[00:01:46] halfak: and there's a hammock [00:07:42] https://upload.wikimedia.org/wikipedia/commons/8/8d/Man_in_the_Hight_Castle_Dick.png <- wow this would have made geography easier [11:04:04] ToAruShiroiNeko: https://gist.github.com/Ladsgroup/cc22515f55ae3d868f47#file-itwiki [11:04:16] we need an Italian native speaker [12:51:27] Amir1, I'll see if we can get DarTar or Elitre [12:52:46] Hey! [12:52:53] I'm waiting for the meeting [12:53:14] Yup. Just getting my thoughts together. [12:55:34] Hi [12:55:42] * YuviPanda is up early [12:55:52] hi YuviPanda :) [12:56:04] have to go, back in two or three min. [12:56:19] o/ YuviPanda [12:56:25] Hi :) [12:56:35] Just preparing to talk to some researchers about vandalism detection in Wikidata. [12:56:46] Nice [12:57:22] Saw a bunch of pull requests from familiar names but new to revscoring [12:57:24] That's great [12:59:40] YuviPanda, +1 :) [13:00:03] larger contributor base == more better [13:00:16] Indeed [13:00:42] I asked some of them to add translations and bad words [16:05:52] Amir1, joining us for sync call? [16:05:59] We should probably switch to skype. [16:06:06] I'm in the hangout with ToAruShiroiNeko [16:32:50] halfak, sorry I'm late. I wasn't sure if we were meeting today. [16:38:33] Yes please join us :) [16:38:36] aetilley, ^ [17:55:54] actually can we get wikibugs here posting only about wiki-ai stuff [17:55:59] namely our phabricator? [17:58:39] ToAruShiroiNeko, that'd get kind of spammy. [17:58:58] but itd be just our phab [17:59:05] we have github so I dont know [17:59:24] I would prefer those to be msgs rather than notices [17:59:30] +1 [17:59:32] (which appear on every window) [17:59:36] And the username much shorter [18:18:36] hey, how quickly do you think you can set up campaigns for the newer languages [18:18:53] namely german and dutch as well as others where we can have language features [18:19:41] I do think we should start with a sample of 20,000 random where 2000 would be provided for human review as we have discussed. [18:20:09] I do think we should start with a sample of 50,000 then downsize it to 18,000 autolabeled good & random where 2000 would be provided for human review as we have discussed. [18:47:48] ls [19:16:40] http://www.quora.com/What-is-a-coders-worst-nightmare [19:16:51] "Internet Explorer (if you are a Web developer)." [19:17:04] When I saw the pull request of Robin [20:39:41] halfak you have a quick 5 mins [20:39:41] http://www.pastebin.ca/3083526 [20:39:51] seems like a conflict? [20:42:27] hmm [20:42:31] I thnk its solved [20:42:33] nevermind [20:44:45] ? anything for me to do here? [20:53:05] No its ok [20:53:11] users own config was the issue [20:53:19] YuviPanda, Do we have backups of labels? [20:53:52] my IT hat suggests I should ask for fgs model [21:16:34] +1 for that question: do we have backups? [21:37:52] ToAruShiroiNeko, I keep encrypted backups from the last 7 days on my home server. [21:38:50] glad to hear that but I would also like to have some on server thing [21:38:52] I use them for testing :) [21:38:53] on site I mean [21:38:59] On "site"? [21:39:01] corruption can creep [21:39:06] on wmf cluster somewhere [21:39:07] I'd like them off site thank you very much! [21:39:14] Maybe a different cluster [21:39:16] Sure [21:39:32] some place that will be maintained even if all of us gets hit by a hover bys [21:39:35] *bus [22:09:58] I have a to drive across the state tonight. I won't be on for the hackathon tomorrow. [22:10:15] Just packing up now. I should be back on around 2200 UTC on Sunday. [22:10:18] o/