Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
.Net: Agents - Getting Started Step4 Improvement: Cross Model (#6044)
### Motivation and Context <!-- Thank you for your contribution to the semantic-kernel repo! Please help reviewers and future users, providing the following information: 1. Why is this change required? 2. What problem does it solve? 3. What scenario does it contribute to? 4. If it fixes an open issue, please link to the issue here. --> [Step4_KernelFunctionStrategies](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/GettingStartedWithAgents/Step4_KernelFunctionStrategies.cs) was either: 1. Failing with null selection 2. Exceeding model token limit 3. Not performing the requested task ### Description <!-- Describe your changes, the overall approach, the underlying design. These notes will help understanding how your code works. Thanks! --> A couple things at play here: 1. The selection prompt was copied from AutoGen. While it worked on `GPT-4`, it wasn't generating a reliable result for `GPT-3.5-Turbo` 2. The agent instructions were never tuned outsidre of `GPT-4` 3. No sensible default / fall-through (can't trust model 100%) 4. 4k token limit may be too low for history evaluation. Tested with `gpt-35-turbo-16k` > Note: When formal support for history manipulation comes on line, we should be able to optimize around last(n) messages. ### Contribution Checklist <!-- Before submitting this PR, please make sure: --> - [X] The code builds clean without any errors or warnings - [X] The PR follows the [SK Contribution Guidelines](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md) and the [pre-submission formatting script](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md#development-scripts) raises no violations - [X] All unit tests pass, and I have added new tests where possible - [X] I didn't break anyone 😄
- Loading branch information