|
Yes it indeed requires Torch. I was thinking more about reading it to
get some ideas than using it as is in your case. ;)
And I don't think that Torch does not modify luajit binary. It is a lua
library as any other, just a little bit complicated. The `th` command
that it provides is just calling `luajit` with some header like loading
`torch` module and prompting a message and providing a nice REPL.
Élie
On 16.10.2015 02:46, Paco Zamora Martínez wrote:
> Thanks for the info Élie.
>
> I'm aware of iTorch, but if I'm not wrong, it has been developed for
> luajit and Torch is embeded into its core. For instance, you cannot do
> plots using other objects than tensors, so IMHO it is not as general to
> play well with plain Lua. Howecer I have to say that iTorch has a very
> well grounded code base, and a very large community.
>
> My intention is to integrate into this kernel the ability and
> flecibility to be ectended easily by other Lua modules, as scilua and
> similars which also implement kind of matrix/tensor libraries.
> --
> PakoZM :-)
> cafre.dsic.upv.es <http://cafre.dsic.upv.es>:8080/~pako/index.html
>
> El 16 oct 2015, a las 1:00, Elie Michel <elie.michel@exppad.com
> <mailto:elie.michel@exppad.com>> escribió:
>
>> Hi all,
>>
>> Facebook researchers released an IPython kernal (well, we should say
>> Jupyter now) for Torch:
>>
>> https://github.com/facebook/iTorch
>>
>> Torch is a lua library so as a by-product, this is a lua kernel. You
>> might be interested in having a look at it!
>>
>>
>> Élie
>>
>> On 10/15/2015 06:55 PM, Paco Zamora Martínez wrote:
>>> I would like to announce some achivements with this IPython kernel:
>>>
>>> https://github.com/pakozm/iPyLua
>>>
>>> I have added basic code for plotting allowing to be extended using new
>>> metamethods (ipylua_show and ipylua_toseries functions declared at the
>>> metatable). You can see this kernel interacting with APRIL-ANN (a
>>> toolkit for pattern recognition we are developing for 10 years now):
>>>
>>> http://nbviewer.ipython.org/github/pakozm/IPyLua/blob/7c181c41651f982e98423c9e683944ba2e11869d/demo.ipynb
>>>
>>> The documentation stills inexistent, I would like to document the
>>> current state of this kernel as soon as possibl. As always, any help
>>> will be wellcome :-) The code can be improved in several ways (sure)
>>>
>>> Thanks
>>>
>>> Pako ZM :)
>>> http://pakozm.hol.es
>>>
>>> On 5 October 2015 at 10:18, Paco Zamora Martínez <pakozm@gmail.com
>>> <mailto:pakozm@gmail.com>
>>> <mailto:pakozm@gmail.com>> wrote:
>>>
>>> I have a mistake in my last email, IPyLua is being implemented for
>>> IPython v2.
>>>
>>> So what exactly does this do? Let Lua use Python modules?
>>>
>>>
>>> Before Jupyter, IPython was the named of a frontend for notebooks
>>> allowing mixing literate sentences with code and fancy outputs (as
>>> images, plots, etc.) This frontend was agnostic of the programming
>>> language, however its first kernel was for Python and they called it
>>> IPython. Currently, the frontend has been moved to project Jupyter.
>>> What I'm doing is continuing the work of Neomantra
>>> (lua_ipython_kernel) in order to make it work with IPython v2.
>>>
>>> So, this project is a Lua kernel for IPython (not Jupyter), besides
>>> a simple API which allow Lua modules to extend the fancy
>>> functionality of this kernel, allowing plots, images, etc.
>>> Nevertheless, the project stills in development. You can see an
>>> example of the notebooks here:
>>>
>>> https://github.com/pakozm/IPyLua/blob/master/demo.ipynb
>>>
>>> --
>>> Sent from my Game Boy.
>>>
>>>
>>>
>>