#alphatensor 搜索结果

Turns out everything is a matrix multiplication from computer graphics to training neural networks :) Our latest front cover @Nature paper on #AlphaTensor here: dpmd.ai/nature-alpha-t…

demishassabis's tweet image. Turns out everything is a matrix multiplication from computer graphics to training neural networks :) Our latest front cover @Nature paper on #AlphaTensor here: dpmd.ai/nature-alpha-t…

Happy to be at #NeurIPS2022 presenting #AlphaTensor, work with my @DeepMind colleagues. - Friday 13:30 in New Frontiers in Graph Learning - Saturday 13:30 in RL for Real Life

matejbalog's tweet image. Happy to be at #NeurIPS2022 presenting #AlphaTensor, work with my @DeepMind colleagues.

- Friday 13:30 in New Frontiers in Graph Learning
- Saturday 13:30 in RL for Real Life

I'm at #NeurIPS2022; DM me if you want to chat about RL, AlphaZero, and ML for algorithm discovery. We'll present #AlphaTensor: our recent Nature cover publication beating a 50-year-old record in CS and mathematics –– Mon 28 Nov, 3:30-5pm at @DeepMind's booth.

BarekatainAmin's tweet image. I'm at #NeurIPS2022; DM me if you want to chat about RL, AlphaZero, and ML for algorithm discovery. 

We'll present #AlphaTensor: our recent Nature cover publication beating a 50-year-old record in CS and mathematics –– Mon 28 Nov, 3:30-5pm at @DeepMind's booth.

Today in @Nature: #AlphaTensor, an AI system for discovering novel, efficient, and exact algorithms for matrix multiplication - a building block of modern computations. AlphaTensor finds faster algorithms for many matrix sizes: dpmd.ai/dm-alpha-tensor & dpmd.ai/nature-alpha-t… 1/


Finally got a chance to review the #AlphaTensor paper from @AlhusseinFawzi and team @DeepMind. Interesting piece and trajectory But the response from @ManuelKauers and @jakobmoosbauer in their paper "FBHHRBNRSSSHK-ALGORITHM", beating Deepmind's 5x5 matmul by 1 operation is funny

FelixCLC_'s tweet image. Finally got a chance to review the #AlphaTensor paper from @AlhusseinFawzi and team @DeepMind. Interesting piece and trajectory
But the response from @ManuelKauers and @jakobmoosbauer in their paper "FBHHRBNRSSSHK-ALGORITHM", beating Deepmind's 5x5 matmul by 1 operation is funny

#Nature #AlphaTensor Discovering faster matrix multiplication algorithms with reinforcement learning nature.com/articles/s4158… Artificial intelligence finds faster algorithms for multiplying matrices nature.com/articles/d4158…

emulenews's tweet image. #Nature #AlphaTensor Discovering faster matrix multiplication algorithms with reinforcement learning nature.com/articles/s4158… Artificial intelligence finds faster algorithms for multiplying matrices nature.com/articles/d4158…

Spent some time implementing in #Nim the SIMD optimized version of the 2x2x2 matrix multiplication found by #AlphaTensor Had to fix a mistake in original code on the right... #Rust syntax looks awful.

planetis_m's tweet image. Spent some time implementing in #Nim the SIMD optimized version of the 2x2x2 matrix multiplication found by #AlphaTensor Had to fix a mistake in original code on the right... #Rust syntax looks awful.
planetis_m's tweet image. Spent some time implementing in #Nim the SIMD optimized version of the 2x2x2 matrix multiplication found by #AlphaTensor Had to fix a mistake in original code on the right... #Rust syntax looks awful.

⚡Discovering Novel Algorithms with AlphaTensor⚡ In @DeepMind's paper the team introduces #AlphaTensor, the first artificial intelligence (AI) system for discovering novel, efficient, and provably correct algorithms for fundamental tasks such as matrix multiplication. 🧵👇


Since 1969 Strassen’s algorithm has famously stood as the fastest way to multiply 2 matrices - but with #AlphaTensor we’ve found a new algorithm that’s faster, with potential to improve efficiency by 10-20% across trillions of calculations per day! dpmd.ai/dm-alpha-tensor


Se que estamos muy pendientes de #ChatGPT y algoritmos text to image pero también hay que tener presente a modelos como #alphatensor uno de los modelos que más impacto y disrupción puede presentar para IA!!! 💛🚀


“Pekiştirmeli öğrenme ile daha hızlı matris çarpma algoritmalarını keşfetme” #AlphaTensor @Nature nature.com/articles/s4158…

oktayarslan's tweet image. “Pekiştirmeli öğrenme ile daha hızlı matris çarpma algoritmalarını keşfetme”

#AlphaTensor @Nature

nature.com/articles/s4158…

.@Google's #AlphaTensor #Quantum and #GenAI advance circuit optimization and cancer research. Russia and South Korea plan quantum advancements, while @EuroHPC_JU explores energy-efficient #quantumcomputing. Read more: ow.ly/6XHT50QJpYj

HPCwire's tweet image. .@Google's #AlphaTensor #Quantum and #GenAI advance circuit optimization and cancer research. Russia and South Korea plan quantum advancements, while @EuroHPC_JU explores energy-efficient #quantumcomputing. Read more: ow.ly/6XHT50QJpYj

Na AlphaGo en AlphaZero (dat in resp. go en schaken de wereldkampioen wist te verslaan) presenteert @DeepMind nu het diepe neurale netwerk #AlphaTensor: ideaal voor matrixvermenigvuldigingen. bit.ly/3VlLUKn

de_ingenieur's tweet image. Na AlphaGo en AlphaZero (dat in resp. go en schaken de wereldkampioen wist te verslaan) presenteert @DeepMind nu het diepe neurale netwerk #AlphaTensor: ideaal voor matrixvermenigvuldigingen. bit.ly/3VlLUKn

Starting from scratch, #AlphaTensor gradually improves from finding the “class-room” algorithm to discovering historical algorithms such as Strassen’s - eventually surpassing them. This work is an example of how AI can advance the field of algorithmic discovery. 2/


3 months leading social media at @DeepMind today 🚀 from helping unveil new + innovative AI systems like #AlphaTensor and #DeepNash to covering exciting research at #NeurIPS2022, it's been inspiring to work with such talented teams. can't wait for 2023 💡


“if the traditional algorithm taught in school multiplies a 4x5 by 5x5 matrix using 100 multiplications, and this number was reduced to 80 w/ human ingenuity, #AlphaTensor has found algorithms that do the same operation using just 76 multiplications. ” buff.ly/3VgyLlM #AI

woojinrad's tweet image. “if the traditional algorithm taught in school multiplies a 4x5 by 5x5 matrix using 100 multiplications, and this number was reduced to 80 w/ human ingenuity, #AlphaTensor has found algorithms that do the same operation using just 76 multiplications. ” buff.ly/3VgyLlM #AI

TLDR: We developed StrassenNet ~ 4 years ago. It already re-discovered Strassen's algorithm, reduced multiplications by 99.5% while maintaining accuracy of #deeplearning No guarantee #alphatensor is stable while StrassenNet explicitly trains for it. proceedings.mlr.press/v80/tschannen1…


未找到 "#alphatensor" 的结果

#AlphaTensor is the first #AI system for discovering the fastest way to multiply two matrices.

dumbledrip's tweet image. #AlphaTensor is the first #AI system for discovering the fastest way to multiply two matrices.

Turns out everything is a matrix multiplication from computer graphics to training neural networks :) Our latest front cover @Nature paper on #AlphaTensor here: dpmd.ai/nature-alpha-t…

demishassabis's tweet image. Turns out everything is a matrix multiplication from computer graphics to training neural networks :) Our latest front cover @Nature paper on #AlphaTensor here: dpmd.ai/nature-alpha-t…

i thk the UFO are a multiverse bug. (like the photoshop layer) they are in another universe and we thk they are in our universe. God is twisted we have to define a matrix multiverve law. Like #alphatensor of deepmind a faster matrix multiplication algorithms + blackhole data

blackice747's tweet image. i thk the UFO are a multiverse bug.
(like the photoshop layer) 
they are in another universe and we thk they are in our universe.
God is twisted
we have to define a matrix multiverve law. 
Like #alphatensor of deepmind a  faster matrix multiplication algorithms + blackhole data
blackice747's tweet image. i thk the UFO are a multiverse bug.
(like the photoshop layer) 
they are in another universe and we thk they are in our universe.
God is twisted
we have to define a matrix multiverve law. 
Like #alphatensor of deepmind a  faster matrix multiplication algorithms + blackhole data
blackice747's tweet image. i thk the UFO are a multiverse bug.
(like the photoshop layer) 
they are in another universe and we thk they are in our universe.
God is twisted
we have to define a matrix multiverve law. 
Like #alphatensor of deepmind a  faster matrix multiplication algorithms + blackhole data
blackice747's tweet image. i thk the UFO are a multiverse bug.
(like the photoshop layer) 
they are in another universe and we thk they are in our universe.
God is twisted
we have to define a matrix multiverve law. 
Like #alphatensor of deepmind a  faster matrix multiplication algorithms + blackhole data

.@Google's #AlphaTensor #Quantum and #GenAI advance circuit optimization and cancer research. Russia and South Korea plan quantum advancements, while @EuroHPC_JU explores energy-efficient #quantumcomputing. Read more: ow.ly/6XHT50QJpYj

HPCwire's tweet image. .@Google's #AlphaTensor #Quantum and #GenAI advance circuit optimization and cancer research. Russia and South Korea plan quantum advancements, while @EuroHPC_JU explores energy-efficient #quantumcomputing. Read more: ow.ly/6XHT50QJpYj

Finally got a chance to review the #AlphaTensor paper from @AlhusseinFawzi and team @DeepMind. Interesting piece and trajectory But the response from @ManuelKauers and @jakobmoosbauer in their paper "FBHHRBNRSSSHK-ALGORITHM", beating Deepmind's 5x5 matmul by 1 operation is funny

FelixCLC_'s tweet image. Finally got a chance to review the #AlphaTensor paper from @AlhusseinFawzi and team @DeepMind. Interesting piece and trajectory
But the response from @ManuelKauers and @jakobmoosbauer in their paper "FBHHRBNRSSSHK-ALGORITHM", beating Deepmind's 5x5 matmul by 1 operation is funny

And 12 pages of @ManuelKauers & @jakobmoosbauer's 14 page paper look like this. Take that, #AlphaTensor!

dajmeyer's tweet image. And 12 pages of @ManuelKauers & @jakobmoosbauer's 14 page paper look like this. Take that, #AlphaTensor!

TensorGame in AlphaTensor for 2x2 matrix multiplication #AlphaTensor #RL #ML #AI

steve_z_seattle's tweet image. TensorGame in AlphaTensor for 2x2 matrix multiplication #AlphaTensor  #RL #ML #AI

Happy to be at #NeurIPS2022 presenting #AlphaTensor, work with my @DeepMind colleagues. - Friday 13:30 in New Frontiers in Graph Learning - Saturday 13:30 in RL for Real Life

matejbalog's tweet image. Happy to be at #NeurIPS2022 presenting #AlphaTensor, work with my @DeepMind colleagues.

- Friday 13:30 in New Frontiers in Graph Learning
- Saturday 13:30 in RL for Real Life

(3/8)2022: AlphaTensor lands in Nature! Navigated massive search spaces to decompose tensors, discovering faster matrix multiplication than the classic Strassen algorithm. Boosts AI training and scientific computing efficiency! 📈 #AlphaTensor

David6LIANG8's tweet image. (3/8)2022: AlphaTensor lands in Nature! Navigated massive search spaces to decompose tensors, discovering faster matrix multiplication than the classic Strassen algorithm. Boosts AI training and scientific computing efficiency! 📈
#AlphaTensor

Spent some time implementing in #Nim the SIMD optimized version of the 2x2x2 matrix multiplication found by #AlphaTensor Had to fix a mistake in original code on the right... #Rust syntax looks awful.

planetis_m's tweet image. Spent some time implementing in #Nim the SIMD optimized version of the 2x2x2 matrix multiplication found by #AlphaTensor Had to fix a mistake in original code on the right... #Rust syntax looks awful.
planetis_m's tweet image. Spent some time implementing in #Nim the SIMD optimized version of the 2x2x2 matrix multiplication found by #AlphaTensor Had to fix a mistake in original code on the right... #Rust syntax looks awful.

Multiplying matrices is a fundamental operation, yet the optimal algorithm is unknown even for matrices as small as 3x3! Here is an algorithm that #AlphaTensor discovered for multiplying a 4x5 and 5x5 matrix. It uses 76 scalar multiplications (previous best used 80). (2/n)

matejbalog's tweet image. Multiplying matrices is a fundamental operation, yet the optimal algorithm is unknown even for matrices as small as 3x3!

Here is an algorithm that #AlphaTensor discovered for multiplying a 4x5 and 5x5 matrix. It uses 76 scalar multiplications (previous best used 80). (2/n)

#AlphaTensor is a substantial extension of #AlphaZero, which was developed to master Go, chess & other games. A mathematical game like ours poses major additional challenges, and we incorporated multiple ingredients that I think can generally apply to many other problems: (5/n)

matejbalog's tweet image. #AlphaTensor is a substantial extension of #AlphaZero, which was developed to master Go, chess & other games.

A mathematical game like ours poses major additional challenges, and we incorporated multiple ingredients that I think can generally apply to many other problems:

(5/n)

“Pekiştirmeli öğrenme ile daha hızlı matris çarpma algoritmalarını keşfetme” #AlphaTensor @Nature nature.com/articles/s4158…

oktayarslan's tweet image. “Pekiştirmeli öğrenme ile daha hızlı matris çarpma algoritmalarını keşfetme”

#AlphaTensor @Nature

nature.com/articles/s4158…

Bernardino Romera, 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗦𝗰𝗶𝗲𝗻𝘁𝗶𝘀𝘁 𝗮𝘁 𝗚𝗼𝗼𝗴𝗹𝗲 DeepMind, will be on 𝗗𝗲𝗰𝗲𝗺𝗯𝗲𝗿 𝟭𝟱 at the @etsinfupv Assembly Hall, talking about #AlphaTensor, a reinforcement learning agent based on AlphaZero. More info ➡️ linkedin.com/feed/update/ur…

fvalgrai's tweet image. Bernardino Romera, 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗦𝗰𝗶𝗲𝗻𝘁𝗶𝘀𝘁 𝗮𝘁 𝗚𝗼𝗼𝗴𝗹𝗲 DeepMind, will be on 𝗗𝗲𝗰𝗲𝗺𝗯𝗲𝗿 𝟭𝟱 at the @etsinfupv  Assembly Hall, talking about #AlphaTensor, a reinforcement learning agent based on AlphaZero. 

More info ➡️ linkedin.com/feed/update/ur…

The biggest eureka moment was when #AlphaTensor discovered this algorithm for multiplying 4x4 matrices modulo 2 using 47 (rather than 7x7=49) scalar multiplications. It improved upon the 2-level version of Strassen’s famous algorithm for the first time in 50 years. (11/n)

matejbalog's tweet image. The biggest eureka moment was when #AlphaTensor discovered this algorithm for multiplying 4x4 matrices modulo 2 using 47 (rather than 7x7=49) scalar multiplications. It improved upon the 2-level version of Strassen’s famous algorithm for the first time in 50 years. (11/n)

How #AlphaTensor works: 1) We framed the discovery of matrix multiplication algorithms as a single-player game with a 3D board. 2) We trained an agent using Deep Reinforcement Learning (DRL) to achieve superhuman performance in this game. (4/n)

matejbalog's tweet image. How #AlphaTensor works:

1) We framed the discovery of matrix multiplication algorithms as a single-player game with a 3D board.
2) We trained an agent using Deep Reinforcement Learning (DRL) to achieve superhuman performance in this game.

(4/n)

Ingredient #1: generalist agent We trained a single network for all matrix sizes, rather than a separate network for each matrix size. It was cool to see #AlphaTensor found ways of transferring knowledge between algorithms for different matrix sizes in non-obvious ways. (6/n)

matejbalog's tweet image. Ingredient #1: generalist agent

We trained a single network for all matrix sizes, rather than a separate network for each matrix size.

It was cool to see #AlphaTensor found ways of transferring knowledge between algorithms for different matrix sizes in non-obvious ways. (6/n)

Loading...

Something went wrong.


Something went wrong.


United States Trends