[04:21:58] https://fi.wikipedia.org/wiki/Toiminnot:ORESModels [04:22:04] ^ what's up with the ??? [04:32:26] Also what's going on when the intervals don't overlap? [04:32:35] https://lv.wikipedia.org/wiki/Special:ORESModels [08:40:41] o/ [08:40:54] groceryheist: not yet, there are some plans for futre [12:42:51] I go eat something, will be back [13:12:47] back now [13:47:37] I've got to hop offline to head to the U in a bit. Should be back online for my 9:30 meeting. [15:37:36] hare: https://phabricator.wikimedia.org/T122676 :) [15:39:34] groceryheist, actually I'm in IRC now because we were fast with standup. [15:39:58] This is a great task/task-tree Amir1, thank you. [15:40:02] Nice documentation work :) [15:42:57] AFK for a bit. [15:57:53] 10MediaWiki-extensions-ORES, 10ORES, 10Scoring-platform-team: Why are there three Q-marks (???) in threshholds in Special:ORESModels? - https://phabricator.wikimedia.org/T221870 (10Groceryheist) [16:00:30] 10MediaWiki-extensions-ORES, 10ORES, 10Scoring-platform-team: Non-overlapping threshholds in ORESModels on lvwiki - https://phabricator.wikimedia.org/T221871 (10Groceryheist) [16:01:53] back [16:03:59] halfak: do those tasks seem ok? [16:04:59] 10MediaWiki-extensions-ORES, 10ORES, 10Scoring-platform-team, 10Growth-Team: Non-overlapping threshholds in ORESModels on lvwiki - https://phabricator.wikimedia.org/T221871 (10Halfak) [16:22:08] 10MediaWiki-extensions-ORES, 10ORES, 10Scoring-platform-team, 10Growth-Team: Why are there three Q-marks (???) in threshholds in Special:ORESModels? - https://phabricator.wikimedia.org/T221870 (10Halfak) [17:06:22] 10MediaWiki-extensions-ORES, 10ORES, 10Scoring-platform-team, 10Growth-Team: Why are there three Q-marks (???) in threshholds in Special:ORESModels? - https://phabricator.wikimedia.org/T221870 (10Halfak) In conversation with @Ladsgroup, we couldn't figure out whether unused thresholds should be invisible (... [17:09:03] groceryheist, just made some updates to the task. [17:09:09] *tasks [17:09:14] But otherwise, great. Thank you [17:09:32] Also, we should get you into the next WMF Research Group meeting to talk about your work. [17:10:05] I wasn't able to make it today because of annual planning craziness --> tons of meetings. [17:27:46] o/ I'm thinking of grabbing lunch. [17:28:02] Anyone need anything before I run away for a bit? Looking at you, groceryheist :) [17:36:25] halfak: (semi-urgent) do you think we should ask for DB resources for next fiscal year, what with Jade? [17:37:18] Good Q. Could you reach out to Corey Floyd to find out if he needs us to coordinate with him. As it stands, I don't think that *we* need and CapEx for DB, but Ops/Core platform might. [17:37:28] Missed a "?" in there. [17:38:56] OK lunch! [17:46:41] halfak|Lunch: if you have any DB-related requests for next years that are not Jade related, I need to know about them. Otherwise, I'm going to pursue this thread with Core Platform as the Jade product owner and coordinate with you to the extent it concerns your budget authority or technical leadership, but otherwise leave you alone if this is something that Core Platform can handle on their own. Sound good? [18:02:15] hare, sounds about right to me. [18:02:35] Only other concerns I have about DB are novel engineering around embeddings. [18:02:45] I can also bring that up [18:02:49] I still need to reach out to dsaez to start building some estimates. [18:03:17] hare, that sounds good. [18:03:52] Gist is that we're looking into strategies for doing approximate nearest neighbors type queries on dense data. [18:04:30] Eventually we want to index *everything* but in the next FY, I think we're probably going to be looking into a few target wikis. [18:06:06] do you think Jade will lead to 1% database growth over the next fiscal? I remember that being Adam's estimate [18:36:29] hare, it will if we deploy to enwiki and other big wikis [18:36:44] Which I'd like to wait for revision sharding for [18:46:07] halfak: things are going good [18:47:14] Good :) [18:47:37] for research team meeting next week. If we sent out the outline a few days in advance do you think folks would take a look? [18:48:20] update to damaging fpr plot: [18:48:20] https://teblunthuis.cc/outgoing/Damaging_fpr.png [18:53:11] Another Q: What's up with ORES and dewiki? [18:53:37] dewiki isn't sure that they like us. So we're waiting on someone to carry the burden of getting community buy in there. [18:54:09] groceryheist, probably not going to get most of the people in the meeting to view the outline, but maybe some. [18:54:13] i see but we built the model already [18:54:19] I bet J-Mo would be interested. [18:54:23] Right groceryheist [18:54:43] got it [18:54:44] If there weren't already a ton of stuff on hare's plate, I'd ask him to reach out to the people who did labeling work. [18:55:11] that should be made as a phabricator ticket [18:55:15] Usually the Growth team would handle this part, but they have moved onto new projects -- so it falls on us. [18:55:30] hare, we have one for "deploy ORES filters to dewiki" or something like that. [19:37:05] https://phabricator.wikimedia.org/T215354 [19:37:19] hare, rcfilters isn't our product but we do support it. [19:37:47] Should we push the Growth team to do their own community outreach? [19:38:50] See https://phabricator.wikimedia.org/T164331 [19:39:04] I think it is clear from that task that Growth should be handling this. [19:48:16] 10Scoring-platform-team, 10Edit-Review-Improvements-Integrated-Filters, 10Growth-Team, 10Patch-For-Review: Enable ORES RCFilters for German Wikipedia (dewiki) - https://phabricator.wikimedia.org/T215354 (10Halfak) @JTannerWMF, per {T164331}, I believe we'd decided that it is the #Growth team's responsibili... [19:54:27] hare, let me know when you have some time to sync up on annual planning stuff. I'm around for 2 more hours. [19:54:39] I think we'll need about 30 minutes. [22:32:46] I'm off. Have a good one, folks! [22:33:37] later