Missing required configuration to use LangChain integration: '_llm_type' parameter not found in PANDASAI LangChain LLM settings

[jasongonin]

System Info

similar to "Unfortunately, I was not able to get your answers, because of the following error:\n\n'HuggingFaceLLM' object has no attribute '_llm_type'\n" #1322

pandasai==3.0.0b10
Python 3.11
pandasai-langchain==0.1.4
langchain==0.1.16
langchain-community==0.0.38
langchain-core==0.1.53
langchain-openai==0.1.3

as described below
as simple code (to exec in Jupyter notebook)

using as a df source https://github.com/datasciencedojo/datasets/blob/master/titanic.csv

🐛 Describe the bug

cell 1 : setup venv one shot

!pip install pandas pandasai==3.0.0b10 pandasai-langchain==0.1.4

!pip install langchain==0.1.16 langchain-community==0.0.38 langchain-core==0.1.53 langchain-openai==0.1.3

 

cell 2 : actual test

import pandas as pd

from langchain_openai import ChatOpenAI

from langchain import LLMChain

from pandasai_langchain import LangchainLLM

from langchain_core.prompts import PromptTemplate, MessagesPlaceholder

from pandasai import SmartDataframe

 

llm= ChatOpenAI(api_key="sk_xxxx")

systemPrompt="my system prompt"

myQueryWrapperPrompt=" my query wrapper prompt"

template= PromptTemplate(

    input_variables=["query"],

    template = systemPrompt + "{query}" + myQueryWrapperPrompt,

)

chain= LLMChain(llm=llm, prompt=template)

 

from pandasai_langchain import LangchainLLM

langchainllm= LangchainLLM(chain)

df = pd.read_csv("/home/[email protected]/Downloads/titanic.csv")

smartdf=SmartDataframe(df,config={"llm":langchainllm})

smartdf.chat(query="How does the age (Age) of passengers affect their likelihood of survival (Survived)?")

error:

AttributeError: 'LangchainLLM' object has no attribute '_llm_type'"

See bellow complete error

{

            "name": "AttributeError",

            "message": "'LangchainLLM' object has no attribute '_llm_type'",

            "stack": "---------------------------------------------------------------------------

AttributeError Traceback (most recent call last)

Cell In[22], line 21

 19 df = pd.read_csv(\"/home/[email protected]/Downloads/titanic.csv\")

 20 smartdf=SmartDataframe(df,config={\"llm\":langchainllm})

---> 21 smartdf.chat(query="question?")

File ~/project/gpt-be-fastapi/jupyter.venv/lib/python3.11/site-packages/pandasai/smart_dataframe/init.py:80, in SmartDataframe.chat(self, query, output_type)

 61 def chat(self, query: str, output_type: Optional[str] = None):

 62     \"\"\"

 63     Run a query on the dataframe.

 64     Args:

(...)

78         ValueError: If the query is empty

 79     \"\"\"

---> 80 return self._agent.chat(query, output_type)

File ~/project/gpt-be-fastapi/jupyter.venv/lib/python3.11/site-packages/pandasai/agent/base.py:92, in Agent.chat(self, query, output_type)

 88 \"\"\"

 89 Start a new chat interaction with the assistant on Dataframe.

 90 \"\"\"

 91 self.start_new_conversation()

---> 92 return self._process_query(query, output_type)

File ~/project/gpt-be-fastapi/jupyter.venv/lib/python3.11/site-packages/pandasai/agent/base.py:247, in Agent._process_query(self, query, output_type)

244 query = UserQuery(query)

245 self._state.logger.log(f\"Question: {query}\")

246 self._state.logger.log(

--> 247 f"Running PandaAI with {self._state.config.llm.type} LLM..."

248 )

250 self._state.output_type = output_type

251 try:

File ~/project/gpt-be-fastapi/jupyter.venv/lib/python3.11/site-packages/pandasai_langchain/langchain.py:60, in LangchainLLM.type(self)

 58 @property

 59 def type(self) -> str:

---> 60 return f"langchain_{self.langchain_llm._llm_type}"

AttributeError: 'LangchainLLM' object has no attribute '_llm_type'"

}