchInput(BaseModel):
16
    query: str = Field(description="Query to search the internet with")
17
internet_search.args_schema = TavilySearchInput
18
19
# LLM
20
llm = ChatCohere(model="command-r-plus", temperature=0)
21
22
# Preamble
23
preamble = """
24
You are an expert who answers the user's question with the most relevant datasource. You are equipped with an internet search tool and a special vectorstore of information about how to write good essays.
25
"""
26
27
# Prompt template
28
prompt = ChatPromptTemplate.from_template("{input}")
29
30
# Create the ReAct agent
31
agent = create_cohere_react_agent(
32
    llm=llm,
33
    tools=[internet_search],
34
    prompt=prompt,
35
)
36
37
agent_executor = AgentExecutor(agent=agent, tools=[internet_search], verbose=True)
38
39
response = agent_executor.invoke({
40
    "input": "I want to write an essay. Any tips?",
41
    "preamble": preamble,
42
})
43
44
print(response['output'])

1
### Router
2
from typing import Literal
3
4
from langchain_core.prompts import ChatPromptTemplate
5
from langchain_core.pydantic_v1 import BaseModel, Field
6
from langchain_cohere import ChatCohere
7
8
# Data model
9
class web_search(BaseModel):
10
    """
11
    The internet. Use web_search for questions that are related to anything else than agents, prompt engineering, and adversarial attacks.
12
    """
13
    query: str = Field(description="The query to use when searching the internet.")
14
15
class vectorstore(BaseModel):
16
    """
17
    A vectorstore containing documents related to agents, prompt engineering, and adversarial attacks. Use the vectorstore for questions on these topics.
18
    """
19
    query: str = Field(description="The query to use when searching the vectorstore.")
20
21
# Preamble
22
preamble = """You are an expert at routing a user question to a vectorstore or web search.
23
The vectorstore contains documents related to agents, prompt engineering, and adversarial attacks.
24
Use the vectorstore for questions on these topics. Otherwise, use web-search."""
25
26
# LLM with tool use and preamble
27
llm = ChatCohere()
28
structured_llm_router = llm.bind_tools(tools=[web_search, vectorstore], preamble=preamble)
29
30
# Prompt
31
route_prompt = ChatPromptTemplate.from_messages(
32
    [
33
        ("human", "{question}"),
34
    ]
35
)
36
37
question_router = route_prompt | structured_llm_router
38
response = question_router.invoke({"question": "Who will the Bears draft first in the NFL draft?"})
39
print(response.response_metadata['tool_calls'])
40
response = question_router.invoke({"question": "What are the types of agent memory?"})
41
print(response.response_metadata['tool_calls'])
42
response = question_router.invoke({"question": "Hi how are you?"})
43
print('tool_calls' in response.response_metadata)