Exploring SemanticKernel
This week, I had the chance to explore the SemanticKernel code base, particularly the core plugins. SemanticKer...
For further actions, you may consider blocking this person and/or reporting abuse
I loved it.
Here is a question, is all the semantic kernel thing, about Azure users? Or could it be used without using azure?
The last times our software broke down, was due to the fact that how LLM models tended to respond, had changed! We solved it by utilizing the "json mode" feature of OpenAI, so that the structure of the response was a bit more deterministic. We couldn't go further in terms of validating the response.
Your posts are tempting my curiosity.
SemanticKernel is not tied to Azure. You can integrate it with any OpenAI service that you have access to. You are right about the importance of unit testing for OpenAI projects. Since the outputs of these projects can vary significantly depending on the prompts and the models, you need a reliable way to test and verify them. That’s where skUnit comes in. It is a testing tool that lets you write and run scenarios for SemanticKernel units, such as plugin functions and kernels.
Thank you Mehran.
It helped alot. I'll dive into it, as soon as I convince others to bring "Validation" tasks up.