Reset metrics before each task starts

[rohitgr7]

What does this PR do?

Fixes #7558

Does your PR introduce any breaking changes? If yes, please list them.

Before submitting

• Was this discussed/approved via a GitHub issue? (not for typos and docs)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes? (if necessary)
• Did you write any new necessary tests? (not for typos and docs)
• Did you verify new and existing tests pass locally with your changes?
• Did you list all the breaking changes introduced by this pull request?
• Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

Anyone in the community is welcome to review the PR.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:

• Is this pull request ready for review? (if not, please submit in draft mode)
• Check that all items from Before submitting are resolved
• Make sure the title is self-explanatory and the description concisely explains the PR
• Add labels and milestones (and optionally projects) to the PR so it can be classified

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #9410 (054f1f2) into master (eb6aa7a) will increase coverage by 0%.
The diff coverage is 100%.

@@          Coverage Diff           @@
##           master   #9410   +/-   ##
======================================
  Coverage      93%     93%           
======================================
  Files         179     179           
  Lines       15307   15312    +5     
======================================
+ Hits        14199   14204    +5     
  Misses       1108    1108           

[tchaton]

LGMT !

[awaelchli]

Given the description of the PR and looking at the implementation, I think it's better these calls go directly into the trainer.run implementation, not the loops.

[rohitgr7]

Given the description of the PR and looking at the implementation, I think it's better these calls go directly into the trainer.run implementation, not the loops.

good point, thanks :)

[rohitgr7]

due to recent fault-tolerant changes, I had to move it back to the trainer. Since logger_connector is sort of handled by the trainer, this seems ok to me.

[carmocca]

due to recent fault-tolerant changes, I had to move it back to the trainer

Can you elaborate on what are the changes and why they force moving it?

[rohitgr7]

due to recent fault-tolerant changes, I had to move it back to the trainer

Can you elaborate on what are the changes and why they force moving it?

Training state is being restored before dispatch so adding these resets inside loops will reset the resultcollection set by resume_from_checkpoint.