Prune deprecated: classification and regression

[ashutoshml]

What does this PR do?

Fixes part of #771

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure to update the docs?
• Did you write any new necessary tests?

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #806 (5ea82b7) into master (8e9eb8e) will increase coverage by 1%.
The diff coverage is 100%.

@@          Coverage Diff          @@
##           master   #806   +/-   ##
=====================================
+ Coverage      95%    95%   +1%     
=====================================
  Files         165    163    -2     
  Lines        6776   6710   -66     
=====================================
- Hits         6410   6384   -26     
+ Misses        366    326   -40     

[ashutoshml]

Failing integration test for python 3.6. Not sure how to resolve this.

[Borda]

@SkafteNicki seems we have the same issue with PL compatibility here too, as F1 was one of the very legacy metrics first implemented in PL itself... 🐰

so shall we break the compatibility with PL v1.3 or preserve this burden till TM v1.0?
cc: @PyTorchLightning/core-metrics @PyTorchLightning/core-lightning

[SkafteNicki]

@Borda this is a though one, but I am in favour of breaking backwards-compatibility with lightning. We need first and foremost to think of metrics as its own package which does not depend on lightning. In practise I do not know how many actually uses metrics with base torch and how many uses it with lightning.
However, I would also be more comfortable with this breaking change if lightning v1.6 is out before v0.8 of metrics. This would mean that TM v0.8 is compatible with two major versions of lightning v1.5 and v1.6. If you are using lightning <v1.5 we recommend sticking with TM v0.7.

[Borda]

However, I would also be more comfortable with this breaking change if lightning v1.6 is out before v0.8 of metrics. This would mean that TM v0.8 is compatible with two major versions of lightning v1.5 and v1.6. If you are using lightning <v1.5 we recommend sticking with TM v0.7.

we have this testing already inside EcoCI - https://github.com/PyTorchLightning/ecosystem-ci/blob/main/configs/PyTorchLightning/metrics_pl-develop.yaml

[awaelchli]

However, I would also be more comfortable with this breaking change if lightning v1.6 is out before v0.8 of metrics. This would mean that TM v0.8 is compatible with two major versions of lightning v1.5 and v1.6. If you are using lightning <v1.5 we recommend sticking with TM v0.7.

+1 for this approach

[ashutoshml]

@Borda: Any updates on the final approach for this issue?

[Borda]

@Borda: Any updates on the final approach for this issue?

it is going to merge as soon as all test passes :)