Session is created by default . Different users , different sessions

since its only 3-4 lines of code i did not maintain it. Rest of the code comes automatically when you crate you streamlit app .... can you try the vide with higher quality. i could see it clearly with higher resolution.

@@techlycan thanks I can see it.can you few more on streamlit with actual real case scenario

Convert the query output in Data frame and then simply display it. Note that my DB here is RDBMS.

plz write to [email protected]

Thanks brother ! Happy to know it could help you to get better understanding :)

import os from app_secrates import OPENAI_API_KEY from langchain.llms import OpenAI from langchain.memory import ConversationBufferMemory from langchain.prompts import PromptTemplate from lamgchain.chains import LLMChain import streamlit as st os.environ["OPENAI_API_KEY"] = OPENAI_API_KEY memory = ConversationBufferMemory() st.title("Conversational Bot") user_input = st.text_input("Enter your message here") if 'chat_history' not in st.session_state: st.session_state.chat_history = [] else: for message in st.session_state.chat_history: memory.save_context({"input": message['human']}, {'output': message['AI']}) promprt_template = PromptTemplate( input_variables= ['history', 'input'], template=""" You are coversational bot. Maintain formal tone in your responses. History: {history} humen: {input} AI: """ ) llm = OpenAI(temprature=0.0) conversation_chain = LLMChain(llm=llm, prompt=promprt_template, memory=memory, verbose=True) if user_input: response = conversation_chain.run(input=user_input) message = {"human": user_input, "AI": response['text']} st.session_state.chat_history.append(message) st.write(response) with st.expander(label="Chat History", expanded=True): st.write(st.session_state.chat_history)

Yes but I think you are missing the use case here.... Tech PPL do not need this as they are proficient in SQL already....it's usually the business PPL who face challenges when they need to interact with tables....and for reporting the tables are usually summarized tables limited in count

Partially true.... It depends what you define as a cluster....... for example the small VW in SF is single cluster XS which has 8 compute nodes (CPUs) with in.... be it vertical or horizontal scaling , you can not add less than a single cluster (8 compute nodes) to VW , that's why it is referred as adding cluster in vertical scaling. In case of horizontal scaling suppose you applied it on VW of M size i.e. a VW with 4 clusters.... in that case what you will get is another set of 4 clusters getting active or suspended as a single unit , you can not add a copy of different cluster size in horizontal scaling hence the another unit of same clusters size (4 clusters i.e 32 compute nodes) i referred as clone of existing VW because they will always be identical in size and behavior. Hope that makes sense to you !

@@techlycanCan you confirm if this is true : XS VWH has 1 node, 4 cores and 8 threads and it can run max 8 queries in parallel? www.google.com/search?q=how+many+nodes+in+xs+snowflake+warehouse&rlz=1C1GCEA_enUS1012US1012&oq=how+many+nodes+in+xs+snowflake+warehouse&gs_lcrp=EgZjaHJvbWUyBggAEEUYOTIHCAEQIRigATIHCAIQIRigATIHCAMQIRigATIHCAQQIRigATIHCAUQIRigATIHCAYQIRifBTIHCAcQIRifBTIHCAgQIRifBTIHCAkQIRifBdIBCTIzNzQzajBqN6gCALACAA&sourceid=chrome&ie=UTF-8

Securityadmin is a role not user

@newg1203 This prompt has a limited scope. If you increase the scope then prompt needs to be modified accordingly. Moreover you may need to pass on further information to LLM to help writing queries. well defined catalogue or RAG could be used to enhance the functionality.

Your API key defines the model which you would have generated in openai portal... This section only define the properties and not model.

@@techlycan ok I see, but when I create a new key in my portal, it just ask my for a name, i dont even have a chocie of a model? i dont get it....

@@renaudgg I have not tried gpt4 key as of yet but I think the portal provides documentation to generate it....it's not free so may be you need to subscribe first and then you may get the option to create it.

Yes that's the context setting but in real case you won't need messages upto that back.... Realistically thinking not more than 3-5 previous messages should be passed and even that depends on what you are using LLM app for...so fix the no of previous messages

That's what ... You can't put full schema ddl in config file and pass as prompt usually for reporting there are limited tables but not always ....there are 2 ways to achieve it.... Train the model on your schema ..highly unlikely....another way is to store the ddl into vector db and then make two calls in iteration ...first to extract the relevant ddls from vector db and second to build SQL using ddl and question...... This is what is displayed here how to extract data from vector db...aka RAG

@@techlycan yea ok, but I just thought of something, i can just create one big view of all important fields I need and there you go..i just solve my problem

@@renaudgg that's interesting ... One Difference ... RDBMS works based on exact match where as VDB based on vectors but you can try this approach and tell us if that works... May be if I have to go your way then I would pass the table names in the prompt to identify which tables would be needed... Extract may be columns needed from view and then build final prompt to pass to LLM .... But that not necessarily be your approach too....Eager to listen if it works for you

@@techlycan I did a big table that regroups many columns of different tables. then once an hour, I append new data to the table (without deleting the rest) so the AI only know that table which I think is perfect.

You can simply create database and schemas using create statements

also how do we import execute_sf_query? is that a predefined function? somewhere in the project?

Yes ...plz join my group for details chat.whatsapp.com/Jz2ltS009Zi9DqXgiQMG0U

Thank you 👍

@@techlycan Could you please share your mail id.

Glad it was helpful!

Keep it up

Complete ? What got left in agenda ?