Preserve-Then-Quantize: Balancing Rank Budgets for Quantization Error Reconstruction in LLMs

Posted by klaurent Stanford

Kavli Affiliate: Hsiao-Mei (Sherry) Cho| First 5 Authors: [#item_custom_name[1, [#item_custom_name[2, [#item_custom_name[3, [#item_custom_name[4, [#item_custom_name[5| Summary:Quantization Error Reconstruction (QER) reduces accuracy loss in Post-Training Quantization (PTQ) by approximating weights as $mathbfW approx mathbfQ + mathbfLmathbfR$, using a rank-$r$ correction to reconstruct quantization error. Prior methods devote the full rank budget to error reconstruction, which is suboptimal when […]


Continue.. Preserve-Then-Quantize: Balancing Rank Budgets for Quantization Error Reconstruction in LLMs

Enhancing Multi-Image Understanding through Delimiter Token Scaling

Posted by klaurent Stanford

Kavli Affiliate: Hsiao-Mei (Sherry) Cho| First 5 Authors: [#item_custom_name[1, [#item_custom_name[2, [#item_custom_name[3, [#item_custom_name[4, [#item_custom_name[5| Summary:Large Vision-Language Models (LVLMs) achieve strong performance on single-image tasks, but their performance declines when multiple images are provided as input. One major reason is the cross-image information leakage, where the model struggles to distinguish information across different images. Existing LVLMs already […]


Continue.. Enhancing Multi-Image Understanding through Delimiter Token Scaling

Physics-Informed Neural Networks for Modeling Galactic Gravitational Potentials

Posted by klaurent MIT

Kavli Affiliate: Lina Necib | Summary:We introduce a physics-informed neural framework for modeling static and time-dependent galactic gravitational potentials. The method combines data-driven learning with embedded physical constraints to capture complex, small-scale features while preserving global physical consistency. We quantify predictive uncertainty through a Bayesian framework, and model time evolution using a neural ODE approach. […]


Continue.. Physics-Informed Neural Networks for Modeling Galactic Gravitational Potentials

Physics-Informed Neural Networks for Modeling Galactic Gravitational Potentials

Posted by klaurent MIT

Kavli Affiliate: Lina Necib | Summary:We introduce a physics-informed neural framework for modeling static and time-dependent galactic gravitational potentials. The method combines data-driven learning with embedded physical constraints to capture complex, small-scale features while preserving global physical consistency. We quantify predictive uncertainty through a Bayesian framework, and model time evolution using a neural ODE approach. […]


Continue.. Physics-Informed Neural Networks for Modeling Galactic Gravitational Potentials

Prefrontal working memory activity partially mediates link between enriched neighborhood environments and episodic memory among 9-13-year-olds

Posted by dbos UCSD / Salk

Kavli Affiliate: Wesley Thompson | Authors: Michael Rosario, Carlos Cardenas-Iniguez, Jennifer V Chavez, Katherine L Bottenhorn, Hedyeh Ahmadi, Megan M Herting and Wesley K Thompson | Summary: Background Enriched environments support neurodevelopment. The pathways linking enrichment to cognitive processes such as episodic memory among youth remain unclear. This study examined whether brain function and structure […]


Continue.. Prefrontal working memory activity partially mediates link between enriched neighborhood environments and episodic memory among 9-13-year-olds

Divergent mitochondrial and metabolic adaptations shape selective vulnerability in ALS

Posted by dbos Rockefeller

Kavli Affiliate: Nathaniel Heintz | Authors: Bianca A Cotto, Lizi Zhang, Benjamin W Fait, Christopher Peralta, Ece Kilic, Henrik Molina, Nathaniel Heintz and Eric F Schmidt | Summary: Neurodegenerative diseases like amyotrophic lateral sclerosis (ALS) exhibit striking cell-type selectivity, yet the basis for this vulnerability remains elusive. Here, we uncover that even closely related neurons […]


Continue.. Divergent mitochondrial and metabolic adaptations shape selective vulnerability in ALS

Cortistatin neurons in the prelimbic cortex regulate seizure susceptibility in female mice via BDNF-TrkB signaling

Posted by dbos Hopkins

Kavli Affiliate: Keri Martinowich | Authors: Aaron J Salisbury, Lourdes Figueroa, Keri Martinowich and Michael Totty | Summary: The risk of developing psychiatric disorders, particularly stress-related disorders such as major depressive disorder (MDD) and post-traumatic stress disorder (PTSD), is increased threefold in patients with epilepsy. While this increased risk may arise as a consequence of […]


Continue.. Cortistatin neurons in the prelimbic cortex regulate seizure susceptibility in female mice via BDNF-TrkB signaling

Pathogenic tau in the mouse locus coeruleus induces noradrenergic hyperactivity and neuropsychiatric phenotypes reminiscent of early Alzheimers disease

Posted by dbos Hopkins

Kavli Affiliate: Keri Martinowich | Authors: Anu Korukonda, Harris E Blankenship, Korey Kam, Dilpreet Kour, Claudia Espinosa-Garcia, Wooyoung Eric Jang, Upsana Srivastava, Bernard Mulvey, Margaret M Tish, Ronit Witztum, Christina C Ramelow, Brittany S Pate, L Cameron Liles, Linyue Lu, Jake Atallah, Eugene Guo, Keri Martinowich, Amanda L Sharpe, Andrew Varga, Srikant Rangaraju, Michael J […]


Continue.. Pathogenic tau in the mouse locus coeruleus induces noradrenergic hyperactivity and neuropsychiatric phenotypes reminiscent of early Alzheimers disease

Modeling the Dynamics and Thermochemistry for the Outer Atmospheres of the Ultra-hot Jupiter WASP-121b

Posted by klaurent Peking

Kavli Affiliate: Lile Wang |Summary:We present three-dimensional simulations of the ultra-hot Jupiter (UHJ) WASP-121b from the planetary surface to extended outflows, coupling hydrodynamics with consistent non-equilibrium thermochemistry, ray-tracing radiative transfer, and hydrodynamics using the GPU-accelerated Kratos framework. The fiducial model exhibits several atmospheric layers, including the lower atmospheres controlled by day-night circulation, and transonic photoevaporative […]


Continue.. Modeling the Dynamics and Thermochemistry for the Outer Atmospheres of the Ultra-hot Jupiter WASP-121b

Modeling the Dynamics and Thermochemistry for the Outer Atmospheres of the Ultra-hot Jupiter WASP-121b

Posted by klaurent Peking

Kavli Affiliate: Lile Wang |Summary:We present three-dimensional simulations of the ultra-hot Jupiter (UHJ) WASP-121b from the planetary surface to extended outflows, coupling hydrodynamics with consistent non-equilibrium thermochemistry, ray-tracing radiative transfer, and hydrodynamics using the GPU-accelerated Kratos framework. The fiducial model exhibits several atmospheric layers, including the lower atmospheres controlled by day-night circulation, and transonic photoevaporative […]


Continue.. Modeling the Dynamics and Thermochemistry for the Outer Atmospheres of the Ultra-hot Jupiter WASP-121b