Convert hans to Trainer

[sgugger]

This follows up from #4854 (@julien-c I took all your comments into account) and will close #4742.

[codecov]

Codecov Report

Merging #5025 into master will decrease coverage by 0.09%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #5025      +/-   ##
==========================================
- Coverage   77.18%   77.09%   -0.10%     
==========================================
  Files         128      128              
  Lines       21877    21877              
==========================================
- Hits        16886    16866      -20     
- Misses       4991     5011      +20     

Impacted Files	Coverage Δ
src/transformers/modeling_xlnet.py	76.35% <0.00%> (-2.30%)	⬇️
src/transformers/modeling_tf_utils.py	85.86% <0.00%> (-1.40%)	⬇️
src/transformers/modeling_utils.py	90.73% <0.00%> (+0.23%)	⬆️
src/transformers/file_utils.py	73.79% <0.00%> (+0.40%)	⬆️

---

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 1bf4098...201fae2. Read the comment docs.

[julien-c]

Looks great! Don't forget to fill a checkmark into the Big Table of Tasks 😉