what is difference between Concepts and Componants of semantic kernel #9600
-
|
I was going through the GitHub repo of Semantic kernel there I found python/samples/concepts folder. It had many concepts like auto function invocation, filters, planner, plugin. So now I am confused between the concept and components of the semantic kernel. Can you please help me to understand this? |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments
-
|
Hello @rakshahulle, concepts/components can be used interchangeably. The The main takeaway is: if you can understand that the Kernel is the central component to Semantic Kernel, that's a great start. The kernel controls the state of an application -- one can add plugins, filters, invoke those functions, etc, via the kernel object. An input to the kernel is an AI service (how you interact with the LLM) and one can define multiple AI services like To help round things out, I'm including a screenshot from the learn site:
You can follow the top purple/blue line from left to right, and then the bottom blue/purple line from right to left. This is a high-level flow using SK. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you so much this was helpful.
…On Thu, Nov 7, 2024 at 7:39 PM Evan Mattson ***@***.***> wrote:
Hello @rakshahulle <https://github.com/rakshahulle>, concepts/components
can be used interchangeably. The python/samples/concepts are
demonstrating various concepts about how to apply different SK components
-- like auto function calling, filters, etc.
The main takeaway is: if you can understand that the Kernel is the central
component to Semantic Kernel, that's a great start. The kernel controls the
state of an application -- one can add plugins, filters, invoke those
functions, etc, via the kernel object. An input to the kernel is an AI
service (how you interact with the LLM) and one can define multiple AI
services like AzureChatCompletion alongside an AnthropicChatCompletion
service. Each service has its own prompt execution settings (which settings
are sent to the model).
To help round things out, I'm including a screenshot from the learn site:
image.png (view on web)
<https://github.com/user-attachments/assets/2085dd0f-8ff6-4d70-8c1f-96e66a3aed71>
You can follow the top purple/blue line from left to right, and then the
bottom blue/purple line from right to left. This is a high-level flow using
SK.
—
Reply to this email directly, view it on GitHub
<#9600 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AXCZPRMSJRPDYO2TLTBJGRTZ7NYDNAVCNFSM6AAAAABRKIVSGWVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCMJXHA2DAMY>
.
You are receiving this because you were mentioned.Message ID:
***@***.***
com>
|
Beta Was this translation helpful? Give feedback.
Hello @rakshahulle, concepts/components can be used interchangeably. The
python/samples/conceptsare demonstrating various concepts about how to apply different SK components -- like auto function calling, filters, etc.The main takeaway is: if you can understand that the Kernel is the central component to Semantic Kernel, that's a great start. The kernel controls the state of an application -- one can add plugins, filters, invoke those functions, etc, via the kernel object. An input to the kernel is an AI service (how you interact with the LLM) and one can define multiple AI services like
AzureChatCompletionalongside anAnthropicChatCompletionservice. Each service has its own prompt ex…