Likely so, see MAHOUT-694.
On May 20, 2011, at 1:39 PM, Sean Owen wrote: > Oh sorry these are the same issue? Great! > On May 20, 2011 5:44 PM, "Jake Mannix" <[email protected]> wrote: >> Looks like Grant got a fix posted? Has anyone else tried it? >> >> -jake >> >> On Fri, May 20, 2011 at 9:32 AM, Sean Owen <[email protected]> wrote: >> >>> I think we definitely need to figure out whether it's a bug or some other >>> confusion. If it's a doesn't-work-at-all bug yes probably the kind of > thing >>> that needs a fix ASAP in which case write up all you know and everyone > will >>> pile in to look at it. >>> >>> On Fri, May 20, 2011 at 5:29 PM, Jeff Eastman <[email protected]> wrote: >>> >>>> Is this an issue that should be fixed before we release? It seems to be >>>> broken to me. >>>> >>>> -----Original Message----- >>>> From: Jeff Eastman [mailto:[email protected]] >>>> Sent: Thursday, May 19, 2011 5:05 PM >>>> To: [email protected] >>>> Subject: Is LDA Broken? >>>> >>>> I'm running build-reuters option 2 and the LDA runs to maxIterations > (20) >>>> without ever producing a non-zero Log Likelihood. This is not the >>> behavior >>>> that I recall from earlier runs and seems quite unlikely to be correct. >>>> >>>> >>>
