Introduction to Memories in the Semantic Kernel SDK
Вставка
- Опубліковано 22 лип 2024
- Memories are a powerful way to provide broader context to your prompts. Semantic memory uses embeddings to represent words or data as vectors which enables semantic memory to perform meaningful comparisons and operations on text data, which is crucial for AI comprehension and processing.
In this video, I show you how Semantic Memory works, and how we can use it with the Semantic Kernel SDK.
0:00 Memories in Semantic Kernel
1:17 How semantic memory works
2:02 Why embeddings are important
3:24 How are embeddings used?
3:50 Where do Vector Databases come in?
5:19 Available connectors to vector databases
5:34 Adding Azure AI Search Memory Store
8:34 Manually adding memories
10:50 Adding memories to kernel arguments
15:52 Adding documents to your memory
19:09 Wrap up
Connect with me!
Twitter: / willvelida
GitHub: github.com/willvelida
Bluesky: bsky.app/profile/willvelida.b... - Наука та технологія
Clicked on the video to learn about it and realized it's the same thing with FastBert and SQLite. The I integrated it as a long- and short-term memory. Guess I did it the hard way, but I learned a lot. lol. Great video either way!!
My sincere thanks go out to you for making this playlist and for explaining the concepts so thoroughly.
I appreciate Will for this content
Love these videos. Really helpful in getting up to speed with semantic kernel
Thank you so much! I love what you do for the .NET and Dapr community! Hope you are well!
I love your videos, mate, and this is great timing, too; I just integrated my agent into semantic search last night, and the performance difference was huge. And highly recommend this approach for anyone looking for better search performance.
Thank you!
Thanks, mate! Would be great to see more videos with different use cases 😊
Thank you! Do you have anything in particular that you'd like to see? I'm thinking of doing some e2e examples (infra, pipelines etc.)
Is the source for this in a git repo somewhere? I am not finding it in your github account. Thanks for the video!