Record training loss

[XianzheMa]

Per the PR title, this PR solves issue #488.

[codecov]

Codecov Report

Attention: Patch coverage is 87.50000% with 1 line in your changes missing coverage. Please review.

Project coverage is 82.80%. Comparing base (42354cf) to head (328be5a).
Report is 2 commits behind head on main.

Files	Patch %	Lines
...trainer_server/internal/trainer/pytorch_trainer.py	80.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #539      +/-   ##
==========================================
- Coverage   82.81%   82.80%   -0.01%     
==========================================
  Files         217      217              
  Lines       10152    10157       +5     
==========================================
+ Hits         8407     8411       +4     
- Misses       1745     1746       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[github-actions]

^{( % to main)
( % to main)}

[MaxiBoether]

The loss logic itself is ok but I am confused about the gRPC changes since we explicitly did not want to pass the None arguments because in gRPC there is a difference between passing None and no value at all, if I remember correctly. We had issues before with this

[MaxiBoether]

Thanks. i tried to dig into the grpc optional thing again but could not find my old meeting notes with Francesco. I guess it should be ok