[01:53:38] 06Revision-Scoring-As-A-Service, 10revscoring, 07Spike: [Spike] Investigate HashingVectorizer - https://phabricator.wikimedia.org/T128087#2566519 (10Sabya) Shouldn't it be (2^20)*0.0005? It is 524. [03:05:05] 10Revision-Scoring-As-A-Service-Backlog: [Discuss] automatic statement extraction from Wikipedia - https://phabricator.wikimedia.org/T123337#2566569 (10Ladsgroup) I guess this is better to be around but lowest is a proper assessment. [04:08:35] 06Revision-Scoring-As-A-Service, 10revscoring, 07Spike: [Spike] Investigate HashingVectorizer - https://phabricator.wikimedia.org/T128087#2566578 (10Sabya) @Halfak I think I got it wrong about the threshold in my above comment. Now I get it as the threshold is to keep the features whose importance is >= th... [08:57:23] akosiaris: thanks :) [09:01:50] Amir1: o/ [09:02:44] Hey, the result is there already, it's better to wait for 24 hours before implementing them in grafana [09:19:52] 10Revision-Scoring-As-A-Service-Backlog, 10rsaas-editquality, 15User-Ladsgroup: Include specific user groups in the trwiki edit quality model - https://phabricator.wikimedia.org/T140474#2566986 (10Ladsgroup) @Superyetkin: Hey, I'm going to pick this up and do it. I checked and trwiki doesn't have "whitelist"... [09:26:20] 10Revision-Scoring-As-A-Service-Backlog, 10rsaas-editquality, 15User-Ladsgroup: Include specific user groups in the trwiki edit quality model - https://phabricator.wikimedia.org/T140474#2566991 (10Superyetkin) Yes, the correct group name is "autoreview". The change would also work for the "patroller" group.... [10:34:16] 06Revision-Scoring-As-A-Service, 10rsaas-editquality, 15User-Ladsgroup: Include specific user groups in the trwiki edit quality model - https://phabricator.wikimedia.org/T140474#2567065 (10Ladsgroup) [10:41:56] 06Revision-Scoring-As-A-Service, 10ORES, 13Patch-For-Review, 15User-Ladsgroup: Add uwsgi-related metrics to grafana - https://phabricator.wikimedia.org/T143081#2567071 (10Ladsgroup) Who deleted the average response time graph and why? (Why grafana does not have history) [11:56:31] 06Revision-Scoring-As-A-Service, 10ORES, 06Operations, 07Puppet: Clean up puppet & configs for ORES - https://phabricator.wikimedia.org/T142002#2567190 (10akosiaris) >>! In T142002#2565031, @Halfak wrote: > Hi @akosiaris. It seems that you've commented on the naming scheme. Er no. I was commenting on t... [13:22:00] o/ [13:42:28] 06Revision-Scoring-As-A-Service, 10ORES, 06Operations, 07Puppet: Clean up puppet & configs for ORES - https://phabricator.wikimedia.org/T142002#2567443 (10Halfak) @akosiaris, I see. The current state is on the left side of the arrows --> and the proposed state is on the right side of the arrows. I made... [13:44:20] 06Revision-Scoring-As-A-Service, 10rsaas-editquality, 15User-Ladsgroup: Include specific user groups in the trwiki edit quality model - https://phabricator.wikimedia.org/T140474#2567448 (10Ladsgroup) https://github.com/wiki-ai/editquality/pull/44 [13:49:56] 06Revision-Scoring-As-A-Service, 10ORES, 13Patch-For-Review, 15User-Ladsgroup: Add uwsgi-related metrics to grafana - https://phabricator.wikimedia.org/T143081#2567455 (10Halfak) I'm not sure how that got removed, but I did remove the VSZ graph since VSZ is a confusing if not misleading stat. [13:55:03] 06Revision-Scoring-As-A-Service, 10revscoring, 07Spike: [Spike] Investigate HashingVectorizer - https://phabricator.wikimedia.org/T128087#2567468 (10Halfak) +1 for >= threshold. I think we'll want to try training the GradientBoosting model again with 77 + 375 features. You'll likely need to flatten the v... [14:14:22] 06Revision-Scoring-As-A-Service, 10ORES, 06Operations, 07Puppet: Clean up puppet & configs for ORES - https://phabricator.wikimedia.org/T142002#2567511 (10akosiaris) >>! In T142002#2567443, @Halfak wrote: > @akosiaris, I see. The current state is on the left side of the arrows --> and the proposed state i... [14:39:49] Pushing new models to staging [14:40:14] And now getting on bike to head to UMN.. [16:03:37] Hello, does WikiLabels work with python 3 or 2? [16:06:14] bmansurov: python 3 [16:06:20] python 2 is not supported [16:09:05] Amir1: thanks [16:13:01] :) [16:36:47] o/ [16:36:49] forgot to reconnect [16:36:57] am reading a paper quick and then will check on staging and push to labs main [16:37:25] o/ Amir1 [16:37:37] I saw that you worked on the trwiki models and we got a fitness boost with those user rights, so I merged and included that [16:37:47] I think it will go nicely with the version increment :) [17:01:44] Amir1: I've cloned the WikiLabels repo and have followed the instructions. Looks like some things have evolved. Do you have a sample dev config file that I can use? Should I use the 00-main file? [17:02:12] Amir1: I've actually installed the package using pip after cloning didn't work [17:02:25] so now I have the wikilabels command available [17:12:40] o/ bmansurov [17:12:59] I recommend the 00-main.yaml if you want the install to look like labs [17:13:02] o/ [17:13:10] otherwise, there's a dev config in the `wikilabels` repo [17:13:44] halfak: where is that file located? I can't seem to find it. [17:14:16] in the config dir [17:14:28] https://github.com/wiki-ai/wikilabels/tree/master/config ? [17:14:31] yup [17:14:41] Oh! I'd forgotten that we named it 00-main.yaml [17:14:49] it used to be called test.yaml or something like that [17:14:50] :) [17:14:59] So, all the yaml files get merged when you start a servere [17:15:09] You'll need to have psql installed locally. [17:15:28] And run `wikilabels load_schema` [17:15:43] halfak: I've gotten the DB ready by copy & pasting the schema and test data files [17:15:48] Nice [17:15:50] that works [17:15:53] halfak: how do I start the dev server with the above changes? [17:16:02] `wikilabels dev_server` [17:16:06] ok let me try [17:16:50] halfak: I need to use the github repo and not the pip package right? [17:17:04] otherwise I'm getting an error related to yaml files [17:18:36] Yeah. It'll look for config in the local dir. [17:18:41] There's also an argument you can give [17:18:47] run `wikilabels dev_server -h` [17:19:18] ok thanks [20:52:17] o/ [21:02:02] o/ [21:12:10] Hey GhassanMas [21:12:42] Deployment complete! New models available on wmflabs [21:12:42] All looks good for moving to prod [21:12:47] I'll work on a patchset [21:17:22] hey halfak [21:25:50] Woo! Our celery workers are down from 1030MB to 885MB [21:25:53] That's awesome! [21:26:33] Let's see. 40 workers * 1030MB = 41200MB [21:26:53] 40 workers * 885MB = 35400MB [21:27:19] 41200 - 35400 = 5800 [21:27:38] 5800 / 885 = ~6.5MB [21:27:51] So we can safely start up 6 more workers! [21:28:23] s/MB// for that last math [22:49:17] 06Revision-Scoring-As-A-Service, 10ORES: ORES vision blog post - https://phabricator.wikimedia.org/T141275#2568963 (10jeffelder) I do like the human angle. I wonder if we want to save that for the detox. You could do the ORES production post now and do the human angle later. ORES production doesn't have to be...