chore: Update dependencies, logging

[carlosgjs]

• Update dependencies in conda.yaml to match the package dependencies
• Log cuda version / availability when running via Typer app (which doesn't run with __name__=="__main__"
• Map non-quantized version of the model to a copy of llama in our own HF org (avoids the token issue)
• Fix metric logging in eval() (ooops!)

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (5615549) 97.98% compared to head (06af509) 98.80%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #39      +/-   ##
==========================================
+ Coverage   97.98%   98.80%   +0.82%     
==========================================
  Files           6        6              
  Lines         248      251       +3     
==========================================
+ Hits          243      248       +5     
+ Misses          5        3       -2     

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

[lsetiawan]

LGTM