feat: Added QLoRA fine-tuning pipeline

[carlosgjs]

Initial code for model fine-tuning using QLoRA.

This has successfully run on a private A40 GPU (44gb). However, it throws an Out of Memory exception on a 16gb V100 in AzureML, even though it should theoretically fit.

Issue #17

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (4c79762) 98.80% compared to head (93799fc) 98.98%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #40      +/-   ##
==========================================
+ Coverage   98.80%   98.98%   +0.17%     
==========================================
  Files           6        8       +2     
  Lines         252      296      +44     
==========================================
+ Hits          249      293      +44     
  Misses          3        3              

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

[lsetiawan]

Just some stylistic comments, nothing major here. Everything else looks good and makes sense.