Chat Conversation agent seems to be broken and not working even with gpt-4

[cvgoudar]

https://python.langchain.com/docs/modules/agents/agent_types/chat_conversation_agent

Executed the following code and doesn't work with gpt-4 as well:

from langchain.agents import Tool
from langchain.agents import AgentType
from langchain.memory import ConversationBufferMemory
# from langchain import OpenAI
from langchain.chat_models import ChatOpenAI
from langchain.utilities import GoogleSerperAPIWrapper
from langchain.agents import initialize_agent

import os
from dotenv import load_dotenv

env_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../.env")
load_dotenv(dotenv_path=env_path, override=True)

search = GoogleSerperAPIWrapper()
tools = [
    Tool(
        name="Current Search",
        func=search.run,
        description="useful for when you need to answer questions about current events or the current state of the world"
    ),
]

memory = ConversationBufferMemory(memory_key="chat_history")
llm = ChatOpenAI(temperature=0, model_name="gpt-4")
agent_chain = initialize_agent(tools, llm, agent=AgentType.CHAT_CONVERSATIONAL_REACT_DESCRIPTION, verbose=True,
                               memory=memory)

while True:
    input_text = input()
    if input_text == "END":
        break
    agent_chain.run(input=input_text)

Following is the error message:

Hi

Entering new AgentExecutor chain...
Traceback (most recent call last):
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\base.py", line 282, in call
raise e
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\base.py", line 276, in call
self._call(inputs, run_manager=run_manager)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\agents\agent.py", line 1036, in _call
next_step_output = self._take_next_step(
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\agents\agent.py", line 833, in _take_next_step
output = self.agent.plan(
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\agents\agent.py", line 456, in plan
full_output = self.llm_chain.predict(callbacks=callbacks, **full_inputs)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\llm.py", line 255, in predict
return self(kwargs, callbacks=callbacks)[self.output_key]
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\base.py", line 282, in call
raise e
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\base.py", line 276, in call
self._call(inputs, run_manager=run_manager)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\llm.py", line 91, in _call
response = self.generate([inputs], run_manager=run_manager)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\llm.py", line 100, in generate
prompts, stop = self.prep_prompts(input_list, run_manager=run_manager)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\chains\llm.py", line 136, in prep_prompts
prompt = self.prompt.format_prompt(**selected_inputs)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\prompts\chat.py", line 322, in format_prompt
messages = self.format_messages(**kwargs)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\prompts\chat.py", line 564, in format_messages
message = message_template.format_messages(**rel_params)
File "E:\ProgramData\Anaconda3\envs\fastapi\lib\site-packages\langchain\prompts\chat.py", line 100, in format_messages
raise ValueError(
ValueError: variable chat_history should be a list of base messages, got python-BaseException

langchain version :
langchain==0.0.268

[dosubot[bot]]

🤖

Hello,

Thank you for reaching out and providing detailed information about the issue you're facing. The error message you're seeing, "ValueError: variable chat_history should be a list of base messages, got python-BaseException", indicates that the chat_history variable is not being assigned the expected type of data. In LangChain, chat_history is expected to be a list of BaseMessage objects, which can be either HumanMessage or AIMessage instances.

The chat_history variable is used in the create_prompt method of the ConversationalChatAgent class. If the chat_history variable is not a list of base messages when the create_prompt method is called, it would cause the ValueError you are seeing. You can check the relevant code in the conversational_chat/base.py file.

As for the issue with gpt-4, the ChatOpenAI class in the openai.py file has a default model name of "gpt-3.5-turbo". If you are trying to use gpt-4 with LangChain, you would need to explicitly set the model_name to "gpt-4" when creating an instance of the ChatOpenAI class. However, the _get_encoding_model method in the ChatOpenAI class handles the model name for "gpt-4". If the model_name is "gpt-4", it changes it to "gpt-4-0314". This could be the reason why you are facing issues when trying to use gpt-4 with LangChain. You can check the relevant code in the openai.py file.

Please ensure that the chat_history variable is being assigned a list of BaseMessage objects and that the model_name is correctly set to "gpt-4" when creating an instance of the ChatOpenAI class. If you continue to face issues, please provide more details about the steps you're taking and the full error message you're seeing, and I'll be happy to assist you further.

Best,
Dosu

Sources

• libs/langchain/langchain/agents/conversational_chat/base.py
• libs/langchain/langchain/chat_models/openai.py
• libs/langchain/langchain/memory/chat_message_histories/cosmos_db.py
• libs/langchain/langchain/schema/memory.py
• libs/langchain/langchain/memory/chat_message_histories/firestore.py
• libs/langchain/tests/unit_tests/memory/chat_message_histories/test_file.py

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.