--- Log opened Wed May 30 00:00:28 2018 | ||
-!- durovo [~durovo@31.8f.559e.ip4.static.sl-reverse.com] has joined #shogun | 02:02 | |
-!- durovo2 [~durovo@31.8f.559e.ip4.static.sl-reverse.com] has quit [Read error: Connection reset by peer] | 02:03 | |
-!- Netsplit *.net <-> *.split quits: @ChanServ | 02:13 | |
-!- Netsplit over, joins: @ChanServ | 02:34 | |
-!- wuwei1 [wuweilinma@gateway/shell/matrix.org/x-lwvcxvbgbdmrbwiy] has quit [Ping timeout: 255 seconds] | 02:38 | |
-!- wuwei [wuweilinma@gateway/shell/matrix.org/x-yjobwnmvmkogacxz] has joined #shogun | 03:00 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Ping timeout: 248 seconds] | 05:56 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 05:59 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4314 opened by abhishek-1991 | 06:44 |
---|---|---|
wuwei | wiking; very weird that renaming make travis dying and i can reproduce that error locally | 07:50 |
@wiking | wuwei, if you want i can restart that job | 07:59 |
@wiking | see if it's just a glitch | 07:59 |
wuwei | well.. i think that's actually a bug | 08:01 |
wuwei | it's string preprocessor | 08:01 |
wuwei | in python: feats = transformer.transform(feats) | 08:02 |
wuwei | in stringpreproc::transform, in this case, the refcount is one when entering and exiting | 08:02 |
wuwei | and then it's unrefed in python | 08:02 |
wuwei | because of '=', the original feats is unrefed | 08:02 |
wuwei | however in this case they are the same object | 08:03 |
wuwei | meaning that, changing the line to: transformer.transform(feats) | 08:03 |
wuwei | it works | 08:03 |
wuwei | but i don't see any points that this should be happening | 08:03 |
wuwei | ah.. maybe that's related to %newobject in swig, let me check | 08:08 |
-!- sukey1 [~nodebot@ks312251.kimsufi.com] has joined #shogun | 08:27 | |
-!- sukey [~nodebot@ks312251.kimsufi.com] has quit [Read error: Connection reset by peer] | 08:28 | |
wuwei | does this line apply to things outside CMachine? | 09:11 |
wuwei | https://github.com/shogun-toolbox/shogun/blob/7e9699e5a1c7df75271e8dd26fc71e615e8138b4/src/interfaces/swig/Machine.i#L11 | 09:11 |
sukey1 | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4285 synchronized by vinx13 | 09:40 |
@wiking | wuwei, imo yes | 10:02 |
wuwei | yeah | 10:03 |
wuwei | travis is green now :) | 10:03 |
sukey1 | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4285 merged by vigsterkr | 10:08 |
sukey1 | [https://github.com/shogun-toolbox/shogun] vigsterkr pushed 61 commits: | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/6029e94a3cdad6ce20c9a420a11a273bda3d22c2 | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/8a7dcabb7c6b40abcc69b3978e40e2b2378221d9 | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/0d43482a9ee7542851e01baa6d936212cd0b48a1 | 10:08 |
@wiking | boom | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/9a0dcc067336e8d15cd0ebc8df1d650b0d76cff0 | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/b9c7067241e1bb432a513f3393a71c14f187c180 | 10:08 |
@wiking | :D | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/7acd42e985bd85eb686fd204da7a2ae76e8c5c95 | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/c26877aac1abb4735ec0f30c21d07397bdf91583 | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/5c87fdb38feee82f1258649deea8576e88a6890e | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/4840a8caa8300687b06a85ea14e76e3f8b22ae62 | 10:08 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/f93b8525cd58542c69b815a8a9bc562940299f14 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/0155355f09e5395f2a638e2b132d85d4056b6b81 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/f2d9b85d2870da7f61129fb3e64d9d8443d736aa | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/556cb6ec51cb71b89260f7706fd9c8dc532812d1 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/c8fd10c75f42477502c7e7c47ffa814a71f6ff1f | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/363d54aae91e810e75e1f9c8814057805739767b | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/d2df975c05e94fe0d35e8da3ebd8f04957ba5835 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/582721c46c6ac141ea37064b99d1e3357e942599 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/f32f61621299c86fd7cef627416abf176dd78339 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/5f2ac96fe87275d5c26d941b75a6a7161839d7f5 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/6b098585028a139bb128e5877a505b3c052a397d | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/67f02c4f5208d8e8308e8ecaadbeeae59e52856a | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/66015f2fff7f4e707456f1dc4a68ae0dd9cd66d4 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/b9a95dd8212650b4ecd6d253d314e59d6aef711f | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/cd798503ae19141bcccf8f69a9c19f998f53c23d | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/ec66d1d6f3c12ef170097f67c7693f250be6132a | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/f765b525534067a9db7baf826222c0e6dc46dc4d | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/801633caac31ac898e32161fa5fda7409bc83262 | 10:09 |
sukey1 | https://github.com/shogun-toolbox/shogun/commit/b383892c73fab74b3d3458ad3d3b9a923f613c28 | 10:09 |
-!- sukey1 [~nodebot@ks312251.kimsufi.com] has quit [Remote host closed the connection] | 10:09 | |
@wiking | wuwei, great stuff!!! | 10:16 |
wuwei | :D | 10:16 |
@wiking | wuwei, ok so next one? | 10:17 |
@wiking | pipeline? | 10:17 |
wuwei | yeah | 10:17 |
@wiking | ccool | 10:17 |
@wiking | just asec | 10:17 |
@wiking | in a meeting | 10:17 |
wuwei | should i refactor out those old preproc api from features | 10:19 |
-!- HeikoS [~heiko@host81-153-166-11.range81-153.btcentralplus.com] has joined #shogun | 10:56 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 10:56 | |
-!- travis-ci [~travis-ci@ec2-184-73-117-198.compute-1.amazonaws.com] has joined #shogun | 10:59 | |
travis-ci | it's Wuwei Lin's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385569885 | 10:59 |
-!- travis-ci [~travis-ci@ec2-184-73-117-198.compute-1.amazonaws.com] has left #shogun [] | 10:59 | |
wuwei | wiking: ready for today's meeting? :) | 11:06 |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-smarjtpaswqoslpr] has joined #shogun | 11:07 | |
-!- HeikoS [~heiko@host81-153-166-11.range81-153.btcentralplus.com] has quit [Ping timeout: 240 seconds] | 11:37 | |
@wiking | wuwei, yes | 11:49 |
@wiking | i am | 11:49 |
@wiking | i was in a meeting till now | 11:49 |
@wiking | :DDD | 11:49 |
@wiking | wuwei, i'm soooo bad :( | 11:49 |
@wiking | sorry aboutit | 11:49 |
@wiking | still here? | 11:49 |
wuwei | sure no problem | 11:50 |
@wiking | ok so yeah | 11:50 |
@wiking | actually maybe that would be better now | 11:50 |
@wiking | before jumping into pipelines | 11:50 |
@wiking | to clear the air about preprocessors in features | 11:50 |
wuwei | apply preproc , I think this can be removed | 11:51 |
wuwei | But there are on the fly things that should be kept | 11:51 |
wuwei | like apply to vector | 11:52 |
@wiking | ye | 11:52 |
-!- travis-ci [~travis-ci@ec2-54-224-19-64.compute-1.amazonaws.com] has joined #shogun | 11:55 | |
travis-ci | it's Wuwei Lin's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385570465 | 11:55 |
-!- travis-ci [~travis-ci@ec2-54-224-19-64.compute-1.amazonaws.com] has left #shogun [] | 11:55 | |
wuwei | btw about pipeline, if in linear case it should be zero or more transformers, +1 machine as the final stage right ? | 11:56 |
@wiking | yeah that's the idea | 11:57 |
wuwei | then actually we don't need variant :) | 11:58 |
@wiking | well | 11:58 |
@wiking | coz you wanna store them separately? :) | 11:58 |
-!- sukey [~nodebot@ks312251.kimsufi.com] has joined #shogun | 11:59 | |
-!- mode/#shogun [+o sukey] by ChanServ | 12:00 | |
@wiking | sukey, flip | 12:00 |
@sukey | (????)??????/ | 12:00 |
wuwei | yeah | 12:00 |
wuwei | Otherwise we have a vector of variant, and only the last element is machine | 12:01 |
@wiking | yeah | 12:02 |
@wiking | now of course | 12:02 |
@wiking | we should keep in mind | 12:02 |
@wiking | the dag case right? :) | 12:02 |
@wiking | when you can have 2 machines :P | 12:02 |
@wiking | or more | 12:02 |
wuwei | oh yeah | 12:03 |
@wiking | lisitsyn, you're back? :) | 12:03 |
@wiking | wuwei, i mean for the time being we'll just work on linear stuff | 12:03 |
@wiking | but let's do things more flexible way in the bakcend | 12:04 |
@wiking | so we dont need to completely change things :) | 12:04 |
wuwei | sure I see | 12:04 |
-!- travis-ci [~travis-ci@ec2-54-224-19-64.compute-1.amazonaws.com] has joined #shogun | 12:17 | |
travis-ci | it's Wuwei Lin's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385570465 | 12:17 |
-!- travis-ci [~travis-ci@ec2-54-224-19-64.compute-1.amazonaws.com] has left #shogun [] | 12:17 | |
-!- HeikoS [~heiko@eduroam-int-pat-8-0.ucl.ac.uk] has joined #shogun | 12:23 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:23 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4313 synchronized by karlnapf | 12:27 |
-!- HeikoS [~heiko@eduroam-int-pat-8-0.ucl.ac.uk] has quit [Ping timeout: 248 seconds] | 12:27 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4313 synchronized by karlnapf | 12:32 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4315 opened by karlnapf | 12:36 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4315 karlnapf added label: "good first issue" | 12:36 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4315 karlnapf added label: "Cleanups" | 12:36 |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has joined #shogun | 12:42 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:42 | |
@HeikoS | lisitsyn: you around? | 13:22 |
@HeikoS | wiking: can you send me the reimbursement infos for the hackathon in belgrade? | 13:23 |
@HeikoS | I already put in the airbnb, but anything else pls send | 13:24 |
@wiking | sure i'll try to send out asap | 13:24 |
@HeikoS | just need numbers | 13:24 |
@HeikoS | have everything else | 13:24 |
@HeikoS | from previous | 13:24 |
@wiking | sure | 13:24 |
@wiking | meeting | 13:24 |
@wiking | sorry | 13:24 |
@HeikoS | kk | 13:24 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4313 merged by karlnapf | 13:27 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/bf2f54543425a62f36d02061da7d8b8f6005847c by karlnapf | 13:27 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4310 merged by karlnapf | 13:55 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/35449565ac35d72f91a79d60606e7cc43966af08 by karlnapf | 13:55 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4316 opened by vinx13 | 14:52 |
-!- travis-ci [~travis-ci@ec2-54-161-181-102.compute-1.amazonaws.com] has joined #shogun | 14:56 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385643348 | 14:56 |
-!- travis-ci [~travis-ci@ec2-54-161-181-102.compute-1.amazonaws.com] has left #shogun [] | 14:56 | |
@wiking | wuwei, i'm wondering what happens if you change this to CFeatures* https://github.com/vinx13/shogun/blob/d7d61d67aa21c1c12be3d101b19054515550a188/tests/unit/classifier/GaussianProcessClassification_unittest.cc#L152 | 14:57 |
@wiking | same for features_test | 14:57 |
@wiking | is there anything in that unit test that it actually needs that feature type? :) | 14:58 |
@wiking | specifically | 14:58 |
@wiking | i would hope not | 14:58 |
@wiking | this would allow us to drop the bit ugly ->as<CDense...>() part :P | 15:01 |
wuwei | in this case we can actually use CFeatures*, imo | 15:04 |
@wiking | yeah that would be better | 15:04 |
@wiking | could you add those changes into that pr plz | 15:04 |
@wiking | thnx heaps! | 15:04 |
@wiking | HeikoS, here? | 15:05 |
wuwei | as for others, we should either cast to dense after transforming or cast to dense when get_feature_matrix cuz we need to compare the matrix in unit test | 15:06 |
@wiking | where. ?:) | 15:06 |
@wiking | plz copy me a link | 15:06 |
@wiking | sorry my head is all over atm | 15:06 |
wuwei | for example you can check tests for dense preproc, like log plus one | 15:07 |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has joined #shogun | 15:07 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385643348 | 15:07 |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has left #shogun [] | 15:07 | |
@wiking | just a sec | 15:08 |
@wiking | hehe | 15:09 |
@wiking | DotIterator! | 15:09 |
@wiking | :) | 15:09 |
wuwei | yeah :) | 15:09 |
wuwei | cool! | 15:09 |
@wiking | you can either test with dot product :) | 15:10 |
@wiking | or add | 15:10 |
@wiking | :> | 15:10 |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has joined #shogun | 15:16 | |
travis-ci | it's Shubham Shukla's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385653046 | 15:16 |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has left #shogun [] | 15:16 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4316 synchronized by vinx13 | 15:16 |
@wiking | cool | 15:16 |
@wiking | lets see CI but when it's ready i'll merge | 15:17 |
@wiking | ok? | 15:17 |
@wiking | and then lets try to use dotiterator :) | 15:17 |
wuwei | seems i broke integration test | 15:18 |
@wiking | :D | 15:18 |
wuwei | we need update integration data | 15:18 |
wuwei | cuz 'processed' in features is dropped | 15:18 |
@wiking | ah ok | 15:20 |
@wiking | sure thng | 15:20 |
@wiking | just prepare a patch for shogun-data | 15:20 |
wuwei | should i send a pr to a new branch in shogun-data, otherwise that will break develop | 15:23 |
@wiking | yeye | 15:26 |
@wiking | new pr to shogun-data | 15:26 |
wuwei | i'm wondering why there's only one integration test broken locally | 15:27 |
@wiking | : ?\_(?)_/? | 15:28 |
@wiking | sukey, flip | 15:28 |
@sukey | ?(????)? ??? | 15:28 |
-!- Farouk [9cc2876f@gateway/web/freenode/ip.156.194.135.111] has joined #shogun | 15:35 | |
@wiking | Farouk, hellooooo | 15:35 |
@wiking | whatsup | 15:35 |
Farouk | wiking hey | 15:35 |
@wiking | Farouk, can you do a rebase over this https://github.com/shogun-toolbox/shogun/pull/4309 | 15:36 |
@wiking | using latest develop | 15:37 |
@wiking | and then you should be able to use the new api everywhere | 15:37 |
@wiking | https://git-scm.com/book/en/v2/Git-Branching-Rebasing | 15:37 |
@wiking | fyi | 15:37 |
@wiking | thnx | 15:37 |
@wiking | lemme know if you are having troulbes | 15:37 |
@wiking | *troubles | 15:37 |
Farouk | Yea, I saw Heiko's PR yesterday. was waiting for it to be merged. Just woke up and found it merged so will get to it now | 15:37 |
@wiking | kk | 15:38 |
@wiking | cool | 15:38 |
@wiking | just push when ready | 15:38 |
Farouk | sure (y) | 15:38 |
@wiking | Farouk, what's with the other one | 15:38 |
@wiking | ? | 15:38 |
@wiking | the cost function | 15:38 |
@wiking | what's the blocker there atm? | 15:38 |
@wiking | or is there any blocker? | 15:38 |
Farouk | The cost function me and Fernando are meeting tomorrow at 10pm to discuss some design things | 15:38 |
Farouk | CEST time | 15:38 |
Farouk | so he told me to pause on it for now. But no blockers | 15:39 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4314 merged by karlnapf | 15:45 |
@sukey | [https://github.com/shogun-toolbox/shogun] karlnapf pushed 2 commits: | 15:45 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/dac796c2f47ac2321483e488df096afd2067ad93 | 15:45 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/363c0dc853108c25c02d3743536f690bf27d3444 | 15:45 |
wuwei | wiking: actually no problem with integration test :) i checked out the latest shogun-data and tests passed | 15:45 |
wuwei | let's see how ci goes | 15:45 |
@wiking | Farouk, if you want we can talk now | 15:46 |
@wiking | about that | 15:46 |
@wiking | wuwei, ah cool | 15:46 |
Farouk | wiking, yea sure we can talk now. | 15:47 |
@wiking | shoot | 15:47 |
@wiking | what's the design issue there atm | 15:48 |
@wiking | ? | 15:48 |
@wiking | i mean fist of all | 15:48 |
@wiking | what i wanna ask is | 15:48 |
@wiking | why that may pointers? | 15:48 |
@wiking | i.e. why everything is a pointer :) | 15:48 |
Farouk | so Fernando said he has a couple of comments on the patterns I used (that some of them is anti-shogun) and also some comments on pointers :D | 15:48 |
Farouk | hahaha thats one of the things Fernando commented on too :-D | 15:48 |
Farouk | I am not sure, I guess I've mostly been using C-like syntax when it comes to these things | 15:49 |
Farouk | and C has lots of pointers | 15:49 |
Farouk | also he wanted to talk about whether the cost function should hold the data (i.e. X, y) | 15:50 |
Farouk | or we should supplement it | 15:50 |
@wiking | first answer would be that the cost function shouldnt hold it but then again | 15:50 |
@wiking | i have to see the context | 15:50 |
@wiking | but first about the pointers | 15:50 |
@wiking | lemme open the pr | 15:50 |
@wiking | and paste you here liens | 15:50 |
Farouk | sure | 15:50 |
@wiking | https://github.com/shogun-toolbox/shogun/pull/4294/files#diff-09a10f92947e82f417b2b175687c3e6eR19 | 15:52 |
@wiking | does this link bring u to a line? | 15:52 |
Farouk | yes | 15:53 |
@wiking | so the point being is that StanVector* m_trainable_parameters | 15:53 |
@wiking | why? | 15:53 |
@wiking | does it have to be a pointer? | 15:53 |
@wiking | why couldn't you just pass a ref ? | 15:53 |
@wiking | Matrix<function<var(int32_t)>, Dynamic, 1>& cost_for_ith_point | 15:53 |
@wiking | same goes for the cost fucntion | 15:54 |
@wiking | total_cost | 15:54 |
Farouk | I agree with you. I think all of them should be by reference to be honest. I didn't realise that except when writing the unittest | 15:54 |
Farouk | I wasn't creating any objects on the heap, so a reference would've been fine | 15:55 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/4a1299e4b4ed6e05fe467ce063ebebe119e12681 by vigsterkr | 15:56 |
@wiking | k | 15:57 |
@wiking | so go with that change right away | 15:57 |
@wiking | i'm just scrolling around again | 15:57 |
Farouk | alright, and about the data? Should the data be provided when I want to calculate the cost/gradient, or should I keep a reference of the data? | 15:57 |
@wiking | just a sec | 16:00 |
@HeikoS | Farouk: I commented on a few things in your PRs | 16:10 |
Farouk | Hey Heiko, you mean just now? | 16:10 |
Farouk | Ohh year just saw them | 16:11 |
Farouk | will finish cookbook and go there | 16:11 |
@wiking | Farouk, ok so | 16:12 |
@wiking | i would really not like to store data in such a method | 16:13 |
@wiking | but then again | 16:13 |
Farouk | But im not sure how I would keep track of the current sample that I am dealing with | 16:21 |
Farouk | it would have to fall on the user to keep track of that | 16:21 |
wuwei | wiking: I've made some initial implementation of pipeline, should I make the pr to a new branch, or the one for transformer | 16:27 |
@wiking | welll | 16:27 |
@wiking | because we wanna use the transformer api | 16:27 |
@wiking | it would make sense to use the transformer branch | 16:27 |
@wiking | i know that the name of the branch is a bit mixing | 16:27 |
@wiking | but whatever | 16:27 |
@wiking | just go with that branch plz :) | 16:27 |
@wiking | thnx | 16:27 |
@wiking | wuwei, and great! | 16:28 |
wuwei | sure | 16:28 |
@wiking | Farouk, REQUIRE(!X_new.equals(SGMatrix<float64_t>()), "Empty X provided"); :) | 16:32 |
@wiking | so i know this is not the main issue | 16:32 |
@wiking | but if you wanna check whether an SGMatrix is emptuy | 16:32 |
@wiking | just call | 16:32 |
@wiking | sgmatrix.size() | 16:32 |
@wiking | this one actually constructs an matrix | 16:32 |
@wiking | object etc etc | 16:32 |
@wiking | Farouk, in general as fernando mentioned | 16:33 |
@wiking | if you use the word Interface | 16:33 |
@wiking | plz just really create an interface | 16:33 |
@wiking | meaning barely any implementation | 16:33 |
@wiking | or just drop the Interface naming | 16:34 |
@wiking | Farouk, btw the usecase of FirstOrderSAGCostFunctionInterface | 16:34 |
@wiking | do we really need the support | 16:35 |
@wiking | for chaning a cost function's member variables | 16:35 |
@wiking | after the object being created? | 16:35 |
@wiking | so for example if you have a new train data | 16:35 |
@wiking | wouldn't you actually create a new cost function | 16:35 |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has quit [Ping timeout: 260 seconds] | 16:35 | |
@wiking | ? | 16:35 |
@wiking | same goes for set_trainable_parameters | 16:36 |
@wiking | etc? | 16:36 |
@wiking | Farouk, see what i mean? | 16:40 |
@wiking | but again the best would be that actual;lyu | 16:41 |
@wiking | you draft me a code that uses this cost function | 16:41 |
@wiking | in a 'real-life' setup | 16:41 |
@wiking | then we could a littlebit discuss this | 16:41 |
@wiking | pseudo code is fine | 16:41 |
@wiking | currently i'm just checking it from the unit test perspective | 16:41 |
Farouk | wiking, sorry my irc was buggy, I didn't get a notification alert | 16:44 |
@wiking | sure | 16:45 |
@wiking | Farouk, you can use gitter.im | 16:45 |
@wiking | the messages are being relayed by shog | 16:45 |
@wiking | shogitter, | 16:45 |
@wiking | anyhow | 16:45 |
@wiking | do you see my comments/messages? | 16:45 |
Farouk | yea reading them now | 16:45 |
@wiking | kk | 16:45 |
Farouk | Hmm I agree, if a user wants to change the cost function, just create a new cost function, so maybe the setter of cost functions isn't really needed. | 16:46 |
@wiking | yep | 16:46 |
@wiking | but yeah | 16:47 |
@wiking | would be great if you could give me a pseudo code | 16:47 |
@wiking | how you would imagine this now with the current api | 16:47 |
Farouk | so the way I envisioned it to be used is similar to how its used in the unit test | 16:47 |
Farouk | basically you define X, y | 16:47 |
@wiking | so say we have now a NN | 16:47 |
@wiking | i guess this is your usecase right? | 16:47 |
Farouk | then write a function that defines your cost regardless | 16:47 |
Farouk | yes | 16:47 |
@wiking | ok so lets have the big picutre now | 16:47 |
@wiking | :) | 16:47 |
@wiking | let's have a pseudo code for nn | 16:48 |
@wiking | could you write this in a paastebin | 16:48 |
@wiking | or whatever sharing tool | 16:48 |
@wiking | so we can discuss it | 16:48 |
@wiking | and maybe we refactor a bit the whole cost function story? | 16:48 |
Farouk | yea sure, so you mean how the NN will use the cost function? | 16:48 |
Farouk | in a big picture | 16:48 |
Farouk | ? | 16:48 |
@wiking | yeah | 16:48 |
Farouk | sure | 16:49 |
@wiking | i mean we have our current design of NN | 16:49 |
@wiking | best would be to use that in yoru examplke | 16:49 |
@wiking | but it's ok if it's all imaginary | 16:49 |
@wiking | jsut wanna see how this would exactly play out | 16:49 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4317 opened by vinx13 | 16:49 |
@wiking | wuwei, is on fire! | 16:50 |
@wiking | \o/ | 16:50 |
@wiking | wuwei, ehhehe | 16:51 |
@wiking | train_require_labels | 16:51 |
@wiking | :> | 16:51 |
@wiking | yeah the shitty design of train(data) | 16:51 |
@wiking | but not actually having train(data, label) | 16:51 |
@wiking | :( | 16:51 |
wuwei | well | 16:51 |
-!- HeikoS [~heiko@eduroam-int-pat-8-0.ucl.ac.uk] has joined #shogun | 16:52 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:52 | |
@wiking | wuwei, maybe tiiiiime for refactor? :D | 16:52 |
@wiking | lol | 16:52 |
wuwei | u mean fit + predict? :) | 16:53 |
@wiking | yes :DDD | 16:53 |
@wiking | but maybe that's a bit heavy heavy :) | 16:53 |
@wiking | but yeah this is quite shitty :) | 16:53 |
@wiking | train_require_labels | 16:54 |
@wiking | i guess you agree :) | 16:54 |
wuwei | sure | 16:54 |
wuwei | yeah | 16:54 |
wuwei | but how can we know which one to call | 16:55 |
wuwei | fit(features*) or fit (features,labels) | 16:55 |
@wiking | yeah | 16:55 |
@wiking | can we actually use type for this? | 16:56 |
@wiking | i mean both preprocessors | 16:56 |
@wiking | and machines has types | 16:56 |
@wiking | EPreprocessorType | 16:56 |
@wiking | all machines needs labels right? | 16:57 |
wuwei | look up a table by type? | 16:58 |
wuwei | almost | 16:58 |
@wiking | mmmm clustering doesn | 16:58 |
wuwei | except linear latent machine | 16:58 |
@wiking | mmm | 16:58 |
@wiking | virtual bool train_require_labels() const { return false; } | 16:59 |
@wiking | this thing is around since when? :) | 16:59 |
@wiking | ok | 16:59 |
@wiking | nevermind my comments | 16:59 |
@wiking | :))) | 16:59 |
@wiking | cool | 16:59 |
@wiking | :D | 16:59 |
wuwei | i made it public | 16:59 |
@wiking | some minor stuff | 17:02 |
@wiking | ;) | 17:02 |
@wiking | i've just sent i | 17:03 |
@wiking | u | 17:03 |
@wiking | brb | 17:03 |
@wiking | Farouk, i'll be here in 10, just paste me stuff when u hacce it | 17:03 |
@wiking | *have | 17:03 |
Farouk | Sure wiking | 17:05 |
-!- Farouk [9cc2876f@gateway/web/freenode/ip.156.194.135.111] has quit [Ping timeout: 260 seconds] | 17:09 | |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has joined #shogun | 17:25 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385698900 | 17:25 |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has left #shogun [] | 17:25 | |
-!- HeikoS [~heiko@eduroam-int-pat-8-0.ucl.ac.uk] has quit [Ping timeout: 256 seconds] | 17:38 | |
-shogun-buildbot:#shogun- Build deb4 - python3 #437 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/28/builds/437 | 17:45 | |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has joined #shogun | 17:53 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 17:53 | |
-!- TwofTp [67d51504@gateway/web/freenode/ip.103.213.21.4] has joined #shogun | 17:54 | |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-smarjtpaswqoslpr] has quit [Quit: Connection closed for inactivity] | 17:57 | |
-!- TwofTp [67d51504@gateway/web/freenode/ip.103.213.21.4] has quit [Client Quit] | 17:58 | |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has joined #shogun | 18:07 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385704295 | 18:07 |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has left #shogun [] | 18:07 | |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/15624ba7941b174f6ba999012ca6154e95707d9c by vigsterkr | 18:09 |
-shogun-buildbot:#shogun- Build deb4 - python3 #438 is complete: Failure [failed python (failure)] - http://buildbot.shogun-toolbox.org:8080/#builders/28/builds/438 | 18:35 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4206 synchronized by FaroukY | 18:40 |
-!- Farouk [33267174@gateway/web/freenode/ip.51.38.113.116] has joined #shogun | 18:42 | |
-!- HeikoS1 [~heiko@eduroam-int-pat-8-0.ucl.ac.uk] has joined #shogun | 19:01 | |
-!- HeikoS1 [~heiko@eduroam-int-pat-8-0.ucl.ac.uk] has quit [Client Quit] | 19:01 | |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has quit [Ping timeout: 248 seconds] | 19:03 | |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has joined #shogun | 19:34 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385768440 | 19:34 |
-!- travis-ci [~travis-ci@ec2-54-161-208-147.compute-1.amazonaws.com] has left #shogun [] | 19:34 | |
-!- Farouk [33267174@gateway/web/freenode/ip.51.38.113.116] has quit [Ping timeout: 260 seconds] | 19:42 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4309 synchronized by FaroukY | 19:59 |
-!- Farouk [33267174@gateway/web/freenode/ip.51.38.113.116] has joined #shogun | 20:01 | |
-!- travis-ci [~travis-ci@ec2-54-224-19-64.compute-1.amazonaws.com] has joined #shogun | 20:13 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/385768440 | 20:13 |
-!- travis-ci [~travis-ci@ec2-54-224-19-64.compute-1.amazonaws.com] has left #shogun [] | 20:13 | |
Farouk | wiking, you here? | 20:59 |
@wiking | ye | 20:59 |
Farouk | Ohh, I have the pseudocode | 20:59 |
Farouk | send it? | 20:59 |
@wiking | ye | 21:00 |
@wiking | email whatever is good | 21:00 |
Farouk | https://pastebin.com/Yxf6sebY | 21:00 |
@wiking | perfect | 21:00 |
@wiking | lemme check | 21:00 |
@wiking | ok | 21:01 |
@wiking | so the problem is here for me is kind of like | 21:01 |
@wiking | the adhoc dropins | 21:01 |
@wiking | of the train features and labels | 21:02 |
@wiking | meaning | 21:02 |
@wiking | i mean i'm now cirticising the whole not your part per se | 21:02 |
@wiking | first of all how the NN and optimizer being connected? | 21:03 |
Farouk | what do you mean by adhoc dropins? | 21:03 |
Farouk | So the optimizer will operate on the cost function, which has the **weights** of the NN | 21:03 |
Farouk | so it's directly acting on the NN weights | 21:04 |
@wiking | what i mean is | 21:04 |
@wiking | that some way i'd like to have this encapsulated within nn | 21:04 |
@wiking | right? | 21:04 |
Farouk | Hmm, yea, but I was kind of thinking of it like the way in Tensorflow | 21:05 |
Farouk | so in tf the backprogation is not part of the nn itself | 21:05 |
@wiking | ok | 21:05 |
@wiking | i would do this compile time | 21:06 |
@wiking | sorry | 21:06 |
@wiking | in ctor | 21:06 |
@wiking | line 21-23 | 21:06 |
@wiking | i mean the problem is that actually | 21:06 |
@wiking | this is an sgobject (FirstOrderSAGCostFunctionInterface) | 21:06 |
@wiking | so if you would use this from modular (python etc) | 21:07 |
@wiking | then with the new api | 21:07 |
@wiking | it would look like this | 21:07 |
@wiking | loss.put("features", train_features) | 21:07 |
@wiking | loss.put("labels", labels_test) | 21:07 |
@wiking | etc | 21:07 |
@wiking | which is super combersome :D | 21:07 |
@wiking | *cumbersome | 21:08 |
@wiking | but yeah | 21:08 |
@wiking | there's gonna be the problem | 21:08 |
@wiking | of | 21:08 |
@wiking | ith_cost_function and total_cost_function | 21:08 |
@wiking | which is in a way impossible to be mapped into any interface atm | 21:08 |
@wiking | but lets keep that on ice for the time being | 21:08 |
Farouk | Ohh I see where you're coming with this. There isn't a particularly obvious way of definint the total_cost_function and ith_cost_function | 21:09 |
Farouk | like using .put() ? | 21:09 |
Farouk | adding them into the loss that is | 21:09 |
@wiking | yeah | 21:11 |
@wiking | but ok lets put that in the back log | 21:11 |
Farouk | Alright I'll keep that in mind | 21:11 |
Farouk | are you joining us tomorrow? | 21:12 |
Farouk | or is it too late? | 21:12 |
@wiking | i mean one way or another one could say that | 21:12 |
Farouk | I'll make sure to discuss it with Fernando | 21:12 |
@wiking | it'd be great to have all this auto loss = some<FirstOrderSAGCostFunctionInterface>(); | 21:12 |
@wiking | in the ctor | 21:12 |
@wiking | because i dont see the point changing these values | 21:12 |
@wiking | after construction | 21:12 |
@wiking | but yeah | 21:12 |
@wiking | need to think a bit | 21:13 |
Farouk | Hmmm yea I think it makes sense. training data shouldn't change, and cost function shouldn't (we just create a new one if we want) | 21:13 |
@wiking | Farouk, when's the meeting tomo? | 21:36 |
@wiking | Farouk, have you updated in this https://github.com/shogun-toolbox/shogun/pull/4206 the new data? | 21:37 |
@wiking | as the integration test seems to be failing | 21:37 |
@wiking | or idk if that needs change... but definitely the integration test seems to be failing :( | 21:39 |
Farouk | wiking, sorry I am downloading an irc client, it keeps disconnecting/not alerting me so hopefully after I download the client it will be better | 22:39 |
Farouk | the meeting is at 10pm | 22:40 |
@wiking | ok | 22:40 |
@wiking | will try to make it | 22:40 |
@wiking | sent u some feedbak | 22:40 |
Farouk | Yea, saw it and working on it | 22:40 |
Farouk | I think this was before you told me that i need to use obtain_from_generic | 22:41 |
Farouk | so I am changing it | 22:41 |
Farouk | Also regarding: labs? please rename labs to labels otherwise this will fail... as it fails now | 22:41 |
Farouk | if I rename to labels, then if I do: Labels labels= labels("...") | 22:42 |
Farouk | then it causes an error | 22:42 |
Farouk | that I am using labels before using it, so all labels have to have a different name | 22:42 |
@wiking | ok | 22:44 |
@wiking | then use labs | 22:44 |
@wiking | just be consistent | 22:44 |
@wiking | as this way its failing :) | 22:44 |
Farouk | Alright | 22:45 |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-koiqjutynwflbjuq] has joined #shogun | 22:57 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4206 synchronized by FaroukY | 23:11 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4309 synchronized by FaroukY | 23:19 |
-!- Farouk [33267174@gateway/web/freenode/ip.51.38.113.116] has quit [Quit: Page closed] | 23:35 | |
--- Log closed Thu May 31 00:00:29 2018 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!