fix bugs in KLDivergence

[hunto]

Motivation

1. The computation of KLDivergence Loss is wrong (see definitions in PyTorch docs: https://pytorch.org/docs/master/generated/torch.nn.KLDivLoss.html#torch.nn.KLDivLoss).

Modification

1. Use the api provided by PyTorch (torch.nn.functional.kl_div) to compute the KLDiv.
2. Add a reduction parameter. By default, the reduction method is batchnorm.
3. Use F.log_softmax instead of initializing new torch.nn.LogSoftmax module in the forward function.
4. Modify the docs.

BC-breaking (Optional)

No BC-breaking.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here and update the documentation.

Checklist

Before PR:

• Pre-commit or other linting tools are used to fix the potential lint issues.
• Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness.
• The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• CLA has been signed and all committers have signed the CLA in this PR.

[codecov]

Codecov Report

Merging #35 (f58d3ca) into master (57a5549) will decrease coverage by 0.05%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #35      +/-   ##
==========================================
- Coverage   55.59%   55.54%   -0.06%     
==========================================
  Files          81       81              
  Lines        2941     2942       +1     
  Branches      544      544              
==========================================
- Hits         1635     1634       -1     
- Misses       1231     1232       +1     
- Partials       75       76       +1     

Flag	Coverage Δ
unittests	55.54% <100.00%> (-0.06%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
mmrazor/models/losses/kl_divergence.py	100.00% <100.00%> (ø)
mmrazor/models/mutators/one_shot_mutator.py	92.59% <0.00%> (-3.71%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 57a5549...f58d3ca. Read the comment docs.