Langchain KeyError: 'prompt' in ConversationChain

[fahim9778]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

system_template = system_prompt
human_template = "{human_query}"
 
chat_prompt = ChatPromptTemplate.from_messages(
    [
        SystemMessagePromptTemplate.from_template(system_template),
        HumanMessage(content="Hello teacher!"),
        AIMessage(content="Welcome everyone!"),
        HumanMessagePromptTemplate.from_template(human_template),
    ]
)
 
messages = chat_prompt.format_messages(
    user_age = user_age,
    user_edu_scope = user_edu_scope,
    user_interest = user_interest,
    prog_language_to_teach = prog_language_to_teach,
    human_query=human_query
)
print(messages)


```python
from langchain.chains.conversation.memory import ConversationSummaryMemory

conversation_sum = ConversationChain(prompt=messages, llm=llm)

Description

I am trying to run ConversationChain with ChatPromptTemplate, but it's giving me the following KeyError: 'prompt' for the line
conversation_sum = ConversationChain(prompt=messages, llm=llm)

Error log:

KeyError Traceback (most recent call last)
Cell In[17], line 3
1 from langchain.chains.conversation.memory import ConversationSummaryMemory
----> 3 conversation_sum = ConversationChain(prompt=messages, llm=llm)

File /opt/conda/lib/python3.10/site-packages/langchain_core/_api/deprecation.py:203, in deprecated..deprecate..finalize..warn_if_direct_instance(self, *args, **kwargs)
201 warned = True
202 emit_warning()
--> 203 return wrapped(self, *args, **kwargs)

File /opt/conda/lib/python3.10/site-packages/pydantic/v1/main.py:339, in BaseModel.init(pydantic_self, **data)
333 """
334 Create a new model by parsing and validating input data from keyword arguments.
335
336 Raises ValidationError if the input data cannot be parsed to form a valid model.
337 """
338 # Uses something other than self the first arg to allow "self" as a settable attribute
--> 339 values, fields_set, validation_error = validate_model(pydantic_self.class, data)
340 if validation_error:
341 raise validation_error

File /opt/conda/lib/python3.10/site-packages/pydantic/v1/main.py:1100, in validate_model(model, input_data, cls)
1098 continue
1099 try:
-> 1100 values = validator(cls_, values)
1101 except (ValueError, TypeError, AssertionError) as exc:
1102 errors.append(ErrorWrapper(exc, loc=ROOT_KEY))

File /opt/conda/lib/python3.10/site-packages/langchain/chains/conversation/base.py:58, in ConversationChain.validate_prompt_input_variables(cls, values)
53 if input_key in memory_keys:
54 raise ValueError(
55 f"The input key {input_key} was also found in the memory keys "
56 f"({memory_keys}) - please provide keys that don't overlap."
57 )
---> 58 prompt_variables = values["prompt"].input_variables
59 expected_keys = memory_keys + [input_key]
60 if set(expected_keys) != set(prompt_variables):

KeyError: 'prompt'

System Info

langchain==0.2.6
langchain-community==0.2.6
langchain-core==0.2.10
langchain-text-splitters==0.2.2

[dosubot[bot]]

To resolve the KeyError: 'prompt' when trying to run ConversationChain with ChatPromptTemplate in LangChain, ensure that you are using the correct key for the prompt. The provided context shows that the prompt is defined with the key QA_PROMPT.

Here is an example of how you can use it:

from langchain.chains import ConversationChain
from langchain.chains.conversational_retrieval.prompts import QA_PROMPT

# Initialize the ConversationChain with the correct prompt
conversation_chain = ConversationChain(prompt=QA_PROMPT, llm=llm)

# Now you can run the conversation chain without encountering the KeyError
response = conversation_chain.run(input="Hello, how are you?")
print(response)

Make sure that you import QA_PROMPT from the correct module and use it when initializing the ConversationChain. This should resolve the KeyError: 'prompt' ^[1][2].

To continue talking to Dosu, mention @dosu.

[fahim9778]

@dosu
I didn't get it. I want to use it with ChatPromptTemplate according to my existing code, which is:

system_template = system_prompt
human_template = "{human_query}"
 
chat_prompt = ChatPromptTemplate.from_messages(
    [
        SystemMessagePromptTemplate.from_template(system_template),
        HumanMessage(content="Hello teacher!"),
        AIMessage(content="Welcome everyone!"),
        HumanMessagePromptTemplate.from_template(human_template),
    ]
)
 
messages = chat_prompt.format_messages(
    user_age = user_age,
    user_edu_scope = user_edu_scope,
    user_interest = user_interest,
    prog_language_to_teach = prog_language_to_teach,
    human_query=human_query
)

How to do it? I don't need any conversational_retrieval.prompts import QA_PROMPT.

[dosubot[bot]]

To resolve the KeyError: 'prompt' when using ConversationChain with ChatPromptTemplate according to your existing code, you need to ensure that the ConversationChain is initialized correctly with the formatted messages.

Here's how you can modify your code:

1. Define the prompt template correctly.
2. Initialize the ConversationChain with the correct prompt.

Here is the updated code:

from langchain.chains import ConversationChain
from langchain.prompts import ChatPromptTemplate, SystemMessagePromptTemplate, HumanMessagePromptTemplate
from langchain.llms import OpenAI  # or any other LLM you are using

# Define your system and human templates
system_template = system_prompt
human_template = "{human_query}"

# Create the ChatPromptTemplate
chat_prompt = ChatPromptTemplate.from_messages(
    [
        SystemMessagePromptTemplate.from_template(system_template),
        HumanMessagePromptTemplate.from_template(human_template),
    ]
)

# Format the messages
messages = chat_prompt.format_messages(
    user_age=user_age,
    user_edu_scope=user_edu_scope,
    user_interest=user_interest,
    prog_language_to_teach=prog_language_to_teach,
    human_query=human_query
)

# Initialize the ConversationChain with the correct prompt
llm = OpenAI(api_key="your_openai_api_key")  # Replace with your LLM initialization
conversation_chain = ConversationChain(prompt=chat_prompt, llm=llm)

# Use the conversation chain
response = conversation_chain.run({"history": "previous conversation here", "input": "your question here"})
print(response)

Make sure that:

• The ChatPromptTemplate is correctly defined and used.
• The ConversationChain is initialized with the chat_prompt and the LLM instance.

This should resolve the KeyError: 'prompt' and allow you to use the ConversationChain with your ChatPromptTemplate ^[1][2].

To continue talking to Dosu, mention @dosu.

[fahim9778]

@dosu, I need that messages from the from_messages output to be used in ConversationChain input, which you skipped. That's where my error starts. I need to pass those variables into the chat_prompt formatted messages. Please help me to integrate these two without errors.

[dosubot[bot]]

To integrate the ChatPromptTemplate with ConversationChain and use the formatted messages correctly, you need to ensure that the ConversationChain is initialized with the correct prompt and that the formatted messages are used properly.

Here's how you can modify your code to achieve this:

1. Define the prompt template correctly.
2. Format the messages using the ChatPromptTemplate.
3. Initialize the ConversationChain with the correct prompt.

Here is the updated code:

from langchain.chains import ConversationChain
from langchain.prompts import ChatPromptTemplate, SystemMessagePromptTemplate, HumanMessagePromptTemplate
from langchain.llms import OpenAI  # or any other LLM you are using

# Define your system and human templates
system_template = system_prompt
human_template = "{human_query}"

# Create the ChatPromptTemplate
chat_prompt = ChatPromptTemplate.from_messages(
    [
        SystemMessagePromptTemplate.from_template(system_template),
        HumanMessagePromptTemplate.from_template(human_template),
    ]
)

# Format the messages
formatted_messages = chat_prompt.format_messages(
    user_age=user_age,
    user_edu_scope=user_edu_scope,
    user_interest=user_interest,
    prog_language_to_teach=prog_language_to_teach,
    human_query=human_query
)

# Initialize the ConversationChain with the correct prompt
llm = OpenAI(api_key="your_openai_api_key")  # Replace with your LLM initialization
conversation_chain = ConversationChain(prompt=chat_prompt, llm=llm)

# Use the conversation chain
response = conversation_chain.run({"history": formatted_messages, "input": "your question here"})
print(response)

Make sure that:

• The ChatPromptTemplate is correctly defined and used.
• The ConversationChain is initialized with the chat_prompt and the LLM instance.
• The formatted messages are passed correctly to the ConversationChain.

This should resolve the KeyError: 'prompt' and allow you to use the ConversationChain with your ChatPromptTemplate ^[1][2].

To continue talking to Dosu, mention @dosu.