How do I use templates and memory together?

[perezbalen]

Hi.
I'm new here, so I don't know if this is a stufid question, but I've been having trouble with using templates and memory on a chain at the same time.

I have this code, that works:

        # Prompt templates
        # generate message from system_template file
        system_template=  load_file (f"{template_dir}/{system_template_filename}")  
        system_message_prompt = SystemMessagePromptTemplate.from_template(system_template)

        # generate message from human_template file
        human_template_text = load_file (f"{template_dir}/{human_template_filename}") 
        human_message_prompt = HumanMessagePromptTemplate.from_template(human_template_text) 

        # sets up the chat-GPT model
        chat_prompt = ChatPromptTemplate.from_messages( [system_message_prompt, human_message_prompt] )
        
        #.format(role="fantasy writer")
        chat_chain = LLMChain(llm       = chat_model_chain, 
                              prompt    = chat_prompt, 
                              verbose   = True, 
                              memory    = memory,
                              )

        # runs the prompt
        response = chat_chain.predict(input=text_body)

This code also works (note the removal of memory on the LLMChain and the inclusion of keys in the chat_chain.predict):

        ...
        
        #.format(role="fantasy writer")
        chat_chain = LLMChain(llm       = chat_model_chain, 
                              prompt    = chat_prompt, 
                              verbose   = True, 
                              # memory    = memory,
                              )

        # runs the prompt
        response = chat_chain.predict(input=text_body,
                                    role=role,
                                    writing_style=writing_style,
                                    tone=tone,
                                    text_body=text_body )

But this doesn't:

        ...
        
        #.format(role="fantasy writer")
        chat_chain = LLMChain(llm       = chat_model_chain, 
                              prompt    = chat_prompt, 
                              verbose   = True, 
                              memory    = memory,
                              )

        # runs the prompt
        response = chat_chain.predict(input=text_body,
                                    role=role,
                                    writing_style=writing_style,
                                    tone=tone,
                                    text_body=text_body )

It gives an error I don't understand

Message=One input key expected got ['base_author', 'role', 'incoming_chunk', 'outline', 'text_body', 'writing_style', 'input', 'summary', 'profiles', 'tone']
Source=
StackTrace:

4.62s - Error in linecache.getline('V:\Code Projects\repos\LangChain\textSplitter.py', 96, f_globals)
Traceback (most recent call last):
File "_pydevd_bundle/pydevd_cython.pyx", line 546, in _pydevd_bundle.pydevd_cython.PyDBFrame._handle_exception
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 30, in getline
lines = getlines(filename, module_globals)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 46, in getlines
return updatecache(filename, module_globals)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 137, in updatecache
lines = fp.readlines()
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\codecs.py", line 322, in decode
(result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x93 in position 816: invalid start byte
0.00s - Error on build_exception_info_response.
Traceback (most recent call last):
File "c:\program files\microsoft visual studio\2022\community\common7\ide\extensions\microsoft\python\core\debugpy_vendored\pydevd_pydevd_bundle\pydevd_comm.py", line 1452, in build_exception_info_response
line_text = linecache.getline(original_filename, lineno)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 30, in getline
lines = getlines(filename, module_globals)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 46, in getlines
return updatecache(filename, module_globals)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 137, in updatecache
lines = fp.readlines()
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\codecs.py", line 322, in decode
(result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x93 in position 816: invalid start byte
0.02s - Error on build_exception_info_response.
Traceback (most recent call last):
File "c:\program files\microsoft visual studio\2022\community\common7\ide\extensions\microsoft\python\core\debugpy_vendored\pydevd_pydevd_bundle\pydevd_comm.py", line 1452, in build_exception_info_response
line_text = linecache.getline(original_filename, lineno)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 30, in getline
lines = getlines(filename, module_globals)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 46, in getlines
return updatecache(filename, module_globals)
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\linecache.py", line 137, in updatecache
lines = fp.readlines()
File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python39_64\lib\codecs.py", line 322, in decode
(result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x93 in position 816: invalid start byte

[Galkimron]

Hey,

Actually I'm building a tool that will help us with exactly this use case.
let me know if you want more details

[dina-khalid]

Actually, I tried to make smth like that

my_template = """
          

"""

my_prompt = PromptTemplate(
    input_variables=['.......'], template=my_template
)
# langchain example
template = """You are a chatbot having a conversation with a human.

{chat_history}
Human: {human_input}
Chatbot:"""

prompt = PromptTemplate(
    input_variables=["chat_history", "human_input"], template=template
)
memory = ConversationBufferMemory(memory_key="chat_history")

and when I want to get specific answer i use:

new_prompt = meal_prompt.format(...........)
llm_chain.predict(human_input=new_prompt)