There are no words to say thanks for what you've given to the developers community. This video and the code provided has helped me a lot and saved tons of hours researching. Sincerely.
Thanks for making this video. So what is the probability that it won't hallucinate . Well I am using similar kind of prompting approach when using llama model but curious to know about hallucination.
It's difficult to estimate hallucinations due to so many varying factors. However I'll just say anecdotally that the results for me have either generally been correct or just not worked at all (the SQL query doesn't work) - I'd have to setup a more rigorous test plan to estimate hallucinations better.
@@alexthecodewolf I took one approach by first storing the table in the form of JSON and then asking the question from the JSON. but again the things didn't go well . Is it I have to do something with the prompt?
Thanks for this video. It is really helpful; I am wondering if the tokens get consumed when we supply the chat history. What happens if the tokens run out (the limit set by each model)? How can we clear the history when we know that the tokens have ran out?
Thanks! Chat history does count towards token history - anything you pass to the AI generally counts as tokens. The larger your token limit for a given model, the more history you could send, for example GPT-4 32k runs out faster than GPT-4 128k. Luckily the chat history is just a standard list, so whenever you send a prompt you can update your history list to only include the X number of most recent messages, whatever you think will fit in the token limit for your model. I believe the AI model auto-truncates the tokens provided and just ignores the excess but I'd have to double check on that. You could also store all the messages in one list for the user and for display purposes, but then just take the X most recent number of messages to actually send to the AI.
Thank you for these great videos, however it's not working for me. I downloaded the github repo and compiled it successfully, but when I test WebAssembly, I get the error "Upload failed. Error: Object reference not set to an instance of an object." Server Static and WebAssembly both throw the error "Could not find a part of the path." None of the options work.
Thank you for alerting me of these issues - I was able to reproduce them from the GitHub repo. For the Server Static and Server Interactive, the issue was that there's no images folder included in the project by default so the file path for the upload doesn't work - I've added that folder and committed it to the repository. For the webassembly page, there was a missing handler method in the version on GitHub, so I updated that with the latest. these should all work now.
Hi Alex, good video, but Azure is killing the container and so the pot model disappears a few minutes later, do you know how to solve that issue? thanks
Hey, thank you - are you referring to the video that shows you how to chat with a database? ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-hw6oTjw9_Ro.html
@@alexthecodewolf Well , I am looking as how you have connected database using ollama in this video which starts from 10:20 sec. Do you have a github repository for this
Hi! Thank you so much for this! I'm trying to work on a demo for my company where I want to read a table of incidents and summarize the resolution note of each one and send back to Service Now. The incidents table is already a view on our Azure enviroment, so I think I need first to get the query from AI, get the data from the table and then ask again the AI to iterate over the retrivied table and summarize the notes, structure the data again with the summarized notes and send back to Service Now via API maybe. Do you think this architecture makes sense? Really appreciate this knowledge shraring!
@@AbhinavKumar-tx5er Hi, yes! I've getting some pretty good results with the GPT-4o. I followed this tutorial to adapt to my case: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-9ccl1_Wu24Q.html
Excellent video. But unfortunately I'm getting very inconsistent results. As soon as a little bit more complicated SQL needs to be produced - every run returns a different result. Subqueries, unions are too much for it. Or I'm doing something wrong. Probably will have to create views so AI will be able to do simple selects.
Hello sir this is very help full video , i have doubt can i used the 500 tables and then we can used for this retrieving the SQL Query. because i am using the SQL Database so i give the error of Rate Limit Error 429 , where token limit is 60000 and i request the token for the 157656 so its give the error because the lots of table it taken can you give me solution for this, it is very helpfull for me
I believe current gpt models are limited to 128k tokens - I would recommend either breaking apart your schema to only the tables you need if possible, or searching for an AI model with higher token limits. There might also be local or self hosted models for code generation that support higher token inputs. I have some other videos about local ai models that show this database connection scenario
Thanks -2024 update - the method services.AddStackExchangeRedisOutputCache is now in the namespace Microsoft.AspNetCore.OutputCaching.StackExchangeRedis not the one specified in the video
Hey, I probably should have mentioned you have to apply for openai access, but it's fairly easy to get approved. If you can't get access you can also substitute the AI service with gpt-4 similar alternatives like standard openai instead of azure openai, or even local ai models that are specifically trained for code generation
@@Ajmal_Yazdani You can find some examples in the semantic kernel docs at learn.microsoft.com/en-us/semantic-kernel/agents/plugins/using-the-kernelfunction-decorator?tabs=Csharp I would recommend using semantic kernel for local AI dev. Semantic kernel and more involved AI flows with local functions, plugins, services etc is a topic I plan on covering here in the near future. Semantic kernel is a big topic so I'm working on the best way to present tthat.
Why don't you do a complete Blazor course on Pluralsight. Learning blazor in bits is very difficult. I am new to blazor. If you already have the course, please point me in the right direction. Thank you.
Hey, thanks for the comment, there's already complete blazor courses on pluralsight by another author so I wouldn't be eligible to do that. I have thought about making one on RU-vid here of some sort but full courses are very time consuming and already exist on here as well, maybe at some point
@alexthecodewolf Yeah. But you don't get to learn all of it. Learning bit by bit on different RU-vid channels is difficult. Also, bootcams are way out of budget. I am from India, and platforms like Pluralsight help us with the volume of courses and also our pockets. I think this dilema will always remain. Thank you for your reply.
@@JtendraShahani Hey, you are correct that this is an ongoing challenge. As a self taught developer, I can relate. Many of the videos on this channel are oriented towards learners who already know the basics but are looking for guidance around new or specific topics so they don't have to watch additional courses, they can just supplement their knowledge on specific topics in shorter form content. However, I would like to get to the point where I have the time or incentives to create full courses on RU-vid. Thanks for sharing your thoughts on this!
Hey, yes it can work with local DB and on prem. You would just replace the connection string placeholders in the demo app with localdb connection strings. You could then use the same Azure AI service or you could use a local AI model - I just posted a couple videos explaining how to setup and connect to local AI if that's what you're looking for.
Have you tried to connect your local LLM to talk to an on-premise network database without providing the schema in the instructions? Im wondering how well it would function or if this is even feasible. Asking for a friend :)
Also, I want to add that this is great content. Useful real-world examples like this will be golden as more people leverage AI in the enterprise landscape.
Thanks for the feedback! I'm not sure if I understand your original question though - are you asking about essentially just giving the AI a connection string or something so it can access the database directly?
@@justinkrawczak4929 As far as I know it's not possible to just give an AI a connection string and have it understand/explore a database. I think the closest solution I've seen to this is to add some kind of layer between the AI and the database, such as using a RAG solution where you connect the AI to a search service that has crawled your database, or using vector search capabilities to provide the AI with database data. I have a couple videos on my channel that explore these types of topics, but I don't know of any "Direct access" solutions.
Hi @Thecodewolf, Could you please explain how to set up a local laptop to run the .NET code? As I'm not a coding expert, any useful suggestions you can provide would be greatly appreciated. If you can outline the necessary steps, it would make it easier for us to follow along and perform the demo ourselves.
Hey, thanks for watching and providing feedback. This tutorial made some assumptions about coding knowledge with .NET or other languages, but I probably should have mentioned that up front. If you want to get started with .NET to recreate the code sample you can download visual studio here, which is the code editor I'm using that also installs .NET for you: visualstudio.microsoft.com/ You can then either clone the sample app from GitHub and just open it up in Visual Studio and follow along with the tutorial, or you can create a new project using Visual Studio and copy in the code. File -> New Project -> C# console app. Hope this helps a little bit!
I actually just made a video about this using Ollama, hah. It shows this same setup at the end using codellama ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-177qX6mpyMg.html The concepts in this video should work with any model that is skilled at text to code (such as code llama) and a SQL database of some sort. You'll have to experiment with different models and prompts to see which works best.
The AI stuff is only available to a few selected partners, the signup form even states that if you are not a part of the select group you will be denied. Trash.
Big thanks for this great video. I'll be interested in the Aspire but mostly interested in ways to reduce spending too much time catching technology rather than producing. Meaning Applications Builder rather than technology while the App Bilder will take care of the underline technology changes/upgrades.
Hi Code Wolf. Just a quick note, I really appreciate you parsing out the relevant Build 2024 info for .Net devs. Watching Build is like standing in front of a firehose of AI, so thank you very much for a .Net dev oriented summary. Thumbs up & I'm now a new subscriber. Rock on! 👍🏻😎👍🏻
Thanks for this specific feedback - this is actually why I made the video. There was actually a good amount of .NET content at Build but it was sort of overshadowed by all of the AI, Azure, and tooling integration (copilots/analytics) news.