Seems like it failed because sgd is flaky, like ted mentioned its the shuffle issue. Ted, shouldnt sgd be resilient about this. I mean, you can never expect to get randomly distributed data in production. ------ Robin Anil
On Sat, Jun 9, 2012 at 5:52 PM, Robin Anil <[email protected]> wrote: > Yep. Its all good now. > > > On Sat, Jun 9, 2012 at 4:56 PM, Jeff Eastman > <[email protected]>wrote: > >> This Jenkins project is back to normal after I made some logging >> enhancements to debug it further. I conclude this, like the 20 newsgroups >> failure #53, is a Jenkins transient. >> >> >> On 6/9/12 4:00 PM, Jeff Eastman wrote: >> >>> Hi Robin, >>> >>> I can't even get that far running locally. Perhaps this is a >>> configuration issue with my new laptop. The script itself looks right. I've >>> seen it passing intermittently and thought it to be a Jenkins issue: >>> >>> >> >
