You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For Synthesis (and its subclasses), we have two stopping criteria: either you reach max_iter or your (absolute) loss decreases by less than loss_thresh over the past loss_change_iter iterations. But this is an absolute number, which is going to differ wildly depending on the magnitude of your loss.
Would like to add support for a relative threshold, rel_loss_thresh, which checks whether (absolute) loss has decreased by rel_loss_thresh * loss_prev over the past loss_change_iter iterations. On each iteration check if loss < loss - rel_loss_thresh * loss_prev and, if so, update loss_prev = loss. Keep going until there have been loss_change_iter iterations without that, and then break.
This would go into Synthesis._check_for_stabilization, and would need to do a similar check with coarse_to_fine
The text was updated successfully, but these errors were encountered:
For
Synthesis
(and its subclasses), we have two stopping criteria: either you reachmax_iter
or your (absolute) loss decreases by less thanloss_thresh
over the pastloss_change_iter
iterations. But this is an absolute number, which is going to differ wildly depending on the magnitude of your loss.Would like to add support for a relative threshold,
rel_loss_thresh
, which checks whether (absolute) loss has decreased byrel_loss_thresh * loss_prev
over the pastloss_change_iter
iterations. On each iteration check ifloss < loss - rel_loss_thresh * loss_prev
and, if so, updateloss_prev = loss
. Keep going until there have beenloss_change_iter
iterations without that, and then break.This would go into
Synthesis._check_for_stabilization
, and would need to do a similar check withcoarse_to_fine
The text was updated successfully, but these errors were encountered: