DOC: Clarify DeviceStatsMonitor logged metrics #20895
Open
Conversation
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
MrAnayDongre
requested review from
lantiga,
Borda,
tchaton,
justusschock and
ethanwharris
as code owners
Borda
changed the title
Borda
reviewed
Jun 12, 2025
| @@ -45,6 +45,23 @@ class DeviceStatsMonitor(Callback): | |||
| ModuleNotFoundError: | |||
| If ``psutil`` is not installed and CPU stats are monitored. | |||
|
|
|||
| Logged Metrics: | |||
There was a problem hiding this comment.
Raises: or Args: are Sphinx-specific keywords compared to Logged Metrics:, so pls let's move it just to the top of this docstring
There was a problem hiding this comment.
Thanks, @Borda! I've moved the 'Logged Metrics' section to the top of the docstring as requested in commit dcd1042.
There was a problem hiding this comment.
well, I still see it without change
MrAnayDongre
force-pushed
the
docs/fix-20807-device-stats-metrics
branch
4 times, most recently
from
cf7e36a to
798f9c9
Compare
MrAnayDongre
force-pushed
the
docs/fix-20807-device-stats-metrics
branch
from
798f9c9 to
2461f52
Compare
for more information, see https://pre-commit.ci
Borda
reviewed
Jun 18, 2025
Comment on lines
+41
to
+51
| CPU (via ``psutil``): Logs ``cpu_percent``, ``cpu_vm_percent``, ``cpu_swap_percent``. | ||
| All are percentages (%). | ||
| CUDA GPU (via :func:`torch.cuda.memory_stats`): Logs detailed memory statistics from | ||
| PyTorch's allocator (e.g., ``allocated_bytes.all.current``, ``num_ooms``; all in Bytes). | ||
| GPU compute utilization is not logged by default. | ||
| Other Accelerators (e.g., TPU, MPS): Logs device-specific stats: | ||
|
|
||
| - TPU example: ``avg. free memory (MB)``. | ||
| - MPS example: ``mps.current_allocated_bytes``. | ||
|
|
||
| Observe logs or check accelerator documentation for details. |
There was a problem hiding this comment.
'Let''s make this a complete list and you can validate the compiled docs in readthedocs link 📚: pytorch-lightning--20895.org.readthedocs.build/en/20895
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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 does this PR do?
This PR addresses issue #20807 by adding detailed documentation for the metrics logged by
DeviceStatsMonitor.The key clarifications include:
psutil, CUDA GPU viatorch.cuda.memory_stats, and other accelerators viaaccelerator.get_device_stats()).DeviceStatsMonitor.{hook_name}/{base_metric_name}.torch.cuda.memory_stats()for the full list of memory metrics.profiler_basic.rstto align with these clarifications and link to the API docs.This documentation aims to help users understand what statistics to expect when using
DeviceStatsMonitorwith different hardware configurations.Fixes #20807
No breaking changes are introduced by this documentation update.
Before submitting
PR review
Anyone in the community is welcome to review the PR.
Before you start reviewing, make sure you have read the review guidelines. In short, see the following bullet-list:
Reviewer checklist
📚 Documentation preview 📚: https://pytorch-lightning--20895.org.readthedocs.build/en/20895/