[08:08:18] 10Scoring-platform-team-Backlog, 10ORES, 10Operations, 10Graphite, 10User-fgiunchedi: Regularly purge old ores graphite metrics - https://phabricator.wikimedia.org/T169969#3429182 (10fgiunchedi) cc @Halfak @Ladsgroup [08:57:34] 10Scoring-platform-team-Backlog, 10ORES, 10Operations, 10Graphite, 10User-fgiunchedi: Regularly purge old ores graphite metrics - https://phabricator.wikimedia.org/T169969#3429269 (10Ladsgroup) @fgiunchedi : Hey, How we can purge scores? I couldn't find anything in wikitech. [09:40:29] 10Scoring-platform-team-Backlog, 10ORES, 10Operations, 10Graphite, 10User-fgiunchedi: Regularly purge old ores graphite metrics - https://phabricator.wikimedia.org/T169969#3429416 (10fgiunchedi) @Ladsgroup this would be all old graphite metrics for ores not just scores, anyways what we do is setup a cron... [13:36:41] 10Scoring-platform-team, 10WMF-Communications, 10Wikimedia-Blog-Content: Announce new team: "Scoring Platform" - https://phabricator.wikimedia.org/T169755#3407584 (10MelodyKramer) Editing now. [13:56:21] 10Scoring-platform-team, 10WMF-Communications, 10Wikimedia-Blog-Content: Announce new team: "Scoring Platform" - https://phabricator.wikimedia.org/T169755#3430539 (10MelodyKramer) Edits in, @Halfak [15:44:53] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10ORES, 10Collaboration-Team-Triage (Collab-Team-Q3-Jan-Mar-2017), and 2 others: Enable the ORES good faith and damaging UI by default, on wikis that have these ORES models available (instead of b... - https://phabricator.wikimedia.org/T158225#3431011 [15:46:43] 10Scoring-platform-team-Backlog: ORES UI could accept more human-readable inputs such as pasted article links - https://phabricator.wikimedia.org/T170168#3421485 (10Halfak) Not sure that we *should*. I think it's best to leave this kind of thing to 3rd party tools. Maybe instead, we should advocate for an arti... [15:46:52] 10Scoring-platform-team-Backlog, 10ORES: ORES UI could accept more human-readable inputs such as pasted article links - https://phabricator.wikimedia.org/T170168#3431038 (10Halfak) [15:47:58] 10Scoring-platform-team-Backlog, 10Wikimania-Hackathon-2017: ORES @ the Wikimania Hackathon - https://phabricator.wikimedia.org/T170015#3431041 (10Halfak) One idea that came up is designing a tool to visualize the quality of articles historically. It would be cool if we could make something that looked like h... [15:56:34] 10Scoring-platform-team, 10DBA, 10Operations, 10cloud-services-team: Labsdb* servers need to be rebooted - https://phabricator.wikimedia.org/T168584#3431085 (10madhuvishy) [15:57:57] o/ Nettrom [15:58:11] o/ halfak [15:58:12] Good morning! [15:58:18] likewise! [15:58:32] I just CC'd you on an email from Andrew Lih about the article quality model [15:58:44] Apparently we have some redirects in our dataset that are labeled FA :) [15:58:48] ouch [15:59:08] haha. So I was thinking that maybe we could think about how to filter these kinds of datasets [15:59:20] Maybe even upload an update of your labeled data [15:59:30] yeah, I did that kind of thing for the importance project [15:59:44] let me find the writeup [15:59:46] Checking for "#REDIRECT" seems like one thing. Maybe there's something else we want to filter. [15:59:48] kk [16:00:01] here’s the writeup: https://meta.wikimedia.org/wiki/Research:Automated_classification_of_article_importance/Gathering_importance_data [16:00:29] and yes, you have to check if the article page is a redirect, and it’s also useful to check if the talk page path contains “/Archive” and whether the corresponding article page exists [16:00:52] Gotcha. [16:00:59] lastly, we should filter out disambiguation pages [16:01:26] I was thinking about the historical extractor too. Currently we are using your dataset for enwiki that only includes most recent classifications (with historical rev_ids) [16:01:27] I think those are the main paths to things that should be “Not an article" [16:01:51] Maybe we should use the historical extractor, but only use labels that were applied somewhat recently -- well after the scale stabilized. [16:02:04] And then apply these filters and see if we can beat the current fitness level. [16:02:35] hmm, do we know the distribution of labels in the dataset I gathered? I can do the analysis on Friday, in case you haven’t already [16:03:12] but having a cutoff date for how recent a label should have been applied does sound like a sensible heuristic to add [16:03:17] Good Q. I don't. [16:04:34] I’ll add a task to check the timestamp of the revision IDs in my dataset then, shouldn’t take more than maybe an hour or two to gather stats on that, and I can do it on Friday [16:05:10] Cool! I'm glad you'll have time to dig into this. It sounds fun but I don't think I can :( [16:05:35] BTW, if you didn't know, Andrew Lih is a huge fan of that model. Have you talked to him about it before? [16:06:09] I should hopefully be able to dig into the extractor code next week and look at implementing heuristics for gathering better data [16:07:13] I haven’t really talked much to Andrew, great to hear he likes it! :) [16:08:18] He uses it with his class. He said it's a game changer for teaching students to write Wikipedia articles. I've been asking him to blog about it. [16:11:34] That’s a great way to use the model, cool! And yeah, a blog post about that would be neat [16:13:16] 10Scoring-platform-team, 10Analytics, 10EventBus, 10ORES, and 3 others: Emit revision-score event to EventBus and expose in EventStreams - https://phabricator.wikimedia.org/T167180#3431152 (10Fjalapeno) @mobrovac cool… sounds good to me 👍 [16:16:34] halfak: did you make a Phab task based on the email from Andrew, or should I do that? [16:17:18] not yet. Please do! [16:17:22] on it [16:24:42] 10Scoring-platform-team-Backlog, 10ORES: Improve cleaning of article quality assessment datasets - https://phabricator.wikimedia.org/T170434#3431179 (10Nettrom) [16:24:50] {{done}} :) [16:24:50] You rule, Nettrom! [16:25:52] 10Scoring-platform-team-Backlog, 10Wikilabels, 10User-Zppix: Change the skip button's confirmation message on Wikilabels - https://phabricator.wikimedia.org/T168185#3431194 (10Zppix) 05Open>03Resolved Merged sorry for late update [16:40:34] OK initial round of emails done. [16:40:58] Next managery task is getting a periodic update posted. I attempted that yesterday but got side-tracked. [16:57:22] 10Scoring-platform-team, 10MediaWiki-extensions-ORES: Make list of features and locations of ORES Review Tool for handoff - https://phabricator.wikimedia.org/T167911#3431390 (10awight) [16:57:36] 10Scoring-platform-team, 10ORES, 10Cloud-VPS (Quota-requests), 10User-bd808, 10cloud-services-team (Kanban): Decrease quota for ores project to 80GB ram & 40 CPUs - https://phabricator.wikimedia.org/T170348#3431393 (10bd808) a:03bd808 [16:58:31] 10Scoring-platform-team, 10MediaWiki-extensions-ORES: Make list of features and locations of ORES Review Tool for handoff - https://phabricator.wikimedia.org/T167911#3349300 (10awight) @Catrope @Halfak @Ladsgroup When you have a minute, please double-check the "current agreements" in the task description, and... [17:04:40] * halfak runs to lunch [17:04:44] back in a bit [17:12:27] 10Scoring-platform-team, 10ORES, 10Cloud-VPS (Quota-requests), 10User-bd808: Request increase quota for ores-staging to 52GB RAM - https://phabricator.wikimedia.org/T169811#3431482 (10bd808) [17:12:30] 10Scoring-platform-team, 10ORES, 10Cloud-VPS (Quota-requests), 10User-bd808, 10cloud-services-team (Kanban): Decrease quota for ores project to 80GB ram & 40 CPUs - https://phabricator.wikimedia.org/T170348#3431480 (10bd808) 05Open>03Resolved ``` $ nova quota-show --tenant ores +---------------------... [17:46:06] 10Scoring-platform-team, 10ORES, 10revscoring, 10artificial-intelligence: Why don't timeouts work during long regular expression matching? - https://phabricator.wikimedia.org/T168965#3382708 (10ssastry) >>! In T168965#3413258, @awight wrote: > I found that the regex can be reduced to this and still causes... [18:11:18] OOf, that “]*)*” construct in that regex… not that I should be reading bug reports… IIRC that’s something that’s mentioned in Mastering Regular Expressions, but my copy of that book got lost somewhere [18:11:34] lol :) [18:16:45] Looks like I missed fajne [18:27:12] so, I'm nearly ready to pull labeled lists for fiwiki.flaggedrevs [18:27:57] apparently the file format is just json_lines, {"rev_id": 123, "your_label_name": "classification"} ? [18:27:57] The "done" column is really messy. [18:28:14] Yup, that's right [18:28:17] kk [18:28:27] I don't understand how the "done" column got so messy. [18:28:28] hehe "done" column is a nice problem to have. [18:28:31] The price of success. [18:28:37] I can't even find what we have been doing recently. [18:28:43] Well, this isn't a nice problem [18:28:48] It looks like it got shuffled. [18:29:08] Is there a system for, moving tasks to the top or bottom of each column? [18:29:48] Oh! I think I see. It looks like a bunch of non-open tasks got mover into there over the last couple of weeks. [18:29:59] I moved a ton of closed tasks [18:30:03] [18:30:07] busted. [18:30:22] It looks shuffly. But I guess there must be a point at which I'll make it past the pile of old tasks [18:31:18] The way I moved closed tasks was haphazard [18:35:24] s'ok. I think I've made sense of it now :) [18:36:27] Any idea why I wouldn't be able to log into ores-misc-01.ores.eqiad.wmflabs ? [18:36:44] Hmm... I have an idea, yeah. [18:36:46] * halfak checks [18:37:10] oh! [18:37:24] I know! Try ores-misc-01.ores-staging.eqiad.wmflabs [18:37:33] Or just ores-misc-01.eqiad.wmflabs [18:37:59] Seems to not be in this project fwiw https://tools.wmflabs.org/openstack-browser/project/ores [18:38:28] ores-misc-01.eqiad.wmflabs ftw, thanks! [18:38:54] \o/ :) [18:45:28] halfak: would you add me to the wikilabels openstack admins? [18:45:55] currently, I'm just trying to run an explain query against a labs replica db, but having auth issue [18:46:05] Woops. Yes. [18:46:08] So I was gonna steal service user creds :) [18:46:59] {{done}} [18:47:00] You rule, halfak! [18:47:05] Not sure what they have to do with one another [18:47:15] ty [18:47:18] o/ fajne [18:49:16] halfak, so, the question is probably what am i supposed to see when i make some target from a Makefile, say, $make datasets/ruwiki.labeled_revisions.20k_2015.json ? [18:49:59] Gotcha! Yeah, there should be a lot of output to the terminal but it looks like you were maybe getting a NoneType error in your output, right? [18:51:11] lots of: TypeError: 'NoneType' object is not iterable .a?....br.?...a...........?.........a...aa.a.....?ra.r......a......ra?.......a......r...r..rr.aa...r.?b?.r.Traceback (most recent call last): [18:51:52] OK so that looks OK [18:51:58] Can you show what happens for that traceback? [18:52:09] Best to use a past service like pastebin.com [18:52:13] *paste [18:53:14] that's the first time i managed to cause a segmentation fault in php [18:53:23] it gets caused in phabricator's daemons [18:54:45] halfak, https://pastebin.com/embed_js/1EcUYMUM [18:55:35] WTF [18:55:40] that's a weird error. [18:57:04] halfak: unable to login to ores-compute-01, any changes recently? [18:57:21] codezee, yes! we moved everything over to ores-misc-01 [18:57:24] (finally!) [18:57:28] And it has way more memory :) [18:57:29] oh... [18:57:47] nice.... [18:57:58] I copied your stuff into /srv/ores-compute-01-20170711/(home|srv)/ [18:58:01] halfak, let me give you some more. the beginning https://pastebin.com/embed_js/dsHMVytL [18:58:13] fajne, does the process continue to run after that error? [18:58:26] Ahh yes. it does. [18:58:26] yep [18:58:32] and ended fine [18:58:37] fwiw T168592 [18:58:37] T168592: badid_rvstartid error during autolabel - https://phabricator.wikimedia.org/T168592 [18:58:40] So here's what I propose: continue and disregard the error for now. [18:58:51] awight, oh good there's already a bug. [18:58:55] or not so fine.. [18:59:50] halfak, this is the very end https://pastebin.com/embed_js/iKZyJXLh [19:00:20] this is OK [19:00:29] It's expected. [19:00:47] We've gotten better at flagging things for review and so old data throws warnings like this. [19:01:22] halfak, it there a convenient way to look through the actual edits? [19:03:37] take the rev_id and put it at the end of this URL: [19:03:58] https://ru.wikipedia.org/wiki/?diff= [19:04:07] e.g. https://ru.wikipedia.org/wiki/?diff=2342342 [19:10:28] halfak, yes, this one i know.. [19:34:31] halfak: are reverts a type of edits? [19:35:10] Sort of yes. Reverts aren't officially tracked by the MediaWiki software so we have to do some processing to figure out which edits are reverts. [19:35:59] i am asking because i am looking at "67965161 has no labels, but was flagged for review" and it's essentially a revert of a vandal edit [19:36:11] ..wit only one word added [19:40:48] right. Not sure what you are asking me. [19:43:56] you answered) [19:53:25] fwiw, I'm dealing with this awkward query in order to get reversions. Not sure if "rv" is always present in tag_summary? https://quarry.wmflabs.org/query/20201 [20:37:57] 10Scoring-platform-team, 10editquality-modeling, 10User-Ladsgroup, 10artificial-intelligence: Flagged revs approve model to fiwiki - https://phabricator.wikimedia.org/T166235#3432729 (10awight) [20:48:31] 10Scoring-platform-team, 10editquality-modeling, 10User-Ladsgroup, 10artificial-intelligence: Flagged revs approve model to fiwiki - https://phabricator.wikimedia.org/T166235#3432759 (10awight) @Zache I could use more eyeballs on: https://quarry.wmflabs.org/query/20200 https://quarry.wmflabs.org/query/20... [20:53:04] halfak: ^ if you have spare eyeballs as well [21:25:01] 10Scoring-platform-team, 10ORES, 10editquality-modeling, 10artificial-intelligence: ORES deployment - Mid July, 2017 - https://phabricator.wikimedia.org/T170485#3432972 (10Halfak) [21:30:15] 10Scoring-platform-team-Backlog, 10ORES, 10editquality-modeling, 10Tamil-Sites, 10artificial-intelligence: Deploy reverted model for tawiki - https://phabricator.wikimedia.org/T166048#3433009 (10Halfak) [21:30:17] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train/test damaging & goodfaith models for Albanian Wikipedia - https://phabricator.wikimedia.org/T163009#3433010 (10Halfak) [21:30:19] 10Scoring-platform-team, 10editquality-modeling, 10revscoring, 10artificial-intelligence: Build damaging/goodfaith models for Romanian Wikipedia - https://phabricator.wikimedia.org/T156503#3433011 (10Halfak) [21:30:21] 10Scoring-platform-team, 10ORES, 10editquality-modeling, 10artificial-intelligence: ORES deployment - Mid July, 2017 - https://phabricator.wikimedia.org/T170485#3433008 (10Halfak) [21:31:57] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train damaging/goodfaith model for English Wiktionary - https://phabricator.wikimedia.org/T170487#3433018 (10Halfak) [21:32:04] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train damaging/goodfaith model for English Wiktionary - https://phabricator.wikimedia.org/T170487#3433018 (10Halfak) [21:32:06] 10Scoring-platform-team-Backlog, 10Wikilabels: Complete edit quality campaign in English Wiktionary - https://phabricator.wikimedia.org/T166910#3433031 (10Halfak) [21:33:07] 10Scoring-platform-team-Backlog, 10Wikilabels: Complete edit quality campaign in English Wiktionary - https://phabricator.wikimedia.org/T166910#3311630 (10Halfak) [21:33:09] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train damaging/goodfaith model for English Wiktionary - https://phabricator.wikimedia.org/T170487#3433018 (10Halfak) [21:33:15] 10Scoring-platform-team-Backlog, 10Wikilabels: Complete edit quality campaign in English Wiktionary - https://phabricator.wikimedia.org/T166910#3311630 (10Halfak) [21:33:16] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train damaging/goodfaith model for English Wiktionary - https://phabricator.wikimedia.org/T170487#3433018 (10Halfak) [21:33:37] awight, will have a look before I leave today [21:34:46] great thx [21:38:44] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10ORES, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Define a process for adding ORES filters to new wikis when ORES is enabled on those wikis - https://phabricator.wikimedia.org/T164331#3433075 (10jmatazzoni) @Halfak @Ladsgroup @aw... [21:40:35] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train reverted model for Bengali Wikipedia - https://phabricator.wikimedia.org/T170490#3433085 (10Halfak) [21:41:32] 10Scoring-platform-team, 10Bad-Words-Detection-System, 10revscoring, 10Bengali-Sites, and 2 others: Add language support for Bengali - https://phabricator.wikimedia.org/T162620#3433100 (10Halfak) [21:41:35] 10Scoring-platform-team, 10revscoring, 10artificial-intelligence: Greek language assets - https://phabricator.wikimedia.org/T122727#3433101 (10Halfak) [21:41:37] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train reverted model for Bengali Wikipedia - https://phabricator.wikimedia.org/T170490#3433085 (10Halfak) [21:41:41] 10Scoring-platform-team, 10revscoring, 10artificial-intelligence: Greek language assets - https://phabricator.wikimedia.org/T122727#1912177 (10Halfak) [21:41:43] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train reverted model for Bengali Wikipedia - https://phabricator.wikimedia.org/T170490#3433085 (10Halfak) [21:42:03] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train reverted model for Greek Wikipedia - https://phabricator.wikimedia.org/T170491#3433108 (10Halfak) [21:42:26] 10Scoring-platform-team, 10revscoring, 10artificial-intelligence: Greek language assets - https://phabricator.wikimedia.org/T122727#1912177 (10Halfak) [21:42:28] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Train reverted model for Greek Wikipedia - https://phabricator.wikimedia.org/T170491#3433108 (10Halfak) [21:50:53] I'm trying to find what it looks like when a FlaggedRevs change involved multiple diffs. I'm not convinced it's a real thing, yet. [21:51:19] 10Scoring-platform-team, 10ORES, 10revscoring, 10artificial-intelligence: Why don't timeouts work during long regular expression matching? - https://phabricator.wikimedia.org/T168965#3433217 (10Halfak) [21:51:21] 10Scoring-platform-team, 10ORES, 10Easy: ORES 500's on integers that can't be processed - https://phabricator.wikimedia.org/T168920#3433218 (10Halfak) [21:51:23] 10Scoring-platform-team, 10ORES, 10Easy, 10User-Ladsgroup: Switch ores precache to use new EventStreams - https://phabricator.wikimedia.org/T166046#3433219 (10Halfak) [21:51:25] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: ORES UI doesn't handle API errors - https://phabricator.wikimedia.org/T149118#3433220 (10Halfak) [21:51:28] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: ORES UI is broken - https://phabricator.wikimedia.org/T149117#3433221 (10Halfak) [21:51:31] 10Scoring-platform-team, 10ORES, 10editquality-modeling, 10artificial-intelligence: ORES deployment - Mid July, 2017 - https://phabricator.wikimedia.org/T170485#3433216 (10Halfak) [21:58:30] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10MediaWiki-extensions-ORES, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Reduce very long search times on RC Page when using ORES for rare combos - https://phabricator.wikimedia.org/T164796#3433257 (10jmatazzoni) [22:00:32] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10MediaWiki-extensions-ORES, 10Wikidata, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Support ORES for propagated Wikidata edits - https://phabricator.wikimedia.org/T158025#3433289 (10jmatazzoni) [22:18:25] 10Scoring-platform-team, 10editquality-modeling, 10User-Ladsgroup, 10artificial-intelligence: Flagged revs approve model to fiwiki - https://phabricator.wikimedia.org/T166235#3433383 (10awight) I haven't been able to find any multi-diff approvals. Using the condition `AND r1.rev_parent_id != r2.rev_id`, I... [22:21:12] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10MediaWiki-extensions-ORES, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Reduce very long search times on RC Page when using ORES for rare combos - https://phabricator.wikimedia.org/T164796#3433406 (10jmatazzoni) a:03Catrope [22:28:38] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10MediaWiki-extensions-ORES, 10Wikidata, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Support ORES for propagated Wikidata edits - https://phabricator.wikimedia.org/T158025#3433462 (10jmatazzoni) @Mattflaschen-WMF, is this blocked? [22:36:53] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10MediaWiki-extensions-ORES, 10Wikidata, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Support ORES for propagated Wikidata edits - https://phabricator.wikimedia.org/T158025#3433476 (10Mattflaschen-WMF) @jmatazzoni There's nothing... [22:45:12] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10ORES, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Define a process for adding ORES filters to new wikis when ORES is enabled on those wikis - https://phabricator.wikimedia.org/T164331#3433567 (10awight) Thanks for thinking about... [22:48:18] I'm going to take advantage of this spam to make more spam. [22:48:23] Resolved tasks incoming! [22:48:33] spam, spam, spam, spam, spam, spam, spam, spam, eggs, and spam [22:48:35] * awight braces against the tractor beam [22:48:50] 10Scoring-platform-team: Add test to ensure timeout of functions taking too long - https://phabricator.wikimedia.org/T170205#3433574 (10Halfak) 05Open>03Resolved [22:48:50] Here we go [22:48:52] 10Scoring-platform-team, 10ORES, 10revscoring, 10artificial-intelligence: Why don't timeouts work during long regular expression matching? - https://phabricator.wikimedia.org/T168965#3433575 (10Halfak) [22:48:53] * awight switches to pounding the table and chanting with Nettrom [22:48:53] OMG [22:48:57] _o/ [22:49:02] 10Scoring-platform-team, 10ORES, 10Services (watching), 10User-Ladsgroup: ORES POST precaching always fails with 500 - https://phabricator.wikimedia.org/T168674#3433590 (10Halfak) 05Open>03Resolved [22:49:04] /o\ [22:49:04] 10Scoring-platform-team, 10editquality-modeling, 10artificial-intelligence: Remove other non-badwords from huwiki model. - https://phabricator.wikimedia.org/T167231#3433595 (10Halfak) 05Open>03Resolved [22:49:06] 10Scoring-platform-team, 10ORES: Add API tests to ORES CI - https://phabricator.wikimedia.org/T168007#3433594 (10Halfak) 05Open>03Resolved [22:49:10] _o_ [22:49:11] 10Scoring-platform-team, 10ORES: Mid June 2017 ORES deployment - https://phabricator.wikimedia.org/T168099#3356127 (10Halfak) 05Open>03Resolved [22:49:13] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: Early June 2017 ORES prod deploy - https://phabricator.wikimedia.org/T167223#3433596 (10Halfak) [22:49:15] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: Early June 2017 ORES prod deploy - https://phabricator.wikimedia.org/T167223#3321234 (10Halfak) [22:49:17] 10Scoring-platform-team, 10articlequality-modeling, 10artificial-intelligence: Implement page_level (OCR) model for frwikisource - https://phabricator.wikimedia.org/T167196#3433598 (10Halfak) 05Open>03Resolved [22:49:19] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: Early June 2017 ORES prod deploy - https://phabricator.wikimedia.org/T167223#3321234 (10Halfak) 05Open>03Resolved [22:49:19] Spam swimming is the best [22:49:21] 10Scoring-platform-team, 10Wikilabels: [Discuss] Wikilabels campaign names should reflect number of tasks - https://phabricator.wikimedia.org/T166479#3433600 (10Halfak) 05Open>03Resolved a:03Halfak [22:49:23] 10Scoring-platform-team, 10ORES, 10revscoring, 10artificial-intelligence: Language assets for Tamil - https://phabricator.wikimedia.org/T166052#3433602 (10Halfak) 05Open>03Resolved [22:49:25] 10Scoring-platform-team, 10ORES, 10editquality-modeling, 10Tamil-Sites, and 2 others: Train/test reverted model for tawiki - https://phabricator.wikimedia.org/T166051#3433604 (10Halfak) [22:49:28] 10Scoring-platform-team, 10ORES, 10editquality-modeling, 10artificial-intelligence: ORES deployment - Mid July, 2017 - https://phabricator.wikimedia.org/T170485#3432972 (10Halfak) [22:49:29] 10Scoring-platform-team, 10ORES, 10Easy, 10User-Ladsgroup: Switch ores precache to use new EventStreams - https://phabricator.wikimedia.org/T166046#3433603 (10Halfak) 05Open>03Resolved [22:49:31] 10Scoring-platform-team, 10MediaWiki-extensions-ORES, 10User-Ladsgroup: No new data on ores_classification on beta labs since march memory issue - https://phabricator.wikimedia.org/T165716#3433607 (10Halfak) 05Open>03Resolved [22:49:37] * Nettrom goes to look for a viking costume to fit in awight’s scene [22:49:50] Oh yeah. Look at that empty "done" column. So sweet. https://phabricator.wikimedia.org/project/board/1306/ [22:49:53] 10Scoring-platform-team, 10articlequality-modeling, 10artificial-intelligence: Implement wp10 model for trwiki - https://phabricator.wikimedia.org/T164671#3433610 (10Halfak) 05Open>03Resolved [22:49:53] 10Scoring-platform-team, 10ORES, 10Epic: Using TensorFlow in ORES - https://phabricator.wikimedia.org/T161375#3433626 (10Halfak) [22:49:55] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: Early June 2017 ORES prod deploy - https://phabricator.wikimedia.org/T167223#3321234 (10Halfak) [22:49:56] Must fill it with things! [22:49:57] 10Scoring-platform-team, 10Wikilabels: Complete Romanian Wikipedia edit quality campaign - https://phabricator.wikimedia.org/T156517#3433623 (10Halfak) 05Open>03Resolved a:03Halfak [22:49:59] 10Scoring-platform-team, 10editquality-modeling, 10revscoring, 10artificial-intelligence: Build damaging/goodfaith models for Romanian Wikipedia - https://phabricator.wikimedia.org/T156503#3433625 (10Halfak) [22:50:01] 10Scoring-platform-team, 10revscoring, 10Easy, 10artificial-intelligence: Store the detailed system information inside of model files. - https://phabricator.wikimedia.org/T160223#3433622 (10Halfak) 05Open>03Resolved [22:50:02] Oh no there's more! [22:50:03] 10Scoring-platform-team: Scoring platform team FY17-18 Q4 goals - https://phabricator.wikimedia.org/T164210#3225672 (10Halfak) [22:50:05] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: Initial assessment of TensorFlow on ORES data - https://phabricator.wikimedia.org/T161376#3433619 (10Halfak) 05Open>03Resolved [22:50:07] 10Scoring-platform-team, 10ORES, 10Patch-For-Review, 10User-Ladsgroup: ORES swagger doc based API requests do not work - https://phabricator.wikimedia.org/T162184#3433618 (10Halfak) 05Open>03Resolved [22:50:09] 10Scoring-platform-team-Backlog, 10MediaWiki-extensions-ORES: Deploy ORES Review Tool on Romanian Wikipedia - https://phabricator.wikimedia.org/T156518#3433633 (10Halfak) [22:50:12] 10Scoring-platform-team, 10MediaWiki-extensions-ORES, 10Patch-For-Review, 10User-Ladsgroup: Special:Contribution should have javascript highlighting as well - https://phabricator.wikimedia.org/T155930#3433629 (10Halfak) 05Open>03Resolved [22:50:15] 10Scoring-platform-team, 10editquality-modeling, 10revscoring, 10artificial-intelligence: Build damaging/goodfaith models for Romanian Wikipedia - https://phabricator.wikimedia.org/T156503#2977066 (10Halfak) 05Open>03Resolved [22:50:17] 10Scoring-platform-team, 10ORES, 10editquality-modeling, 10artificial-intelligence: ORES deployment - Mid July, 2017 - https://phabricator.wikimedia.org/T170485#3432972 (10Halfak) [22:50:19] 10Scoring-platform-team-Backlog, 10MediaWiki-extensions-ORES: Enable ORES Review Tool in Romanian Wikipedia - https://phabricator.wikimedia.org/T156501#3433635 (10Halfak) [22:50:22] 10Scoring-platform-team, 10MediaWiki-extensions-ORES, 10MW-1.30-release-notes (WMF-deploy-2017-06-27_(1.30.0-wmf.7)), 10Patch-For-Review, 10User-Ladsgroup: Some bad changes are not highlighted in enhanced recent changes or watchlist - https://phabricator.wikimedia.org/T155903#3433630 (10Halfak) 05Open>... [22:50:25] 10Scoring-platform-team, 10editquality-modeling, 10Epic, 10artificial-intelligence: [Epic] Edit quality models (damaging/goodfaith) - https://phabricator.wikimedia.org/T130213#3433636 (10Halfak) [22:50:27] 10Scoring-platform-team, 10ORES, 10editquality-modeling, 10artificial-intelligence: ORES deployment - Mid July, 2017 - https://phabricator.wikimedia.org/T170485#3432972 (10Halfak) [22:50:29] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: ORES UI is broken - https://phabricator.wikimedia.org/T149117#3433634 (10Halfak) 05Open>03Resolved [22:50:31] 10Scoring-platform-team, 10ORES, 10User-Ladsgroup: ORES UI doesn't handle API errors - https://phabricator.wikimedia.org/T149118#3433632 (10Halfak) 05Open>03Resolved [22:50:33] 10Scoring-platform-team, 10editquality-modeling, 10Bengali-Sites, 10artificial-intelligence: Train reverted model for Bengali Wikipedia - https://phabricator.wikimedia.org/T170490#3433641 (10Liuxinyu970226) [22:50:55] 10Scoring-platform-team, 10Edit-Review-Improvements-RC-Page, 10ORES, 10Collaboration-Team-Triage (Collab-Team-Q1-Jul-Sep-2017): Add a new config setting to enable ORES UI features - https://phabricator.wikimedia.org/T170500#3433642 (10awight) [22:50:55] OMG I think it ended [22:51:12] Back to your regular scheduled spam [22:51:43] halfak: btw I think I need a phone call to make any more progress on fiwiki.flaggedrevs, so I may drag another task into the workboard. [22:52:01] I'm just about ready to take a look [22:52:09] Not quite sure I understand where you are stuck [22:52:17] I need to copy the status update into an email first [22:52:34] this can wait until tomorrow, n.b.d. [22:56:31] OK I won't strain myself :) [22:56:38] mandatory spam video: https://youtu.be/anwy2MPT5RE [22:56:52] apologies for the off-topic post :D [22:57:34] lol [22:58:00] seems appropriate [22:58:11] * awight stamps "OK for import" [22:59:56] halfak: quick link to my questions, when you're ready, https://phabricator.wikimedia.org/T166235#3432759 [23:01:05] There's some other stuff I might not understand well enough, but I was going to run into that later. Such as, whether I can provide disjoint labels to training. I currently have one data set with "approved": 1, and the other has "reverted": 1. [23:01:34] Sure! we'll just shuffle them together when training. [23:01:50] Actually it doesn't matter what order they come in. We just need to shuf for selecting a test set [23:02:09] What label would I have the training use, though? [23:02:44] Oh! I see. So there are two separate labels not two sides of a boolean [23:02:59] Or are there three labels. [23:03:04] right, two labels so far [23:03:05] approves, ???, reverted [23:03:11] *approved [23:03:19] approved, reverted [23:03:51] For the "reverted" is this actually recorded as structured data? [23:04:29] I think I should look at your queries. :) [23:04:47] I had a question about that, too. zache's query was looking for rows with tag_summary.ts_tags =~ /\brv\b/ [23:05:00] ... and I haven't been able to reverse engineer where that comes from [23:05:02] Oh. shit [23:05:09] so donno if it's specific to flaggedrevs or not [23:10:01] Is there an .. unapprove? [23:10:09] no! [23:10:12] damn [23:10:17] There is, but it means "removed approval" [23:10:19] afaict [23:10:35] Oh OK [23:10:43] yeah I would be sad if we were just cloning the reverted data set [23:10:46] So, what does "approval" mean [23:10:53] We might be. [23:11:09] Maybe "I reviewed this and if it was bad, I reverted it" [23:11:12] This is the best guide I have, https://phabricator.wikimedia.org/T166235#3389498 [23:11:54] If you really want to be confused, https://www.mediawiki.org/wiki/Help:Extension:FlaggedRevs [23:12:18] 'all can be thinked as "checked" reviews.' [23:12:20] Oh! [23:12:35] So "approved" means something like "patrolled" [23:12:43] afaict yes [23:12:43] "I looked at this and did a thing" [23:13:07] So, here's what I propose: We try building the revert dataset only on "approved" revisions. [23:13:08] and approved2 is another editor confirming the original decision, I believe [23:13:21] mm? So approved which was then unapproved? [23:13:31] or just ditch the negative set? [23:13:44] Yeah. Just build the negative set with our autolabeler. [23:13:54] * awight blinks. [23:14:08] ok I saw some of that in the makefile, I'll figure it out. [23:14:09] editquality autolabel -h [23:14:50] https://github.com/wiki-ai/editquality/blob/master/editquality/utilities/autolabel.py#L233 [23:14:59] oh fwiw, that's right, zache says the "2" is not a double-check, it's something about the quality level. [23:15:29] "reverted_for_damage" == not a self revert or reverted_back_to by someone else or matching a comment_exclusion [23:15:35] We only use comment exclusions in Wikidata [23:15:41] Because client edits are stupid [23:16:05] Any insight into "multi-diff" flagged revs? [23:16:23] I can't convince myself that these exist. [23:17:41] Yeah. I still haven't figured out WTF. [23:18:11] kk, I'll wait for feedback from zache [23:18:42] Yeah. Where did he bring that up? [23:18:59] In the task description, > In a single approved diff there can be one or more edits and editors [23:19:36] FlaggedRevs is an impressive piece of software--unfortunately, it's impenetrable to reverse engineering. [23:20:05] awight: Platform Team, congrats! [23:20:15] \o/ [23:20:27] err Scoring Team I mean. [23:20:27] (how does he know ;) [23:20:51] halfak just emailed about it to the AI list. [23:21:31] I am proud to be toiling alongside a fine eater of toast [23:22:08] no idea what that means, but I'm happy for you! [23:22:33] me neither! I was cribbing from halfak's signature [23:23:03] I think it's like https://en.wikipedia.org/wiki/One-Punch_Man [23:23:27] Was just feeling silly as I was putting titles in there :D [23:24:30] OK I'm off. Have a good one folks! [23:24:35] o/ [23:44:21] oops, missed halfak. can anyone here help me with running this: $ make models/ruwiki.damaging.gradient_boosting.model [23:46:30] fajne: sure--afaik that command should do what you need [23:46:36] are the dependencies failing to build? [23:54:43] awight: yes. i am slowly pip3 installing the packages (because setup didn't work for me) but stack with enchant [23:55:12] fwiw, what does "which pip" give you? [23:55:53] I don't know what enchant is [23:56:08] I've had luck with this sequence, [23:56:17] virtualenv ~/.env -p python3 [23:56:21] the dictionary they use. vvvery inportant [23:56:28] source ~/.env/bin/activate [23:56:35] pip install -r requirements.txt [23:56:41] oh, i am in ma virualenv of cource [23:56:48] kk [23:58:34] did requirements, but still: ImportError: No module named 'enchant' [23:59:27] sounds like the requirements broke halfway, that's an awkward state to be in