ConversationChain Not Running Properly

[akshatbahety]

This is the code snippet I am trying to run

`from langchain.schema import AIMessage, HumanMessage, SystemMessage
from langchain.prompts import ChatPromptTemplate
import os
from apikey import apikey
from langchain.chat_models import ChatOpenAI
from langchain.chains import ConversationChain

os.environ["OPENAI_API_KEY"] = apikey

chat_model = ChatOpenAI(openai_api_key=apikey)
system_message = SystemMessage(content="Act as a Recruiter for a Big Tech Startup")

prompt = ChatPromptTemplate(messages=[system_message])

conversation = ConversationChain(llm=chat_model,prompt=prompt, verbose=True)

def input(human_response):
return conversation.run(human_response)
`

And I am facing the below error

ValidationError: 1 validation error for ConversationChain __root__ Got unexpected prompt input variables. The prompt expects [], but got ['history'] as inputs from memory, and input as the normal input key. (type=value_error)

The documentation is also not super robust for conversation chain, and I am really not able to figure out what can I do to resolve this as conversation chain has auto memory handling, really not sure where I am going wrong,

It works fine without using the prompt and system message.

[akshatbahety]

My code is not displaying properly above and for some reason I am not able to edit the formatting
here is the code

from langchain.schema import AIMessage, HumanMessage, SystemMessage
from langchain.prompts import ChatPromptTemplate
import os
from apikey import apikey
from langchain.chat_models import ChatOpenAI
from langchain.chains import ConversationChain

os.environ["OPENAI_API_KEY"] = apikey

chat_model = ChatOpenAI(openai_api_key=apikey)
system_message = SystemMessage(content="Act as a Recruiter for a Big Tech Startup")

prompt = ChatPromptTemplate(messages=[system_message])

conversation = ConversationChain(llm=chat_model,verbose=True)


def input(human_response):
    return conversation.run(human_response)

[dosubot[bot]]

🤖

Hello,

Thank you for reaching out and providing detailed information about the issue you're facing. The error message you're seeing, ValidationError: 1 validation error for ConversationChain __root__ Got unexpected prompt input variables. The prompt expects [], but got ['history'] as inputs from memory, and input as the normal input key. (type=value_error), is typically raised when there's a mismatch between the input variables expected by the prompt and the ones provided.

In your case, it seems like the prompt is not expecting any input variables ([]), but it's receiving ['history'] from the memory. This discrepancy is causing the validation error.

The ChatPromptTemplate you're using only contains a SystemMessage, which doesn't require any input variables. However, the ConversationChain is designed to handle a conversation with multiple turns, and it uses the memory to keep track of the conversation history. This history is then used as an input variable for the prompt.

To resolve this issue, you could modify your ChatPromptTemplate to include a HumanMessage that uses the conversation history as an input variable. Here's an example of how you could do this:

from langchain.schema import AIMessage, HumanMessage, SystemMessage
from langchain.prompts import ChatPromptTemplate, HumanMessagePromptTemplate
import os
from apikey import apikey
from langchain.chat_models import ChatOpenAI
from langchain.chains import ConversationChain

os.environ["OPENAI_API_KEY"] = apikey

chat_model = ChatOpenAI(openai_api_key=apikey)
system_message = SystemMessage(content="Act as a Recruiter for a Big Tech Startup")
human_message = HumanMessagePromptTemplate.from_template("{history}")

prompt = ChatPromptTemplate(messages=[system_message, human_message])

conversation = ConversationChain(llm=chat_model, prompt=prompt, verbose=True)

def input(human_response):
    return conversation.run(human_response)

In this code, the HumanMessagePromptTemplate is created with a template that uses history as an input variable. This HumanMessagePromptTemplate is then added to the ChatPromptTemplate. Now, the prompt expects the history input variable, which matches what's provided by the ConversationChain.

Please try this solution and let me know if it resolves your issue or if you have any other questions.

Sources

• libs/langchain/tests/unit_tests/chains/test_conversation.py
• libs/langchain/tests/unit_tests/prompts/test_prompt.py
• libs/langchain/langchain/chains/qa_generation/prompt.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.

[asikaim]

I'm facing the same problem and can't get prompts working with ConversationChain. Adding history into HumanMessagePromptTemplate leads to same error.

Got unexpected prompt input variables. The prompt expects ['history'], but got ['history'] as inputs from memory, and input as the normal input key. (type=value_error)

EDIT: HumanMessagePromptTemplate requires both {history} and {input}. That was the culprit.

[fahim9778]

@maasikai Can you please explain what you mean by the following? I am getting the same error.

EDIT: HumanMessagePromptTemplate requires both {history} and {input}. That was the culprit.

[mtluczek]

Just wanted to point out that this issue is another example that Langchain looks cool on the outside but if you try to do something more complex then the nasty things blow up. It'd be nice if anyone from Langchain tracks and responds to such bugs.

[akshatbahety]

Hey with any new tech that's the case, still far better than others, I have not been working on the above side project anymore but will be coming back to it in some time, if I find a solution will update this