[theano-users] MILA and the future of Theano

[Pascal Lamblin]

Dear users and developers,

After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.

The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.

We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.

In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.

MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.

Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.

-- Yoshua Bengio,
Head of MILA

[Neil Girdhar]

Thanks for all your hard work over the years!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Soumith Chintala]

Dear Pascal, Fred, Arnaud, Ian and the rest of the team,

Your contributions to the field of deep learning have been revolutionary
and deeply inspiring.

Thank you for powering a whole generation of deep learning research.
Thank you for new ideas and being bold.

Most importantly, thank you for always pushing us on the Torch side.
Best,
a torch maintainer

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Philippe Vandermersch]

Thanks for your contribution to the Deep Learning community !
Your feedback has greatly contributed to the development of cuDNN.

Best,
Phil.

[Emad Barsoum]

Dear Theano team,

Thank you for your contribution, pretty much all modern frameworks are
based on Theano ideas and flexibility.

Thanks,
Emad

[Rahul S]

Thank you all so much for Theano!
We'll always remember Theano as the first major library for DL development.
All the best everyone!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Ragav Venkatesan]

Thank you so much for all the work! Particular thanks to Pascal and Fred in
this group, without you guys I would found my PhD much more difficult.
Theano will be missed!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[d***@gmail.com]

Thanks for all your contributions.

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Anmol Jawandha]

Thank you very much for all your work! It may not be obvious but you helped
expose so many people to deep learning - your contribution is invaluable!

- someone who learned a lot from implementing things in Theano

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Lucas Beyer]

I cannot thank you enough for all your hard work. IMO, Theano is the single
best thing that has happened to the field machine learning in the last
decade.
Also, good call stopping now.

[Vishrut Talekar]

I started my journey of deep learning with Theano. Thanks a lot for all the
contribution!!!!

On Thursday, September 28, 2017 at 9:53:05 PM UTC+5:30, Pascal Lamblin

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[石墨眸]

Thanks for all your work , which unifies all the researchers even from us
china. Thanks for this journey.
In past decade, theano is for everyone !

圚 2017幎9月29日星期五 UTC+8䞊午12:23:05Pascal Lamblin写道

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Joseph Martz]

thank you for everything <3

[Fariz Rahman]

Thanks for years of hard work on behalf of the Keras community.


On Thursday, September 28, 2017 at 9:53:05 PM UTC+5:30, Pascal Lamblin

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Thomas Wiecki]

On behalf of the PyMC3 community -- thanks for all the hard work. A shame,
Theano is still leading in many areas compared to other packages like
TensorFlow (e.g. graph optimizations).

Is no one here interested in carrying on maintenance and development of
this amazing package?

Post by Fariz Rahman
Thanks for years of hard work on behalf of the Keras community.
On Thursday, September 28, 2017 at 9:53:05 PM UTC+5:30, Pascal Lamblin

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.

[Anatoly Belikov]

Well, I definitely would like to continue using Theano and be able to
contribute and merge code after MILA lab abandons it's repository.

Post by Thomas Wiecki
On behalf of the PyMC3 community -- thanks for all the hard work. A shame,
Theano is still leading in many areas compared to other packages like
TensorFlow (e.g. graph optimizations).
Is no one here interested in carrying on maintenance and development of
this amazing package?

Post by Fariz Rahman
Thanks for years of hard work on behalf of the Keras community.
On Thursday, September 28, 2017 at 9:53:05 PM UTC+5:30, Pascal Lamblin

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.

[Paul Wohlhart]

Dear theano team,

Thanks for all the awesome work!
It certainly jump started my venture into modern day deep learning, having
complex models training on GPU within days
without having derived a single gradient or written a line of CUDA code
myself.
And special thanks to everybody who contributed to the super responsive
support!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

['Valentin' via theano-users]

Thank you to all the team maintaining theano and to all the contributeurs.
As a student, theano was my first deep learning framework and I can't thank
you enough for making it available to me when I was discovering neural nets
and machine learning in general.

All the best,

Valentin

[Karthik Narayan]

Thanks very much for your hard work and dedication to the field. I'm
eternally grateful :)

Karthik

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Paul Rigor]

Thank you & the theano team and community for all of the great work!

Paul

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA
--
---
You received this message because you are subscribed to the Google Groups
"theano-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.

[Max]

Theano is where I started my journey in DL. Thank you for all the hard work!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Gabriel Perdue]

I'd like to thank the Theano team for their service to the community.
Theano was and is a wonderful project - I always describe it to new people
as "magic programming from the future." I really hope the community is able
to rally and find a new home for development and support, but either way,
thank you for your work!

pax
Gabe Perdue

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[g***@jan-schlueter.de]

Thank you for all your work! I still remember how disappointed I was when I
switched from cudamat to Theano and found that Theano was not even a little
bit slower than my own carefully-crafted code :) Thank you for your
contribution to deep learning, and for pushing the accompanying software
ecosystem. Thank you for continuing to provide a university-driven deep
learning framework while more and more corporate-driven frameworks
appeared. It's sad to see Theano go, but it's not a scientific endeavour
anymore to maintain a deep learning framework, and we will be able to
pursue independent research with other frameworks now. Well done!

[Michaeel Kazi]

Echoing the thanks of many here. I've long considered Theano the best math expression package, since it was very intuitive to implement many techniques beyond the standard deep learning structures.

It looks like Tensorflow has caught up, though I admit it was strange to see it declared 'the winner' over a year ago, before it had even supported a non-bucketed approach to recurrence...

[Sameer Mahabole]

Thank you for your hard work and commitment for Theano for almost a decade.

I hope it will be picked up by others, I would love to contribute but I'm
very new to it.

On Thursday, September 28, 2017 at 9:53:05 PM UTC+5:30, Pascal Lamblin

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Bhagesh Sr]

Thank you for making our ML-DL life easy for so many years. I started my DL
journey with Theano and loved every bit of it.

On Thursday, September 28, 2017 at 9:53:05 PM UTC+5:30, Pascal Lamblin

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Alex V]

I also started with theano before other symbolic computation libraries
suitable for DL existed... Thanks so much for all the work your group put
in.

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Juan Camilo Gamboa Higuera]

I understand that MILA has found that it doesn't need to continue
supporting Theano. However, I believe that the statement "MILA will
discontinue support of Theano" is not equivalent to "Theano is dead". I
feel that pytorch and tensorflow, even though they're backed by large
companies, they're still lagging behind Theano as general machine learning
framewroks (i.e. not just deep learning). Personally, I've benefited from
all the effort that people have put in theano, and will continue to use it
as long as it is the best tool that suits my needs (bayesian methods for
model based reinforcement learning). For this, I am very grateful towards
the Theano development team.

THanks!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Frédéric Bastien]

Just a historical note.

Before Theano, we had another framework. It took 1 year to prototype
Theano, then it took 1 year to have most people use Theano internally, but
the previous framework was still used for 10 year. So I won't be surprised
it Theano continue to be used for a long time.



On Wed, Oct 25, 2017 at 1:58 PM Juan Camilo Gamboa Higuera <

Post by Juan Camilo Gamboa Higuera
I understand that MILA has found that it doesn't need to continue
supporting Theano. However, I believe that the statement "MILA will
discontinue support of Theano" is not equivalent to "Theano is dead". I
feel that pytorch and tensorflow, even though they're backed by large
companies, they're still lagging behind Theano as general machine learning
framewroks (i.e. not just deep learning). Personally, I've benefited from
all the effort that people have put in theano, and will continue to use it
as long as it is the best tool that suits my needs (bayesian methods for
model based reinforcement learning). For this, I am very grateful towards
the Theano development team.
THanks!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.

[Ganesh Iyer]

Thank you so much for the years of development of support. i ventured into
deep learning with theano and I continue to use the transparent and
simplistic style for all my models, in any library and framework of my
choice.

All the best for future ventures :)

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[Carlos Guerrero-Mosquera]

Thank you for all guys! Your contribution to the deep learning community
has been remarkable!!

Post by Pascal Lamblin
Dear users and developers,
After almost ten years of development, we have the regret to announce
that we will put an end to our Theano development after the 1.0 release,
which is due in the next few weeks. We will continue minimal maintenance
to keep it working for one year, but we will stop actively implementing
new features. Theano will continue to be available afterwards, as per
our engagement towards open source software, but MILA does not commit to
spend time on maintenance or support after that time frame.
The software ecosystem supporting deep learning research has been
evolving quickly, and has now reached a healthy state: open-source
software is the norm; a variety of frameworks are available, satisfying
needs spanning from exploring novel ideas to deploying them into
production; and strong industrial players are backing different software
stacks in a stimulating competition.
We are proud that most of the innovations Theano introduced across the
years have now been adopted and perfected by other frameworks. Being
able to express models as mathematical expressions, rewriting
computation graphs for better performance and memory usage, transparent
execution on GPU, higher-order automatic differentiation, for instance,
have all become mainstream ideas.
In that context, we came to the conclusion that supporting Theano is no
longer the best way we can enable the emergence and application of novel
research ideas. Even with the increasing support of external
contributions from industry and academia, maintaining an older code base
and keeping up with competitors has come in the way of innovation.
MILA is still committed to supporting researchers and enabling the
implementation and exploration of innovative (and sometimes wild)
research ideas, and we will keep working towards this goal through other
means, and making significant open source contributions to other projects.
Thanks to all of you who for helping develop Theano, and making it
better by contributing bug reports, profiles, use cases, documentation,
and support.
-- Yoshua Bengio,
Head of MILA

[s***@gmail.com]

Dear Theano Team,

I would like to say thank you very much to your entire team for developing
a awesome deep learning library.
I hoping that you guys will again get stared the halted campaign.




Regards,
Sunny Saxena

[Pat York]

That fking sucks.

It's been a hell of a ride though - thank you.

['Rabah Nory' via theano-users]

Dear MILA

Thank you too much for the wonderful Theano DL Framework
in my opinion it is the best DL I worked in my research .
just I wonder why you killed the lovely Theano
I am sure that all the researcher around the world are very sad to see the
end of charming Theano .
Frankly speaking I hate TensorFlow and wish to see Theano Alive again
with my best regards

[Thomas Wiecki]

Dear all,

The PyMC developers will take over maintenance (primarily in service of
PyMC3) of Theano. See here for the full announcement:
https://medium.com/@pymc_devs/theano-tensorflow-and-the-future-of-pymc-6c9987bb19d5

If anyone here wants to help, please reach out! Evaluating the various
alternatives to Theano we found that it still holds its own and is more
mature and feature complete than any single alternative.

Best,
Thomas

On Tue, May 22, 2018 at 10:44 PM, 'Rabah Nory' via theano-users <

Dear MILA
Thank you too much for the wonderful Theano DL Framework
in my opinion it is the best DL I worked in my research .
just I wonder why you killed the lovely Theano
I am sure that all the researcher around the world are very sad to see the
end of charming Theano .
Frankly speaking I hate TensorFlow and wish to see Theano Alive again
with my best regards
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.