Aldo Santiago
@alsanty.bsky.social
1.4K followers
8.9K following
100 posts
Engineering | Finances | Complex System Dynamics
HPC | AI | QC | DACS
Infrastructure, Construction, Energy, Defense, FinTech, BioTech.
DACS .- Dynamically Adaptive Complex Systems
Posts
Media
Videos
Starter Packs
Aldo Santiago
@alsanty.bsky.social
· Oct 2
GitHub - StanfordBDHG/OpenTSLM: OpenTSLM: Time-Series Language Models for Reasoning over Multivariate Medical Text- and Time-Series Data
OpenTSLM: Time-Series Language Models for Reasoning over Multivariate Medical Text- and Time-Series Data - StanfordBDHG/OpenTSLM
share.google
Aldo Santiago
@alsanty.bsky.social
· Sep 28
Aldo Santiago
@alsanty.bsky.social
· Sep 28
Non-linear dynamical approaches for characterizing multi-sector climate impacts under irreducible uncertainty - npj Climate and Atmospheric Science
npj Climate and Atmospheric Science - Non-linear dynamical approaches for characterizing multi-sector climate impacts under irreducible uncertainty
share.google
Aldo Santiago
@alsanty.bsky.social
· Sep 26
Parallel-R1: Towards Parallel Thinking via Reinforcement Learning
Parallel thinking has emerged as a novel approach for enhancing the reasoning capabilities of large language models (LLMs) by exploring multiple reasoning paths concurrently. However, activating such ...
arxiv.org
Aldo Santiago
@alsanty.bsky.social
· Sep 3
Ultrabroadband on-chip photonics for full-spectrum wireless communications - Nature
Adaptive wireless communication over an unprecedented frequency range spanning over 100 GHz can be achieved by a thin-film lithium niobate photonic wireless system, which can process a large flux of i...
www.nature.com
Aldo Santiago
@alsanty.bsky.social
· Aug 10
Pure quantum state without the need for cooling
Even large objects with several hundred million atoms can exhibit quantum mechanical behaviour – without cooling and at room temperature, as researchers at ETH Zurich have shown. This yields exciting ...
ethz.ch
Reposted by Aldo Santiago
Sung Kim
@sungkim.bsky.social
· Aug 9
Compute Better Spent: Replacing Dense Layers with Structured Matrices
Dense linear layers are the dominant computational bottleneck in foundation models. Identifying more efficient alternatives to dense matrices has enormous potential for building more compute-efficient...
arxiv.org
Aldo Santiago
@alsanty.bsky.social
· Aug 9
How Attention Sinks Keep Language Models Stable
We discovered why language models catastrophically fail on long conversations: when old tokens are removed to save memory, models produce complete gibberish. We found models dump massive attention ont...
hanlab.mit.edu
Aldo Santiago
@alsanty.bsky.social
· Aug 9
On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification
We present a simple yet theoretically motivated improvement to Supervised Fine-Tuning (SFT) for the Large Language Model (LLM), addressing its limited generalization compared to reinforcement learning...
arxiv.org
Aldo Santiago
@alsanty.bsky.social
· Aug 3
Existence of smooth solutions of the Navier-Stokes equations in three-dimensional Euclidean space
Based on the essential connection of the parabolic inertia Lamé equations and Navier-Stokes equations, we prove the existence of smooth solutions of the incompressible Navier-Stokes equations in three...
arxiv.org
Aldo Santiago
@alsanty.bsky.social
· Aug 3
Persona Vectors: Monitoring and Controlling Character Traits in Language Models
Large language models interact with users through a simulated 'Assistant' persona. While the Assistant is typically trained to be helpful, harmless, and honest, it sometimes deviates from these ideals...
arxiv.org
Reposted by Aldo Santiago
Tim Kellogg
@timkellogg.me
· Aug 1
Persona Vectors: Monitoring and Controlling Character Traits in Language Models
Large language models interact with users through a simulated 'Assistant' persona. While the Assistant is typically trained to be helpful, harmless, and honest, it sometimes deviates from these ideals...
arxiv.org
Aldo Santiago
@alsanty.bsky.social
· Jun 27
NIST and Partners Use Quantum Mechanics to Make a Factory for Random Numbers
Broadcast as a free public service, the beacon can be used anywhere an independent source of random numbers would be useful, such as selecting jury candidates or assigning resources through a lottery
www.nist.gov
Reposted by Aldo Santiago
Ross Duncan
@rossquantum.bsky.social
· Jun 27
Breaking even with magic: demonstration of a high-fidelity logical non-Clifford gate
Encoding quantum information to protect it from errors is essential for performing large-scale quantum computations. Performing a universal set of quantum gates on encoded states demands a potentially...
arxiv.org
Aldo Santiago
@alsanty.bsky.social
· Jun 3
AlphaOne: Reasoning Models Thinking Slow and Fast at Test Time
This paper presents AlphaOne ($α$1), a universal framework for modulating reasoning progress in large reasoning models (LRMs) at test time. $α$1 first introduces $α$ moment, which represents the scale...
arxiv.org
Aldo Santiago
@alsanty.bsky.social
· Jun 2
RareFold: Structure prediction and design of proteins with noncanonical amino acids
Protein structure prediction and design have traditionally been limited to the 20 canonical amino acids. Expanding this space to include noncanonical amino acids (NCAAs) offers new opportunities for p...
www.biorxiv.org
Reposted by Aldo Santiago
