The Research Desk.

The most upvoted and starred AI research crossing the community today.

Last Brew Time: Oct 17, 2026, 11:44 AM PT

X.com Research Buzz

Computer Vision
Analytic toroidal 3D MHD equilibria and steady Euler flows with invariant surfaces
X.com
4691

Analytic toroidal 3D MHD equilibria and steady Euler flows with invariant surfaces

Matt Landreman

University of Maryland, College Park

AlphaXiv Trending

Machine Learning
AlphaXiv
103

Neural Networks Are Already Clean

4k, Views

Are We Overthinking One-Step Generation?
AlphaXiv
81

Are We Overthinking One-Step Generation?

1k, Views

Affiliation: Seoul National University, Seoul National University, Harvard University

NLP
Generative Representation Learning for Scalable Latent Tokenization
AlphaXiv
70

Generative Representation Learning for Scalable Latent Tokenization

1k, Views

Affiliation: Stanford University, Affiliation: Massachusetts Institute of Technology, Affiliation: University of Michigan, Stanford University, Massachusetts Institute of Technology

Machine Learning
Training Language Models to Fail Gracefully
AlphaXiv
34

Training Language Models to Fail Gracefully

Views

Affiliation: Carnegie Mellon University, Carnegie Mellon University

Robotics
T2-z: Test-Time Optimization of Environment Latents for Robotic World Models
AlphaXiv
23

T2-z: Test-Time Optimization of Environment Latents for Robotic World Models

Views

Stanford University CMU, Toyota Research Institute University of Southern California

Stay CALM: Calibrated Online Aggregation for Multi-Horizon Forecasting Under Delayed, Overlapping Feedback
AlphaXiv
7

Stay CALM: Calibrated Online Aggregation for Multi-Horizon Forecasting Under Delayed, Overlapping Feedback

HuggingFace Daily Papers

OpenSAE
Behavioral Compositionality of Sparse Autoencoder Representations
HuggingFace
104

Behavioral Compositionality of Sparse Autoencoder Representations

Haotian Ma, Ziye Mei, Frederic Sala

OpenSAE

Reinforcement Learning
LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents
HuggingFace
6

LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents

Qingni Wang, Changdae Oh, Yuzhe Yang, Sharon Li, Jiachen Yang

Safety Alignment
Mine, Sample, Render: Test-Time Diversity Alignment Of Generative Models
HuggingFace
5

Mine, Sample, Render: Test-Time Diversity Alignment Of Generative Models

Kostas Triaridis, Morris E. Sharp, Nebojsa Jojic

Computer Vision
LeanPRM: Lean-Grounded Process Supervision for Natural Language Proof Verification
HuggingFace
4

LeanPRM: Lean-Grounded Process Supervision for Natural Language Proof Verification

Rodney Lafuente-Mercado, Salman Rahman, Jingyan Shen, James Shiffer, Saadia Gabriel

Reinforcement Learning
Scalable Weak-to-Strong Verifier Hardening for Agent Environments
HuggingFace
3

Scalable Weak-to-Strong Verifier Hardening for Agent Environments

Chong Zeng, Ziqian Zhong, Kexun Zhang, Ivan Bercovich, Ivgeni Segal

Stanford AI

Diminishing Returns of Data in Sparse Autoencoders: Functional Behavior does not Guarantee Feature Reproducibility
HuggingFace
1

Diminishing Returns of Data in Sparse Autoencoders: Functional Behavior does not Guarantee Feature Reproducibility

Sambit Ghosh, Madelyn Mathai, R. Venkatesh Babu, Chirag Agarwal

Aikyam Lab