Fix - Avoid two softmax calls in LogisticRegression

[pierrenodet]

Avoid two softmax calls in LogisticRegression

What does this PR do?

Fixes #654

[codecov]

Codecov Report

Merging #655 (fe2b0c2) into master (50969a8) will not change coverage.
The diff coverage is 0.00%.

@@           Coverage Diff           @@
##           master     #655   +/-   ##
=======================================
  Coverage   25.54%   25.54%           
=======================================
  Files         118      118           
  Lines        7110     7110           
=======================================
  Hits         1816     1816           
  Misses       5294     5294           

Flag	Coverage Δ
cpu	25.54% <0.00%> (ø)
pytest	25.54% <0.00%> (ø)

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

Impacted Files	Coverage Δ
pl_bolts/models/regression/logistic_regression.py	23.80% <0.00%> (ø)

---

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 50969a8...fe2b0c2. Read the comment docs.

[tchaton]

LGTM !

[garryod]

Hello @pierrenodet, could we make the same changes in validation_step() and test_step() such that the losses are the same? (Will then need to add a 'F.softmax()' inside the 'accuracy()' call as it expects probabilities)