undefined | Better HN

0 pointsericpauley8d ago0 comments

True. I’d be curious whether a combination of matching comp/training cutoff and censoring web searches could yield a more precise evaluation.

0 comments

chaisan8d ago

as its from 2024 (MaxSAT was not held in 2025), its quite likely all the solvers are in the training data. so the interesting part here is the instances for which we actually got better costs that what is currently known (in the best-cost.csv) file.

ericpauleyOP8d ago

As GP noted the issue is that even better versions than competed in MaxSAT are likely in the training data or web resources.

j / k navigate · click thread line to collapse

0 pointsericpauley8d ago0 comments

True. I’d be curious whether a combination of matching comp/training cutoff and censoring web searches could yield a more precise evaluation.

0 comments

chaisan8d ago

ericpauleyOP8d ago

As GP noted the issue is that even better versions than competed in MaxSAT are likely in the training data or web resources.

j / k navigate · click thread line to collapse