[01:02:17] Hi there. [01:03:33] halfak and YuviPanda ^ [01:18:01] Had some trouble yesterday installing the scikit-neuralnetwork. Didn't realize that everything back to scikit is dependant on scipy and cython. And it took a while to install cython - an overnight - but luckily when I awoke cython was there. [01:19:04] Dunno if it's the web interface, my internet, slow install (over net) or slow servers. Any ideas? [01:19:38] The install is over net I guess [01:25:55] I'd like to try and experiment with some real data - msg me at cobe011(at)gmail.com on how to obtain it. I will come by when I get the chance... I am not used to chatting but I guess this is the fastest way to contact you. [01:29:50] Most close to a live communication is this IRC (to which I am not very familiar as I said) but if this is not the preferred way I would like to at least know which one is :) [01:30:37] Bye for now. [02:12:30] YuviPanda: hey, if you're around. Can we put this GUI in ores.wmflabs.org/ui/ ? [02:13:01] https://github.com/wiki-ai/ORES-GUI [02:40:01] (03CR) 10Ladsgroup: [C: 032] Don't show the r flag when edit is patrolled [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264293 (https://phabricator.wikimedia.org/T121589) (owner: 10Ladsgroup) [02:43:48] (03CR) 10jenkins-bot: [V: 04-1] Don't show the r flag when edit is patrolled [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264293 (https://phabricator.wikimedia.org/T121589) (owner: 10Ladsgroup) [02:48:21] (03PS3) 10Ladsgroup: Don't show the r flag when edit is patrolled [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264293 [02:55:14] (03CR) 10Ladsgroup: [C: 032] Don't show the r flag when edit is patrolled [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264293 (owner: 10Ladsgroup) [02:56:33] (03Merged) 10jenkins-bot: Don't show the r flag when edit is patrolled [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264293 (owner: 10Ladsgroup) [09:46:02] Hello halfak, YuviPanda and others [09:46:25] hey pipivoj [09:46:33] this is a bad time (middle of the night for most of us) [09:46:43] pipivoj: if IRC doesn't have good timing, have you considered posting to the mailing list? [09:47:20] No, but hey you're right. I'll try it there also [09:47:52] :D [09:48:16] https://lists.wikimedia.org/mailman/listinfo/ai [09:56:24] Yeah, I know that it's just that I didn't want that when I post sth for my ignorance to be so much public, transparent.I guess I will have to reconsider the cost/benefit ratio of IRC and mailman :) [09:57:03] pipivoj: no such thing as an ignorant question :D [09:59:16] Yeah maybe, but then there's that phrase: "a stupid question", which I equate with "ignorant question". [10:02:31] I've played with scikit-neuralnetwork for almost the whole night and I guess that I missed making arguments of a fit function to the right form - type. It frustrated me and then I slept for only 4 hours, so it feels like hell now to decide whether to post this to the mailing list [10:07:18] only stupid question is one that isn't asked :D [10:08:12] It sure is buggy, or at least the docs are not in sync with my version - maybe it's the testing branch and not stable - and now I have to think about that too. I'm thinking trying smth else completely, but given the time and energy in deciphering the simple "Parameters: [10:08:12] X : array-like, shape (n_samples, n_features) [10:08:12] Training vectors as real numbers, where n_samples is the number of samples and n_inputs is the number of input features. [10:08:14] " I am really dumbfounded about what to do [10:09:59] And the example given is simple "from sknn.mlp import Classifier, Layer nn = Classifier( [10:09:59] layers=[ [10:09:59] Layer("Maxout", units=100, pieces=2), [10:09:59] Layer("Softmax")], [10:09:59] learning_rate=0.001, [10:09:59] n_iter=25) [10:10:01] nn.fit(X_train, y_train)" [10:10:21] uff, I didn't format it right [10:10:41] so: from sknn.mlp import Classifier, Layer [10:10:45] nn = Classifier( [10:10:58] layers=[ [10:10:59] Layer("Maxout", units=100, pieces=2), [10:10:59] Layer("Softmax")], [10:11:14] learning_rate=0.001, [10:11:15] n_iter=25) [10:11:15] I've no idea about sklearn or machine learning so am unlikely to be able to help :) [10:11:21] nn.fit(X_train, y_train) [10:11:21] posting to the list is definitely going to be better [11:24:56] Hello Amir1 [11:25:17] hey pipivoj [11:25:19] :) [11:26:01] halfak told me that you've been using a neural network on Wikipedia [11:26:35] yeah [11:26:58] I built Kian, it harvests data based on categories and add them to wikidata [11:27:27] the core is entirely self-made for several reasons [11:27:57] Have you used a kit of some kind for help? [11:28:41] I've tried with scikit-neuralnetwork but I'm stuck with arguments of some fit function. Seems I can't get the type and/or size of the matrix right. [11:29:52] I'm not that versed in Python, and it seems there's a tendency to use scikit for classifying and such. [11:30:04] I am willing to learn tho. [11:31:55] Have you built Kian with numpy and similiar, or sth entirely else. [11:32:02] ? [11:33:15] Phew, sorry. Google is faster and better than to bother you. :/ [11:33:16] it's uses only math library of python [16:13:33] (03CR) 10Awight: "Great, thank you for adding this! Agreed that the hook is hard to work with, and the non-enhanced RC view is inconsistent with the new vi" (031 comment) [extensions/ORES] - 10https://gerrit.wikimedia.org/r/263184 (https://phabricator.wikimedia.org/T122535) (owner: 10Ladsgroup) [16:47:30] (03PS1) 10Awight: Explain function [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264594 [16:47:44] (03CR) 10Awight: [C: 032] Explain function [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264594 (owner: 10Awight) [17:03:51] (03Merged) 10jenkins-bot: Explain function [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264594 (owner: 10Awight) [17:18:41] ello [17:45:59] (03PS1) 10Awight: Update authors += ladsgroup [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264607 [17:46:02] (03PS1) 10Awight: [WIP] Filter out non-damaging changes in Special:Contributions [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264608 (https://phabricator.wikimedia.org/T122537) [17:46:44] (03CR) 10Awight: "@ladsgroup: Please confirm this is how you'd like the attribution to read. Thank you!" [extensions/ORES] - 10https://gerrit.wikimedia.org/r/264607 (owner: 10Awight) [17:49:28] halfak o/ [18:09:44] o/ [19:00:21] halfak check pm if you have the time [20:10:23] halfak o/ [20:14:57] hey :)