kernel-memory
kernel-memory copied to clipboard
RAG architecture: index and query any data using LLM and natural language, track sources, show citations, asynchronous memory patterns.
At times there are documents we are uploading that contain dozens or hundreds of pages. It would be nice to include the page number or approximate page number, maybe a...
There was one earlier that allowed seamless deployment of kernel memory. https://devblogs.microsoft.com/semantic-kernel/how-to-deploy-semantic-kernel-to-azure-as-a-web-application-service/ Looks like it is not working now. Can we have an updated guidance on this please? Merely deploying...
I'm trying to ingest a 100+GB file of legal data into a Kernel Memory service. The data I would like to access are the "opinions" files from this link (https://com-courtlistener-storage.s3-us-west-2.amazonaws.com/list.html?prefix=bulk-data/)....
I think we can use one model for summarization (when we build data for vector db) like gpt3.5 and anoter like gpt4 for actual text writeng
The `DataPipeline.CustomData` property currently only supports inter-handler communication. It would be useful to be able to "bootstrap" with data from the caller (telemetry correlation identifiers, etc). Proposal, add a `Dictionary?...
I think it's a great thing when we can recognize the voice and then add the results to the index
The OpenAI/AzureOpenAI `TextGeneration` classes use the legacy Completions API ([deprecation from OpenAI](https://platform.openai.com/docs/deprecations/2023-06-13-updated-chat-models), [announcement](https://openai.com/blog/gpt-4-api-general-availability) and [Azure model compatibility](https://learn.microsoft.com/azure/ai-services/openai/concepts/models#gpt-35-models)). This poses a challenge when working with Azure OpenAI Service in particular as...
Original issue: https://github.com/microsoft/semantic-kernel/issues/2168 from @Nurgo > Hi, > > I've been working with SemanticKernel and I appreciate all the hard work you've put into this amazing project. > > I'm...
It's ok for now that if I provide an index that doesn't exist (when uploading a file) it creates the index, but if I want to create an 'admin' app...