Why is all messages still being passed when LLM is called? Trim_message isn't working. Here is my code

OSS Product Help LangChain
1 
from langchain.messages import RemoveMessage

from langgraph.graph.message import REMOVE_ALL_MESSAGES

from langgraph.checkpoint.memory import InMemorySaver

from langchain.agents import create_agent, AgentState

from langchain.agents.middleware import before_model

from langgraph.runtime import Runtime

from langchain_core.runnables import RunnableConfig

from typing import Any

from localLLMConfiger import office_chat_model as model

import fiddlerConfiger

from langchain_core.messages.utils import trim_messages

from langchain.messages import SystemMessage,HumanMessage,AIMessage

messages = [

SystemMessage("you're a good assistant, you always respond with a joke."),

HumanMessage("i wonder why it's called langchain"),

AIMessage(

            'Well, I guess they thought "WordRope" and "SentenceString" just '

            "didn't have the same ring to it!"

        ),

HumanMessage("and who is harrison chasing anyways"),

AIMessage(

            "Hmmm let me think.\\n\\nWhy, he's probably chasing after the last "

            "cup of coffee in the office!"

        ),

HumanMessage("what's my name")\]

print(trim_messages(messages,

                max_tokens=3,

                token_counter=len,

                strategy="last",

                start_on="human",

                include_system=True))

@before_model

def __trim_messages(state: AgentState, runtime: Runtime) → dict[str, Any] | None:

"""Keep only the last few messages to fit context window."""

messages = state\["messages"\]

msg = trim_messages(messages,

                max_tokens=3,

                token_counter=len,

                strategy="last",

                start_on="human",

                include_system=True

    )

return {

    "messages": msg

    

}

agent = create_agent(

model,

middleware=\[\__trim_messages\],

checkpointer=InMemorySaver(),

)

config: RunnableConfig = {“configurable”: {“thread_id”: “1765”}}

final_response = agent.invoke({“messages”: messages}, config)

final_response[“messages”][-1].pretty_print()
2

Hi @hopegood

could you format you post again please? Right now I’m having a hard time trying to read and understand it :cry:

And it would be helpful if you shared more code on how the graph is built :slight_smile:

3

Sorry, the formatting was incorrect when I copied the code. I have now resubmitted the code.

1 Like
4

In @wrap_model_call, use the following code:
request.messages = trim_messages(
request.messages,
max_tokens=3,
token_counter=len,
strategy=“last”,
start_on=“human”,
include_system=True
)
return handler(request)`

1 Like