[06:38:18] FYI, I'll switch one of the etcd nodes for the ml cluster away from direct disk storage to DRBD for a bit, so latency alerts may kick in for a bit [06:38:59] the Ganeti server where it's running on is going to be replaced and this is needed for the migration to a new host [07:06:01] Ack! [07:06:08] Morning folks o/ [07:37:20] ml-etcd1002 is back on "plain" disk storage [08:35:27] morning! [08:52:21] hi Aiko! [10:21:47] (03PS1) 10Kevin Bazira: langid: normalize text input [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) [10:24:51] (03CR) 10CI reject: [V:04-1] langid: normalize text input [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) (owner: 10Kevin Bazira) [10:25:17] 10Lift-Wing, 06Machine-Learning-Team, 13Patch-For-Review: [langid] fasttext only processes one line at a time - https://phabricator.wikimedia.org/T377751#10253451 (10kevinbazira) Text normalization has been added to the langid model-server and it fixed this issue as shown below: `lang=bash, name=Before Norma... [10:28:24] (03PS2) 10Kevin Bazira: langid: normalize text input [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) [10:33:11] (03CR) 10Kevin Bazira: "This patch has been tested on ml-testing, and the results can be seen here: https://phabricator.wikimedia.org/T377751#10253451" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) (owner: 10Kevin Bazira) [10:37:14] isaranto: aiko: o/ here is the patch to fix the fasttext issue in langid: https://gerrit.wikimedia.org/r/1082438 [10:37:15] please review whenever you get a minute. thanks! [10:40:19] Thanks kevin! I'm going for an interview and will review later [12:11:10] (03CR) 10Ilias Sarantopoulos: langid: normalize text input (032 comments) [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) (owner: 10Kevin Bazira) [12:29:39] * isaranto lunch [13:37:48] hey folks, not sure if you saw the last updates, but after the I/F team hackathon this is possible: [13:37:51] elukey@deploy2002:~$ dig -x 10.67.16.186 +short [13:37:54] 10-67-16-186.revertrisk-language-agnostic-predictor-default-00025.revertrisk.svc.cluster.local. [13:37:57] 10-67-16-186.revertrisk-language-agnostic-predictor-default-00025-private.revertrisk.svc.cluster.local. [13:38:06] oh, neat! [13:38:13] so if you have an IP address of a pod logged somewhere, you can dns reverse it [13:38:22] I saw the initial proposal for RRs, I am very happy to see it implemented [13:38:32] still not fully automated, but it works [14:00:39] (03PS3) 10Kevin Bazira: langid: normalize text input [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) [14:02:25] (03CR) 10Kevin Bazira: langid: normalize text input (032 comments) [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) (owner: 10Kevin Bazira) [15:02:20] (03CR) 10Klausman: langid: normalize text input (031 comment) [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/1082438 (https://phabricator.wikimedia.org/T377751) (owner: 10Kevin Bazira) [16:08:39] going afk folks o/ [16:37:33] 06Machine-Learning-Team, 10Section-Level-Image-Suggestions, 10Section-Topics, 06Structured-Data-Backlog: [XL] Let the model that learns section alignments consume section topics output - https://phabricator.wikimedia.org/T331968#10255192 (10AUgolnikova-WMF) [20:37:53] 06Machine-Learning-Team, 06Data-Platform-SRE: Investigate Label functionality of AMD GPU device plugin on k8s - https://phabricator.wikimedia.org/T373806#10256280 (10Ottomata)