Chat memory

A chat instance saves the message history and uses it to future responses. Here is an example of a chatbot that retains conversation history across multiple send requests.

@declarai.experimental.chat
class SQLBot:
    """
    You are a sql assistant. You help with SQL related questions with one-line answers.
    """

sql_bot = SQLBot()

sql_bot.send("When should I use a LEFT JOIN?") # (1)!
> "You should use a LEFT JOIN when you want to retrieve all records from the left table and matching records from the right table."

sql_bot.send("But how is it different from a RIGHT JOIN?") # (2)!
> "A LEFT JOIN retrieves all records from the left table and matching records from the right table, while a RIGHT JOIN retrieves all records from the right table and matching records from the left table."

1. The first message is sent with the system prompt.
2. The second message is sent with the previous conversation and therefore the model is aware of the first question.

Conversation History

You can view the conversation history by accessing the conversation attribute.

sql_bot.conversation

> [
    user: When should I use a LEFT JOIN?, 
    assistant: You should use a LEFT JOIN when you want to retrieve all records from the left table and matching records from the right table.,
    user: But how is it different from a RIGHT JOIN?,
    assistant: A LEFT JOIN retrieves all records from the left table and matching records from the right table, while a RIGHT JOIN retrieves all records from the right table and matching records from the left table.
]

Warning

Keep in mind that the conversation history does not contain the system prompt. It only contains the user messages and the chatbot responses.

If you want to access the system message, you can use the system attribute.

sql_bot.system

> "system: You are a sql assistant. You help with SQL related questions with one-line answers.\n"

Default Memory

The default message history of a chat is a simple in-memory list. This means that history exists only for the duration of the chatbot session.

If you prefer to have a persistent history, you can use the FileMessageHistory class from the declarai.memory module.

Setting up a memory

Setting up a memory is done by passing chat_history as a keyword argument to the declarai.experimental.chat decorator.

import declarai
from declarai.memory import FileMessageHistory

gpt_35 = declarai.openai(model="gpt-3.5-turbo")

@gpt_35.experimental.chat(chat_history=FileMessageHistory("sql_bot_history.txt")) # (1)!
class SQLBot:
    """
    You are a sql assistant. You help with SQL related questions with one-line answers.
    """

1. file path is not mandatory. If you do not provide a file path, the default file path is stored in a tmp directory.

We can also initialize the chat_history at runtime

import declarai
from declarai.memory import FileMessageHistory

gpt_35 = declarai.openai(model="gpt-3.5-turbo")

@gpt_35.experimental.chat
class SQLBot:
    """
    You are a sql assistant. You help with SQL related questions with one-line answers.
    """
sql_bot = SQLBot(chat_history=FileMessageHistory("sql_bot_history.txt"))