How do I pass session id in when using @chain

[1100111GTH]

Langserve return runnable object, and i .stream it on the client, but can't find id

[1100111GTH]

i am not sure what kind object type bing send with invoke two dict, so may be may source: Dict problem, but i don't know how to fix it

[eyurtsev]

I suspect that configuration isn't propagated to the @chain

Could you paraphrase the code to use a runnable sequence and serve that?

If you're able to share minimal reproducible code (with imports etc.), so i can copy and paste and get it working, it'll be easier for me to help

[1100111GTH]

server code:

from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder, HumanMessagePromptTemplate
from langchain_core.runnables.history import RunnableWithMessageHistory
from langchain_community.chat_message_histories.redis import RedisChatMessageHistory
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import chain
from typing import Any, List, Optional, Tuple, Dict

llm = ChatOpenAI(
    api_key="EMPTY", 
    base_url="http://127.0.0.1:8000/v1", 
    temperature=0,
    max_tokens=2048,
    model="qwen-14b-chat"
)

@chain
def base_answer(source: Dict):
  prompt_b = ChatPromptTemplate.from_messages(
      [
          ("system", "You are a helpful assiant", 
          MessagesPlaceholder(variable_name="history"),
          ("human", "{input}")
      ]
  )
  
  b_a = RunnableWithMessageHistory(
      prompt_b | llm, 
      RedisChatMessageHistory, 
      input_messages_key="input", 
      history_messages_key="history"
  ) | StrOutputParser()

  return b_a

server.py as usual...

client code ( same device for test ):

from langserve import RemoteRunnable

base_answer = RemoteRunnable("http://localhost:2031/base_answer/")
base_answer.invoke({"input": "Something to say"}, {"configurable": {"session_id": "id"}})  # or .stream

I know the func def or not is same, but igone it. and i need streaming it on the gradio, so i really need to return the object, otherwise i will return b_a.invoke({"input": "Something to say"}, {"configurable": {"session_id": "id"}}) this work well.

[1100111GTH]

Whether the problem is solved or not, thank you for your help first.

[1100111GTH]

Hi, did you find any solution？