MILA and the future of Theano

52902 views
Skip to first unread message

Pascal Lamblin

unread,
Sep 28, 2017, 12:23:05 PM9/28/17
to theano-announce, theano-dev, theano-users
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

unread,
Sep 28, 2017, 4:41:21 PM9/28/17
to theano-users
Thanks for all your hard work over the years!

Soumith Chintala

unread,
Sep 28, 2017, 6:56:39 PM9/28/17
to theano-users
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


On Thursday, September 28, 2017 at 12:23:05 PM UTC-4, Pascal Lamblin wrote:

Philippe Vandermersch

unread,
Sep 28, 2017, 8:40:34 PM9/28/17
to theano-users

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

Best,
Phil.


Emad Barsoum

unread,
Sep 28, 2017, 9:02:20 PM9/28/17
to theano-users
Dear Theano team,

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

Thanks,
Emad 

Rahul S

unread,
Sep 28, 2017, 11:32:30 PM9/28/17
to theano-users
Thank you all so much for Theano! 
We'll always remember Theano as the first major library for DL development. 
All the best everyone!


On Thursday, September 28, 2017 at 12:23:05 PM UTC-4, Pascal Lamblin wrote:

Ragav Venkatesan

unread,
Sep 28, 2017, 11:46:31 PM9/28/17
to theano-users
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!

digvi...@gmail.com

unread,
Sep 29, 2017, 12:02:06 AM9/29/17
to theano-users
Thanks for all your contributions.


On Thursday, September 28, 2017 at 9:23:05 AM UTC-7, Pascal Lamblin wrote:

Anmol Jawandha

unread,
Sep 29, 2017, 12:28:17 AM9/29/17
to theano-users
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


 On Thursday, September 28, 2017 at 12:23:05 PM UTC-4, Pascal Lamblin wrote:

Lucas Beyer

unread,
Sep 29, 2017, 5:13:12 AM9/29/17
to theano-users
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

unread,
Sep 29, 2017, 11:53:26 AM9/29/17
to theano-users
I started my journey of deep learning with Theano. Thanks a lot for all the contribution!!!!

石墨眸

unread,
Sep 29, 2017, 12:44:23 PM9/29/17
to theano-users
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 !

在 2017年9月29日星期五 UTC+8上午12:23:05,Pascal Lamblin写道:

Joseph Martz

unread,
Sep 29, 2017, 4:32:16 PM9/29/17
to theano-users
thank you for everything <3

Fariz Rahman

unread,
Sep 29, 2017, 4:33:18 PM9/29/17
to theano-users
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 wrote:

Thomas Wiecki

unread,
Sep 29, 2017, 5:02:56 PM9/29/17
to theano...@googlegroups.com
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?

--

---
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 email to theano-users+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Paul Wohlhart

unread,
Sep 29, 2017, 7:21:44 PM9/29/17
to theano-users
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!

Valentin

unread,
Sep 30, 2017, 4:29:45 PM9/30/17
to 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

unread,
Sep 30, 2017, 9:11:55 PM9/30/17
to theano-users
Thanks very much for your hard work and dedication to the field. I'm eternally grateful :)

Karthik


On Thursday, September 28, 2017 at 12:23:05 PM UTC-4, Pascal Lamblin wrote:

Paul Rigor

unread,
Oct 1, 2017, 2:56:15 PM10/1/17
to theano-announce, thean...@googlegroups.com, theano-users
Thank you & the theano team and community for all of the great work!

Paul

--

---
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 email to theano-dev+...@googlegroups.com.

Max

unread,
Oct 2, 2017, 10:00:31 AM10/2/17
to theano-users
Theano is where I started my journey in DL. Thank you for all the hard work!


On Thursday, September 28, 2017 at 12:23:05 PM UTC-4, Pascal Lamblin wrote:

Gabriel Perdue

unread,
Oct 2, 2017, 11:05:02 AM10/2/17
to theano-users
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

goo...@jan-schlueter.de

unread,
Oct 2, 2017, 1:23:44 PM10/2/17
to theano-users
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!

Anatoly Belikov

unread,
Oct 2, 2017, 3:45:15 PM10/2/17
to theano-users
Well, I definitely would like to continue using Theano and be able to contribute and merge code after MILA lab abandons it's repository.

суббота, 30 сентября 2017 г., 0:02:56 UTC+3 пользователь Thomas написал:
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users...@googlegroups.com.

Michaeel Kazi

unread,
Oct 4, 2017, 12:42:14 PM10/4/17
to theano-users
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

unread,
Oct 7, 2017, 5:33:04 AM10/7/17
to theano-users
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 wrote:

Bhagesh Sr

unread,
Oct 17, 2017, 10:05:35 AM10/17/17
to theano-users
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 wrote:

Alex V

unread,
Oct 24, 2017, 1:29:27 PM10/24/17
to theano-users
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.

Juan Camilo Gamboa Higuera

unread,
Oct 25, 2017, 1:58:47 PM10/25/17
to theano-users

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!

Frédéric Bastien

unread,
Oct 25, 2017, 3:15:26 PM10/25/17
to theano...@googlegroups.com
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.



--

Ganesh Iyer

unread,
Oct 29, 2017, 12:56:09 PM10/29/17
to theano-users
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 :)

Carlos Guerrero-Mosquera

unread,
Nov 20, 2017, 3:49:36 AM11/20/17
to theano-users
Thank you for all guys! Your contribution to the deep learning community has been remarkable!!

sunnysaxen...@gmail.com

unread,
Nov 29, 2017, 2:03:24 AM11/29/17
to theano-users
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

unread,
Dec 31, 2017, 1:54:13 AM12/31/17
to theano-users
That fking sucks.

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

Rabah Nory

unread,
May 22, 2018, 4:44:45 PM5/22/18
to 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

unread,
Jun 1, 2018, 2:28:59 AM6/1/18
to theano...@googlegroups.com
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

--

---
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 email to theano-users+unsubscribe@googlegroups.com.

Buruk Aregawi

unread,
Jun 4, 2018, 2:09:40 PM6/4/18
to theano-users
R.I.P Theano. Thank you for all your work, Theano team!
Reply all
Reply to author
Forward
0 new messages