Skip to content
By using Twitter’s services you agree to our Cookies Use. We and our partners operate globally and use cookies, including for analytics, personalisation, and ads.
  • Home Home Home, current page.
  • About

Saved searches

  • Remove
  • In this conversation
    Verified accountProtected Tweets @
Suggested users
  • Verified accountProtected Tweets @
  • Verified accountProtected Tweets @
  • Language: English
    • Bahasa Indonesia
    • Bahasa Melayu
    • Català
    • Čeština
    • Dansk
    • Deutsch
    • English UK
    • Español
    • Filipino
    • Français
    • Hrvatski
    • Italiano
    • Magyar
    • Nederlands
    • Norsk
    • Polski
    • Português
    • Română
    • Slovenčina
    • Suomi
    • Svenska
    • Tiếng Việt
    • Türkçe
    • Ελληνικά
    • Български език
    • Русский
    • Српски
    • Українська мова
    • עִבְרִית
    • العربية
    • فارسی
    • मराठी
    • हिन्दी
    • বাংলা
    • ગુજરાતી
    • தமிழ்
    • ಕನ್ನಡ
    • ภาษาไทย
    • 한국어
    • 日本語
    • 简体中文
    • 繁體中文
  • Have an account? Log in
    Have an account?
    · Forgot password?

    New to Twitter?
    Sign up
fchollet's profile
François Chollet
François Chollet
François Chollet
Verified account
@fchollet

Tweets

François CholletVerified account

@fchollet

Deep learning @google. Creator of Keras. Author of 'Deep Learning with Python'. Opinions are my own.

United States
fchollet.com
Joined August 2009

Tweets

  • © 2021 Twitter
  • About
  • Help Center
  • Terms
  • Privacy policy
  • Cookies
  • Ads info
Dismiss
Previous
Next

Go to a person's profile

Saved searches

  • Remove
  • In this conversation
    Verified accountProtected Tweets @
Suggested users
  • Verified accountProtected Tweets @
  • Verified accountProtected Tweets @

Promote this Tweet

Block

  • Tweet with a location

    You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more

    Your lists

    Create a new list


    Under 100 characters, optional

    Privacy

    Copy link to Tweet

    Embed this Tweet

    Embed this Video

    Add this Tweet to your website by copying the code below. Learn more

    Add this video to your website by copying the code below. Learn more

    Hmm, there was a problem reaching the server.

    By embedding Twitter content in your website or app, you are agreeing to the Twitter Developer Agreement and Developer Policy.

    Preview

    Why you're seeing this ad

    Log in to Twitter

    · Forgot password?
    Don't have an account? Sign up »

    Sign up for Twitter

    Not on Twitter? Sign up, tune into the things you care about, and get updates as they happen.

    Sign up
    Have an account? Log in »

    Two-way (sending and receiving) short codes:

    Country Code For customers of
    United States 40404 (any)
    Canada 21212 (any)
    United Kingdom 86444 Vodafone, Orange, 3, O2
    Brazil 40404 Nextel, TIM
    Haiti 40404 Digicel, Voila
    Ireland 51210 Vodafone, O2
    India 53000 Bharti Airtel, Videocon, Reliance
    Indonesia 89887 AXIS, 3, Telkomsel, Indosat, XL Axiata
    Italy 4880804 Wind
    3424486444 Vodafone
    » See SMS short codes for other countries

    Confirmation

     

    Welcome home!

    This timeline is where you’ll spend most of your time, getting instant updates about what matters to you.

    Tweets not working for you?

    Hover over the profile pic and click the Following button to unfollow any account.

    Say a lot with a little

    When you see a Tweet you love, tap the heart — it lets the person who wrote it know you shared the love.

    Spread the word

    The fastest way to share someone else’s Tweet with your followers is with a Retweet. Tap the icon to send it instantly.

    Join the conversation

    Add your thoughts about any Tweet with a Reply. Find a topic you’re passionate about, and jump right in.

    Learn the latest

    Get instant insight into what people are talking about now.

    Get more of what you love

    Follow more accounts to get instant updates about topics you care about.

    Find what's happening

    See the latest conversations about any topic instantly.

    Never miss a Moment

    Catch up instantly on the best stories happening as they unfold.

    1. François Chollet‏Verified account @fchollet 15 Mar 2019

      "Ever bigger steam engines burning ever more coal is the superior approach to build faster trains", they said. More data + scaling up simple algos will get you better models, obviously. But there are diminishing returns -- meanwhile, new approaches can be transformational

      10 replies 130 retweets 520 likes
      Show this thread
    2. François Chollet‏Verified account @fchollet 15 Mar 2019

      I guarantee you that the dominant AI algorithm of the 2050s will not be a stack of differentiable layers trained with SGD.

      11 replies 62 retweets 326 likes
      Show this thread
      François Chollet‏Verified account @fchollet 15 Mar 2019

      At this point we are pretty close to training our language models on the entirety of text corpora that humanity has produced so far. Their language understanding capabilities are still close to zero.

      10:14 AM - 15 Mar 2019
      • 116 Retweets
      • 432 Likes
      • andrew ilachinski paras kaushik Josu Gómez 🇪🇺 Matthew Miller Artopal Carlos del Cacho e'lena 'alβ̞aɾeð me'ʝ̞að̞o Francisco Soares Namespace
      14 replies 116 retweets 432 likes
        1. New conversation
        2. Ryan Chesler‏ @ryan_chesler 15 Mar 2019
          Replying to @fchollet

          But in 2050 imagine how much more text data we'll have and how many layers we can train 🤣. Jokes aside things will need to seriously change to get to the sci-fi levels everyone thinks are around the corner.

          1 reply 0 retweets 3 likes
        3. Ryan Chesler‏ @ryan_chesler 15 Mar 2019
          Replying to @ryan_chesler @fchollet

          To think we can sum up the meaning of a word with 300 numbers we find by checking what words are next to it is kind of absurd.

          1 reply 0 retweets 6 likes
        4. Show replies
        1. This Tweet is unavailable.
        2. Thomas G. Dietterich‏ @tdietterich 15 Mar 2019
          Replying to @Kufat @fchollet @jerrykuch

          Studies of the source of speedup across many areas has shown that roughly half of the improvement is due to hardware and half due to software. In some cases, algorithm advances have been much more important than hardware.pic.twitter.com/Zh3BOHsHra

          1 reply 0 retweets 6 likes
        3. Show replies
        1. Matt Ranger‏ @vhranger 15 Mar 2019
          Replying to @fchollet

          I think the breakthrough might be more at the theoretical level. Most of NLP is driven by the distributional hypothesis at the moment. Similar low level research might yield fruit, instead of stacking more effort at the higher levels.

          0 replies 0 retweets 4 likes
          Thanks. Twitter will use this to make your timeline better. Undo
          Undo
        1. New conversation
        2. Miles Brundage‏ @Miles_Brundage 15 Mar 2019
          Replying to @fchollet

          No we aren’t (re: closeness to maxing out the data). A few dozen gigs were used for GPT-2.

          1 reply 0 retweets 18 likes
        3. Florian Laurent‏ @MasterScrat 15 Mar 2019
          Replying to @Miles_Brundage @fchollet

          True that. GPT-2 was trained on 8M documents. Wayback Machine hosts 388B web pages.

          0 replies 0 retweets 6 likes
        4. End of conversation
        1. New conversation
        2. cepcam‏ @cepcam 15 Mar 2019
          Replying to @fchollet

          This , I do think, because understanding human language serve no purpose for IA yet I mean, what could the loss function be for «understanding » human language ?

          1 reply 0 retweets 0 likes
        3. cepcam‏ @cepcam 15 Mar 2019
          Replying to @cepcam @fchollet

          Mostov our langage modeling is based on cross entropy aka retreiving same distribution of words of the original corpus Even a seq2seq just find the most probable answer given distribution of the input sequence

          0 replies 0 retweets 0 likes
        4. End of conversation
        1. Rohit Roy Chowdhury‏ @RyChwdhryRohit 15 Mar 2019
          Replying to @fchollet

          @LordSomen True. I feel AI is so much more than current applications.

          0 replies 0 retweets 1 like
          Thanks. Twitter will use this to make your timeline better. Undo
          Undo

      Loading seems to be taking a while.

      Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

        Promoted Tweet

        false

        • © 2021 Twitter
        • About
        • Help Center
        • Terms
        • Privacy policy
        • Cookies
        • Ads info