François Chollet on Twitter: "My 4-month old and I are teaching things to each other, but he's learning much faster than I do."

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

1. David BM‏ @dabetm Aug 30
  Replying to @fchollet
  
  neuroplasticity
  
  4 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Vladimir Lukin‏ @alscor1966 Aug 30
  Replying to @fchollet
  
  It is likely that you are not as focused on learning as your son.
  
  6 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Jason Phang‏ @zhansheng Aug 30
  Replying to @fchollet
  
  You're just further down on your learning rate schedule
  
  15 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Kyle McDonald‏ @kcimc Aug 30
  Replying to @fchollet
  
  last few % accuracy is always hard earned. especially when the model has been placed in production and is vulnerable to regression upon iteration.
  
  8 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Mat Fournier‏ @mat4nier Aug 30
  Replying to @fchollet
  
  The 24 month mark is insane right now. Enjoy it! It is a journey!
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Digital Physics Film‏ @DigPhysics Aug 30
  Replying to @fchollet
  
  Learning is monotonic but logarithmic
  
  1 like
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Ibrahim Alshubaily‏ @IbraAlshubaily Aug 30
  Replying to @fchollet
  
  The simulation have learning rate decay. I hear lifting and running reduce the decay rate.
  
  2 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Swaroop Nath‏ @SwaroopNath3 Aug 30
  Replying to @fchollet
  
  His neural net is still in the early stages of the loss function - the gradients are steep.
  
  9 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Shubham Panchal (He/Him)‏ @im_mobileML_dev Aug 31
  Replying to @fchollet
  
  I was just about to tweet about his learning rate. Well, everybody has written that before me
  
  2 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Trademaker‏ @mr_trademaker Aug 31
  Replying to @fchollet
  
  Well, your child still has got a lot of low hanging fruit to exploit...
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo