Hi,
On Thu, Jul 12, 2018 at 04:20:05PM +0200, Georges Khaznadar wrote:
> 4- if no commercial machine exists at a "fair price", to train the
>neural network in a reasonable amount of time, the publisher of Y
>must lend a machine to fulfill the GPL requirements.
I don't think so.
On Thu, Jul 12, 2018 at 8:35 PM, Lumin wrote:
> I just noticed that one of us tries to package deep-learning based
> application[1], specifically it is AlphaGo-Zero[2] based. However, this
> raised my concern about software freedom. Since mankind relys on artificial
> intelligence more and more,
Ian Jackson writes:
> Things in Debian main shoudl be buildable *from source* using Debian
> main. In the case of a pretrained neural network, the source code is
> the training data.
>
> In fact, they are probably not redistributable unless all the training
> data is supplied, since the GPL's
BTW, please keep me on CC, I am not subscribed to debian-devel.
Ximin Luo:
> Lumin:
>> [..]
>>
>> My core concern is:
>>
>> Even if upstream releases their pretrained model under GPL license,
>> the freedom to modify, research, reproduce the neural networks,
>> especially "very deep" neural
Lumin:
> [..]
>
> My core concern is:
>
> Even if upstream releases their pretrained model under GPL license,
> the freedom to modify, research, reproduce the neural networks,
> especially "very deep" neural networks is de facto controled by
> PROPRIETARIES.
>
> [..]
I think in general
On Fri, Jul 13, 2018 at 06:44:39PM +0200, Jonas Smedegaard wrote:
> I therefore believe there is no license violation, as long as the code
> is _possible_ to compile without non-free code (e.g. blobs to activate
> GPUs) - even if ridiculously expensive in either time or hardware.
>
> We have a
Quoting Lumin (2018-07-13 18:13:26)
> > Seems you elaborated only that it is ridiculously slow so use CPUs
> > instead of [non-free blob'ed] GPUs - not that it is *impossible to
> > use CPUs.
> >
> > If I am mistaken and you addressed the _possibility_ (not
> > popularity) of
Hi Jonas,
> Seems you elaborated only that it is ridiculously slow so use CPUs
> instead of [non-free blob'ed] GPUs - not that it is *impossible to use
> CPUs.
>
> If I am mistaken and you addressed the _possibility_ (not popularity) of
> reproducing/modifying/researching with CPUs, then I
Hi Russell,
> On Thu, 2018-07-12 at 18:15 +0100, Ian Jackson wrote:
> > Compare neural networks: a user who uses a pre-trained neural network
> > is subordinated to the people who prepared its training data and set
> > up the training runs.
>
> In Alpha-Zero's case (it is Alpha-Zero the original
Quoting Lumin (2018-07-13 16:34:44)
> > Perhaps I am missing something, but if _possible_ just 100x slower
> > to use CPUs instead of GPUs, then I fail to recognize how it cannot
> > be reproduced, modified, and researched 100x slower.
[...]
>CPU takes ridiculously long time, compared to
Hi Ian,
> Lumin writes ("Concerns to software freedom when packaging deep-learning
> based appications."):
> > 1. Is GPL-licended pretrained neural network REALLY FREE? Is it really
> > DFSG-compatible?
>
> No. No.
>
> Things in Debian main shoudl be buildable *from source* using Debian
Hi Jonas
> Perhaps I am missing something, but if _possible_ just 100x slower to
> use CPUs instead of GPUs, then I fail to recognize how it cannot be
> reproduced, modified, and researched 100x slower.
>
> Quite interesting question you raise!
I can provide at least two data points:
1. The
Russell Stuart writes:
> Apart from the "non-human" intelligence bit none of this is different to
> what we _already_ accept into Debian. It's very unlikely I could have
> sensible contributions to the game engines of the best chess, backgammon
> or Go programs Debian has now.
For what it's
On Thu, 2018-07-12 at 18:15 +0100, Ian Jackson wrote:
> Compare neural networks: a user who uses a pre-trained neural network
> is subordinated to the people who prepared its training data and set
> up the training runs.
In Alpha-Zero's case (it is Alpha-Zero the original post was about)
there is
Ian Jackson writes:
> Taking a step back: the point of this exercise is to preserve user
> freedom. That is, a user should be able to make their computer serve
> their interests, and should not be subordinated to upstreams (nor to
> Debian or to one of our derivatives0.
> A user who uses
Russ Allbery writes ("Re: Concerns to software freedom when packaging
deep-learning based appications."):
> I'm not sure I disagree, but I think it's worth poking at this a bit and
> seeing if it holds up in extension by analogy to other precomputed data.
Perhaps.
> Suppose
Ian Jackson writes:
> Lumin writes:
>> 1. Is GPL-licended pretrained neural network REALLY FREE? Is it really
>> DFSG-compatible?
> No. No.
> Things in Debian main shoudl be buildable *from source* using Debian
> main. In the case of a pretrained neural network, the source code is
> the
On Thu, Jul 12, 2018 at 8:35 PM, Lumin wrote:
> I just noticed that one of us tries to package deep-learning based
> application[1], specifically it is AlphaGo-Zero[2] based. However, this
> raised my concern about software freedom. Since mankind relys on artificial
> intelligence more and more,
Georges Khaznadar writes ("Re: Concerns to software freedom when packaging
deep-learning based appications."):
> The question is, what are sources of Y? I presume that any sane judge
> would agree that in that case, sources are the set of materials and
> immaterials which a
Lumin writes ("Concerns to software freedom when packaging deep-learning based
appications."):
> 1. Is GPL-licended pretrained neural network REALLY FREE? Is it really
> DFSG-compatible?
No. No.
Things in Debian main shoudl be buildable *from source* using Debian
main. In the case of a
On Thu, Jul 12, 2018 at 02:16:01PM +0100, Ian Campbell wrote:
> > so what? Debian runs on non-free CPUs too, how is this any different?
> AIUI there is a pile of non-free libraries/tools on the host side too,
thanks for this clarification!
--
cheers,
Holger
signature.asc
On Thu, 2018-07-12 at 19:03 +0500, Andrey Rahmatullin wrote:
> On Thu, Jul 12, 2018 at 02:16:01PM +0100, Ian Campbell wrote:
> > > > (3) CUDA Deep Neural Network library (cuDNN)[4] is NVIDIA's
> > > > **PROPRIETARY**,
> > > > stacked on CUDA, and requires NVIDIA GPU exclusively.
> > >
> > >
Hello,
Lumin a écrit :
> [...]
> Even if upstream releases their pretrained model under GPL license,
> the freedom to modify, research, reproduce the neural networks,
> especially "very deep" neural networks is de facto controled by
> PROPRIETARIES.
In my opinion, a company releasing a
On Thu, Jul 12, 2018 at 02:16:01PM +0100, Ian Campbell wrote:
> > > (3) CUDA Deep Neural Network library (cuDNN)[4] is NVIDIA's
> > > **PROPRIETARY**,
> > > stacked on CUDA, and requires NVIDIA GPU exclusively.
> >
> > so what? Debian runs on non-free CPUs too, how is this any different?
>
On Thu, 2018-07-12 at 13:09 +, Holger Levsen wrote:
> On Thu, Jul 12, 2018 at 12:35:24PM +, Lumin wrote:
> > (3) CUDA Deep Neural Network library (cuDNN)[4] is NVIDIA's
> > **PROPRIETARY**,
> > stacked on CUDA, and requires NVIDIA GPU exclusively.
>
> so what? Debian runs on
On Thu, Jul 12, 2018 at 12:35:24PM +, Lumin wrote:
> (3) CUDA Deep Neural Network library (cuDNN)[4] is NVIDIA's **PROPRIETARY**,
> stacked on CUDA, and requires NVIDIA GPU exclusively.
so what? Debian runs on non-free CPUs too, how is this any different?
(and yes, I'd be happy to
Hi Lumin,
Quoting Lumin (2018-07-12 14:35:24)
> My core concern is:
>
> Even if upstream releases their pretrained model under GPL license,
> the freedom to modify, research, reproduce the neural networks,
> especially "very deep" neural networks is de facto controled by
>
27 matches
Mail list logo
