kbeaugrand / semantickernel.assistants Goto Github PK

Why not too take some code from LittleLittleCloud and implement some of the Autogen abstractions? ;)
PR for the Autogen .NET implementation: https://github.com/microsoft/autogen/pull/924/files

I don't like the handling of multi-model but if we do this first and delegate anything on the models to the kernel there we go... we can start bringing in the ConversableAgent, and all the others, including the super cool implementation of the advanced analysis (self coding skill) implemented with dotnetinteractive and source generators :)

The plan:

1. We got those abstractions working on this codebase.
   2, We bring them to the PR in SK.
2. We bring them into Autogen (once PR of .NET is approved).

i get following error when i try to run
Unhandled exception. (Line: 6, Col: 1, Idx: 340) - (Line: 6, Col: 16, Idx: 355): Property 'input_parameter' not found on type 'SemanticKernel.Assistants.Models.AssistantModel'.

Desription

Assistant declaration parameters has changed. However samples are not up to date.

Older input_parameter paramter is now changed to:

input_parameters: 
     - name: <name>
       is_required: True|False
       default_value: ""
       description: |

Need to reflect this change to sample provided and readme.

The kernel is supporting more models, but we keep creating and doing everything by hand on the AgentBuilder (line 117 on AgentBuilder.cs for the PR https://github.com/microsoft/semantic-kernel/pull/3912/files) WithAzureOpenAIChatCompletion() for example.
Similar to how the current version of SK is doing it: https://github.com/microsoft/semantic-kernel/blob/main/dotnet/src/Experimental/Agents/AgentBuilder.cs line 73, WithOpenAIChatCompletion.

But... the kernel supports already multi model so, let's use it shall we?

Suggestion: Add only a "WithKernel(Kernel kernel, string serviceId)" we could either copy the configuration of the kernel or clone the kernel (even more simple). I understand we want to decouple the kernel, but maybe we can reuse it if it allows parallel calls...
If we look at the example AIServiceSelector, https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/KernelSyntaxExamples/Example62_CustomAIServiceSelector.cs

        var builder = Kernel.CreateBuilder()
            .AddAzureOpenAIChatCompletion(
                deploymentName: TestConfiguration.AzureOpenAI.ChatDeploymentName,
                endpoint: TestConfiguration.AzureOpenAI.Endpoint,
                apiKey: TestConfiguration.AzureOpenAI.ApiKey,
                serviceId: "AzureOpenAIChat",
                modelId: TestConfiguration.AzureOpenAI.ChatModelId)
            .AddOpenAIChatCompletion(
                modelId: TestConfiguration.OpenAI.ChatModelId,
                apiKey: TestConfiguration.OpenAI.ApiKey,
                serviceId: "OpenAIChat");

We could add a third or fourth local models, Ollama, LM Studio... through a service id.
And then reuse the existing kernel for this (or create a new one if needed, unsure on the internals and if SK kernel allows calls in parallel)

Description

When using AutoGen from NuGet Package, I'm facing this issue :

Method not found: 'System.Threading.Tasks.Task`1<Microsoft.SemanticKernel.Planning.FunctionCallingStepwisePlannerResult> Microsoft.SemanticKernel.Planning.FunctionCallingStepwisePlanner.ExecuteAsync(Microsoft.SemanticKernel.Kernel, System.String, System.Threading.CancellationToken)'.

I can step over functions, but this fails when reaching the Semantic Kernel Core capabilities.