Theo Van Dinter writes: > On Tue, Dec 26, 2006 at 04:30:25PM +0000, Justin Mason wrote: > > I don't think we should add PBL to 3.1 -- (a) there's no data there > > yet, and (b) it'll take a rescore run to figure out appropriate > > scoring. > > We can set the score low (like 0.001), but we'll want live data in the corpus > for reuse to work during net runs. I think I left it in there as a default 1, > but I'll lower it in a minute.
ah ok. note that there's still no data in there, though, so that should probably wait until they've published something worth reusing.... > Speaking of which, if there's any other network tests which haven't been put > in the 3.1 rules yet, we'll want to do that so we can start using reuse during > the weekly and scoregen runs. unless you're like me and have been dogfooding 3.2.0 for several months ;) --j.
