SQL Chain Example | 🦜🔗 LangChain UseCases

[@foysalmamun5106]

Many Thanks

[@AO-ip9hi]

Tapai ko video haru Sarai ramro cha bro! Keep up the good work.

[@datasciencebasics]

dhanyabaad :)

[@praveenc2248]

Awesome, Thanks bro

[@veeru18]

how we could get answer in json format please advice

[@python360]

Thank you, would querying a large database with say 1million row use up a lot of tokens / is there any for of caching that can be done with ChatGPT ?

[@datasciencebasics]

Hello, haven’t tested it out but you could try it out and have a look at OpenAI usage to see how it behaves.

[@ИгорьТ-х6м]

How does SQLDatabaseChain knows about the database structure (tables, etc)?

[@hadjersa28]

I made a function to compare between LLMs results and humans results. is sql Query already exist in the database or is it you who makes them ?

[@datasciencebasics]

SQL queries are created by LLMs when we ask the natural langauge.

[@Anna00722]

Can we plot a graph of thr output data

[@datasciencebasics]

this is quite old video and I haven’t tried myself yet. Give a try with latest models.

[@AliQureshi-n6g]

Hi, thanks for the video. I have 2 questions: Question 1: what is the difference between "SQLDatabaseChain" and "SQLDatabaseToolkit"? Question 2: I get an error when i try using the create_sql_agent in the below code. toolkit = SQLDatabaseToolkit(db=db) agent_executor = create_sql_agent( llm=OpenAI(temperature=0), toolkit=toolkit, verbose=True ) ValidationError: 1 validation error for SQLDatabaseToolkit llm field required (type=value_error.missing) Can you please help here? Appretiate your help

[@rajivraghunathan9104]

Excellent info ,,, pls do one video in querying JSON Documents . No one has done any video

[@datasciencebasics]

Hello, Created one, hope you find it helpful :)

[@roberthuff3122]

Thank you. FUBAR! LOL

[@colofthedead6101]

Unfortunately, in reality this functionality is very early days and doesn't work very well, generating joins across non-existent columns and SQL that is engine specific. It's only good for playing around with at the moment.

[@datasciencebasics]

I agree, hopefully more functionality in the future as llms and usecases around it are quite in early stage.

[@colofthedead6101]

@@datasciencebasics I've persisted with this library and the more information about my database that I include in the prompt, the better the results. Less, denormalised tables help too. You can see the potential though and I've gotten some 'wow factor' from demoing it to management.

[@datasciencebasics]

@@colofthedead6101 Great findings. Eventually, everything in LLM ends in good prompt :)

[@mariocuezzo8027]

nice video! i try this with ggml-gpt4all-l13b-snoozy.bin and i have this problem : SQLQuery:The prompt size exceeds the context window size and cannot be processed. The prompt size exceeds the context window size and cannot be processed. any ideas?