[00:35:41] 10Scoring-platform-team (Current), 10Release-Engineering-Team, 10Patch-For-Review: ORES deployment submodules should point to phabricator HTTPS repos. - https://phabricator.wikimedia.org/T179009#4170079 (10mmodell) @demon: At least for now, since the swift backend I've been working on for phabricator's git-l... [09:44:58] 10Scoring-platform-team (Current), 10ChangeProp, 10ORES, 10Services (done): Change ORES rules to send all events to new "/precache" endpoint - https://phabricator.wikimedia.org/T158437#4170579 (10mobrovac) [09:45:04] 10Scoring-platform-team (Current), 10Analytics, 10ChangeProp, 10EventBus, and 2 others: Drop "non bot" condition from ORES changeprop rules - https://phabricator.wikimedia.org/T187927#4170577 (10mobrovac) 05Open>03Resolved [14:02:07] o/ [14:52:37] 10Scoring-platform-team, 10Analytics, 10Analytics-Wikistats, 10ORES: Discuss Wikistats integration for ORES - https://phabricator.wikimedia.org/T184479#4171115 (10Halfak) One that I think would be interesting for Wikistats is a count of the number of non-redirect main namespace articles that fall into a se... [15:58:52] 10Scoring-platform-team (Current), 10Research, 10Research-outreach, 10Epic, 10Paper: [Epic] Write paper about ORES as a socio-technical probe - https://phabricator.wikimedia.org/T121719#1886102 (10Halfak) https://commons.wikimedia.org/wiki/File:ORES_-_Facilitating_re-mediation_of_Wikipedia%27s_socio-tech... [15:59:03] 10Scoring-platform-team (Current), 10Research, 10Research-outreach, 10Epic, 10Paper: [Epic] Write paper about ORES as a socio-technical probe - https://phabricator.wikimedia.org/T121719#4171310 (10Halfak) Submitted to CSCW'18 second-round [16:45:38] * halfak looks around for awight [16:45:53] * halfak runs to lunch [17:34:47] 10Scoring-platform-team, 10ORES, 10Reading-Admin: Announce presence of "oresscores" in api.php - https://phabricator.wikimedia.org/T153688#4171620 (10dr0ptp4kt) Any update here? [17:43:53] Should have mentioned, I'm working a late shift today to make it more convenient to go to my class afterwards. [17:50:41] o/ awight [17:51:07] I got surprised by quarterly checkin work. So I should unlick some cookies. [17:51:14] ah hehe [17:51:29] I've filled my plate with drafttopic cleanup for the next day or two, so no rush. [17:52:00] Weird stuff. It hangs at "revscoring extract", even with the observations stripped down to just a few lines. [17:52:23] awight, it might be hanging on loading up the word2vec stuff. [17:52:43] that all seems fine, I'm writing a ipynb that vectorizes sample words... [17:53:01] I believe it's stuck doing something parallel with feature aggregation [17:53:17] possibly a synchronization deadlock [17:54:25] Weird. There is some block-io that happens. [17:54:34] If you turn on debug mode, you might get some more insights. [17:55:04] I saw more activity, but it still hangs w/o clues. I'll probably add logging etc. [17:59:23] I'm probably making this sounds too fun. Good luck with the paperwork :p [17:59:26] *sound [18:09:02] omg extraction works fine under linux [18:09:06] Cannot explain. [18:11:07] 10Scoring-platform-team: Investigate why revscoring extract hangs on MacOS - https://phabricator.wikimedia.org/T193514#4171781 (10awight) [18:11:16] 10Scoring-platform-team: Investigate why revscoring extract hangs on MacOS - https://phabricator.wikimedia.org/T193514#4171791 (10awight) p:05Triage>03Low [18:33:08] halfak: Random question, do you think we should continue to demonstrate traning and testing separately in reverted_detection_demo.ipynb, or combine using cv_train? Maybe it's more clear to do the steps explicitly, but the tradeoff seems to be complexity... [18:33:34] cv is just multiple train/test cycles [18:34:19] right [18:34:36] it hides the complexity of splitting into different data sets, is all [18:36:52] Not a big deal either way. I'm only making small changes to support the new APIs, won't tackle any structural changes in this pass. [18:40:23] I kind of like the idea of explicitly doing a testing pass on the date. [18:40:25] *data [18:40:31] It shows what a "test" is. [18:41:05] Also lol about extraction just working on linux. [18:41:16] Linux(TM) it just works ;) [18:41:19] lololol [18:41:19] cool [18:41:24] yeah... not always the case [19:19:10] Ack! Gotta run away to pick up my sick pet. Back in... 45 mins or so. [19:19:19] ohno! [19:19:34] S'ok. Planned visit. [19:19:41] ferret got a tumor removed :D [19:19:44] This is the first I've heard of Aaron's pet, sick or otherwise. [19:19:57] I've got a ferret and a dog. [19:19:58] :) [19:19:59] o/ [19:20:09] I think you've discussed the dog before. [19:21:11] we need pet single-payer... [19:22:20] That will definitely be a California state referendum in 10 years. [19:22:46] I think we need human single-payer before we can get state single-payer. [19:25:10] +1 for human-first! btw, I think this might be real: https://twitter.com/jules_su/status/989510998649397249 [20:22:20] back. Ferret did well. I just need to fabricate a tiny little cone of shame for her and I'll be back [20:24:03] \o/ [20:43:27] (03Abandoned) 10Awight: LFS enabled, word2vec upload [scoring/ores/assets] - 10https://gerrit.wikimedia.org/r/419637 (https://phabricator.wikimedia.org/T180627) (owner: 10Awight) [20:44:49] OK I failed on the cone of shame (to attempt again later, but she seems to have no interest in her stitches, so I'm giving her a break [20:44:56] Back to working on quarterly checkin stuff. [20:49:14] The open PR gets drafttopic building at least, so I'm optimistically running on ores-misc-01 [20:50:03] editquality notebooks are looking good, but I need to rebuild the demo data. [20:51:01] I don't like the transformations between our nice, JSON observations format and the values_labels tuples; maybe there's a revscoring utility that I overlooked... [20:58:15] 10Scoring-platform-team, 10Scap, 10Patch-For-Review, 10Release-Engineering-Team (Kanban): Support git-lfs - https://phabricator.wikimedia.org/T180627#4172761 (10mmodell) @awight: I think the latest problem is that we don't run `git lfs install` automatically on targets. I've ran the command manually on or... [21:02:41] 10Scoring-platform-team (Current), 10Gerrit, 10ORES, 10Operations, 10Patch-For-Review: Plan migration of ORES repos to git-lfs - https://phabricator.wikimedia.org/T181678#4172772 (10mmodell) @awight: `git lfs install` needs to be executed on each target and that isn't happening, currently. I can add a ho... [21:35:28] 10Scoring-platform-team (Current), 10Gerrit, 10ORES, 10Operations, 10Patch-For-Review: Plan migration of ORES repos to git-lfs - https://phabricator.wikimedia.org/T181678#4172851 (10awight) @mmodell I'm reading some strange stuff here, https://github.com/git-lfs/git-lfs/wiki/Installation Apparently, `git... [21:52:42] 10Scoring-platform-team (Current), 10Gerrit, 10ORES, 10Operations, 10Patch-For-Review: Plan migration of ORES repos to git-lfs - https://phabricator.wikimedia.org/T181678#4172903 (10mmodell) @awight: I don't think `git lfs install --local` will take care of the submodules. I suppose I could do `git submo... [22:02:20] 10Scoring-platform-team, 10Scap, 10Patch-For-Review, 10Release-Engineering-Team (Kanban): Support git-lfs - https://phabricator.wikimedia.org/T180627#4172945 (10awight) Good news! We've done an initial LFS deployment of the 1.6GB word2vec binary and it landed successfully on ores1001! There's one last de... [22:05:16] 10Scoring-platform-team (Current), 10Gerrit, 10ORES, 10Operations, 10Patch-For-Review: Plan migration of ORES repos to git-lfs - https://phabricator.wikimedia.org/T181678#4172949 (10awight) I'm happy with either solution, either a redundant `git lfs install` or the submodule foreach. It would be very su... [22:15:49] halfak: Maybe I don't understand drafttopic. Here's the first training revision, https://en.wikipedia.org/?diff=604413609&diffmode=source [22:16:03] It's a removal of talk page content... wat? [22:16:20] No idea what's going on there. [22:16:25] Are you looking at page_id? [22:16:36] hmmm that'd be impossible [22:17:06] The talk page is where the templates are [22:17:09] hmmm [22:17:22] The wikiproject templates. [22:17:29] Ah, one thing I just realized is that these are *revision* scores (actually, *page* scores) and not diff scores [22:17:36] right [22:18:03] um. [22:19:07] Conceptually, I thought this model would be scoring article page content at its first revision. [22:19:18] right [22:19:27] That is what is supposed to be happening [22:20:00] all the training data is... not that. [22:20:07] https://en.wikipedia.org/?diff=434813547 [22:20:21] Where are you getting training data [22:20:36] datasets/enwiki.labeled_wikiprojects.json [22:20:40] comes from a figshare [22:20:51] * awight rubs eyes [22:22:03] Right. Could it be that there's another step that gets the article text? [22:22:06] My guess is that we've blindly pulled pages that include project templates, but we should be excluding Talk pages [22:22:20] Maybe. I'll keep digging [22:22:43] The makefile implies that no [22:22:45] :| [22:22:46] WTF [22:22:58] I guess the good news is that our health will be WAY higher once we straighten this out. [22:23:05] Right [22:23:13] Either way, I'm AFK for the day. Will look more in the morning. Want to throw your notes somewhere for me? [22:23:14] bad news is only mildly bad, that I boiled a small river training locally and on ores-misc this morning [22:23:18] kk [22:23:25] I'll use the PR [22:23:33] Great [22:23:45] We might have to retract our paper :( [22:23:54] ohwat [22:24:07] I'll be sure to finish this digging, then. [22:24:12] Thanks [23:10:37] 10Scoring-platform-team (Current), 10ORES, 10drafttopic-modeling, 10artificial-intelligence: Check drafttopic model memory usage - https://phabricator.wikimedia.org/T192293#4173198 (10awight) a:05Sumit>03awight [23:26:52] 10Scoring-platform-team (Current), 10Gerrit, 10ORES, 10Operations, 10Patch-For-Review: Plan migration of ORES repos to git-lfs - https://phabricator.wikimedia.org/T181678#4173222 (10mmodell) {D1039} [23:27:19] 10Scoring-platform-team, 10Scap, 10Patch-For-Review, 10Release-Engineering-Team (Kanban): Support git-lfs - https://phabricator.wikimedia.org/T180627#4173223 (10mmodell)