Build Your RAG-based ChatGPT Web App with Azure: LawGPT Use Case Tutorial

No video

Build Your RAG-based ChatGPT Web App with Azure: LawGPT Use Case Tutorial

Рет қаралды 26,608

Күн бұрын

In this comprehensive tutorial, I'll guide you through the step-by-step process of creating your very own ChatGPT-like web application using Azure services. We'll leverage the power of Azure OpenAI Service, specifically the gpt-3.5-turbo model, and Azure Cognitive Search to build an intelligent information retrieval system.
🔗 Explore the code and sample data in the repository for a hands-on experience.
⚖️ We'll use the lawGPT use case to showcase its capabilities in handling law-based documents.
Whether you're a developer, AI enthusiast, or just curious about the possibilities of AI and Azure AI Services, this tutorial will provide you with the knowledge and tools to create your own customized AI-driven web app. Dive in and start building today! 💡🚀
Don't forget to like, share, and subscribe for more tutorials on AI, Azure, and technology!
Find the GitHub here: github.com/AIA...
Find the command used below:
Install Powershell 7+ (pwsh)
Important: Ensure you can run pwsh.exe from a PowerShell terminal. If this fails, you likely need to upgrade PowerShell.
Install Azure Developer CLI
$ winget install microsoft.azd
Login to Azure
$ azd auth login
Initialize and clone the repository
$ azd init -t azure-search-openai-demo
Provision Azure resources and deploy this sample to those resources, including building the search index based on the files found in the ./data folder.
$ azd up
#ai #generativeai #chatbot

Пікірлер: 76

@Harsh_Kaju 11 ай бұрын

I am really impressed with what this channel has become. Probably the best content on KZfaq right now 👍❤

@AIAnytime 11 ай бұрын

Glad you enjoy it!

@user-rd9dy9uw4n 11 ай бұрын

Thanks! Ai Anytime! AI Anytime Dive deep into the world of AI with meticulously crafted content that informs, inspires, and empowers. Stay ahead in the AI revolution with this essential guide

@heist7539 11 ай бұрын

was really waiting on this vid time to study

@AIAnytime 11 ай бұрын

Thanks. Hope you liked it.

@jorgerios4091 11 ай бұрын

As usual, always great stuff in your channel. I can't wait for the future open source versions!

@AIAnytime 11 ай бұрын

Glad you like them! A few open source videos of similar nature can be found in my LLM playlists.

@Ikkii94 11 ай бұрын

Hi, I was wondering, you put 3 files in the data folder as information for your app to explore as information, what if later, you want to add more documents, or else, some documents add more information/update information from previous documents (exemple: EUR money rate change every x time, so it needs update) how do you do that ? Do you put it all together in the data folder ? Is there a command line to do to update/recreate the database ? thank you for your video, very interesting

@kaspah 5 ай бұрын

azd up

@antonioribeiro1988 5 ай бұрын

Parabéns pelo trabalho! Só temos a agradecer por essa montanha de conhecimentos! Se você conseguir fazer um vídeo com esse propósito, porém com ferramentas open-source, seria muito bom! Parabéns!

@paritoshk990 7 ай бұрын

Great Stuff! Could you please also share how can we change/add the data after deployment?

@micbab-vg2mu 11 ай бұрын

Thank you for the video.I will try to build it and test the accuricy of resonses.

@AIAnytime 11 ай бұрын

Pls let me know your findings.

@Scooterboy_and_others109 11 ай бұрын

At 5:10 you mentioned about K=3 (number of documents to be scanned) If the answer can result from say 25 documents, how does it decide which of the 3 it should pick from 25? Is there any specific SORT logic it uses that can be parameterized?

@AIAnytime 11 ай бұрын

Very good point. Loved the intricacies here, Prasanna. Yes there is some sort of mechanism placed for example Similarity search or Max marginal relevance search algorithms. You can look at this link to get more clarity on search kwargs: python.langchain.com/docs/use_cases/question_answering/how_to/vector_db_qa

@ochanabondhu 11 ай бұрын

This is such a brilliant video. Thank you so much. Just one question. If I upload the blob storage manually with new files while the app is still running, will the new data be vectorized automatically?

@michaelgarrigan56 10 ай бұрын

Thank you so much for the video, it was incredibly informative. What is the best way to update the data later on down the line? Let's say a judge or lawyer asked for an additional set of case studies - how would you go about uploading them to the bot? Could you just add it to your blob storage instance created during configuration?

@iainhmunro 8 ай бұрын

Almost got it - it failed with the following mssage - Deployment Error Details: InvalidTemplateDeployment: The template deployment 'openai' is not valid according to the validation procedure. The tracking id is 'f845b724-4380-47ef-9e91-1d12cb67cc44'. See inner errors for details. SpecialFeatureOrQuotaIdRequired: The subscription does not have QuotaId/Feature required by SKU 'S0' from kind 'OpenAI' or contains blocked QuotaId/Feature. Any ideas ?

@yassinebakhti5590 2 ай бұрын

can you share this repo please the new one is a bit more complex

@ehteshamnehal7024 11 ай бұрын

What can be the cost I can expect to build this for production? (Assuming I have 100Gb of data)

@akki_the_tecki 9 ай бұрын

Did you get the answer?

@kevennguyen3507 8 ай бұрын

How can we add an extension to the app? for example, we want to ask users for feedback on the bot's answer, if it's helpful, accurate, and complete. We wanted to integrate that into the conversational aspect of the chatbot. After answer the question, the bot would ask user "would you like to rate my response?" and user would be selecting Yes or No. If yes, then the bot would ask if it's accurate, user would rate... at the end of the survey, all these will be logged into a spreadsheet (question, answer, rating...) for review.

@SnehaRoy-xf3zv 11 ай бұрын

Not slowing down 😂... Liked.

@AIAnytime 11 ай бұрын

Thanks

@user-uy4sm1ei9q 11 ай бұрын

Great stuff, I have been going through your videos recently and loved all of it. Question: can I opt for free tier or basic plan for Cognitive search by making changes to the bicep file? Will it support for less 50mb size? I understand the limitation, however still want to carry out POC with low cost.

@AIAnytime 11 ай бұрын

Thanks.... basic plans will be a great start! Less than 50mb, go ahead and build the PoC and show it to clients. Then scale it later....

@SuperLittle35 8 ай бұрын

So, did the basic plan work?

@CaesarEduBiz-lz2cg 10 ай бұрын

Basically Azure Cognitive Search acts like Llama Index in this tech stack?

@jatinkashyap1491 8 ай бұрын

Can it also process scanned text documents? Thanks for the content.

@SuperLittle35 8 ай бұрын

I have been working with Azure OpenAi and i wanted to confirm something. For every follow up question, do you have to send all the previous chats as well?

@pentall6791 10 ай бұрын

Thank you for this video. Azure is charging a lot of money using this method even during off hours and using it only for a few minutes. Can you help resolve this issue. TY.

@robertbriana4482 9 ай бұрын

Thank you for the video. Question, I'm not a coder so don't laugh but how do I add UI?

@datafakts 11 ай бұрын

How can you connect this to an Elasticsearch Index instead of using Azure Cognitive Search?

@AIAnytime 11 ай бұрын

Very good idea. I am working on something similar, will release that video soon.

@ParthPatel-vv1bg 8 ай бұрын

Hello Bhavyabhai I tried this steps, when I start deployment using " azd up " i got auth error while creating some services, some of them are created on my azure. can you please guide me, in code where exactly I find the service creation part? Thank you so much for this amazing informative video.

@dushankarawita2110 10 ай бұрын

Hi, impressive work. When I tried to execute "azd up" I'm getting the below error. ERROR: Could not build wheels for aiohttp, tiktoken, which is required to install pyproject.toml-based projects Any indicator to solve this would be appreciated.

@user-iu4id3eh1x 11 ай бұрын

Very helpful video.... thank you

@AIAnytime 11 ай бұрын

Glad it was helpful!

@user-qi4jw1lf9i 4 ай бұрын

Please tell me how to with aws mistral model

@JavierTorres-st7gt 2 ай бұрын

How to protect a company's information with technology ?

@Sebastian-di6sj 10 ай бұрын

So this is just for retrieving info? or can u make this also solve cases with similar situations?

@AIAnytime 10 ай бұрын

Can you explain in more detailed.

@Sebastian-di6sj 10 ай бұрын

@@AIAnytime If a user has a similar case, as in the databank. with maybe slight differences, can the databank then output a possible solution to the case based off lots and lots of previous data? Thanks a lot for your fast reply! Greetings from Austria:)

@akworld0017 11 ай бұрын

Awesome content as always - How to get rid of hallucinations in response?

@AIAnytime 11 ай бұрын

I didn't see Hallucination with the questions I had asked. This has been the best performance so far. Let me know if you see any hallucinations with this solution.

@razexamvs8756 11 ай бұрын

What to do if i ask this to simplify the pdf and download it again the simplified version

@AIAnytime 11 ай бұрын

You have to design the prompt that way and a file post-processing functions need to be added.

@akki_the_tecki 9 ай бұрын

bhayya, make a video dedicating to Microsoft Azure and its Payment Subsricption. Ppl are getting confused and We need to know is our data is safe or not

@AIAnytime 9 ай бұрын

Sure Akhil. Will create one.

@akki_the_tecki 9 ай бұрын

@@AIAnytime all of your videos are top notch bhayya 🙌🏻

@RedCloudServices 11 ай бұрын

Streamlit is so much simpler than all of these typescript and react files for an app

@AIAnytime 11 ай бұрын

Yes but sometimes you can't show a Streamlit app to a client. They demand more.

@kundetivamsi1079 11 ай бұрын

I had run this project with Azure student starter account ,but I was end up with "Failed: App Service plan:" error 😌😔

@AIAnytime 11 ай бұрын

This probably might not work with a student account. Let me do a double check anyway.

@futurebeing7927 11 ай бұрын

@@AIAnytime hey have you checked about student account? please let us know, that will be very helpful. thank you.

@siddharthghuge8505 11 ай бұрын

sir to create this project is it free to use or does it require open Ai api key

@AIAnytime 11 ай бұрын

It requires Azure access.

@MkmeOrg 11 ай бұрын

Very neat

@AIAnytime 11 ай бұрын

Thank you! Cheers!

@user-gg9ls2nk9k 11 ай бұрын

what to do if we want to extract exact data as in our document for any query?

@AIAnytime 11 ай бұрын

Temperature needs to be 0. Semantic search algorithms need to be tweaked. The retriever I mean.

@specializedboy6271 7 ай бұрын

awesome

@orlandocastellanos9263 11 ай бұрын

What if a want to use whisper in case the user wants to send an audio what can I do ?

@AIAnytime 11 ай бұрын

Changes need to be done in backend. Put whisper logic into the python file. On frontend, you would need an icon for the microphone or something.

@andreduval7359 11 ай бұрын

How can I contact you?

@AIAnytime 11 ай бұрын

Plz find my contact details or social media through my KZfaq banner or about us channel.

@walberc 11 ай бұрын

Cog search is to expensive

@AIAnytime 11 ай бұрын

Yes but Enterprises who are already on Azure cloud, they prefer this sort of solutions.... cost, performance and scalability all Trade-offs.

@orlandocastellanos9263 11 ай бұрын

Is Christmas already ?

@AIAnytime 11 ай бұрын

Haha 🤣... Gifted by father.

@fabianaltendorfer11 9 ай бұрын

are you on x?

@AIAnytime 9 ай бұрын

No, I am on LinkedIn and GitHub. Don't use X which spreads fake news.

@vladimirshapran7286 Ай бұрын

Hello good sir, your video helped me a lot, but have you separatly instlled npm, cause I am getting this error: Packaging services (azd package) (x) Failed: Packaging service backend ERROR: error executing step command 'package --all': failed packaging service 'backend': failed invoking event handlers for 'prepackage', 'prepackage' hook failed with exit code: '1', Path: 'C:\Users\volsh\AppData\Local\Temp\azd-prepackage-4040297706.ps1'. : exit code: 1, stdout: , stderr: npm: C:\Users\volsh\AppData\Local\Temp\azd-prepackage-4040297706.ps1:4 Line | 4 | cd ../frontend;npm install;npm run build | ~~~ | The term 'npm' is not recognized as a name of a cmdlet, function, script file, or executable program. Check the | spelling of the name, or if a path was included, verify that the path is correct and try again. Thanks and have a good time