Related Posts
Jummah Mubarak Fam ☪️💚
Additional Posts in Data Science and Machine Learning
Hey Folks,
Help me to unlock DM!
Thanks.
New to Fishbowl?
Download the Fishbowl app to
unlock all discussions on Fishbowl.
unlock all discussions on Fishbowl.




Hi,
I am currently working on AWS bedrock and pretty much ready for deploying the app in the prod.
We are using AWS RDS and Open search for vector database.
We found RDS is performing better with our RAG and also we don't have much options as per company policy to move to Pinecone or Redis.
RDS can also be used as vector databases and there are other options as well such as DynamoDB.
The advantage with Open search is you don't need to explicitly create the columns/indexs for your Embeddings. When you create a KnowledgeBase in bedrock and choose open search as a Vector DB - The Kb itlsef will create all the indexes and necessary columns in open search.
This makes life more easy when you even have Metadata files in your KB along with your original documents.
Also RDS uses euclidean distance which is working pretty good in our use case. I can't vouch it for other use cases.
Pinecone
Any particular reason to use it over others?
Faiss: because it works great with cosine similarities.
We have that in ec2. We host it there.
Azure search
Databricks Vector DB.
Our team has expertise on Databricks, hence, it was our first choice. Also, it integrated seamlessly with Azure OpenAI service.
We didn't try anything else yet. But Azure Cognitive Search also looks like a viable option.
Elasticsearch can also be used
Hello seniors,
Ruchi here.
I am new to gen ai. Using it in my first project. We are using openai chat completions to categories the users feedback (through prompt engineering) . Can you please help me understand at what stage of gen ai learning i fall. Is it very novice of gen ai use case. What else can i learn next.
Yes