Profile picture
, 3 tweets, 1 min read
New release of multi-backend Keras: 2.3.0

github.com/keras-team/ker…

- First release of multi-backend Keras with full TF 2 support
- Continued support for Theano/CNTK
- Will be the last major release of multi-backend Keras

We recommend you switch your Keras code to tf.keras.
Both Theano and CNTK are out of development. Meanwhile, as Keras backends they represent less than 4% of Keras usage. The other 96% of users (of which more than half are already on tf.keras) are better served with tf.keras.

Keras development will focus on tf.keras going forward.
Importantly, we will seek to start developing tf.keras in its own standalone GitHub repository at keras-team/keras in order to make it much easier for 3rd party folks to contribute.

Keras has never been moving faster than now :)
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to François Chollet
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

More from @fchollet see all

Profile picture
François Chollet
@fchollet
I had always assumed that the similarity between the Japanese word for "typhoon" (台風, pronounced "taifū") and the English and French versions (in French it's "typhon", pronounced "teefon") was accidental.

But is it? This word turns out to have a strange, thousand-year history
French "typhon" comes from Latin "typhon" (strong wind), itself from ancient greek Τυφῶν (tuphon). It enters the French vocabulary in the 16th century via the Portuguese word "tufão", brought back to Europe by Portuguese sailors who got it from South Asian or East Asia
It turns out that the greek word had previously made its way into Persian (tūfān) and from there, to Arabic and Hindi. It may have travelled as far east as Malaysia. And that's the word the Portuguese brought back.

Like genealogy, etymology isn't a tree -- it's a graph.
Read 5 tweets
Profile picture
François Chollet
@fchollet
I think the notion that there's a "deep learning framework war raging" is pretty mistaken. It's not like there are different clans that don't talk to each other. The majority of people use multiple tools, often within the same project.
And what they use now may not be what they use in 6 months. This is especially true in research, where each project starts from scratch and lasts 2-3 months. For production applications, there is more of a need to settle on one framework and accrete capabilities over time.
So don't believe the FUD. It's typically just a marketing ploy to drive a message. Conflict sells I suppose.
Read 3 tweets
Profile picture
François Chollet
@fchollet
I think if you do have freedom of speech, you should use it, and sometimes you should use it on behalf of those who don't have the same privilege. Kowtowing to autocrats when your safety is not at stake is pretty cowardly
The thing with authoritarians is this: they don't respect any norms or rights. They only respect power. Everything they do is a trial balloon: if it works, they take it as a sign that they can keep pushing further. They won't take a step back until they encounter real resistance.
You can't just hope that authoritarians will spontaneously become more reasonable if you start appeasing them. It's the exact opposite. They will see you as weak and they will keep repeating the same playbook, on a larger scale.
Read 3 tweets

Related threads

Profile picture
Freddy Snijder
@Visionscaper
(1/7) My experience today to migrate my code from #Keras to Keras for @TensorFlow (tf.keras) has been a bit painful unfortunately. Some issue I encountered (@fchollet maybe this is of interest to you?):
(2/7) I found out that 1.13 doesn't have support for float16 in linear algebra operators, luckily this pull request now landed in 1.14-RC0 github.com/tensorflow/ten…
(3/7) It seems that Dropout on an input with a partially unknown shape, and a noise_shape of (None, 1, None), does not work in tf.keras. I could work around it but in the original Keras this works just fine.
Read 7 tweets
Profile picture
CZEdwards
@CZEdwards
Last sewing tip for the night.

YES, YOU HAVE TO TRACE YOUR PATTERNS.

Unless you bought them at a Joann loss leader sale, where you got them for $1 or $2 each. And you bought at least 2 copies.

WHY?

Because you alter the paper pattern.
It’s a lot cheaper to buy a roll of medical exam paper, trace & grade your sizes per the back of the envelope & finished garment measurements, cut THAT shit out, pin it together & find out it doesn’t fit than doing it with fabric you can’t replace.
I’m saving you time & tears.
Your fitting book: Pick one:

Gertie’s New Book for Better Sewing:
goodreads.com/book/show/1359…

Colette Sewing Handbook goodreads.com/book/show/1151…

Palmer Pletsch Complete Guide to Fitting [those who have boobies & hips, commonly called women by the industry] amazon.com/Palmer-Pletsch…
Read 6 tweets
Profile picture
Richard James
@RichJamesBits
I have reached a whole new level of #PISSEDOFF.

I didnt think such a #dark and #scary state, or level, could be reached. It may top 12-14-17's causation. It's. That. Bad. So, I've prepared a statement...

docs.google.com/document/d/1_J…

#PascoSheriff #HaveYouAllNoShame @CigarCityBeer
Dear @FBI,
I must implore you ask your colleauges in @FBITampa to help research this new level of #PISSEDOFFNESS. As locals, they may offer a special indepth knowledge of this new level I have reached. This is not fair & law enforcement must #investigate. No one deserves this.
This is my #apogee of #Pissedoffness.
If #motherfuckers & #tears werent a noticable level.
If #HUSH or #INTIMIDATION tactics being deployed didnt strike a chord.
If #EVIDENCE, PROOF, or 30k Bonds didnt #WOW YOU.
If posting mom with no shred of investigatable value wasnt enough...
Read 463 tweets
Profile picture
Frankly Speaking
@occupiedbypak
Though #Pakistan was created on religious line, #Jinnah was most liberal Muslim who never wanted #Pakistan a religious firm!
Till 1977 , Religion was just a personal choice in #Pakistan , General #Zia not only pushed religion but also took country decades backward !
#SZAB
Zia knew he had zero support from People hence he decided to take support from Mullahs and he ensured that hardcore Islam penetrate each institution in #Pakistan including #Army & #ISI !
#Soviet invaded #Afghanistan in 1979, #Zia used this opportunity to the fullest .
#US wanted to work on Proxy war, Zia got excellent platform to build these Proxies in Madrassa's of Pakistan.
Read 279 tweets
Profile picture
Gaurav Kaushik
@gaurav_bio
The January issue of @NatureMedicine is all about data and machine learning in biomedicine! Here's a quick roundup:
.@fdna and collaborators published DeepGestalt, which predicts genetic syndromes from images of faces. No code shared, but they used a deep convolutional neural network with #keras and @TensorFlow backend and discussed params in detail!

Paper: nature.com/articles/s4159…
.@EricTopol reviewed applications of artificial intelligence in medicine, including some basics of deep neural nets, their history, existing applications + performance, and potential for the future.

Paper: nature.com/articles/s4159…
Read 10 tweets
Profile picture
👩‍💻 DynamicWebPaige @ #OpenAIFive 🤖✨
@DynamicWebPaige
✨🧠 The ecosystem that has grown up around @TensorFlow in the last few years blows my mind. There's just so much functionality, compared to some of the other, newer frameworks.

👉Consider this an ever-expanding thread for me to take notes + wrap my brain around products. Ready?
1) @TensorFlow Extended (TFX)

It's no secret that I 💕 #TFX and all of its tooling for deploying machine learning models into production. If you care about keeping your models up-to-date and monitoring them, you should check out the product + its paper.

tensorflow.org/tfx/?hl=zh-cn
2) @TensorFlow Hub

If you want to train your model on a small data set, or improve generalization, you'll need to use something called transfer learning. #TFHub modules make it easy—and are available in an #OSS marketplace: tfhub.dev.

site: tensorflow.org/hub/
Read 40 tweets

Trending hashtags

#Hafizsaeed #MLKDay #AI #HuJI #Hindus #FFS #Python #FRAUD #Alleged #RWPT9 #Jaish #Sheraton #China #LIVE #ALF #Creatability #Masood #Part1 #LBJ #Geelani #dunnit #brexit #FBI #MAJORLEAD #docker
Follow Us on Twitter!

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!