Devis Lucato

Hi @zsogitbe thanks for the feedback. The use of the word "serverless" is in the context of Kernel Memory, where the default deployment requires to stand up a "Kernel Memory...

There's an example here: https://github.com/microsoft/kernel-memory/blob/main/examples/105-dotnet-serverless-llamasharp/Program.cs For embeddings though we don't recommend using LLama because it doesn't provide sufficient quality for RAG, leading to no results or incorrect ones.

The problem is actually LLama embeddings ability to capture the meaning of text, which is very low particularly with mixed content. When it comes to using cosine similarity to find...

Could you provide a file to reproduce the problem?

TODO: work with tokens, chunking/counting only once. To be part of semantic chunker feature.

@kill136 the comment contained something that looked like a Bing API Key (`//var bingConnector = new BingConnector("...`. I removed the value. You might want to invalidate the key on your...

sorry but I don't see where `{{$chatMessages }}` is used, and looks like your `SearchQuery` is not the same class used by KM, probably your custom class. I see also...

Spent some time inspecting the chunker in KM, which we took from SK. The initial version I wrote in 2022 didn't have the overlapping tokens logic, which was introduced later...

New chunkers just merged. Note: OverlappingTokens is part of MaxTokensPerParagraph. For instance, if MaxTokensPerParagraph = 1000 and OverlappingTokens = 300, a chunk will contain 300 tokens from the previous chunk...

@coryisakson I suspect the code is not picking the settings from your configuration file. I could be due to the typo `appsettings.Developmnet.json` => `appsettings.Development.json` or because the `ASPNETCORE_ENVIRONMENT` env var...