-
Notifications
You must be signed in to change notification settings - Fork 487
Description
Is this a new feature, an improvement, or a change to existing functionality?
New Feature
How would you describe the priority of this feature request
Critical (currently preventing usage)
Please provide a clear description of problem this feature solves
Often times, larger than necessary LLMs are used in production agentic applications. This has latency/throughput, hardware, and cost implications. I want NeMo Agent Toolkit to transmit runtime traces to NeMo Data Flywheel Blueprint to leverage the NeMo Microservices Platform to enable model distillation from runtime traces. This feature will allow the deployment of leaner agentic applications without sacrificing the quality of outputs.
Describe your ideal solution
The ideal solution will put minimal burden on NeMo Agent Toolkit developers to take advantage of this feature. Turning on the feature on should be as simple as adding a new entry to the general.telemetry.tracing section of the YAML configuration file (similar to other telemetry exporters). Additionally, fine grained workload scoping should be possible without requiring additional registered function implementations to allow more targeted model distillation.
Additional context
No response
Code of Conduct
- I agree to follow this project's Code of Conduct
- I have searched the open feature requests and have found no duplicates for this feature request