[01:28:18] Here comes the spam! [01:28:24] 06Revision-Scoring-As-A-Service, 10rsaas-editquality: Increment ruwiki editquality models - https://phabricator.wikimedia.org/T144855#2613487 (10Halfak) 05Open>03Resolved [01:28:26] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 15User-Ladsgroup: Improve technical documentation in Extension:ORES in mediawiki.ore - https://phabricator.wikimedia.org/T144676#2613488 (10Halfak) 05Open>03Resolved [01:28:27] >--|o [01:28:28] 06Revision-Scoring-As-A-Service, 10rsaas-editquality: Fix makefile entry for enwiktionary.rev_reverted.20k_2016.tsv - https://phabricator.wikimedia.org/T144605#2613489 (10Halfak) 05Open>03Resolved [01:28:30] 06Revision-Scoring-As-A-Service, 06Research-and-Data, 10Research-management: ORES and Product: resourcing discussion - https://phabricator.wikimedia.org/T144517#2613490 (10Halfak) 05Open>03Resolved [01:28:32] \o/ [01:28:32] 06Revision-Scoring-As-A-Service, 10revscoring: Update yamlconf so that import_path can handle deep attributes - https://phabricator.wikimedia.org/T144430#2613491 (10Halfak) 05Open>03Resolved [01:28:34] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 13Patch-For-Review, 15User-Ladsgroup, 05WMF-deploy-2016-09-06_(1.28.0-wmf.18): Redundant results in ORES review tool - https://phabricator.wikimedia.org/T144233#2613492 (10Halfak) 05Open>03Resolved [01:28:37] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 13Patch-For-Review, 15User-Ladsgroup, 05WMF-deploy-2016-09-06_(1.28.0-wmf.18): Get model version needs to invalidate cache. - https://phabricator.wikimedia.org/T144196#2613493 (10Halfak) 05Open>03Resolved [01:28:40] 06Revision-Scoring-As-A-Service, 10ORES: Set max-age header to 0 seconds for ORES to quiet secondary caches - https://phabricator.wikimedia.org/T144193#2613495 (10Halfak) 05Open>03Resolved [01:28:42] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 13Patch-For-Review, 15User-Ladsgroup: Check model version replaces every time it runs. - https://phabricator.wikimedia.org/T144195#2613494 (10Halfak) 05Open>03Resolved [01:28:44] \o_ [01:28:45] 06Revision-Scoring-As-A-Service, 10rsaas-editquality: Extend user group features - https://phabricator.wikimedia.org/T143909#2613496 (10Halfak) 05Open>03Resolved [01:28:47] 06Revision-Scoring-As-A-Service, 10revscoring: Implement abstraction for Sparse Feature Vectors - https://phabricator.wikimedia.org/T132580#2613497 (10Halfak) 05Open>03Resolved [01:28:50] * halfak swims in the spam stream [01:37:30] OK... So, PCFG description formats are bad [01:37:33] like really bad [01:37:43] Like, did they not think of escaping? [01:38:05] You need to process the entire file before you know what is a terminal and what is a variable [01:42:23] This isn't OK. I'm going to fix this. [04:40:21] 06Revision-Scoring-As-A-Service, 10revscoring, 07Spike: [Spike] Investigate HashingVectorizer - https://phabricator.wikimedia.org/T128087#2613649 (10Sabya) @Halfak: Here are the results from the grid search: ``` Best ROC AUC Score: 0.910445174634 Best Params: {'max_depth': 5, 'n_estimators': 1100, 'learnin... [10:43:23] 06Revision-Scoring-As-A-Service, 10ORES, 13Patch-For-Review, 15User-Ladsgroup: Move from mediawiki/services/ores/deploy to research/ores/deploy or research/ores/deploy-prod - https://phabricator.wikimedia.org/T139008#2614170 (10akosiaris) This unfortunately failed. After moving aside the old repo, puppet c... [13:52:42] o/ [13:54:23] halfak: hey [13:54:32] I'm at "work" [13:54:42] \o/ [13:55:41] but if there's anything I can help [13:57:25] Nothing urgent. When you're done, I was hoping to talk to you about where we're still cloning an old repo. https://phabricator.wikimedia.org/T139008#2614170 [13:58:23] I saw it today and got surprised [13:58:35] Let me ping the proper person [13:59:21] 06Revision-Scoring-As-A-Service, 10ORES, 13Patch-For-Review, 15User-Ladsgroup: Move from mediawiki/services/ores/deploy to research/ores/deploy or research/ores/deploy-prod - https://phabricator.wikimedia.org/T139008#2614561 (10Ladsgroup) @mmodell Do you know why it's empty? [13:59:25] https://www.mediawiki.org/wiki/Topic:Tb34n8tdmpv4vc38 [13:59:33] We need someone to answer this [13:59:44] halfak: thanks for the weekly update [14:02:33] Amir1, will respond. Gonna put together some phab tasks [14:02:42] okay cool [14:22:19] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES: Visually report damaging confidence - https://phabricator.wikimedia.org/T144922#2614647 (10Halfak) @Pginer-WMF is exploring some design concepts for representing confidence in T138935. In this case, he's looking to include a flag widget that would... [14:26:40] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES: Visually report damaging confidence - https://phabricator.wikimedia.org/T144922#2614652 (10Halfak) [14:29:59] 10Revision-Scoring-As-A-Service-Backlog, 06Collaboration-Team-Triage, 10MediaWiki-extensions-ORES, 10Notifications: Notify users when a page on their watchlist has been damaged - https://phabricator.wikimedia.org/T144926#2614690 (10Halfak) [15:05:00] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES: Visually report damaging confidence - https://phabricator.wikimedia.org/T144922#2614593 (10Sadads) Thanks for making this a phabricator item! Looking forward to the update on how this works! [15:56:41] o/ sabya [15:57:58] I was just replying to your recent notes. [16:00:42] 06Revision-Scoring-As-A-Service, 10revscoring, 07Spike: [Spike] Investigate HashingVectorizer - https://phabricator.wikimedia.org/T128087#2615255 (10Halfak) OK. I have a weird proposal that is going to be more work. It looks like we're solidly doing well with `learn_rate=0.01`, but that increasing the esti... [16:22:48] 10Revision-Scoring-As-A-Service-Backlog, 10revscoring: Implement PCFG features - https://phabricator.wikimedia.org/T144636#2615317 (10Halfak) I've been working from https://github.com/halfak/pcfg I've been thinking about file formats. I have *a lot* of concerns. E.g. the files use spaces to delimit so any s... [17:47:20] 10Revision-Scoring-As-A-Service-Backlog, 10revscoring: Implement PCFG features - https://phabricator.wikimedia.org/T144636#2615882 (10Halfak) Some notes from "pintoch" in #wikimedia-research: > if you want pre-trained models for stochastic CFG parsers, you can use (for instance) these: http://nlp.stanford.edu... [17:53:55] 10Revision-Scoring-As-A-Service-Backlog, 10revscoring: Implement PCFG features - https://phabricator.wikimedia.org/T144636#2615904 (10Halfak) I've been thinking about how to represent parse trees. The standard (???) format looks like this: ``` (S (NP (DET Every) (NN cat)) (VP (VT loves) (NP (DET a) (NN dog)))... [17:56:42] 06Revision-Scoring-As-A-Service, 10ORES, 13Patch-For-Review, 15User-Ladsgroup: Move from mediawiki/services/ores/deploy to research/ores/deploy or research/ores/deploy-prod - https://phabricator.wikimedia.org/T139008#2615929 (10akosiaris) >>! In T139008#2614561, @Ladsgroup wrote: > @mmodell Do you know why... [18:01:17] 06Revision-Scoring-As-A-Service, 10ORES, 13Patch-For-Review, 15User-Ladsgroup: Move from mediawiki/services/ores/deploy to research/ores/deploy or research/ores/deploy-prod - https://phabricator.wikimedia.org/T139008#2615964 (10Ladsgroup) The whole work around this in gerrit was to move this repo. So we wi... [20:27:03] 06Revision-Scoring-As-A-Service, 10rsaas-editquality: Extend user group features - https://phabricator.wikimedia.org/T143909#2616626 (10Iniquity) @Halfak sorry for long time answer. I dont know what the problem with editors and autoeditors, but our RecentChanges page has been having only anonimous "red" edits. [20:43:41] (03PS1) 10Ladsgroup: Get results when the score is not stored too [extensions/ORES] - 10https://gerrit.wikimedia.org/r/309142 (https://phabricator.wikimedia.org/T144999) [20:51:05] (03CR) 10Ladsgroup: [C: 032] Get results when the score is not stored too [extensions/ORES] - 10https://gerrit.wikimedia.org/r/309142 (https://phabricator.wikimedia.org/T144999) (owner: 10Ladsgroup) [20:52:11] (03Merged) 10jenkins-bot: Get results when the score is not stored too [extensions/ORES] - 10https://gerrit.wikimedia.org/r/309142 (https://phabricator.wikimedia.org/T144999) (owner: 10Ladsgroup) [21:25:39] Amir1: Repeating Greg's question from Phabricator: that patch probably needs to be deployed ASAP? [21:26:06] RoanKattouw: I haven't seen the Greg's question [21:26:17] It was introduced in wmf.18 [21:26:18] https://phabricator.wikimedia.org/T144999#2616874 [21:26:36] we can backport it in the SWAT window or now [21:26:41] OK yeah so then it needs to be backported to wmf.18 [21:26:41] I can do it [21:26:49] Should probably just put it in the SWAT window [21:26:58] yeah, Okay [21:27:11] I'm waiting for the window right now (I was planning to sleep) [21:27:59] I can supervise it if you like [21:28:30] It would be awesome. [21:28:33] Thanks [21:37:52] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 13Patch-For-Review, 15User-Ladsgroup, 05WMF-deploy-2016-09-13_(1.28.0-wmf.19): User contribs seems to be empty when ores enabled - https://phabricator.wikimedia.org/T144999#2616975 (10Ladsgroup) [21:41:41] (03PS1) 10Catrope: Get results when the score is not stored too [extensions/ORES] (wmf/1.28.0-wmf.18) - 10https://gerrit.wikimedia.org/r/309174 (https://phabricator.wikimedia.org/T144999) [21:42:09] (03PS2) 10Ladsgroup: Get results when the score is not stored too [extensions/ORES] (wmf/1.28.0-wmf.18) - 10https://gerrit.wikimedia.org/r/309174 (https://phabricator.wikimedia.org/T144999) (owner: 10Catrope) [21:42:35] RoanKattouw: oops, we backported at the same time [21:42:38] :D [21:42:59] lol, doesn't matter, yours just overwrites mine because it's the same Change-Id + branch [21:43:18] yeah, sorry about that :D [21:44:10] No worires [21:44:13] I've put it on the deployments page [21:45:42] awesome thanks [22:16:28] 10Revision-Scoring-As-A-Service-Backlog, 10revscoring: Implement PCFG features - https://phabricator.wikimedia.org/T144636#2617241 (10Halfak) I looked into sentence parsers and AFAICT, the state of the art is BAD. I installed the bllippparser, and got some segmentation faults: ``` (3.5) [halfak@graphite: ~/... [22:33:04] 10Revision-Scoring-As-A-Service-Backlog, 10revscoring: Implement PCFG features - https://phabricator.wikimedia.org/T144636#2617470 (10Halfak) Now I'm looking into spaCy and it's not clear that it even can produce trees like what we want. ``` >>> from spacy.en import English >>> from spacy import parts_of_sp... [22:37:12] 10Revision-Scoring-As-A-Service-Backlog, 10revscoring: Implement PCFG features - https://phabricator.wikimedia.org/T144636#2617496 (10Halfak) Here's another way to look at that: * `ROOT` * `nsubj` * `DET` Every * `NOUN` cat * `VERB` loves * `dobj` * `DET` a * `NOUN` dog [22:42:04] 10Revision-Scoring-As-A-Service-Backlog, 10revscoring: Implement PCFG features - https://phabricator.wikimedia.org/T144636#2617508 (10Halfak) [23:07:34] (03CR) 10Catrope: [C: 032] Get results when the score is not stored too [extensions/ORES] (wmf/1.28.0-wmf.18) - 10https://gerrit.wikimedia.org/r/309174 (https://phabricator.wikimedia.org/T144999) (owner: 10Catrope) [23:08:25] (03Merged) 10jenkins-bot: Get results when the score is not stored too [extensions/ORES] (wmf/1.28.0-wmf.18) - 10https://gerrit.wikimedia.org/r/309174 (https://phabricator.wikimedia.org/T144999) (owner: 10Catrope) [23:37:15] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 13Patch-For-Review, 15User-Ladsgroup, 05WMF-deploy-2016-09-13_(1.28.0-wmf.19): User contribs seems to be empty when ores enabled - https://phabricator.wikimedia.org/T144999#2617849 (10Ladsgroup) 05Open>03Resolved