Chris
Chris
`TerminationStrategy.MaximumIterations` defines the maximum number of turns for agents to take in a chat. This only affects when calling: ```c# public async IAsyncEnumerable InvokeAsync([EnumeratorCancellation] CancellationToken cancellationToken = default) ``` and...
Open AI has recently added streaming responses to the Assistant API: https://platform.openai.com/docs/changelog/mar-14th-2024 https://platform.openai.com/docs/assistants/overview?context=with-streaming Add this to the initial Assistant agents: https://github.com/microsoft/semantic-kernel/blob/main/dotnet/src/Experimental/Agents/IAgentThread.cs
This property is currently not-serializable and intended to take an reference to an `Agent` (or other) object. It may be preferable to serialize a value for `Source` that is useful...
In agent-space, there's an open question as to what is the expected order of the converstation history...and where. Logically, when a (say a long) conversation is accessed, its the most...
Adopting this patterns is in alignment with SK practices and general best practices: - https://learn.microsoft.com/en-us/dotnet/core/extensions/logger-message-generator - https://github.com/microsoft/semantic-kernel/blob/main/dotnet/src/SemanticKernel.Abstractions/Functions/KernelFunctionLogMessages.cs
`AggregatorAgent` currently always provides last message for nested chat. Summarization by LLM is also a valid alternative.
https://blog.langchain.dev/how-to-build-the-ultimate-ai-automation-with-multi-agent-collaboration/ Need to provide an abstraction that supports the following patterns: - LangGraph - Crew AI - Guided Conversations - Flow Orchestrator
AutoGen supports some form of parallel execution. Analyze how this is supported (heirarchical chat? etc..). We know that an Open AI Assistant API does not allow updating a thread (creating...
In support of direct invocation of ChatCompletionAgent, add support for streaming.
While agent related serialization has been identified and iscussed, it hasn't been formally addressed and expicitly deprioritized in previous effort. The need exists to serialize and deserialize an agent conversation...