Hi David, Yes, I think including NuPIC into NuPIC Studio installer makes perfect sense. Nvidia has just released DIGITS package <http://devblogs.nvidia.com/parallelforall/digits-deep-learning-gpu-training-system/> (GUI frontend for Caffe), where they opted to include both Caffe and CUDA into the installer. DIGITS seems to be conceptually similar to Nupic Studio (Caffe plays the role of NuPIC), so you might want to take a look at it.
Regarding installation issues: one thing you can try to help new users is to take a brand new installation of Ubuntu or Centos, with only the default tools installed, and record every step you make to get from the first time you opened the terminal to the point of a working NuPIC studio. By recording every step I don't mean saying something like "install PyQt ", I mean you paste the actual command you used, and the output you received (especially any error messages). This way if I get stuck trying to install your code on my existing system, I can always spin a new clean vm, and be sure that if I follow all your steps exactly, it will work. To be honest, I'm not even sure if it makes sense to announce your code until all sections on this page <https://github.com/nupic-community/nupic.studio/wiki> are complete. On Wed, Mar 18, 2015 at 12:42 PM, Robert Smith <[email protected]> wrote: > David, > > Sure, if you need any Linux Mint testing let me know. Changing the > dependency for pyqtgraph to 0.9.10 and removing that for scipy still left > me with a numpy 1.8.1 install problem. I can look into that further if you > want, but as a temporary measure I changed the numpy dependency to version > 1.9.2 (ie what I have already for nupic) and the build completes and I can > run nustudio, or at least I could load up the three provided projects, step > through them and see pretty pictures. I guess next I should learn how to > use the studio, but that'll have to wait until tomorrow. > > Regards, Robert > > On 18 March 2015 at 11:30, David Ragazzi <[email protected]> wrote: > >> Hi Robert, >> >> Unfortunately Linux Mint/Ubuntu is the OS that most presented problems >> with NuPIC Studio. But finally a good news: PyQtGraph 0.9.10 was released. >> Tthe current version used by NuPIC Studio is 0.9.8 which requires the >> problematic SciPy, while 0.9.10 not requires it! >> >> Next week I'll test this new version in order to check if these >> instalation issues were solved. But if you want help me and accelerate >> this, you could update locally setup.py by update pyqtgraph to 0.9.10 ( >> https://github.com/nupic-community/nupic.studio/blob/master/setup.py#L55) >> and remove scipy dependency ( >> https://github.com/nupic-community/nupic.studio/blob/master/setup.py#L59). >> Once this done, run install again and let me know if this solve your >> problem. >> >> Regard, >> >> On 18 March 2015 at 07:21, Robert Smith <[email protected]> >> wrote: >> >>> Hi David, >>> >>> The recent messages prompted me to try installing the Studio again, so >>> here are my own experiences. The starting point is a 64-bit Linux Mint >>> 17.1, freshly installed a few weeks ago when I came across NuPIC for the >>> first time. Initially I wasn't using virtualenv, indeed my first avenue of >>> exploration was htm.java. I didn't have much time to devote to the studio, >>> although it did seem to offer just what I wanted to explore and understand >>> nupic better. I gave up at missing libraries when installing scipy. >>> >>> The second attempt (ongoing as I write) was with virtaulenv. On first >>> running "python setup.py build" it whinged about the above scipy prroblem. >>> Doing "pip install scipy" finally worked after adding the following >>> packages: libopenbas-dev, liblapack-dev, gfortan. >>> >>> This got me as far as falling over a problem with PyQt4, which I don't >>> think I would have had but for using virtualenv. I couldn't find a nice way >>> of fixing this, so I just symlinked the PyQt4 directory and some sip*.py >>> files to package installed versions. >>> >>> That allowed things to progress further, but after 10 minutes it seems >>> to be having problems with a library from >>> http://math-atlas.sourceforge.net/ although I don't have time now to >>> look at the underlying cause. Indeed it looks like it's bulding numpy. >>> That's, err, numpty. Oh well, too much time spent on this now. >>> >>> Regards, Robert >>> >>> >>> On 18 March 2015 at 01:37, David Ragazzi <[email protected]> wrote: >>> >>>> Hi Michael, >>>> >>>> Sorry for the delay.. I'm running into problems because currently I'm >>>> without my notebook.. :-( >>>> >>>> Well.. Matt is correct in this points (thanks Matt!).. Although NuPIC >>>> Studio use good third software like PyQt and PyQtGraph, sometimes they are >>>> tricky to install (not always, but happens with some users). :-( >>>> Hopefully we will have new versions of such dependencies and so these >>>> issues are solved soon. >>>> >>>> As Matt said, NuPIC Studio is not for "to create applications, but for >>>> experimenting with network setups".. The real intention is experiment >>>> things, but a big advantage is that you can copy to clipboard the code >>>> generated for your network setup in order to you re-use the code in your >>>> application! This decrease a lot of work/learning curve for you. >>>> >>>> Answering your question about NuPIC include NuPIC Studio, I have >>>> another idea: How about the reverse, i.e. include NuPIC library in NuPIC >>>> Studio dependencies? So when you install NuPIC Studio, automatically it >>>> install NuPIC (via pip)! I did not this before because NuPIC was needing >>>> some improvements related to setup.py, but maybe this is possible now.. >>>> >>>> Let me know your thoughts, they seem interesting... >>>> >>>> On 12 March 2015 at 19:09, Matthew Taylor <[email protected]> wrote: >>>> >>>>> A few reasons, from my perspective. >>>>> >>>>> 1. It doesn't seem to be meant to create applications, but for >>>>> experimenting with network setups >>>>> 2. It is not easy to install and get running >>>>> 3. We do not own the copyright for the code >>>>> >>>>> I think David Ragazzi has taken great initiative in creating NuPIC >>>>> Studio, and created a very interesting toolbox for NuPIC. But we've >>>>> been trying to make NuPIC easier to install, so #2 above is my biggest >>>>> concern. Bundling NuPIC Studio with NuPIC would introduce more >>>>> dependencies that are hard for people to install, making NuPIC harder >>>>> to install as well. >>>>> >>>>> >>>>> --------- >>>>> Matt Taylor >>>>> OS Community Flag-Bearer >>>>> Numenta >>>>> >>>>> >>>>> On Thu, Mar 12, 2015 at 2:59 PM, Michael Klachko >>>>> <[email protected]> wrote: >>>>> > Depending on the purpose of Nupic Studio, my next question might be: >>>>> why is >>>>> > it not included into Nupic binary? Why is it not the primary >>>>> interface to >>>>> > Nupic for new users? >>>>> > >>>>> > On Thu, Mar 12, 2015 at 2:38 PM, Matthew Taylor <[email protected]> >>>>> wrote: >>>>> >> >>>>> >> On Thu, Mar 12, 2015 at 2:27 PM, Michael Klachko >>>>> >> <[email protected]> wrote: >>>>> >> > What is the purpose of Nupic Studio, and how does it relate to >>>>> your main >>>>> >> > priority ("to help users build successful applications with >>>>> Nupic")? >>>>> >> >>>>> >> NuPIC studio is not Numenta code. It was build by David Ragazzi, a >>>>> >> NuPIC committer, and he maintains the codebase. I'll let him answer >>>>> >> questions about this project. >>>>> >> >>>>> >> Regards, >>>>> >> --------- >>>>> >> Matt Taylor >>>>> >> OS Community Flag-Bearer >>>>> >> Numenta >>>>> >> >>>>> > >>>>> >>>>> >>>> >>>> >>>> -- >>>> David Ragazzi >>>> MSc in Sofware Engineer (University of Liverpool) >>>> OS Community Commiter at Numenta.org >>>> -- >>>> "I think James Connolly, the Irish revolutionary, is right when he says >>>> that >>>> the only prophets are those who make their future. So we're not >>>> anticipating, we're working for it." >>>> >>> >>> >> >> >> -- >> David Ragazzi >> MSc in Sofware Engineer (University of Liverpool) >> OS Community Commiter at Numenta.org >> -- >> "I think James Connolly, the Irish revolutionary, is right when he says that >> the only prophets are those who make their future. So we're not >> anticipating, we're working for it." >> > >
