add customized step collector into dlrover #1352
Open
Conversation
This file contains 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
--training_log_file specifies which logfile we collect step info --step_rank specifies the rank which will output the step info into logfile --step_pattern specifies the regex pattern of step info matching each log line
majieyue
requested review from
workingloong,
samplise and
BalaBalaYi
as code owners
added 6 commits
November 20, 2024 20:13
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #1352 +/- ##
==========================================
- Coverage 81.17% 81.16% -0.01%
==========================================
Files 231 231
Lines 21988 22272 +284
==========================================
+ Hits 17849 18078 +229
- Misses 4139 4194 +55 ☔ View full report in Codecov by Sentry. |
BalaBalaYi
reviewed
Nov 27, 2024
| @@ -238,6 +238,13 @@ class GlobalStep(Message): | |||
| elapsed_time_per_step: float = 0.0 | |||
|
|
|||
|
|
|||
| @dataclass | |||
| class UserStep(Message): | |||
There was a problem hiding this comment.
An advice:
- Use
DiagnosisReportDatainstead of the new definition here. - Use 'type' to tag it as 'UserStep' info and do resolving in master.
| @@ -177,15 +177,21 @@ class ElasticLaunchConfig(LaunchConfig): | |||
| tee: Union[Std, Dict[int, Std]] = Std.NONE | |||
| training_log_file: str = "" | |||
| failure_node_errors: str = "" | |||
| step_rank: int = -1 | |||
There was a problem hiding this comment.
Use a class object here to unify step info. Or u can define a more common class to include these fields(including: 'training_log_file' and 'failure_node_errors')
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.
usage: add such parameters into agent command line, e.g. dlrover-run
--training_log_file specifies which logfile we collect step info
--step_rank specifies the rank which will output the step info into logfile
--step_pattern specifies the regex pattern of step info matching each log line
This patch support user defined step log format, so dlrover can collect training steps info from log file. It is useful for
us to detect training errors, e.g. hang or stepback
What changes were proposed in this pull request?
step pattern and file match
Why are the changes needed?
add customized step collector into dlrover
Does this PR introduce any user-facing change?
Yes
How was this patch tested?
UT