Modifying train pipeline to be more memory efficient - Githubissues

AI-sandbox / gnomix

A fast, scalable, and accurate local ancestry method.

Other

81 stars 13 forks source link

Modifying train pipeline to be more memory efficient #37

Closed arvind0422 closed 2 years ago

arvind0422 commented 2 years ago

Save gen0 for all splits - train1, train2 and val
Force train1 split's generations to be [0]
Add Gnomix.retrain_base() as a separate function

weekend37 commented 2 years ago

nice