Our RAG live course is coming up soon, and as a way of giving back to our amazing community, we're offering you 15% off. Just use this link: maven.com/angelina-yang/mastering-rag-systems-a-hands-on-guide-to-production-ready-ai?promoCode=TwoSetAI We'd love to see you there! 🎉 In the course, you'll have the chance to connect directly with Professor Mehdi (just like I do 😉 in the videos), and you can even ask him your questions 1:1. Bring your real work projects, and during our office hours, we'll help you tackle your day-to-day challenges. This course is for: 01 👇 𝗔𝗜 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝘀 & 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝘀: For AI engineers/developers looking to master production-ready RAG systems combining search with AI models. 02 👇 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝘁𝗶𝘀𝘁𝘀: Ideal for data scientists seeking to expand into AI by learning hands-on RAG techniques for real-world applications. 03 👇 𝗧𝗲𝗰𝗵 𝗟𝗲𝗮𝗱𝘀 & 𝗣𝗿𝗼𝗱𝘂𝗰𝘁 𝗠𝗮𝗻𝗮𝗴𝗲𝗿𝘀: Perfect for tech leads/product managers wanting to guide teams in building and deploying scalable RAG systems
Thanks! We have a RAG live course coming up soon, come join us and bring your work projects and we'll help you tackle any of your obstacles! maven.com/angelina-yang/mastering-rag-systems-a-hands-on-guide-to-production-ready-ai?promoCode=TwoSetAI
Indeed, thank you for this introduction. Looks like you don't need a GPU to do the embeddings when "training". Very timely for my POC, much appreciated.
Just curious, at 15:15, for the user question "what are the top 3 cities with the most customers" did Vanna formulate the SELECT based the DDL & DML that is in the vector store on its own or was there a view that already existed that it used. If I had just tables and I loaded that schema in the vector store would Vanna be able to figure out the joins? That would be impressive. Of course, I could experiment myself but thought I would ask.
My only question on this is about the Data Sceurity. Exposing database directly to LLM might be risky. As we have seen many times that certain prompts can some time leak crucial data. So LLM having all the access to the DB without Row Level Security or in this case, any kind of security will be a big big risk to the Organizations
That's a good point. Of course security is a big deal in every company. There are multiple solutions. One is to have your own LLM, rather than using gpt4, etc.
Thanks for this. How about more complex/indirect questions that require more reasoning over the schema beyond standard defined relationships for example: 1. Tell me something interesting about my process/data 2. Which of my employees have shown the best the best collaboration practices and what are they doing right?
By training they are referring to RAG in docs. Vanna is not a separate model, it's a implementation, which means depending on the model you are using you will get varying results. 'sqlcoder' is one such model. You will have to benchmark it urself.
have you gotten an answer for this? I work with proprietary data so just making a call to an endpoint of any kind on commercial internet is a NO for me.
2 minutes of forced advertising with no ability to skip is horrid and not on anything else I watch. If I see this again, I will never come back to this channel.
We are sorry to hear that, it’s probably driven by platform algorithms. We can’t access or make any changes from our end as far as we understand, but tips welcome! Thank you!
I gave it a shot, but unfortunately, it didn't perform well. I wanted to know the number of orders in the database, but it tried to query SELECT COUNT(*) FROM orders instead of SELECT COUNT(*) FROM public.order. It seems it's not fully reliable yet.
Our RAG live course is coming up soon, and as a way of giving back to our amazing community, we're offering you 15% off. Just use this link: maven.com/angelina-yang/mastering-rag-systems-a-hands-on-guide-to-production-ready-ai?promoCode=TwoSetAI
We'd love to see you there! 🎉
In the course, you'll have the chance to connect directly with Professor Mehdi (just like I do 😉 in the videos), and you can even ask him your questions 1:1. Bring your real work projects, and during our office hours, we'll help you tackle your day-to-day challenges.
This course is for:
01 👇
𝗔𝗜 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝘀 & 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝘀: For AI engineers/developers looking to master production-ready RAG systems combining search with AI models.
02 👇
𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝘁𝗶𝘀𝘁𝘀: Ideal for data scientists seeking to expand into AI by learning hands-on RAG techniques for real-world applications.
03 👇
𝗧𝗲𝗰𝗵 𝗟𝗲𝗮𝗱𝘀 & 𝗣𝗿𝗼𝗱𝘂𝗰𝘁 𝗠𝗮𝗻𝗮𝗴𝗲𝗿𝘀: Perfect for tech leads/product managers wanting to guide teams in building and deploying scalable RAG systems
Thanks for this introduction. This is evolving rapidly, which is great, but it's hard to keep up. :)
You two are doing great job! Keep sharing your knowledge
salut ! Thx for all your vids, i'm french and i want to use rag in research, i learn a lot with you.
Thanks! We have a RAG live course coming up soon, come join us and bring your work projects and we'll help you tackle any of your obstacles! maven.com/angelina-yang/mastering-rag-systems-a-hands-on-guide-to-production-ready-ai?promoCode=TwoSetAI
Indeed, thank you for this introduction. Looks like you don't need a GPU to do the embeddings when "training". Very timely for my POC, much appreciated.
Just curious, at 15:15, for the user question "what are the top 3 cities with the most customers" did Vanna formulate the SELECT based the DDL & DML that is in the vector store on its own or was there a view that already existed that it used. If I had just tables and I loaded that schema in the vector store would Vanna be able to figure out the joins? That would be impressive. Of course, I could experiment myself but thought I would ask.
My only question on this is about the Data Sceurity. Exposing database directly to LLM might be risky. As we have seen many times that certain prompts can some time leak crucial data. So LLM having all the access to the DB without Row Level Security or in this case, any kind of security will be a big big risk to the Organizations
That's a good point. Of course security is a big deal in every company. There are multiple solutions. One is to have your own LLM, rather than using gpt4, etc.
@@MehdiAllahyariThay is not a problem today. And that is the essense of RAG - you can integrate with local, Open source LLM.
The keypoint in this demo is the pertained model using gpt-3.5 and must be online.
how can i keep
the data base in synchronized with my LOB app say ERP
How to chunk it, can you share how to do with local LLM and SQL server database
is the .ipynb files from this video sourced somewhere for use?
Here's the link to the code:
github.com/mallahyari/twosetai/blob/main/05_sqlite-openai-vanna-vannadb.ipynb
Congrats on the content! 🇧🇷 🇧🇷 🇧🇷 🇧🇷
Thanks for this. How about more complex/indirect questions that require more reasoning over the schema beyond standard defined relationships for example:
1. Tell me something interesting about my process/data
2. Which of my employees have shown the best the best collaboration practices and what are they doing right?
Good job. If documentation calls ‘train’ loosely, I get disappointed. How good or bad it is compared to sqlcoder?
By training they are referring to RAG in docs. Vanna is not a separate model, it's a implementation, which means depending on the model you are using you will get varying results. 'sqlcoder' is one such model. You will have to benchmark it urself.
Great job 🎉🎉🎉
Can integrate with local open source LLM using llama?
have you gotten an answer for this? I work with proprietary data so just making a call to an endpoint of any kind on commercial internet is a NO for me.
@@atrocitus777 not yet
Yes you can. They support Ollama, Hugginface models. We have done this integration already. If you want, I can help you with ur doubts.
I am working on the SQL chatbot and have used Vanna AI and trained my own model, sorry to say but response are not that good and as expected.
Do you mind sharing more details please ?
1. LLM
2. SQL DB
3. Vector DB
Hi am building same lol, am noob can u share more what u have done
Excellent. An extension of this would be RAG to SPARQL.
That would be amazing.
Hi can I get training documentation which you mentioned in video
Sorry for late response. Here's the repo of all the codes so far: github.com/mallahyari/twosetai
Excellent. Loved it.
Does this mean data analysts will be out of job?
Probably not in the near term, but it should be able to empower DS/DAs' work. If you are in DA/DS roles, we recommend trying these tools out!
wow, 2nd ad is 2:20 long. GOODBYE
😂😂😂
What a thumbnail
2 minutes of forced advertising with no ability to skip is horrid and not on anything else I watch. If I see this again, I will never come back to this channel.
I am not sure what you mean by forced advertising! but we never added any ads to the video! It could be TH-cam or Google!
We are sorry to hear that, it’s probably driven by platform algorithms. We can’t access or make any changes from our end as far as we understand, but tips welcome! Thank you!
I gave it a shot, but unfortunately, it didn't perform well. I wanted to know the number of orders in the database, but it tried to query SELECT COUNT(*) FROM orders instead of SELECT COUNT(*) FROM public.order. It seems it's not fully reliable yet.
You can fine-tune the SQL within Vanna as per your needs. Basically, you can train the model with custom sql for your needs.
Is there code to this tutorial?