Very nice explanation. Short & easy to understand. It helped me understanding how AI search works.
@ganirban826 күн бұрын
Nice session.. Can we boot up that vector - qardent db in windows 10 - os environment with docker sw ? Can we connect qardent - docker service with paython code in locally setting up ?
@akankshamakkar853816 күн бұрын
So here we need 3 services, right? azure openai, azure ai search and azure blob storage?
@HazemAzim18 күн бұрын
Great explanation and simple . Good that you focused on search without the hurdles of CosmosDB
@Melophile_Gaurav22 күн бұрын
Hey, this was helpful! :)
@nosa-osazeeomoruyi655227 күн бұрын
Thanks...this simplified langchain. but code is not in the shared github link.
@ambarishg26 күн бұрын
Thanks for the comment. I have updated the GITHUB code link in the video.
@nilesh859527 күн бұрын
This is perfect sir, you are already ahead of industry implementation era.. great stuff..
@shivaprasad1277Ай бұрын
I have created an index for a CSV file and i had few custom fields and also i have enabled Semantic reranker for the AI Search. Now i want to use this in my code but unable to get any fruitfl results. Can you please help ? And I am using Langchain as orchestrator
@kaustavroy6361Ай бұрын
Very informative.
@Momentum_Option_BuyerАй бұрын
1:01 Is Azure search is or acts like a vector database that keeps the knowledge? Please clarify I have seen several videos but confused on this question.
@Momentum_Option_BuyerАй бұрын
You didn't use vector database in this specific video (use case) to check data into vectors?
@ambarishgАй бұрын
Please watch the following for VECTORS 1. kzfaq.info/get/bejne/eLWAictelbSxaIk.html - Vector , Hybrid, Semantic search with Azure AI Search in 11 minutes 2. kzfaq.info/get/bejne/p7CcZqyVqLHKpnk.html -RAG With Azure AI Search using Vector/ Hybrid / Exhaustive KNN / Semantic Reranker in 15 minutes
@gowrinagh896Ай бұрын
thanks for the info, great explanation
@saideepj6998Ай бұрын
Thanks for such a detailed content. Is the uploaded data going to private ? will that data be trained with open source LLM's . How to add extra layer of security to make sure that the uploaded data is within my private GPT
@shashanksagar4878Ай бұрын
Do we not require any embeddings and vector database here? Also, is vector search not required? If so when is it required? Thank you
@mirakamali-jb1szАй бұрын
can you do one between llama-index and azure ai search?
@jeevajilifeАй бұрын
Thank you for this lovely video. I am interested in how you created those indexes within Azure AI Search. What if I have data in json format?
@KenGarff-zp1mhАй бұрын
Thanks for the video!
@abhinavanand6660Ай бұрын
hey in the section of env where could i find the index name of the ai search
@ambarishgАй бұрын
index = "<SEARCH SERVICE INDEX NAME>" This is in github.com/ambarishg/AZURE-AI-SEARCH/blob/main/.env.sample
@pooblock40922 ай бұрын
Does this work with files with size large than 16mb?
@learnwithengineer662 ай бұрын
below is my vector store creation code without using RBAC but if i want to use RBAC what should i have to change in below code vector_store: AzureSearch = AzureSearch( azure_search_endpoint=os.environ.get("SEARCH_ENDPOINT"), azure_search_key=os.environ.get("SEARCH_API_KEY"), index_name=index_name, embedding_function=embeddings.embed_query, )
@RamyahManoharan2 ай бұрын
Really nice stuff to start with.. than you
@abhilashnamdeo81962 ай бұрын
Is it acceptable to perform all these tasks directly from the portal, using the no-code option?
@draco.the.voyager2 ай бұрын
Useful stuff. Thanks for the video!!
@pjIIIEYE2 ай бұрын
Do you mind sharing the repo for this talk ?
@ambarishg2 ай бұрын
github.com/ambarishg/AZURE-AI-VECTOR-SEARCH
@user-ik6rg4mm6h2 ай бұрын
Do you know how can we retrieve access token from AWS secret manager instead putting it in profile.yml ?
@revischea71143 ай бұрын
When I do Vector/Hybrid search, the content that is returned are just references. I want it to answer the question also. Am I doing something wrong?
@UmairMateenKhan3 ай бұрын
Awesome tutorial Ambarish. Keep up the good work.
@AnkurKumar-xh6yw3 ай бұрын
A quick question....Using a single project ....can we write tables in different catalogs in databricks? I see that catalog is defined at a project level and wondering if we can use different catalogs for a single project
@subhrangshudas55843 ай бұрын
It's really helpful. I'm facing a problem local image in accessible through localhost, but pods are not accessible through local host. does the port need to be same a local image?
@neerajr15823 ай бұрын
Hi Sir any tutorial for recommendation service using azure search
@mayanksingh53753 ай бұрын
Love that you were recording the video just a few minutes before a new year😅
@swatishiriyannavar30653 ай бұрын
Can you do data drift for LLmops on azure
@tintintintin5763 ай бұрын
thank you so much for this. this sis very helpful. am gonna modify this a little in order to add to my portfolio. big thanks again , sir.
@ambarishg3 ай бұрын
Glad it was helpful!
@tintintintin5763 ай бұрын
@@ambarishg 💚🙏
@ratnasagar96933 ай бұрын
Hi bro please respond
@ambarishg3 ай бұрын
Connect me on LinkedIn
@creativeideas20734 ай бұрын
i dont know weather it is right to ask or not ,can you please tell keys in microsoft azure about this project for my final year project
@creativeideas20734 ай бұрын
i wanted to know that give text and images as input for this project at a time please help me me also doing similar project in jupyter notebook,but we are unable give new images as input only predefined images it takes. can you help me more
@creativeideas20734 ай бұрын
bro how can i contact you bro for this project more details
@user-wf2uy5wi7i4 ай бұрын
service_endpoint = os.getenv("AZURE_SEARCH_SERVICE_ENDPOINT") index_name = os.getenv("AZURE_SEARCH_INDEX_NAME") api_version = os.getenv("AZURE_SEARCH_API_VERSION") key = os.getenv("AZURE_SEARCH_ADMIN_KEY") aiVisionApiKey = os.getenv("AZURE_AI_VISION_API_KEY") aiVisionRegion = os.getenv("AZURE_AI_VISION_REGION") aiVisionEndpoint = os.getenv("AZURE_AI_VISION_ENDPOINT") Sir, couldn't find the service_endpoint, index_name,api_version ,key Can you please help me out
@BalaMurugan-cm6ev4 ай бұрын
Thanks for the awesome video. When we hit the cognitive Search using langchain. do we get the semantic reranker results? for RAG
@limjuroy70784 ай бұрын
Here is my little 2 cents', can you do an end-to-end tutorial on how you do this, including creating Azure AI Service, Azure Storage as well as the coding part? 😅😅😅
@ambarishg4 ай бұрын
Great suggestion!
@xyz-vv5tg3 ай бұрын
Please. I need it. I have no idea on Azure or any of its services. I'm just exploring and getting confused
@JohnPatrickGallagherАй бұрын
I’ll second this, I’m getting Started and it’s a great tutorial so far!
@limjuroy70784 ай бұрын
What if I want the app "also" has the ability to upload our own PDF file and query the content of the just uploaded file through UI?
@ambarishg4 ай бұрын
Yes it is possble
@Rajdeep64524 ай бұрын
Man I am tired of these half-baked videos
@Tiger-Tippu4 ай бұрын
Hi Ambarish ,please let me know how to get these codes
@ambarishg4 ай бұрын
github.com/ambarishg/dp100
@Tiger-Tippu4 ай бұрын
@@ambarishg thank you
@murtuza.chawala4 ай бұрын
Awesome Video I tried it too wanted to know any way we can do reinforcement learning making the bot train on our feedback
@ASLI-nh3zb4 ай бұрын
Does scoring not seem to work when using hybrid search?
@indrayne18405 ай бұрын
Sir I have one doubt, like we are using Azure Cognitive search and the index updates everyday midnight. So we have both new content and old content with us. Now what I want is to retrieve the most recent content first and than old content. How can I implement it?
@shoaibpatel26315 ай бұрын
Very informative video, please let me know the version of Azure search document library used?
@ambarishg5 ай бұрын
The latest Azure Search Document library [ azure-search-documents ]
@oliva828218 күн бұрын
I guess is 11.4.0 then, right?
@SamSan.085 ай бұрын
Have you created those fields( filename, line and embeddings) in the azure index manually or through code(python)
@ambarishg5 ай бұрын
This is done thru code. We have demonstrated of how we have done it in the video. Thank you
@SamSan.085 ай бұрын
Is that possible to Covent every column in CSV file to fields in azure without writing each column name in the code to convert it into field. Just we need to input a CSV file and it's columns shd be converted into fields in azure
@WinstonCodesOn5 ай бұрын
What advantage do you get from using the bi_encoder that is different from the basic vector hybrid search used in your previous video [RAG with Azure AI Search and Azure Open AI](kzfaq.info/get/bejne/jLF0osSJqbqXYHU.html)?
@ambarishg5 ай бұрын
Thanks for the question. The concept of the basic vector search is the same as the bi_encoder. While the HYBRID Search is combining Text Search and the Vector Search .
@WinstonCodesOn5 ай бұрын
@@ambarishg So is this just showcasing another coding method of implementing the same kind of search algo?
@WinstonCodesOn5 ай бұрын
It would be nice to have a video that goes into more detail about the quality of results in vector vs hybrid vs E-KNN. Also, why use Langchain? I'm unclear what it's doing that is better than makign the API calls directly.
@ambarishg5 ай бұрын
Ans 1 : Great suggestion! Ans 2 : For Langchain used ConversationalMemory which can be coded also but having a good tested framework helps to implement ConversationalMemory easier
@pulkitthapar2 ай бұрын
The sequential API calls to retrieve and then generate the response will result in a higher turn-around time. Langchain runs on LCEL that allows "runnables" or chains to run parallelly, achieving the same functionality but improving time. Also, it'll be easier to implement context-aware retrieving via Agent Tools using the framework. The only downside I see right now is a bug in the community integration for Azure AI Search that doesn't allow Hybrid or Semantic-Hybrid searches on a custom schema.