fix: small changes to MNIST benchmarks

[gumityolcu]

Hello,

MNIST benchmarks are updated slightly. precommit config has been corrected. Tiny Imagenet benchmarks presented some difficulties so they will have their own PR.

one thing that should be fixed: For MixedDataset, adversarial datasets should have test train splits and the eval_dataset should only have the adversarial test split.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.11%. Comparing base (e746691) to head (4a3af4a).
Report is 6 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #240   +/-   ##
=======================================
  Coverage   92.11%   92.11%           
=======================================
  Files          55       55           
  Lines        2345     2345           
=======================================
  Hits         2160     2160           
  Misses        185      185           

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