Add report generation for Nemotron Ultra (Uses Megatron-Run workload in CloudAI)#797
Draft
srivatsankrishnan wants to merge 6 commits intoNVIDIA:agent-devfrom
Draft
Add report generation for Nemotron Ultra (Uses Megatron-Run workload in CloudAI)#797srivatsankrishnan wants to merge 6 commits intoNVIDIA:agent-devfrom
srivatsankrishnan wants to merge 6 commits intoNVIDIA:agent-devfrom
Conversation
More robust bench execution for Dynamo over k8s
Contributor
|
Important Review skippedDraft detected. Please check the settings in the CodeRabbit UI or the You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This PR adds a new MegatronRunReportGenerationStrategy class that parses Megatron-LM training logs to extract performance metrics for Design Space Exploration (DSE) and reporting purposes.
The existing MegatronRun workload only had CheckpointTimingReportGenerationStrategy which parses checkpoint save/load timing information. There was no support for extracting core training metrics like step time and GPU throughput from Megatron-LM iteration logs.
Test Plan
Additional Notes