Rados{\l}aw Nowak, Anna Bielawska, Bogusz Stefa\'nczyk, Maciej Sanocki, Pawe{\l} Wawrzy\'nski: GeoGAE: Scalable Graph-Level Autoencoding via Hyperball Cloud Representations https://arxiv.org/abs/2609.35527 https://arxiv.org/pdf/2609.35527 https://arxiv.org/html/2609.35527
September 30, 2026 at 6:53 AM
Rados{\l}aw Nowak, Anna Bielawska, Bogusz Stefa\'nczyk, Maciej Sanocki, Pawe{\l} Wawrzy\'nski: GeoGAE: Scalable Graph-Level Autoencoding via Hyperball Cloud Representations https://arxiv.org/abs/2609.35527 https://arxiv.org/pdf/2609.35527 https://arxiv.org/html/2609.35527
Rados{\l}aw Nowak, Anna Bielawska, Bogusz Stefa\'nczyk, Maciej Sanocki, Pawe{\l} Wawrzy\'nski: GeoGAE: Scalable Graph-Level Autoencoding via Hyperball Cloud Representations https://arxiv.org/abs/2609.35527 https://arxiv.org/pdf/2609.35527 https://arxiv.org/html/2609.35527
September 29, 2026 at 6:53 AM
Rados{\l}aw Nowak, Anna Bielawska, Bogusz Stefa\'nczyk, Maciej Sanocki, Pawe{\l} Wawrzy\'nski: GeoGAE: Scalable Graph-Level Autoencoding via Hyperball Cloud Representations https://arxiv.org/abs/2609.35527 https://arxiv.org/pdf/2609.35527 https://arxiv.org/html/2609.35527
Jinghui Yuan, Hongtao Zhang, Jade Zou, Tianyu Li, Wenjie Zhou, Tianyu He, Wei Chen: HyperTransfer: Understanding the Equivalence between Base Optimizer and Hyperball https://arxiv.org/abs/2609.07017 https://arxiv.org/pdf/2609.07017 https://arxiv.org/html/2609.07017
September 9, 2026 at 6:44 AM
Jinghui Yuan, Hongtao Zhang, Jade Zou, Tianyu Li, Wenjie Zhou, Tianyu He, Wei Chen: HyperTransfer: Understanding the Equivalence between Base Optimizer and Hyperball https://arxiv.org/abs/2609.07017 https://arxiv.org/pdf/2609.07017 https://arxiv.org/html/2609.07017
This pin used the extra backboxes from Williams Hyperball.
August 29, 2026 at 2:11 AM
This pin used the extra backboxes from Williams Hyperball.
#hyperball #sketch spatial tension, layered dimension, color anchors
#abstractart #expressionism #art #visualart #digitalart
#abstractart #expressionism #art #visualart #digitalart
August 26, 2026 at 2:24 PM
#hyperball #sketch spatial tension, layered dimension, color anchors
#abstractart #expressionism #art #visualart #digitalart
#abstractart #expressionism #art #visualart #digitalart
Pokemon Gen 4 - Hyperball - TikTok Spiel 🖤
August 20, 2026 at 8:53 AM
Pokemon Gen 4 - Hyperball - TikTok Spiel 🖤
Bisasam Fusion Pokemon - Hyperball - TikTok Spiel 🖤
August 8, 2026 at 7:29 AM
Bisasam Fusion Pokemon - Hyperball - TikTok Spiel 🖤
Instead of hyperbole it should be called hyperball
July 29, 2026 at 11:50 PM
Instead of hyperbole it should be called hyperball
I wish I had video of myself playing the arcades in 70's and early 80's. Defender, galaxies, robotron, hyperball. .. I used to return home with burst blisters on the palms of my hands.
July 28, 2026 at 8:53 AM
I wish I had video of myself playing the arcades in 70's and early 80's. Defender, galaxies, robotron, hyperball. .. I used to return home with burst blisters on the palms of my hands.
Yihao Xiao, Jialong Sun, Zitian Gao, Zeming Wei, Chutian Wang, Ran Tao, Jiaye Teng, Bryan Dai: Hyperball May Not Be a Free Lunch https://arxiv.org/abs/2607.22444 https://arxiv.org/pdf/2607.22444 https://arxiv.org/html/2607.22444
July 27, 2026 at 6:43 AM
Yihao Xiao, Jialong Sun, Zitian Gao, Zeming Wei, Chutian Wang, Ran Tao, Jiaye Teng, Bryan Dai: Hyperball May Not Be a Free Lunch https://arxiv.org/abs/2607.22444 https://arxiv.org/pdf/2607.22444 https://arxiv.org/html/2607.22444
Backed up and booming! Love seeing hyperball stuff especially when it goes overboard (bursting). Teasing and peril is fun too (like when they're getting too tight and don't wanna burst but do)
July 26, 2026 at 6:54 AM
Backed up and booming! Love seeing hyperball stuff especially when it goes overboard (bursting). Teasing and peril is fun too (like when they're getting too tight and don't wanna burst but do)
Rayquaza Fusion Pokemon - Hyperball - TikTok Spiel 🖤
July 12, 2026 at 4:51 PM
Rayquaza Fusion Pokemon - Hyperball - TikTok Spiel 🖤
Hyperball is a new optimizer that ensures consistent speedups for large language models, yielding a 20-30% tokenequivalent speedup over traditional methods. It fixes weight norms, enhancing learning rate transfer across varying model sizes. https://arxiv.org/abs/2606.16899
Fantastic Pretraining Optimizers and Where to Find Them II: Hyperball Optimization
ArXiv link for Fantastic Pretraining Optimizers and Where to Find Them II: Hyperball Optimization
arxiv.org
June 16, 2026 at 4:20 PM
Hyperball is a new optimizer that ensures consistent speedups for large language models, yielding a 20-30% tokenequivalent speedup over traditional methods. It fixes weight norms, enhancing learning rate transfer across varying model sizes. https://arxiv.org/abs/2606.16899
Kaiyue Wen, Xingyu Dang, Kaifeng Lyu, Tengyu Ma, Percy Liang: Fantastic Pretraining Optimizers and Where to Find Them II: Hyperball Optimization https://arxiv.org/abs/2606.16899 https://arxiv.org/pdf/2606.16899 https://arxiv.org/html/2606.16899
June 16, 2026 at 6:45 AM
Kaiyue Wen, Xingyu Dang, Kaifeng Lyu, Tengyu Ma, Percy Liang: Fantastic Pretraining Optimizers and Where to Find Them II: Hyperball Optimization https://arxiv.org/abs/2606.16899 https://arxiv.org/pdf/2606.16899 https://arxiv.org/html/2606.16899
..It's Hyperball.
June 15, 2026 at 11:14 PM
..It's Hyperball.
Next time you're in the Loop I recommend visiting The Flip pinball museum which just opened on the lower level/pedway of Block 37. A well curated collection with a variety of vintage tables and an art installation table. Plus two Aztec tables for the @techconnectify.bsky.social fans...
June 13, 2026 at 1:14 AM
Next time you're in the Loop I recommend visiting The Flip pinball museum which just opened on the lower level/pedway of Block 37. A well curated collection with a variety of vintage tables and an art installation table. Plus two Aztec tables for the @techconnectify.bsky.social fans...
Oh gz! Ich habe das damals mit Team Rocket geschafft, das war so anstrengend, aber lustig. Aktuell bin ich, glaube ich, Hyperball oder so lol
May 29, 2026 at 10:44 AM
Oh gz! Ich habe das damals mit Team Rocket geschafft, das war so anstrengend, aber lustig. Aktuell bin ich, glaube ich, Hyperball oder so lol
I've literally started ranked doubles yesterday (new season) and Mega Crabominable already got me into Hyperball Rank in less than 10 battles. 💪🦀💪
Really slept upon pokemon in my opinion!
(Gotta start drawing gijinkas of my mons some day & give my crab girl big abs & wild hair)
Really slept upon pokemon in my opinion!
(Gotta start drawing gijinkas of my mons some day & give my crab girl big abs & wild hair)
May 14, 2026 at 7:27 PM
I've literally started ranked doubles yesterday (new season) and Mega Crabominable already got me into Hyperball Rank in less than 10 battles. 💪🦀💪
Really slept upon pokemon in my opinion!
(Gotta start drawing gijinkas of my mons some day & give my crab girl big abs & wild hair)
Really slept upon pokemon in my opinion!
(Gotta start drawing gijinkas of my mons some day & give my crab girl big abs & wild hair)
Pokemon Gen 6 - Hyperball - TikTok Spiel 🖤
May 13, 2026 at 9:26 AM
Pokemon Gen 6 - Hyperball - TikTok Spiel 🖤
There were two red flags in the runs from that initial sweep: suspiciously large weight norms and learning rates!
For the first, I switched to Kaiyue Wen's Hyperball optimizers.
For the second, I found the answer in Apple's Complete(d)P paper: arxiv.org/abs/2512.22382
For the first, I switched to Kaiyue Wen's Hyperball optimizers.
For the second, I found the answer in Apple's Complete(d)P paper: arxiv.org/abs/2512.22382
Completed Hyperparameter Transfer across Modules, Width, Depth, Batch and Duration
Hyperparameter tuning can dramatically impact training stability and final performance of large-scale models. Recent works on neural network parameterisations, such as $μ$P, have enabled transfer of o...
arxiv.org
May 11, 2026 at 7:59 PM
There were two red flags in the runs from that initial sweep: suspiciously large weight norms and learning rates!
For the first, I switched to Kaiyue Wen's Hyperball optimizers.
For the second, I found the answer in Apple's Complete(d)P paper: arxiv.org/abs/2512.22382
For the first, I switched to Kaiyue Wen's Hyperball optimizers.
For the second, I found the answer in Apple's Complete(d)P paper: arxiv.org/abs/2512.22382
May 9, 2026 at 10:53 PM
May 5, 2026 at 2:14 PM
J'accroche toujours beaucoup à Pokemon Pocket, j'ai réussi à finir le dernier tournoi au rang Hyperball 2. Un jour peut-être Masterball ? Mais je suis surtout surprise par le nombre de mises à jour et de cadeaux, certes numériques, qu'on reçoit même quand on ne dépense rien ou presque dans ce jeu.
April 29, 2026 at 7:11 PM
J'accroche toujours beaucoup à Pokemon Pocket, j'ai réussi à finir le dernier tournoi au rang Hyperball 2. Un jour peut-être Masterball ? Mais je suis surtout surprise par le nombre de mises à jour et de cadeaux, certes numériques, qu'on reçoit même quand on ne dépense rien ou presque dans ce jeu.
Auch wenn das Spiel seine performance Probleme hat (die hoffentlich gefixt werden) hab ich echt spaß an Pokemon Champions. VGC ist so ein spannendes, spaßiges Format und der Zugang ist jetzt einfacher denn je.
Bin zwar erst Hyperball-Rang, aber das ohne Home und ohne einen Euro ausgegeben zu haben.
Bin zwar erst Hyperball-Rang, aber das ohne Home und ohne einen Euro ausgegeben zu haben.
April 26, 2026 at 1:42 PM
Auch wenn das Spiel seine performance Probleme hat (die hoffentlich gefixt werden) hab ich echt spaß an Pokemon Champions. VGC ist so ein spannendes, spaßiges Format und der Zugang ist jetzt einfacher denn je.
Bin zwar erst Hyperball-Rang, aber das ohne Home und ohne einen Euro ausgegeben zu haben.
Bin zwar erst Hyperball-Rang, aber das ohne Home und ohne einen Euro ausgegeben zu haben.