langgraph/examples/streaming-tokens.ipynb

Streaming Tokens

In this example we will focus on explaining how to stream tokens from a language model that is powering an agent. We will use a chat agent executor as an example. There a few specific things we need to do in order to properly stream tokens. They are:

1. Set streaming=True when creating the LLM
2. Create nodes with async methods - this is best practice because in order to stream tokens we will use the async_log method.

we will call them out with the STREAMING tag below (if you just want to search for those).

Setup

First we need to install the packages required

In [1]:

!pip install --quiet -U langchain langchain_openai tavily-python

[notice] A new release of pip is available: 23.3.1 -> 23.3.2
[notice] To update, run: pip install --upgrade pip

Next, we need to set API keys for OpenAI (the LLM we will use) and Tavily (the search tool we will use)

In [2]:

import os
import getpass

os.environ["OPENAI_API_KEY"] = getpass.getpass("OpenAI API Key:")
os.environ["TAVILY_API_KEY"] = getpass.getpass("Tavily API Key:")

OpenAI API Key: ········
Tavily API Key: ········

Optionally, we can set API key for LangSmith tracing, which will give us best-in-class observability.

In [ ]:

os.environ["LANGCHAIN_TRACING_V2"] = "true"
os.environ["LANGCHAIN_API_KEY"] = getpass.getpass("LangSmith API Key:")

Set up the tools

We will first define the tools we want to use. For this simple example, we will use a built-in search tool via Tavily. However, it is really easy to create your own tools - see documentation here on how to do that.

In [1]:

from langchain_community.tools.tavily_search import TavilySearchResults

tools = [TavilySearchResults(max_results=1)]

We can now wrap these tools in a simple ToolExecutor. This is a real simple class that takes in a ToolInvocation and calls that tool, returning the output. A ToolInvocation is any class with tool and tool_input attribute.

In [2]:

from langgraph.prebuilt import ToolExecutor

tool_executor = ToolExecutor(tools)

Set up the model

Now we need to load the chat model we want to use. Importantly, this should satisfy two criteria:

1. It should work with messages. We will represent all agent state in the form of messages, so it needs to be able to work well with them.
2. It should work with OpenAI function calling. This means it should either be an OpenAI model or a model that exposes a similar interface.

Note: these model requirements are not requirements for using LangGraph - they are just requirements for this one example.

STREAMING

Here, we set streaming=True when creating the model.

In [3]:

from langchain_openai import ChatOpenAI

# We will set streaming=True so that we can stream tokens
# See the streaming section for more information on this.
model = ChatOpenAI(temperature=0, streaming=True)

After we've done this, we should make sure the model knows that it has these tools available to call. We can do this by converting the LangChain tools into the format for OpenAI function calling, and then bind them to the model class.

In [4]:

from langchain.tools.render import format_tool_to_openai_function

functions = [format_tool_to_openai_function(t) for t in tools]
model = model.bind_functions(functions)

Define the nodes

We now need to define a few different nodes in our graph. In langgraph, a node can be either a function or a runnable. There are two main nodes we need for this:

1. The agent: responsible for deciding what (if any) actions to take.
2. A function to invoke tools: if the agent decides to take an action, this node will then execute that action.

We will also need to define some edges. Some of these edges may be conditional. The reason they are conditional is that based on the output of a node, one of several paths may be taken. The path that is taken is not known until that node is run (the LLM decides).

1. Conditional Edge: after the agent is called, we should either: a. If the agent said to take an action, then the function to invoke tools should be called b. If the agent said that it was finished, then it should finish
2. Normal Edge: after the tools are invoked, it should always go back to the agent to decide what to do next

Let's define the nodes, as well as a function to decide how what conditional edge to take.

STREAMING

We define each node as an async function.

In [5]:

from langgraph.prebuilt import ToolInvocation
import json
from langchain_core.messages import FunctionMessage


# Define the function that determines whether to continue or not
def should_continue(messages):
    last_message = messages[-1]
    # If there is no function call, then we finish
    if "function_call" not in last_message.additional_kwargs:
        return "end"
    # Otherwise if there is, we continue
    else:
        return "continue"


# Define the function that calls the model
async def call_model(messages):
    response = await model.ainvoke(messages)
    # We return a list, because this will get added to the existing list
    return response


# Define the function to execute tools
async def call_tool(messages):
    # Based on the continue condition
    # we know the last message involves a function call
    last_message = messages[-1]
    # We construct an ToolInvocation from the function_call
    action = ToolInvocation(
        tool=last_message.additional_kwargs["function_call"]["name"],
        tool_input=json.loads(
            last_message.additional_kwargs["function_call"]["arguments"]
        ),
    )
    # We call the tool_executor and get back a response
    response = await tool_executor.ainvoke(action)
    # We use the response to create a FunctionMessage
    function_message = FunctionMessage(content=str(response), name=action.tool)
    # We return a list, because this will get added to the existing list
    return function_message

Define the graph

We can now put it all together and define the graph!

In [6]:

from langgraph.graph import MessageGraph, END

# Define a new graph
workflow = MessageGraph()

# Define the two nodes we will cycle between
workflow.add_node("agent", call_model)
workflow.add_node("action", call_tool)

# Set the entrypoint as `agent`
# This means that this node is the first one called
workflow.set_entry_point("agent")

# We now add a conditional edge
workflow.add_conditional_edges(
    # First, we define the start node. We use `agent`.
    # This means these are the edges taken after the `agent` node is called.
    "agent",
    # Next, we pass in the function that will determine which node is called next.
    should_continue,
    # Finally we pass in a mapping.
    # The keys are strings, and the values are other nodes.
    # END is a special node marking that the graph should finish.
    # What will happen is we will call `should_continue`, and then the output of that
    # will be matched against the keys in this mapping.
    # Based on which one it matches, that node will then be called.
    {
        # If `tools`, then we call the tool node.
        "continue": "action",
        # Otherwise we finish.
        "end": END,
    },
)

# We now add a normal edge from `tools` to `agent`.
# This means that after `tools` is called, `agent` node is called next.
workflow.add_edge("action", "agent")

# Finally, we compile it!
# This compiles it into a LangChain Runnable,
# meaning you can use it as you would any other runnable
app = workflow.compile()

Streaming LLM Tokens

You can access the LLM tokens as they are produced by each node. In this case only the "agent" node produces LLM tokens. In order for this to work properly, you must be using an LLM that supports streaming as well as have set it when constructing the LLM (e.g. ChatOpenAI(model="gpt-3.5-turbo-1106", streaming=True))

In [7]:

from langchain_core.messages import HumanMessage

inputs = [HumanMessage(content="what is the weather in sf")]
async for event in app.astream_events(inputs, version="v1"):
    kind = event["event"]
    if kind == "on_chat_model_stream":
        content = event["data"]["chunk"].content
        if content:
            # Empty content in the context of OpenAI means
            # that the model is asking for a tool to be invoked.
            # So we only print non-empty content
            print(content, end="|")
    elif kind == "on_tool_start":
        print("--")
        print(
            f"Starting tool: {event['name']} with inputs: {event['data'].get('input')}"
        )
    elif kind == "on_tool_end":
        print(f"Done tool: {event['name']}")
        print(f"Tool output was: {event['data'].get('output')}")
        print("--")

/Users/harrisonchase/workplace/langchain/libs/core/langchain_core/_api/beta_decorator.py:86: LangChainBetaWarning: This API is in beta and may change in the future.
  warn_beta(

--
Starting tool: tavily_search_results_json with inputs: {'query': 'weather in San Francisco'}
Done tool: tavily_search_results_json
Tool output was: [{'url': 'https://www.whereandwhen.net/when/north-america/california/san-francisco-ca/january/', 'content': 'Best time to go to San Francisco? Weather in San Francisco in january 2024  How was the weather last january? Here is the day by day recorded weather in San Francisco in january 2023:  Seasonal average climate and temperature of San Francisco in january  8% 46% 29% 12% 8% Evolution of daily average temperature and precipitation in San Francisco in januaryWeather in San Francisco in january 2024. The weather in San Francisco in january comes from statistical datas on the past years. You can view the weather statistics the entire month, but also by using the tabs for the beginning, the middle and the end of the month. ... 23-01-2023 47°F to 61°F. 24-01-2023 43°F to 58°F. 25-01-2023 47°F to ...'}]
--
I|'m| sorry|,| but| I| couldn|'t| find| the| current| weather| in| San| Francisco|.| However|,| you| can| check| the| weather| in| San| Francisco| for| the| month| of| January| on| this| website|:| [|San| Francisco| Weather| in| January|](|https|://|www|.where|and|when|.net|/|when|/n|orth|-|amer|ica|/cal|ifornia|/s|an|-fr|anc|isco|-ca|/j|an|uary|/|).|

In [ ]: