--- Log opened Wed Nov 09 00:00:08 2016 | ||
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun | 03:56 | |
-!- lambday [31cf37a7@gateway/web/freenode/ip.49.207.55.167] has quit [Ping timeout: 260 seconds] | 06:59 | |
@sukey | Pull Request #3427 "get distance matrix for euclidean distance" synchronized by Saurabh7 - https://github.com/shogun-toolbox/shogun/pull/3427 | 07:29 |
---|---|---|
@sukey | New Comment on Pull Request "Yep, I have updated with that now." by Saurabh7: https://github.com/shogun-toolbox/shogun/pull/3427#discussion_r87138671 | 07:30 |
@sukey | Received a new pull_request_review event, just so you know. | 07:30 |
@sukey | New Comment on Pull Request "Hey @Saurabh7 sorry for being annoying again but just a minor comment. Isn't it possible to use the `SGMatrix<float64_t> result` matrix from here itself and then use an Eigen map for it for operations that follow? If operator `=` does a deep copy for Eigen (required in the expression `map_result = (map_l.transpose()*map_r).template cast<double>();`, then we're good, right? Saves an extra alloc and copy. Can you please check | 07:41 |
@sukey | doable?" by lambday: https://github.com/shogun-toolbox/shogun/pull/3427#discussion_r87139393 | 07:41 |
@sukey | Received a new pull_request_review event, just so you know. | 07:41 |
lisitsyn | hhahah | 07:50 |
@sukey | Pull Request #3427 "get distance matrix for euclidean distance" synchronized by Saurabh7 - https://github.com/shogun-toolbox/shogun/pull/3427 | 08:04 |
@sukey | Received a new pull_request_review event, just so you know. | 08:08 |
@sukey | New Comment on Pull Request "@lambday yep Its possibe, I have updated, can u stop previous travis maybe?" by Saurabh7: https://github.com/shogun-toolbox/shogun/pull/3427#discussion_r87141513 | 08:08 |
@sukey | Received a new pull_request_review event, just so you know. | 08:31 |
@sukey | New Comment on Pull Request "Hey @Saurabh7 ! Sorry again for being a PITA (I should have commented on this before) but doesn't `map_result = map_result.array().sqrt();` create a temporary matrix and then do a deep copy? In that case, would it be possible to have an in-place operation for element-wise sqrt for the `result` SGMatrix? Maybe use our `linalg` methods?" by lambday: https://github.com/shogun-toolbox/shogun/pull/3427#discussion_r87143361 | 08:31 |
@sukey | Received a new pull_request_review event, just so you know. | 08:33 |
@sonney2k | lisitsyn, trump will be president. So not for long and you will have no issues to come to the states :=) | 08:45 |
lisitsyn | sonney2k: absolutely | 08:45 |
lisitsyn | hahah | 08:45 |
@sonney2k | lisitsyn, might be that you have to wear a putin mask | 08:46 |
@sonney2k | and ride a whale or horse or sth | 08:46 |
lisitsyn | sonney2k: the greatest victory of russian foreign politics :D | 08:47 |
@sonney2k | lisitsyn, absolutely | 08:47 |
@sonney2k | I am thinking of changing my profile picture to look like a certain blond guy ;P | 08:48 |
@sonney2k | lisitsyn, is putin also more popular among woman? | 08:49 |
lisitsyn | sonney2k: not sure | 08:49 |
@sonney2k | lisitsyn, maybe he didn't find the right words... | 08:50 |
lisitsyn | sonney2k: well he is like 85-90% | 08:50 |
lisitsyn | it doesn't matter who are these 10% | 08:50 |
@sonney2k | looks like trump is superior after all in finding the words attracting woman | 08:50 |
lisitsyn | sonney2k: yeah grab them | 08:50 |
@sonney2k | grab them by.. | 08:50 |
lisitsyn | that's what you do | 08:51 |
@sonney2k | yeah maybe irrelevant then | 08:51 |
@sonney2k | the world just needs the best teachers it can get | 08:51 |
lisitsyn | heh | 08:52 |
@sonney2k | lisitsyn, I certainly need to learn horse riding, submarine diving,... | 08:53 |
@sonney2k | what else did he do? | 08:53 |
@sonney2k | the brain we leave to trump | 08:53 |
lisitsyn | sonney2k: grab some amphora | 08:53 |
lisitsyn | sonney2k: fly su-27 or mig-29 | 08:53 |
@sonney2k | lisitsyn, easy. doing that freediving down to 50m | 08:54 |
lisitsyn | a lot of things | 08:54 |
@sonney2k | lisitsyn, don't we have something more advanced like flying autopiloted space ships? | 08:54 |
lisitsyn | not yet! | 08:54 |
@sonney2k | or google autonomous cars w/o a steering wheel and causing an accident | 08:54 |
@sonney2k | man I need challenges! | 08:54 |
@sonney2k | real ones | 08:54 |
@sonney2k | those migs are too easy too steer anywyas | 08:55 |
lisitsyn | sonney2k: well try to manipulate us elections | 08:55 |
lisitsyn | he managed to | 08:55 |
lisitsyn | :D | 08:55 |
@sonney2k | ok I kind of admit this is an advanced subject | 08:56 |
@sonney2k | how many machine learners where involved on either side? | 08:56 |
@sonney2k | did he hack clinton's NN? | 08:56 |
@sonney2k | adversial learning? | 08:56 |
lisitsyn | sonney2k: she has good old svm | 08:57 |
@sonney2k | lisitsyn, btw at NIPS this year? | 08:57 |
lisitsyn | poison attack | 08:57 |
lisitsyn | sonney2k: for sure! | 08:57 |
lisitsyn | you? | 08:57 |
@sonney2k | no. sold out | 08:57 |
@sonney2k | and too late | 08:57 |
lisitsyn | meh | 08:57 |
lisitsyn | I registered early for this very reason | 08:57 |
@sonney2k | didn't expect that nor was I ready for it mind wise | 08:57 |
lisitsyn | sonney2k: I observe kinda shitstorm in twitter | 09:00 |
lisitsyn | :D | 09:00 |
lisitsyn | it's funny | 09:00 |
-!- praisethemoon [~praisethe@41.230.52.174] has joined #shogun | 09:02 | |
-!- praisethemoon [~praisethe@41.230.52.174] has quit [Changing host] | 09:03 | |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has joined #shogun | 09:03 | |
@sukey | Received a new pull_request_review event, just so you know. | 09:05 |
@sonney2k | lisitsyn, hehe | 09:20 |
@sonney2k | well partly funny this world.. | 09:20 |
@sonney2k | okido! lets say putin and trump will pave the way for german world domination o_O | 09:21 |
* sonney2k off | 09:21 | |
@sukey | New Comment on Issue #3513 by sudk1896: "@karlnapf : [Here](https://gist.github.com/sudk1896/1074c6c74ea6cfceac0d975b94cd19f5) is the gist. It will open as an ipython notebook. I have added some text and the code for time comparison for `KDTREE`. Let me know what else you want. " - https://github.com/shogun-toolbox/shogun/issues/3513#issuecomment-259358768 | 09:33 |
-!- sonne|work [~sonnenbu@88.134.176.135] has joined #shogun | 10:14 | |
-!- protino [0e8bc20c@gateway/web/freenode/ip.14.139.194.12] has joined #shogun | 10:22 | |
-!- protino [0e8bc20c@gateway/web/freenode/ip.14.139.194.12] has quit [Client Quit] | 10:22 | |
@sukey | New Comment on Issue #3513 by sudk1896: "@karlnapf : [Here](https://gist.github.com/sudk1896/1074c6c74ea6cfceac0d975b94cd19f5) is the gist. It will open as an ipython notebook. I have added some text and the code for time comparison for `KDTREE`. Let me know what else you want. | 10:25 |
@sukey | P.S. The changes I made are in the `Accelerating KNN` subsection. Also, you would want the PR to contain the ipython (.ipynb) file right ?" - https://github.com/shogun-toolbox/shogun/issues/3513#issuecomment-259358768 | 10:25 |
-!- zero1hac [~zerooneha@139.59.16.180] has quit [Ping timeout: 258 seconds] | 10:36 | |
-!- zero1hac [~zerooneha@139.59.16.180] has joined #shogun | 10:39 | |
-!- zero1hac [~zerooneha@139.59.16.180] has quit [Ping timeout: 244 seconds] | 10:48 | |
-!- zero1hac [~zerooneha@139.59.16.180] has joined #shogun | 10:50 | |
-!- praisethemoon_ [~praisethe@41.226.248.123] has joined #shogun | 10:56 | |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has quit [Ping timeout: 258 seconds] | 11:00 | |
-!- praisethemoon_ is now known as praisethemoon | 11:02 | |
-!- praisethemoon [~praisethe@41.226.248.123] has quit [Changing host] | 11:02 | |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has joined #shogun | 11:02 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 11:08 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:08 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 260 seconds] | 11:36 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 11:45 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:45 | |
sonne|work | HeikoS: where do we put the old demos? | 12:14 |
sonne|work | (on the website I mean) | 12:14 |
@HeikoS | sonne|work: nowhere | 12:14 |
@HeikoS | they are broken | 12:14 |
sonne|work | and how do we get all the examples converted? | 12:14 |
@HeikoS | suggestions? | 12:14 |
sonne|work | wiking showed my one and it was working.. | 12:15 |
@HeikoS | example converting happens by gsoc students mostly | 12:15 |
@HeikoS | and by us hopefully | 12:15 |
@HeikoS | we still have the python examples | 12:15 |
@HeikoS | I want to put a link somewhere for the meantime | 12:15 |
@HeikoS | good point in fact | 12:15 |
@HeikoS | ill do that today | 12:16 |
sonne|work | http://demos.shogun-toolbox.org/regression/svr/ | 12:16 |
sonne|work | that stuff is still nice.. | 12:16 |
@HeikoS | oh yeah | 12:16 |
@HeikoS | I agree | 12:16 |
@HeikoS | should put it back | 12:16 |
@HeikoS | problem is maintainance | 12:16 |
@HeikoS | I tried a few a while ago | 12:16 |
@HeikoS | and all were broken | 12:16 |
sonne|work | yes but they still work.. | 12:16 |
@HeikoS | which is worse than not having any online | 12:16 |
@HeikoS | you got an idea how to check whether they work? | 12:17 |
sonne|work | maybe we need to add a test for them then | 12:17 |
sonne|work | yeah separate the in this case svr regression call | 12:17 |
sonne|work | and just call it in some test | 12:17 |
@HeikoS | I guess yes | 12:17 |
sonne|work | with dummy data | 12:17 |
sonne|work | shouldn't be hard | 12:17 |
@HeikoS | yeah cool, will check that out | 12:17 |
@HeikoS | we want to put a link back in the showroom | 12:18 |
sonne|work | http://www.shogun-toolbox.org/page/about/contributions is broken | 12:18 |
sonne|work | (the link in) See our list of integrated external libraries. | 12:18 |
@HeikoS | is that linked from anywhere? | 12:18 |
@HeikoS | ah cool | 12:18 |
@HeikoS | thanks! | 12:18 |
@HeikoS | can you file an issue on the web2 repo? | 12:18 |
@HeikoS | ah now | 12:18 |
@HeikoS | its in main repo | 12:18 |
@HeikoS | the page is now in wiki | 12:18 |
@HeikoS | I will fix | 12:19 |
sonne|work | yeah I fiddled with it yesterday | 12:19 |
sonne|work | notebooks have a title now | 12:19 |
sonne|work | and api lists the full class list | 12:19 |
@HeikoS | ah cool! | 12:19 |
@HeikoS | nice one! | 12:19 |
@HeikoS | didnt see yet | 12:19 |
@HeikoS | thats really good | 12:19 |
@HeikoS | for examples | 12:19 |
@HeikoS | there is two components of a cookbook page | 12:19 |
@HeikoS | 1.) code listing | 12:19 |
@HeikoS | 2.) markdown page | 12:19 |
@HeikoS | what takes time to write is 2 | 12:19 |
@HeikoS | what is important internally is 1 (since testing=) | 12:20 |
@HeikoS | so I was thinking of just putting the plain listing if not markdown page exists | 12:20 |
@HeikoS | and have a list in the examples page "all" | 12:20 |
@HeikoS | sonne|work: btw you realise that now | 12:20 |
@HeikoS | we have integration testing across all modular targets? | 12:20 |
@HeikoS | they are all compared against the cpp output | 12:20 |
sonne|work | sry off for food | 12:21 |
@HeikoS | enjoy! | 12:21 |
Saurabh7 | gp demos still working :) | 13:11 |
Saurabh7 | need some tests | 13:11 |
@wiking | wooooooooooooooooho | 13:16 |
@wiking | we won :> | 13:16 |
@wiking | :PPPPPPPPPPPP | 13:16 |
@wiking | ok i was just joking .... | 13:19 |
@wiking | sonne|work, here? | 13:19 |
sonne|work | wiking: yes | 13:21 |
@wiking | ok jsut sent u an emai | 13:22 |
@wiking | l | 13:22 |
sonne|work | wiking: then years later after the NSA and various spam filters did proper spell checking ... it will eventually arrive | 13:23 |
@wiking | :> | 13:24 |
sonne|work | HeikoS: but cool stuff regarding the integration tests | 13:25 |
@HeikoS | sonne|work: yeah now we can really claim that we are multi interface since we know we get the same thing | 13:25 |
sonne|work | yeah that is outstanding... | 13:26 |
@HeikoS | now we need to port examples | 13:26 |
@HeikoS | as I said | 13:26 |
@HeikoS | not need to write the fancy markdown pages in the cookbook | 13:26 |
@HeikoS | but just copy pasting the examples | 13:26 |
@HeikoS | very easy from python | 13:26 |
@HeikoS | and that extends test coverage big time | 13:26 |
sonne|work | don't really understand how... | 13:27 |
@HeikoS | what how? | 13:27 |
@HeikoS | how to port? | 13:27 |
@HeikoS | or why it extends test cov? | 13:28 |
@HeikoS | for testing, tell in some sense it doesnt as results should be covered by unit tests and most typemaps are already tested with the existing examples | 13:28 |
@HeikoS | but still, it is good to explicitly know that everything gives same result | 13:29 |
sonne|work | yes how to port examples | 13:34 |
sonne|work | I mean there is just this small set now.. | 13:34 |
sonne|work | this and the demos and a fully working website would be nice at least.. | 13:34 |
-!- Saurabh7 [Saurabh7@gateway/shell/panicbnc/x-upcolwjrlomkbegm] has quit [Ping timeout: 260 seconds] | 15:14 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 244 seconds] | 15:30 | |
@sukey | Pull Request #3521 "Notebook updated to v3 version and formatted #3517" opened by amarlearning - https://github.com/shogun-toolbox/shogun/pull/3521 | 15:35 |
@sukey | Pull Request #3521 "Notebook updated to v3 version and formatted #3517" - https://github.com/shogun-toolbox/shogun/pull/3521 | 15:35 |
@wiking | heiko | 15:42 |
@wiking | where the fuck are you? :D | 15:42 |
sonne|work | wiking: he is british now.. tea time :P | 15:46 |
@wiking | :) | 15:47 |
@wiking | lol | 15:47 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 17:08 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 17:08 | |
sonne|work | wiking: there you go. victim online. | 17:12 |
@wiking | HeikoS, yoyoyo | 17:12 |
@wiking | what the fuck i wanted from you | 17:12 |
@wiking | ;) | 17:12 |
@wiking | ah yeah | 17:12 |
@wiking | we have some funky shit on osx | 17:12 |
@wiking | :( | 17:12 |
@wiking | currently debugging http://pastebin.com/8H7hkbc0 | 17:13 |
@wiking | but this blocks us on homebrew | 17:13 |
@wiking | ;( | 17:13 |
@wiking | we should try maybe hotfix once this is out ?D:DDD | 17:13 |
@wiking | 5.0.1 | 17:13 |
@wiking | ? | 17:13 |
@wiking | :D | 17:13 |
@wiking | but this is not why i was looking for you | 17:16 |
@sukey | Issue #3522 "fix links in introduction notebook" opened by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3522 | 17:16 |
@sukey | Issue #3522 "fix links in introduction notebook"- https://github.com/shogun-toolbox/shogun/issues/3522 | 17:16 |
@sukey | Issue #3522 "fix links in introduction notebook" karlnapf added label: "Documentation" - https://github.com/shogun-toolbox/shogun/issues/3522 | 17:16 |
@sukey | New Comment on Pull Request #3521 by karlnapf: "Thanks! | 17:17 |
@sukey | Can you post a gist of the notebook with outputs? | 17:17 |
@sukey | Also, can you mention what changed? The diff is not helpful here" - https://github.com/shogun-toolbox/shogun/pull/3521#issuecomment-259454302 | 17:17 |
@HeikoS | wiking: jojo | 17:18 |
@HeikoS | wiking: yeah good point :) | 17:18 |
@HeikoS | wiking: let me do the 5.0.1 I wanna see how it works | 17:19 |
@wiking | :> | 17:19 |
@wiking | ok | 17:19 |
@wiking | lemme first try to fix that shit | 17:19 |
@HeikoS | kk | 17:19 |
@HeikoS | from now on release every week :) | 17:19 |
@sukey | New Comment on Issue #3520 by karlnapf: "Yeah, it is not really a solve but rather a workaround :) We should deal with this properly...." - https://github.com/shogun-toolbox/shogun/issues/3520#issuecomment-259455038 | 17:20 |
@sukey | New Comment on Issue #3513 by karlnapf: "Can we moive the discussion about your changes into a PR?" - https://github.com/shogun-toolbox/shogun/issues/3513#issuecomment-259455264 | 17:21 |
-!- Hieronymous [~sudarshan@122.169.162.187] has joined #shogun | 17:21 | |
@sukey | New Comment on Pull Request #3521 by karlnapf: "BTW can we please split this patch into two: | 17:22 |
@sukey | 1. Just updating the format, no changes at all | 17:22 |
@sukey | 2. The changes you did | 17:22 |
@sukey | Thanks!" - https://github.com/shogun-toolbox/shogun/pull/3521#issuecomment-259455420 | 17:22 |
@sukey | New Comment on Issue #3513 by sudk1896: "Sure, I will submit the `ipynb` file and we can edit them as we go. Assuming you want me to commit the `ipynb` file. Right ?" - https://github.com/shogun-toolbox/shogun/issues/3513#issuecomment-259455496 | 17:22 |
@sukey | New Comment on Issue #3513 by karlnapf: "Yes" - https://github.com/shogun-toolbox/shogun/issues/3513#issuecomment-259455774 | 17:23 |
@sukey | New Comment on Pull Request #3521 by amarlearning: "> BTW can we please split this patch into two: | 17:23 |
@sukey | > Just updating the format, no changes at all | 17:23 |
@sukey | > The changes you did | 17:23 |
@sukey | @karlnapf ??" - https://github.com/shogun-toolbox/shogun/pull/3521#issuecomment-259455819 | 17:23 |
@sukey | New Comment on Pull Request #3521 by amarlearning: "> BTW can we please split this patch into two: | 17:24 |
@sukey | > Just updating the format, no changes at all | 17:24 |
@sukey | > The changes you did | 17:24 |
@sukey | @karlnapf ??" - https://github.com/shogun-toolbox/shogun/pull/3521#issuecomment-259455819 | 17:24 |
@wiking | arghfuck | 17:26 |
@wiking | why the fuck the external project not being triggered? | 17:26 |
@sukey | Pull Request #3523 "[#3513] Added time comparison for KNN_KDTREE." opened by sudk1896 - https://github.com/shogun-toolbox/shogun/pull/3523 | 17:29 |
@sukey | New Comment on Pull Request #3521 by amarlearning: "@karlnapf | 17:29 |
@sukey | 1. I have updated the `Author (GitHub ID: githubid)` | 17:29 |
@sukey | 2. I also formatted the reference section according to the new [template](https://github.com/shogun-toolbox/shogun/blob/develop/doc/ipython-notebooks/template.ipynb) | 17:29 |
@sukey | 3. I also updated the Ipython Notebook to the latest v3. | 17:29 |
@sukey | You can see the preview of changed notebook [here](https://github.com/amarlearning/shogun/blob/amarlearning-NoteBook/doc/ipython-notebooks/classification/HashedDocDotFeatures.ipynb)" - https://github.com/shogun-toolbox/shogun/pull/3521#issuecomment-259457462 | 17:29 |
@sukey | New Comment on Pull Request #3521 by amarlearning: "@karlnapf | 17:30 |
@sukey | 1. I have updated the `Author (GitHub ID: githubid)` | 17:30 |
@sukey | 2. I also formatted the reference section according to the new [template](https://github.com/shogun-toolbox/shogun/blob/develop/doc/ipython-notebooks/template.ipynb) | 17:30 |
@sukey | 3. I also updated the Ipython Notebook to the latest v3. | 17:30 |
@sukey | You can see the preview of changed notebook [here](https://github.com/amarlearning/shogun/blob/amarlearning-NoteBook/doc/ipython-notebooks/classification/HashedDocDotFeatures.ipynb)" - https://github.com/shogun-toolbox/shogun/pull/3521#issuecomment-259457462 | 17:30 |
@sukey | Pull Request #3523 "Added time comparison for KNN_KDTREE." - https://github.com/shogun-toolbox/shogun/pull/3523 | 17:30 |
@sukey | New Comment on Issue #3513 by sudk1896: "@karlnapf : It says conflicting file. I replaced the `KNN.ipynb` file in `/doc/ipython-notebooks/multiclass`. I assumed that was the right thing to do." - https://github.com/shogun-toolbox/shogun/issues/3513#issuecomment-259459240 | 17:35 |
@sukey | Pull Request #3524 "Fix a few links in the notebook" opened by souravsingh - https://github.com/shogun-toolbox/shogun/pull/3524 | 17:58 |
@sukey | New Comment on Pull Request #3521 by karlnapf: "As we cannot see the changes in your diff (all lines are different due to the format change), can you please send a PR where you only update the format (no other changes) first. Then we merge this, and then you can send the changes you made. This way, I can see them in the diff which is important for future bug hunting (knowing in which patch were which changes etc)" - | 18:03 |
@sukey | https://github.com/shogun-toolbox/shogun/pull/3521#issuecomment-259466717 | 18:03 |
@HeikoS | wiking: this sukey guy is kinda noisy :) | 18:04 |
@wiking | ok what should i remove | 18:05 |
@wiking | ? | 18:05 |
@wiking | and it's a she | 18:05 |
-!- sukey [~nodebot@ks312251.kimsufi.com] has quit [Remote host closed the connection] | 18:05 | |
@wiking | export HUBOT_GITHUB_EVENT_NOTIFIER_TYPES=issues,issue_comment,pull_request,push,page_build,gollum,create,commit_comment,pull_request_review_comment,pull_request_review | 18:05 |
@wiking | these | 18:05 |
@wiking | are the current triggers that are happening | 18:05 |
@wiking | HeikoS, ^ | 18:07 |
@wiking | have you seen this btw https://github.com/airbnb/knowledge-repo | 18:07 |
@HeikoS | wiking: yeah, I am wondering | 18:07 |
@HeikoS | nope, checking | 18:07 |
@HeikoS | wiking: you think al PR comments should be in irc | 18:07 |
@wiking | we can just remove shit | 18:07 |
@wiking | like review | 18:07 |
@wiking | and comment | 18:07 |
@wiking | and commit comment | 18:07 |
@wiking | if you want | 18:07 |
@wiking | just have notification | 18:08 |
@wiking | if there's a new pr | 18:08 |
@wiking | or issue | 18:08 |
@wiking | export HUBOT_GITHUB_EVENT_NOTIFIER_TYPES=issues,issue_comment,pull_request,push,page_build,gollum,create | 18:08 |
@wiking | this might be better | 18:08 |
@wiking | less noise | 18:08 |
@wiking | - issue_comment | 18:08 |
@wiking | export HUBOT_GITHUB_EVENT_NOTIFIER_TYPES=issues,pull_request,push,page_build,gollum,create | 18:08 |
-!- sukey [~nodebot@ks312251.kimsufi.com] has joined #shogun | 18:09 | |
-!- mode/#shogun [+o sukey] by ChanServ | 18:09 | |
@wiking | new and improved | 18:09 |
@wiking | :) | 18:09 |
@HeikoS | nice :) | 18:09 |
@HeikoS | yep thats the way | 18:10 |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has quit [Ping timeout: 268 seconds] | 18:13 | |
@wiking | HeikoS, btw https://soundcloud.com/formationrecords/sets/djss-presents-formation-sounds | 18:14 |
@wiking | jungle/dnb | 18:14 |
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Quit: Leaving] | 18:17 | |
@sukey | Pull Request #3525 "Rename interface switches in cmake" opened by souravsingh - https://github.com/shogun-toolbox/shogun/pull/3525 | 18:22 |
@HeikoS | wiking: nice | 18:22 |
@HeikoS | wiking: soooooooo | 18:22 |
@HeikoS | some things | 18:22 |
@HeikoS | wiking: rahul will soon have the feature branch cleaned up | 18:22 |
@HeikoS | And I want to merge it :) | 18:22 |
@HeikoS | might want to make that 5.1 | 18:23 |
@HeikoS | and rename 5.1 milestone to 5.2 | 18:23 |
@HeikoS | thoughts? | 18:23 |
@sukey | Pull Request #3521 "Notebook updated to v3 version and formatted #3517" synchronized by amarlearning - https://github.com/shogun-toolbox/shogun/pull/3521 | 18:23 |
@HeikoS | wiking: second is this parameter stuff | 18:23 |
@HeikoS | wiking: I think before we think about how we do this, we need some discussion on our motivation there | 18:23 |
@HeikoS | wiking: because we are entering probabilistic programming language land with this | 18:24 |
@HeikoS | and the question is whether this is useful for us | 18:24 |
@HeikoS | or whether some simpler things might be sufficient | 18:24 |
@HeikoS | I currently can think of two advantages that parameter domains would have for us | 18:24 |
@HeikoS | 1.) setting paramters has automatic checks .... pretty minor | 18:24 |
@HeikoS | 1a. model selection might be a bit easier compared to these model selection trees. | 18:25 |
@HeikoS | 2.) | 18:25 |
@HeikoS | automatic change of parametrisation under the hood | 18:25 |
@HeikoS | this is really useful when gradients are computed | 18:25 |
@HeikoS | i.e. any optimization or sampling algo | 18:25 |
@HeikoS | the thing I realised yesterday: | 18:26 |
@HeikoS | Shogun is more a collection of hard-coded efficient algos | 18:26 |
@HeikoS | point 2.) only is useful if we have a way to define models in shogun | 18:26 |
@HeikoS | and then want to optimize them (like a probab. prog. lang) | 18:27 |
@HeikoS | the GP stuff could be a use case here | 18:27 |
@HeikoS | but generally, I have doubts | 18:27 |
@HeikoS | we dont do this at many places | 18:27 |
@wiking | but allowing it | 18:27 |
@wiking | wouldn't hurt no | 18:27 |
@wiking | ? | 18:27 |
@HeikoS | no i agree | 18:27 |
@HeikoS | but its a massive undertaking | 18:27 |
@HeikoS | whereas point 1 is simple | 18:27 |
@wiking | yeah but that's like | 18:27 |
@wiking | version 6.0.0 | 18:27 |
@wiking | or 7.0.0. | 18:28 |
@wiking | i mean this is just simple discussion | 18:28 |
@wiking | we should not limit ourselves | 18:28 |
@wiking | to easy stuff | 18:28 |
@wiking | :) | 18:28 |
@HeikoS | I mean if we want to move shogun more towards a modern architechture where all models are defined in a non c++-code way, and inference/optimization be automatic, then it makes sense to do that | 18:28 |
@HeikoS | but do we want to? | 18:28 |
@HeikoS | And closely related: is that feasible? | 18:28 |
@HeikoS | since it would require to rewrite almost all algorithms | 18:29 |
@wiking | yes | 18:29 |
@wiking | :) | 18:29 |
@wiking | and if worst case not | 18:29 |
@HeikoS | we cant just use libsvm anymore | 18:29 |
@HeikoS | and currently, only GP would benefit | 18:29 |
@HeikoS | so I just challenge whether that is a good idea | 18:29 |
@wiking | yeah but we could chagne things right? | 18:29 |
@HeikoS | there is tensorflow, gpflow, etc | 18:29 |
@wiking | i mean i fucking hate libsvm | 18:29 |
@wiking | for the fact | 18:29 |
@HeikoS | these libraries explicitly do that kind of stuff | 18:29 |
@wiking | that there's *Everything* hardcoded | 18:30 |
@wiking | not even a simple linalg thing can be plugged in or out | 18:30 |
@HeikoS | I know | 18:30 |
@HeikoS | Well Shogun is like that | 18:30 |
@HeikoS | if we want to change, that's essentially a re-write | 18:30 |
@HeikoS | I am happy to discuss that | 18:30 |
@HeikoS | but | 18:30 |
@HeikoS | just wanted to point that out | 18:31 |
@wiking | yep | 18:31 |
@wiking | but now we have bigger issues | 18:31 |
@wiking | so lets this discussion run | 18:31 |
@HeikoS | yeah as well | 18:31 |
@wiking | dont kill it just yet | 18:31 |
@wiking | see where it leasds | 18:31 |
@HeikoS | nono | 18:31 |
@wiking | *leads | 18:31 |
@HeikoS | I mean I discuss it at this moment right? :) | 18:31 |
@wiking | ue yep | 18:31 |
@wiking | fuck | 18:32 |
@wiking | i dont get it | 18:32 |
@wiking | why this only fails in osx | 18:32 |
@HeikoS | wiking: man this PR ;) | 18:33 |
@wiking | "In fact I just realised all notebooks will have broken links. | 18:40 |
@wiking | ? | 18:40 |
@wiking | wtf | 18:40 |
@wiking | ??? | 18:40 |
@HeikoS | they used the old websize link | 18:40 |
@HeikoS | and not the alias | 18:40 |
@HeikoS | at least in some of them | 18:40 |
@wiking | what? | 18:40 |
@wiking | can you give an example/ | 18:40 |
@wiking | ? | 18:40 |
@HeikoS | the use explicit links, | 18:41 |
@HeikoS | http://shogun.ml/notebook/latest/Introduction.html | 18:41 |
@wiking | ok | 18:41 |
@wiking | gimme the explicit | 18:41 |
@wiking | link | 18:41 |
@wiking | not where it is | 18:41 |
@wiking | because there are million links there in it | 18:41 |
@HeikoS | http://www.shogun-toolbox.org/api/current/classshogun_1_1CStringFeatures.html | 18:42 |
@wiking | good | 18:42 |
@wiking | do not fucking change these please | 18:42 |
@wiking | :) | 18:42 |
@wiking | just fix the website | 18:42 |
@wiking | i did this already | 18:42 |
@HeikoS | ah ok | 18:42 |
@HeikoS | well best is to use alias | 18:42 |
@HeikoS | no? | 18:42 |
@wiking | https://github.com/shogun-toolbox/shogun-web2/commit/e3a6caca8a256d3451b07977a032a8f44fc5c485 | 18:42 |
@wiking | no | 18:42 |
@wiking | useless commits | 18:42 |
@HeikoS | okok | 18:42 |
@wiking | just fucking fix the links | 18:42 |
@HeikoS | ok | 18:43 |
@wiking | backward compatibility | 18:43 |
@HeikoS | then we can merge that patch | 18:43 |
@wiking | i'll fix the website | 18:43 |
@wiking | if it gets | 18:43 |
@wiking | /doc/current/... | 18:43 |
@wiking | it'll translate it to | 18:43 |
@wiking | /api/latest/... | 18:43 |
@wiking | that's all | 18:43 |
@HeikoS | cool | 18:46 |
@HeikoS | so the string features get solved as well? | 18:46 |
@HeikoS | then Ill merge this patch he sent | 18:46 |
@sukey | Pull Request #3524 "Fix a few links in the notebook" merged by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3524 | 18:46 |
@sukey | New Commit "Merge pull request #3524 from souravsingh/fix-links | 18:46 |
@sukey | Fix a few links in the notebook" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/8c7cd1dec360c9f62136753174c00dc6dc868d05 | 18:46 |
@HeikoS | wiking: ok lets see what else did I want to bring up | 18:47 |
@HeikoS | Shall we announce release on mailing list? | 18:47 |
@HeikoS | and everywhere? | 18:47 |
@HeikoS | or rather wait to 5.1? | 18:47 |
@HeikoS | with all the GSoC stuff in it? | 18:47 |
@HeikoS | wiking: I was thinking of getting into that linalg merge next | 18:48 |
@HeikoS | wiking: other suggestions? | 18:48 |
@HeikoS | wiking: btw there are more broken links in that intro notebook | 18:54 |
@sukey | Issue #3522 "locate broken links in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3522 | 18:55 |
@sukey | Issue #3522 "locate broken links and format conversion errors in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3522 | 18:56 |
@sukey | Issue #3522 "locate broken links and format conversion errors in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3522 | 18:57 |
@wiking | like | 18:59 |
@wiking | can you copy the LINK | 18:59 |
@wiking | not where it is | 18:59 |
@HeikoS | http://www.shogun-toolbox.org/doc/en/current/classshogun_1_1CStringFeatures.html | 19:00 |
@HeikoS | this one we had | 19:00 |
@HeikoS | http://www.shogun-toolbox.org/doc/en/latest/classshogun_1_1CBinaryLabels.html | 19:00 |
@HeikoS | this one works now | 19:00 |
@HeikoS | (notebook renders it wrong) | 19:00 |
@HeikoS | wiking: btw | 19:00 |
@HeikoS | wiking: do you have ideas on | 19:00 |
@wiking | okok | 19:00 |
@wiking | this is the same story | 19:00 |
@wiking | i have a fix for it | 19:00 |
@HeikoS | how to unify cookbook and doxygen | 19:00 |
@HeikoS | would be cool not to write everything twice | 19:01 |
@HeikoS | lisitsyn: ^ | 19:01 |
@wiking | HeikoS, https://github.com/shogun-toolbox/shogun-web2/commit/30d109eea00686df858c033652d68c485da2b354 | 19:01 |
@wiking | this will be deployed in a minute | 19:01 |
@HeikoS | wiking: cool! | 19:01 |
@HeikoS | wiking: so we have the information in cookbook: high-level desription, math, papers | 19:01 |
@HeikoS | and in doxygen we also have: API details, like "this method does this, overload it blabla", that is more c++ related | 19:02 |
@HeikoS | So I wonder | 19:02 |
@HeikoS | whether we can re-use | 19:02 |
@HeikoS | and this way, cookbook polishing also polishes API docs | 19:02 |
@HeikoS | at least parts of it | 19:03 |
@wiking | re-use what way | 19:03 |
@wiking | be less high level | 19:03 |
@HeikoS | ok | 19:03 |
@HeikoS | 1sec | 19:03 |
@HeikoS | http://shogun.ml/cookbook/latest/examples/binary_classifier/kernel_svm.html | 19:04 |
@HeikoS | http://www.shogun-toolbox.org/api/latest/classshogun_1_1CSVM.html | 19:04 |
@HeikoS | redundant! | 19:04 |
@HeikoS | double maintenance, double polishing, double writing | 19:04 |
@HeikoS | Could the cookbook re-use parts of doxygen | 19:05 |
@HeikoS | @brief | 19:05 |
@HeikoS | http://shogun.ml/cookbook/latest/examples/regression/random_forest_regression.html | 19:05 |
@HeikoS | http://www.shogun-toolbox.org/api/latest/classshogun_1_1CRandomForest.html | 19:05 |
@HeikoS | http://shogun.ml/cookbook/latest/examples/multiclass_classifier/random_forest.html | 19:06 |
@HeikoS | does it make sense, or shall we just stick with current approach? | 19:06 |
@HeikoS | better decide (and potentially change) that soon, before it becomes impossible | 19:06 |
@wiking | btw | 19:12 |
@wiking | https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/regression/LeastAngleRegression.h#L201-L202 | 19:13 |
@wiking | why reference of the matriceis? | 19:13 |
@HeikoS | I dont know what you mean | 19:14 |
-!- ssapo [75ee7297@gateway/web/freenode/ip.117.238.114.151] has joined #shogun | 19:14 | |
@HeikoS | wiking: ^ | 19:15 |
@HeikoS | wiking: BTW, rahuls code is faster than solving the same problem with mkl :) | 19:16 |
@HeikoS | (for this mmd stuff | 19:16 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 260 seconds] | 19:23 | |
-!- ssapo [75ee7297@gateway/web/freenode/ip.117.238.114.151] has quit [Quit: Page closed] | 19:28 | |
@sukey | Pull Request #3523 "Added time comparison for KNN_KDTREE." synchronized by sudk1896 - https://github.com/shogun-toolbox/shogun/pull/3523 | 19:32 |
@sukey | Pull Request #3523 "Added time comparison for KNN_KDTREE." synchronized by sudk1896 - https://github.com/shogun-toolbox/shogun/pull/3523 | 19:38 |
@sukey | Pull Request #3523 "Added time comparison for KNN_KDTREE." closed by sudk1896 - https://github.com/shogun-toolbox/shogun/pull/3523 | 19:43 |
@sukey | Pull Request #3523 "Added time comparison for KNN_KDTREE." synchronized by sudk1896 - https://github.com/shogun-toolbox/shogun/pull/3523 | 19:43 |
@sukey | Pull Request #3521 "Notebook updated to v3 version and formatted #3517" synchronized by amarlearning - https://github.com/shogun-toolbox/shogun/pull/3521 | 19:46 |
@sukey | Pull Request #3526 "Adding time comparison for KNN_KDTREE." opened by sudk1896 - https://github.com/shogun-toolbox/shogun/pull/3526 | 19:49 |
@sukey | Pull Request #3521 "Notebook updated to v3 version and formatted #3517" closed by amarlearning - https://github.com/shogun-toolbox/shogun/pull/3521 | 19:50 |
-!- zero1hac [~zerooneha@139.59.16.180] has quit [Ping timeout: 268 seconds] | 20:04 | |
-!- zero1hac [~zerooneha@139.59.16.180] has joined #shogun | 20:08 | |
-!- Hieronymous [~sudarshan@122.169.162.187] has quit [Quit: Leaving] | 20:14 | |
-!- zero1hac [~zerooneha@139.59.16.180] has quit [Ping timeout: 246 seconds] | 20:24 | |
@sukey | Pull Request #3527 "Converted from an older notebook format (v3) to the format (v4)" opened by amarlearning - https://github.com/shogun-toolbox/shogun/pull/3527 | 20:38 |
-!- zero1hac [~zerooneha@139.59.16.180] has joined #shogun | 21:11 | |
-!- Saurabh7 [Saurabh7@gateway/shell/panicbnc/x-hmqprtgyaxjnjouc] has joined #shogun | 21:56 | |
-!- zero1hac [~zerooneha@139.59.16.180] has quit [Ping timeout: 252 seconds] | 23:24 | |
-!- zero1hac [~zerooneha@139.59.16.180] has joined #shogun | 23:31 | |
-!- zero1hac [~zerooneha@139.59.16.180] has quit [Ping timeout: 258 seconds] | 23:42 | |
-!- zero1hac [~zerooneha@139.59.16.180] has joined #shogun | 23:49 | |
--- Log closed Thu Nov 10 00:00:09 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!