Allow definition of Loss and Optimizer in config file.

[drewoldag]

This is an initial implementation that allows defining PyTorch Loss (Criterion) and Optimizers from within the config file.

Two new tables are introduced, [criterion] and [optimizer] and we currently use the relatively familiar libpath approach to avoid having to create a registry of pytorch criterion and optimizer functions which would need to be loaded at run time (kind of heavy and slow)

The keys in the tables need to share the same names as the parameters that would be passed to the loss or optimizer functions. (we just use **config to pass the params in).

Within the __init__ function of either the built in models or in an external class the user will still need to call:

self.optimizer = self._optimizer()
self.criterion = self._criterion()

(I would like to fix that in a followup PR - Issue #134)

Additionally, if the user has a special def _optimizer() or def _criterion() function that is defined in the model class, we will not over write that.

One final followup would be to allow a Learning Rate scheduler (described here) to be defined in the config as well. That won't happen in this PR though.

[codecov]

Codecov Report

Attention: Patch coverage is 42.85714% with 8 lines in your changes missing coverage. Please review.

Project coverage is 39.87%. Comparing base (0467d4a) to head (4d6e778).
Report is 4 commits behind head on main.

Files with missing lines	Patch %	Lines
src/fibad/models/model_registry.py	33.33%	8 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #132      +/-   ##
==========================================
- Coverage   39.97%   39.87%   -0.10%     
==========================================
  Files          22       22              
  Lines        1801     1808       +7     
==========================================
+ Hits          720      721       +1     
- Misses       1081     1087       +6     

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

[github-actions]

Before [0467d4a]	After [9f48f67]	Ratio	Benchmark (Parameter)
1.14±0.8s	2.86±1s	~2.52	benchmarks.time_computation
752	1.03k	1.37	benchmarks.mem_list

Click here to view all benchmarks.

[aritraghsh09]

This implementation looks perfectly in line with what I had requested!

No comments. Looks ready to merge to me!