You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Current multi-domain routing mechanisms are optimized for single-turn, domain-classification scenarios. I propose introducing a dedicated routing/configuration strategy in recipes specifically designed for stateful, multi-step code generation workflows.
Problem Statement:
I have been using the existing multi-domain router for complex coding tasks, but it consistently underperforms.
Open Questions for Discussion
Is there an existing routing pattern or config schema that already addresses this, even partially? I may have missed it.
Would this be better implemented as a separate router type, or as an extension/mode of the existing multi-domain router?
What is the recommended way to pass structured task-state to the router without exceeding token budgets?
Are there known benchmarks or evaluation frameworks for long-horizon code task routing that we should align with?
Has anyone successfully worked around this limitation with prompt engineering or external orchestration? What were the trade-offs?
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Current multi-domain routing mechanisms are optimized for single-turn, domain-classification scenarios. I propose introducing a dedicated routing/configuration strategy in recipes specifically designed for stateful, multi-step code generation workflows.
Problem Statement:
I have been using the existing multi-domain router for complex coding tasks, but it consistently underperforms.
Open Questions for Discussion
Is there an existing routing pattern or config schema that already addresses this, even partially? I may have missed it.
Would this be better implemented as a separate router type, or as an extension/mode of the existing multi-domain router?
What is the recommended way to pass structured task-state to the router without exceeding token budgets?
Are there known benchmarks or evaluation frameworks for long-horizon code task routing that we should align with?
Has anyone successfully worked around this limitation with prompt engineering or external orchestration? What were the trade-offs?
All reactions