[12:42:36] 10ORES, 10Scoring-platform-team, 10Scap: Investigate: why are we getting overload errors during ORES deployments? - https://phabricator.wikimedia.org/T213116 (10Ladsgroup) Thanks for the blogpost @akosiaris I had a complete misunderstanding of what blue/green deployment means. What I wanted to suggest for no... [12:45:54] 10ORES, 10Scoring-platform-team, 10Wikilabels: Upgrade to flask 1.0 - https://phabricator.wikimedia.org/T212002 (10Ladsgroup) ores doesn't respond when I go to flask 1.0 🤔🤔🤔 [13:08:17] 10ORES, 10Scoring-platform-team, 10Wikilabels: Upgrade to flask 1.0 - https://phabricator.wikimedia.org/T212002 (10Ladsgroup) It was this: ` Exception in thread Thread-1: Traceback (most recent call last): File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner self.run() File "/usr/lib... [13:15:20] 10ORES, 10Scoring-platform-team (Current), 10Wikilabels, 10User-Ladsgroup: Upgrade to flask 1.0 - https://phabricator.wikimedia.org/T212002 (10Ladsgroup) a:03Ladsgroup https://github.com/wikimedia/ores/pull/315 [13:19:27] wikimedia/ores#1278 (flask1.0 - 7416af7 : Amir Sarabadani): The build passed. https://travis-ci.org/wikimedia/ores/builds/491620094 [13:30:07] 10ORES, 10Scoring-platform-team (Current), 10Wikilabels, 10User-Ladsgroup: Upgrade to flask 1.0 - https://phabricator.wikimedia.org/T212002 (10Ladsgroup) https://github.com/wikimedia/wikilabels/pull/252 [14:30:05] o/ [15:40:17] 10Scoring-platform-team (Current), 10editquality-modeling, 10artificial-intelligence: Implement better defaults for autolabel utlity - https://phabricator.wikimedia.org/T215671 (10Ladsgroup) Rebuilding everything with the new values, it might take some days to finish. [16:16:42] halfak: hey can you take a look at this? https://github.com/wikimedia/ores/pull/314 I want to do something else after it [16:19:08] looking [16:19:22] Sorry I was AFK was having trouble with the university wifi. [16:19:34] No one is in the office today. Apparently it is a now day ^_^ [16:19:38] I biked in. [16:19:54] basically all of our tests can be ran with docker now \o/ [16:20:20] nice! (no worries about time) [16:22:15] 10Scoring-platform-team, 10revscoring, 10Chinese-Sites, 10artificial-intelligence: Chinese language utilities - https://phabricator.wikimedia.org/T109366 (10Halfak) Flattening could work if there is a 1:1 matching. It looks like this could work: https://pypi.org/project/zhconv/ [16:24:23] Amir1, just a couple of notes. [16:24:49] One question too. I don't see where the API is started before the test_api script is run. [16:27:00] All of that is being done in Dockerfile and docker-compose.yaml [16:27:02] halfak: ^ [16:30:06] Aha. So the other question is why delete docker-compose and re-download it? [16:30:11] Is that a bug/work-around? [16:31:04] Probably. That thing is an exact copy-paste from the manual: https://docs.travis-ci.com/user/docker/#using-docker-compose [16:31:09] halfak: ^ [16:31:46] Interesting. It's weird that you delete it and re-download it, isn't it? Or is this some sort of common pattern with docker? [16:32:31] Yeah, it's probably because it has the old version [16:32:41] the old one is crap [16:35:35] I actually might get away with that version, let me try it [16:37:59] I need to leave [16:38:08] (sent an email) [16:38:17] see you on Wednesday [16:47:22] OK take care. Sorry to delay you! Thanks for answering my questions. [17:35:34] I wonder if we added the AI mailing list to a spam list for every AI and AI-adjacent conference in existence. [18:56:02] 10Scoring-platform-team (Current), 10editquality-modeling, 10artificial-intelligence: Implement better defaults for autolabel utlity - https://phabricator.wikimedia.org/T215671 (10Halfak) Great. Thank you. I put this task together in a hurry so I didn't specify which wikis should have a revert-window. I d... [18:58:02] 10Scoring-platform-team (Current), 10editquality-modeling, 10artificial-intelligence: Implement better defaults for autolabel utlity - https://phabricator.wikimedia.org/T215671 (10Halfak) It looks like we didn't have a window historically. I think this is the last version of the makefile before we switched... [20:51:01] 10Scoring-platform-team (Current), 10Wikilabels, 10articlequality-modeling, 10WMSE-Development-Support-2019 (Automatic article quality assessment), 10artificial-intelligence: Build article quality model for svwiki - https://phabricator.wikimedia.org/T202202 (10Halfak) a:05hoo→03None [20:55:14] 10Scoring-platform-team (Current), 10Wikilabels, 10articlequality-modeling, 10WMSE-Development-Support-2019 (Automatic article quality assessment), 10artificial-intelligence: Build article quality model for svwiki - https://phabricator.wikimedia.org/T202202 (10Halfak) a:03Halfak Looks like hoo can't ta... [21:04:31] srrodlund, do I remember right that you're working to revitalize the brown bag series? [21:05:13] Aha! I see an email now :) [21:12:43] 10ORES, 10Scoring-platform-team: Investigate using MLEAP for serilization of models - https://phabricator.wikimedia.org/T215732 (10Halfak) I don't see anywhere in the readme that talks about model serialization formats for models. It looks like this instead serializes a "pipeline" which seems to convert raw d... [21:15:02] 10ORES, 10Scoring-platform-team (Current): Make ORES articlequality prediction score articles from 0 to 10 in euwiki - https://phabricator.wikimedia.org/T214639 (10Halfak) 05Open→03Resolved @Theklan and I have discussed this extensively on-wiki. See https://eu.wikipedia.org/wiki/Lankide_eztabaida:EpochFai... [21:18:18] 10ORES, 10Scoring-platform-team, 10Continuous-Integration-Config: ORES wheels deployment repo Jenkins job should gate-and-submit - https://phabricator.wikimedia.org/T211041 (10Halfak) p:05Triage→03Normal [21:20:56] 10Scoring-platform-team: Accidentally merged models in editquality; need to rewrite history - https://phabricator.wikimedia.org/T210678 (10Halfak) 05Open→03Declined I'm confused about what the problem is and what needs to be done here. I'm going to "decline" because it seems that the repo is working as expe... [21:27:00] 10ORES, 10Scoring-platform-team: Consider committing all non-private datasets to the repos - https://phabricator.wikimedia.org/T192617 (10Halfak) We can always compress these files, but I agree that it still wouldn't be as good as diff-compression. We probably don't want to store cache files, but we could sto... [21:27:16] halfak: wow that was a coincidence! [21:27:26] any interest in presenting??? [21:27:48] Oh yeah! Totally. Do you have a schedule or should I just propose a talk and we can work out timing? [21:28:43] 10Scoring-platform-team, 10Research, 10Wikilabels: WikiLabels "Labeling Unsourced Statements II" interface is broken - https://phabricator.wikimedia.org/T208337 (10Halfak) @Miriam & @bmansurov, is this labeling campaign still relevant or should we "decline" this task? [21:29:47] 10MediaWiki-extensions-ORES, 10Scoring-platform-team: Write tests for ApiHooksHandler - https://phabricator.wikimedia.org/T210638 (10Halfak) p:05Triage→03Normal [21:30:05] 10MediaWiki-extensions-ORES, 10Scoring-platform-team: Write tests for ApiHooksHandler - https://phabricator.wikimedia.org/T210638 (10Halfak) @Ladsgroup, can we add this to your backlog? [21:31:31] you can propose a talk in phab let me give you the link [21:32:02] https://www.mediawiki.org/wiki/Project:Calendar/How_to_schedule_an_event#Tech_talks [21:32:38] Basically you just fill out the phab task, and I'll reach out to you to schedule a good time (also feel free to put times on there that work for you) [21:33:16] halfak: feel free to share that around, too. Tech talks aren't just for staff. Anyone in the technical community should feel free to propose one. [21:34:00] Great! I'll be working on that soon :) [21:38:25] 10Jade, 10Scoring-platform-team (Current), 10User-Testing, 10Epic: [EPIC] Jade Mockups: Prepare mockups of Jade initial feature set for user testing - https://phabricator.wikimedia.org/T210535 (10Halfak) [21:39:27] 10Scoring-platform-team, 10Wikilabels: Make a wikilabels view to see a single task. - https://phabricator.wikimedia.org/T208239 (10Halfak) [21:44:42] 10Scoring-platform-team, 10Research, 10Wikilabels: WikiLabels "Labeling Unsourced Statements II" interface is broken - https://phabricator.wikimedia.org/T208337 (10bmansurov) 05Open→03Declined Yes, I think we can decline the task as we're no longer collecting data. [23:22:43] This might be relevant https://www.nytimes.com/2019/02/11/business/ai-artificial-intelligence-trump.html [23:23:24] 10MediaWiki-extensions-ORES, 10Scoring-platform-team, 10Wikimania-Hackathon-2018, 10Patch-For-Review: Expose model and filter statistics on a special page - https://phabricator.wikimedia.org/T195199 (10Halfak) Looks like this is done. @Catrope you agree? [23:25:58] 10ORES, 10Scoring-platform-team: Model information UI (graphs and statistics) - https://phabricator.wikimedia.org/T140364 (10Halfak) 05duplicate→03Open Re-opening. This is more about plotting out model statistics. Like fitness and precision/recall curves. [23:26:48] 10ORES, 10Scoring-platform-team: How far is ORES in training? - https://phabricator.wikimedia.org/T207548 (10Halfak) The best way to track completion of the labeling campaigns is to use our support tool: https://tools.wmflabs.org/ores-support-checklist/ There's not a great way to see how much more data ORES n... [23:27:14] 10ORES, 10Scoring-platform-team: How far is ORES in training? - https://phabricator.wikimedia.org/T207548 (10Halfak) [23:27:16] 10ORES, 10Scoring-platform-team: Model information UI (graphs and statistics) - https://phabricator.wikimedia.org/T140364 (10Halfak)