[06:35:40] (03PS1) 10Kevin Bazira: locust: rename data files for reusability [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) [07:09:43] (03PS1) 10Kevin Bazira: locust: rename data files for reusability [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) [07:18:59] (03CR) 10Kevin Bazira: "This is the follow-up patch to this comment:" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) (owner: 10Kevin Bazira) [07:20:40] (03CR) 10Ilias Sarantopoulos: "Is there a reason to not use # of beams anymore? I think that it would be nice to be able to check the difference in performance" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) (owner: 10Kevin Bazira) [07:28:16] (03CR) 10Kevin Bazira: "We are still using # of beams in the json payload in `article_descriptions.py`. The data file had 3 beams that the mobile team uses:" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) (owner: 10Kevin Bazira) [08:58:05] (03CR) 10AikoChou: [C:03+1] "LGTM!" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) (owner: 10Kevin Bazira) [09:10:13] (03CR) 10Ilias Sarantopoulos: [C:03+1] "I meant checking different numbers for the number of beams, but it makes sense for now to check just the ones we need." [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) (owner: 10Kevin Bazira) [09:25:47] (03CR) 10Kevin Bazira: [V:03+2 C:03+2] "Thanks for the reviews :)" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1081309 (https://phabricator.wikimedia.org/T377418) (owner: 10Kevin Bazira) [10:24:37] 06Machine-Learning-Team, 10ORES, 05FY24-25 WE4.2, 10Moderator-Tools-Team (Kanban), 07Spike: [SPIKE] Investigate how to install ORES in idwiki [8HRS] - https://phabricator.wikimedia.org/T374077#10241321 (10kostajh) Tagging with #fy24-25_we4.2 as it's a hypothesis for Q2 work. [10:27:39] 06Machine-Learning-Team, 10ORES: Rename ORES extension - https://phabricator.wikimedia.org/T377563 (10kostajh) 03NEW [11:33:23] 06Machine-Learning-Team, 10ORES: Rename ORES extension - https://phabricator.wikimedia.org/T377563#10241568 (10isarantopoulos) Thanks for opening this task Kosta! Given that the extension provides additional filters for the recent changes page by scoring the revisions I'm thinking something like RevisionScor... [11:59:00] 06Machine-Learning-Team: [ml-lab] Use a (jupyter) notebook and load a LLM from huggingface - https://phabricator.wikimedia.org/T377574 (10isarantopoulos) 03NEW [13:58:59] https://github.com/kserve/kserve/releases/tag/v0.14.0 :D [14:00:32] kevinbazira: `Fix dead links on PyPI by @kevinbazira in #3754` nice to see your name/PR in the release notes! [14:00:56] thanks for taking care of that! [14:02:13] nice work, Kevin! :) [14:10:22] o/ aiko! [14:11:50] 06Machine-Learning-Team: Investigate kserve 0.13.0 upgrade - https://phabricator.wikimedia.org/T367048#10242086 (10isarantopoulos) I just upgraded artilce descriptions model. This is the current state of the services | **service** | **kserve version** | | reference quality| 0.13.1| |revscoring | 0.11.2 | | art... [14:12:52] I played a bit with ml-lab , launching notebooks etc. It seems fine for the initial experimentation. I'll add the info in the task [14:14:36] that's cool!! yayy [14:46:45] isaranto: aiko: super cool! :) [17:11:54] 06Machine-Learning-Team, 10ORES, 10CheckUser, 06Moderator-Tools-Team, and 2 others: Failure in PageTriage extension on CheckUser test GlobalBlockingHandlerWithDatabaseRowsTest::testRetroactiveAutoblockWhenLocalUserNotAttached - https://phabricator.wikimedia.org/T377609 (10Umherirrender) 03NEW [17:23:55] 06Machine-Learning-Team: [ml-lab] Use a (jupyter) notebook and load a LLM from huggingface - https://phabricator.wikimedia.org/T377574#10242945 (10isarantopoulos) Leaving some raw instructions how to do this and will probably use a script to automate this for now: On ml-lab # ssh to ml-lab # navigate to path t... [17:24:09] going afk folks , have an awesome weekend o/