Can we get one more video on this? Like deep diving on the code and how we can just deploy it with our own changes? Also a more detailed look on the embeddings model here. I saw that the model is just splitting the data and indexing them.
How can i use this as an Api if in my website if i have my files? Like do i have to give file on runtime or do ingestion at playground then deploy it then use it as an api?
@@Dewiride Can you please tell me if we clone the repo, make changes to the code and redeploy with the same docker image as given in the repo, will the changes reflect ?
When I try to send a post request with the "custom_prompt" and "custom_temperature" parameters set, I get an error. It works fine with the question and history parameters. I formatted it like this: { "question": "example question", "history": [ [ "example history" ] ], "custom_prompt": "example prompt", "custom_temperature": 0.5 }
What is the average cost for the azure setup given and how much more expensive would it get for a database using 100gb of data? currently azure cognitive search is the most expensive aspect while the blob storage and gpt features are the cheapest. I want a better understanding on pricing/scale since the website price sheet isnt' very helpful..
For cognitive search - it will be $981/month assuming 100 GB Storage + $15 for blob storage assuming 100 Gb + $50 for form recognizer assuming 1000 pages + $72/month for app service assuming production workload. This is approx. Actual cost may wary. For cognitive search, the index size is less compared to the actual raw file size. So, you can experiment with lower pricing tier and slowly upgrade if requirement increases.
I have discovered your channel through the earlier video you did about azure open ai your own data. Great content and smooth explanation. I subbed. Can you please help me solve the 429 error in azure. I did all the steps you did but it sometimes gives me 429 "exceeded quota" or something and other times 404 "resource not found"
Exceeded quota is given if you try to create resources more than the available quota. For example - Azure openai resource currently can only be created once per subscription. If you already have 1 azure openai resource and try to create another one. You will get quota error. Also, you can only have 1 free cognitive search resource per subscription. If tried creating more, it will ask you for the paid resource and give you the error. Hope you don't end up in this scenarios.
@@Dewiride Thanks for your response. I do not have any other resources other than a single OpenAi service + a cognitive search resource + a storage account. I even deleted all of these and created new ones from scratch. I also tried to upload a simple short word document. Now I have 9$ of usage but I did nothing with it lol
Yes, it is not about how many docs you have, it is about the pricing tier your cognitive search is running on. It will keep on billing you if not using as well. You could select free tier and experiment.
I did that. Turns out you cannot call the api from the browser until you turn on CORS. The CORS error can be resolved by removing azure.websites from the white list and adding an *. Which allows the api to be called from the browser. Can you show how to make this bot more responsive as sometimes it crashes after multiple queries. I want to scale it so that it can be used by multiple users at the same time. Almost on an enterprise level.
Anybody can tell me which are the pros/con about using Azure Cognitive instead of consuming direct from openAI? It seems like it is most complex, but I know must be a bunch of reasons to do that. If anyone can help me understand that I would be greatful! thank you for the video!🤓
Only reason - Storing of either text/vectors in index. If you do not have huge data, just a paragraph or a page within a limit of OpenAI's token limit, then you can directly query OpenAI. But, everytime a query comes, you need to pass on your data.
@@Dewiride Humm, thank you! I am building using -> Flowise (Langchain visual) + Pinecone (store vector data) . In fact I will need to use a huuuuge amont of data, so I am searching the best way to do that...
Finally, this is super helpful. This outline is far better than any other tutorials regarding azure open AI that are available right now!
Glad it helped!
Can we get one more video on this? Like deep diving on the code and how we can just deploy it with our own changes? Also a more detailed look on the embeddings model here. I saw that the model is just splitting the data and indexing them.
i am getting error LLM not working please help and trackback is not define
Make sure the models you selected during deployment are deployed in azure openai studio.
How can i use this as an Api if in my website if i have my files? Like do i have to give file on runtime or do ingestion at playground then deploy it then use it as an api?
We have other videos, how to connect Azure OpenAI API with data into your application
How do i change the frontend and add authentication
Clone the repo and make changes to it. Then redeploy.
@@Dewiride Can you please tell me if we clone the repo, make changes to the code and redeploy with the same docker image as given in the repo, will the changes reflect ?
When I try to send a post request with the "custom_prompt" and "custom_temperature" parameters set, I get an error. It works fine with the question and history parameters. I formatted it like this:
{
"question": "example question",
"history": [
[
"example history"
]
],
"custom_prompt": "example prompt",
"custom_temperature": 0.5
}
What error you get?
@@Dewiride I get a 500 Internal Server Error
500 Internal Server error means have issues with the source code. It might also get the stack trace if developing locally.
Can you create this demo in c# ?
like how to upload files from local nd convert into vector embedding
it will be very helpful
What is the average cost for the azure setup given and how much more expensive would it get for a database using 100gb of data? currently azure cognitive search is the most expensive aspect while the blob storage and gpt features are the cheapest. I want a better understanding on pricing/scale since the website price sheet isnt' very helpful..
For cognitive search - it will be $981/month assuming 100 GB Storage + $15 for blob storage assuming 100 Gb + $50 for form recognizer assuming 1000 pages + $72/month for app service assuming production workload.
This is approx. Actual cost may wary. For cognitive search, the index size is less compared to the actual raw file size. So, you can experiment with lower pricing tier and slowly upgrade if requirement increases.
I have discovered your channel through the earlier video you did about azure open ai your own data. Great content and smooth explanation. I subbed. Can you please help me solve the 429 error in azure. I did all the steps you did but it sometimes gives me 429 "exceeded quota" or something and other times 404 "resource not found"
Exceeded quota is given if you try to create resources more than the available quota.
For example - Azure openai resource currently can only be created once per subscription. If you already have 1 azure openai resource and try to create another one. You will get quota error.
Also, you can only have 1 free cognitive search resource per subscription. If tried creating more, it will ask you for the paid resource and give you the error.
Hope you don't end up in this scenarios.
@@Dewiride Thanks for your response. I do not have any other resources other than a single OpenAi service + a cognitive search resource + a storage account. I even deleted all of these and created new ones from scratch.
I also tried to upload a simple short word document. Now I have 9$ of usage but I did nothing with it lol
Yes, it is not about how many docs you have, it is about the pricing tier your cognitive search is running on. It will keep on billing you if not using as well. You could select free tier and experiment.
@@Dewiride i will thanks for the follow up
how can i create a pop up bot on my custom website?
like a qna bot who pops us on the website
You can use any frontend technology and call api. Or, for easy, use bot framework and connect to webchat.
I did that. Turns out you cannot call the api from the browser until you turn on CORS. The CORS error can be resolved by removing azure.websites from the white list and adding an *. Which allows the api to be called from the browser. Can you show how to make this bot more responsive as sometimes it crashes after multiple queries. I want to scale it so that it can be used by multiple users at the same time. Almost on an enterprise level.
awesome. You are doing a great job!!
Thank you so much!
Anybody can tell me which are the pros/con about using Azure Cognitive instead of consuming direct from openAI? It seems like it is most complex, but I know must be a bunch of reasons to do that. If anyone can help me understand that I would be greatful! thank you for the video!🤓
Only reason - Storing of either text/vectors in index. If you do not have huge data, just a paragraph or a page within a limit of OpenAI's token limit, then you can directly query OpenAI. But, everytime a query comes, you need to pass on your data.
@@Dewiride Humm, thank you! I am building using -> Flowise (Langchain visual) + Pinecone (store vector data) . In fact I will need to use a huuuuge amont of data, so I am searching the best way to do that...
@@othon.rodrigues even that is also a way to go if not working on Azure Stack.
very nice!
Thanks!