[06:41:41] 10Machine-Learning-Team, 10Add-Link, 10Growth-Team, 10User-notice: Deploy "add a link" to 8th round of wikis - https://phabricator.wikimedia.org/T308133 (10kevinbazira) [06:46:38] 10Machine-Learning-Team, 10Add-Link, 10Growth-Team, 10User-notice: Deploy "add a link" to 8th round of wikis - https://phabricator.wikimedia.org/T308133 (10kevinbazira) @kostajh, we published datasets for all 20/22 models that passed the evaluation in this round. [09:22:40] 10Machine-Learning-Team, 10Add-Link, 10Growth-Team, 10User-notice: Deploy "add a link" to 8th round of wikis - https://phabricator.wikimedia.org/T308133 (10kostajh) Thanks @kevinbazira! [09:23:03] 10Machine-Learning-Team, 10Add-Link, 10Growth-Team, 10User-notice: Deploy "add a link" to 8th round of wikis - https://phabricator.wikimedia.org/T308133 (10kostajh) [09:42:15] Morning! [10:19:52] Hi Tobias! [10:20:51] So. Many. Mails. [11:31:05] * klausman lunch! [12:31:10] aiko: o/ do you have handy the revision id's you tried and got the errors? I have tried several rev_id(s) so far with the deployed model for arwiki and dewiki and everything seems fine [12:32:13] btw I feel we should invest some time to have some automated testing as it feels like a burden to test :D [12:33:02] not saying something new, what we were discussing last time (have a procedure to get several rev ids for test) is a good first step I think [12:54:47] also, with which models did you test? [12:58:12] aaa, I tested with the wrong image. lemme try again and I'll let you know . sorries [13:15:43] 10Machine-Learning-Team, 10Patch-For-Review: Upgrade python from 3.7 to 3.9 in docker images - https://phabricator.wikimedia.org/T325657 (10isarantopoulos) a:03isarantopoulos [13:16:04] * isaranto afk lunch [13:17:37] I moved the above task to the in progress column since some work has been done/is ongoing on the subject and we can decide how and when we want to proceed with this upgrade in our meeting(s) - as it is a big one and perhaps should be broken down to more tasks [14:05:04] isaranto: o/ yeah definitely we should start thinking about automated testing! [14:33:40] I tested enwiki-goodfaith, arwiki-damaging, jawiki-damaging, wikidatawiki-damaging, dewiki-damaging, frwiki-damaging, and eswiki-goodfaith [14:33:49] four of them (ar, de, fr, es) failed to load the model [14:41:22] ok, I got the same, I am working to see if it can be fixed [15:07:28] Goooood morning all [15:08:57] klausman Tomorrow let’s talk about turning off the AWS staging instance. It is making us go over budget apparently [15:09:12] Sure! [15:09:39] I've been quick-reading the Slack-side discussion. I think using smaller instances may be possible, but it will need some experimenting. [16:23:56] aiko: the issue is with a upgrade done in pyenchant package. I'm trying to find a workaround for that. Can I find the code that trains the deployed model(s)? [16:24:17] logging off folks, cu tomorrow [16:51:28] 10Machine-Learning-Team, 10ORES, 10SRE: [Epic] Deploy ORES in kubernetes cluster - https://phabricator.wikimedia.org/T182331 (10akosiaris) [16:51:55] 10Machine-Learning-Team, 10ORES, 10SRE: Stress test ORES on kubernetes (above 4.5k scores/second) - https://phabricator.wikimedia.org/T214054 (10akosiaris) 05Stalled→03Declined ORES isn't going to ever be on Kubernetes, closing this as Declined. [18:33:16] 10Machine-Learning-Team, 10Data-Engineering-Planning, 10Research: Proposal: deprecate the mediawiki.revision-score stream in favour of more streams like mediawiki-revision-score- - https://phabricator.wikimedia.org/T317768 (10diego) Hi all, we have a use case here T326179. These models are already hos... [19:54:53] (03CR) 10AikoChou: [C: 03+1] "LGTM! I like your refactoring on preprocessing and predict function :)" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/871185 (https://phabricator.wikimedia.org/T323624) (owner: 10Ilias Sarantopoulos)