How can I add memory to my SQLDatabaseChain? Can I embed the data that comes from SQL tables? Can anyone help me out.

[shruthic]

I would like to add memory to my chatbot. Since most of the times it picks the entire dataset for filtering and querying, can I use embeddings here?
If yes how can I incorporate them, please help me.
here is my code.

import pymysql
from langchain import SQLDatabaseChain
from langchain.sql_database import SQLDatabase
import urllib.parse

db_user = "XXX"
db_password = "XXX"
encoded_password = urllib.parse.quote_plus(db_password)
db_host = "XXX.mysql.database.azure.com"
db_name = "XX"
ssl_ca = r"XXX"

db = SQLDatabase.from_uri(database_uri=f"mysql+pymysql://{db_user}:{db_password}@{db_host}/{db_name}?ssl_ca={ssl_ca}")
)

from langchain.chat_models import ChatOpenAI, AzureChatOpenAI

llm = AzureChatOpenAI(deployment_name='gpt-35-turbo',
model_name='gpt-35-turbo',
openai_api_base='XXX',
openai_api_version='2023-05-15',
openai_api_key='XXX',
temperature=0)

db_chain = SQLDatabaseChain.from_llm(llm, db, verbose=True)

sys_msg = """some prompt
Here is the question.
Question: {question}
Answer:
"""

def get_prompt():
print("Type 'exit' to quit")

while True:
    prompt = input("Enter a prompt: ")

    if prompt.lower() == 'exit':
        print('Exiting...')
        break
    else:
        try:
            question = sys_msg.format(question=prompt)
            print(db_chain.run(question))
        except Exception as e:
            print(e)

get_prompt()

[dosubot[bot]]

Answer generated by a 🤖

Answer

I found related issues that might be helpful. I did my best to summarize the solutions, but I recommend looking at the related issues yourself.

Open Issues

How to add memory to SQLDatabaseChain?

The suggested solution is:

For anyone coming in the future. Currently, passing memory directly in SQLDatabaseChain or SQLDatabaseSequentialChain is not possible but I am working on that (PR: #7546) . However, you can create an SQL-agent with memory as follow.

from langchain.agents.agent_toolkits import create_sql_agent,SQLDatabaseToolkit
from langchain.agents.agent_types import AgentType
from langchain.sql_database import SQLDatabase
from langchain.chat_models.openai import ChatOpenAI
from langchain.memory import ConversationBufferMemory

memory = ConversationBufferMemory(memory_key = 'history' , input_key = 'input')
llm = ChatOpenAI(model_name = GPT_MODEL_NAME , temperature = 0)
db = SQLDatabase.from_uri(CONN_STRING)

suffix = """Begin!

Relevant pieces of previous conversation:
{history}
(You do not need to use these pieces of information if not relevant)

Question: {input}
Thought: I should look at the tables in the database to see what I can query.  Then I should query the schema of the most relevant tables.
{agent_scratchpad}
"""
executor = create_sql_agent(
        llm = llm,
        toolkit = SQLDatabaseToolkit(db=db, llm=llm),
        agent_type = AgentType.ZERO_SHOT_REACT_DESCRIPTION,
        input_variables = ["input", "agent_scratchpad","history"],
        suffix = suffix # must have history as variable,
        agent_executor_kwargs = {'memory':memory}
    )

---

This response is meant to be useful, save you time, and share context. 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.

[shruthic]

let me try and get back. Also What about embeddings. Can I embed my SQL data.

[shruthic]

Unfortunately this dosnt work for me,
I keep getting parser issue and it does not understand the context w.r.t the chat history.

[dosubot[bot]]

Answer generated by a 🤖

Answer

I found related issues that might be helpful. I did my best to summarize the solutions, but I recommend looking at the related issues yourself.

Closed Issues

SQLDatabaseChain has no attribute from_llm

No solution found.

Open Issues

How to add memory to SQLDatabaseChain?

The suggested solution is:

For anyone coming in the future. Currently, passing memory directly in SQLDatabaseChain or SQLDatabaseSequentialChain is not possible but I am working on that (PR: #7546) . However, you can create an SQL-agent with memory as follow.

from langchain.agents.agent_toolkits import create_sql_agent,SQLDatabaseToolkit
from langchain.agents.agent_types import AgentType
from langchain.sql_database import SQLDatabase
from langchain.chat_models.openai import ChatOpenAI
from langchain.memory import ConversationBufferMemory

memory = ConversationBufferMemory(memory_key = 'history' , input_key = 'input')
llm = ChatOpenAI(model_name = GPT_MODEL_NAME , temperature = 0)
db = SQLDatabase.from_uri(CONN_STRING)

suffix = """Begin!

Relevant pieces of previous conversation:
{history}
(You do not need to use these pieces of information if not relevant)

Question: {input}
Thought: I should look at the tables in the database to see what I can query.  Then I should query the schema of the most relevant tables.
{agent_scratchpad}
"""
executor = create_sql_agent(
        llm = llm,
        toolkit = SQLDatabaseToolkit(db=db, llm=llm),
        agent_type = AgentType.ZERO_SHOT_REACT_DESCRIPTION,
        input_variables = ["input", "agent_scratchpad","history"],
        suffix = suffix # must have history as variable,
        agent_executor_kwargs = {'memory':memory}
    )

---

This response is meant to be useful, save you time, and share context. 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.