Log ZeRO and parallelism settings to monitor backends - #8401
Open
YeonwooSung wants to merge 4 commits into
Open
Conversation
Record static training topology on WandB/Comet/TensorBoard/CSV at engine init so runs are comparable without digging through logs. Fixes deepspeedai#7494 Signed-off-by: YeonwooSung <neos960518@gmail.com>
YeonwooSung
requested review from
loadams,
tjruwase and
tohtana
as code owners
September 3, 2026 02:51
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f33ad0a6aa
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Pipeline and caller-supplied MPUs keep topology on the MPU object, not the legacy groups globals. Logging those globals made PP/TP runs look like DP=world and TP=1. Signed-off-by: YeonwooSung <neos960518@gmail.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Record static ZeRO, precision, and DP/TP/PP/SP topology on enabled monitor backends when the engine finishes initializing.
Fixes #7494
Why
WandbMonitoronly calledwandb.init()and never wrote run config. The engine already logs per-step scalars (lr,loss_scale, timings), but users could not see ZeRO stage, offload devices, or parallelism sizes on the W&B/Comet/TensorBoard run.Changes
collect_monitor_config()builds a JSON-safe snapshot (stage, offload, precision, batch sizes, DP/TP/PP/SP rank and world size)MonitorMaster.update_config()fans out on rank 0wandb.config.update(..., allow_val_change=True)experiment.log_parameters(...)add_text("DeepSpeed/config", ...)deepspeed_config.csv__init__when monitoring is enableddocs/_tutorials/monitor.mdwas missing commas between backend objects; that is fixed hereThis PR does not add new engine rank getters; those are a separate follow-up to #7423.
Testing
tests/unit/monitor/test_monitor.py(rank-0-only WandB/MonitorMaster, backend update_config, engine collection)pre-commit run --fileson the touched files