Skip to content
/ sinless Public

To avoid some common AI sins, find explicable models using minimal data (via tuned learners).

License

Notifications You must be signed in to change notification settings

timm/sinless

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

85 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Sinless

"A saint is not someone who never sins, but one who sins less and less frequently and gets up more and more quickly."
― St. Bernard of Clairvaux

Here are some AI sins, and their potential cure:

  • If AI generated models are too complex to understand, using symbolic methods to find regions where conclusions hold.
  • If the AI need too much data, use semi-supervised learning to reduce that commissioning cost.
  • If the AI tools are needlessly complicated, use epsilon domination to stop needlessly explore spurious details.
  • If the AI tools need tuning, use the above to tune the models.

For more, read the doc.