The Research Desk.

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

Last Brew Time: Aug 5, 2026, 10:44 AM PT

X.com Research Buzz

NLP
LLMs can't jump
X.com
3706

LLMs can't jump

Valerio Capraro

SITUATIONAL AWARENESS: The Decade Ahead
X.com
3158

SITUATIONAL AWARENESS: The Decade Ahead

Leopold Aschenbrenner

NLP
Inducing language models to assert their own consciousness restores human beliefs and values
X.com
3018

Inducing language models to assert their own consciousness restores human beliefs and values

Junsol Kim, Winnie Street, Roberta Rocca, Diane M. Korngiebel, Adam Waytz, James Evans, Geoff Keeling

Google, University of Chicago, Institute of Philosophy, School of Advanced Study, University of London, University of Washington, Northwestern University, Santa Fe Institute

AlphaXiv Trending

Reasoning
AlphaXiv
193

Qwen3.8-Max: A New Bar for Coding and Cowork

Qwen Team

Robotics
Ego2Robot: Scalable Robot Data Synthesis from Egocentric Human Data
AlphaXiv
41

Ego2Robot: Scalable Robot Data Synthesis from Egocentric Human Data

YW, Ye Wang, Pei Lin, Xiong-Hui Chen

AIM3 Lab, Renmin University of China, Qwen Team, Alibaba Inc, ShanghaiTech University, Beijing Institute for General Artificial Intelligence (BIGAI, Beijing University of Aeronautics and Astronautics

Faster-WAM: Do World Action Models Need Deep Action Modules?
AlphaXiv
40

Faster-WAM: Do World Action Models Need Deep Action Modules?

Liheng Ma, Rui Heng Yang, Zhanguang Zhang

Huawei Noah’s Ark Lab, Huawei Celia Team, Department of Foundation Model, 2012 Labs

Reinforcement Learning
LongHorizon-Harness: Advancing Long-Horizon Agents for Real-World Tasks
AlphaXiv
38

LongHorizon-Harness: Advancing Long-Horizon Agents for Real-World Tasks

Ziyu Ma, Hailang Huang, Shun Zou

DreamX Team, Alibaba Group

Qwen-CUA: Native Computer Use for (almost) Everything
AlphaXiv
24

Qwen-CUA: Native Computer Use for (almost) Everything

Dunjie Lu, Shuai Bai, Tianyi Bai

Qwen Team & Xlang Lab

Machine Learning
SKT: Skill-Use Training at Scale via Verified Synthetic Data Generation
AlphaXiv
23

SKT: Skill-Use Training at Scale via Verified Synthetic Data Generation

Zelin Tan, Yiqun Zhang, Hao Li

HuggingFace Daily Papers

RestoreKV: Recovering Full-Cache Behavior Under Aggressive Query-Agnostic KV Cache Eviction
HuggingFace
5

RestoreKV: Recovering Full-Cache Behavior Under Aggressive Query-Agnostic KV Cache Eviction

Changwoo Baek, Seungjun Shin, Kyeongbo Kong

NLP
ARCHead: Activation-Metric Residual Correction for Large Language Model Output Heads
HuggingFace
2

ARCHead: Activation-Metric Residual Correction for Large Language Model Output Heads

Şuayp Talha Kocabay, Talha Rüzgar Akkuş, Kamer Ali Yuksel

Retrieval
LegalPincite: Multi-level Legal Information Retrieval Dataset
HuggingFace
2

LegalPincite: Multi-level Legal Information Retrieval Dataset

Theresia Veronika Rampisela, Henrik Palmer Olsen, Giovanni Colavizza

NLP
When Many Answers Are Valid, Voting Fails: Symbolic Verification for Best-of-K Causal Reasoning in LLMs
HuggingFace
1

When Many Answers Are Valid, Voting Fails: Symbolic Verification for Best-of-K Causal Reasoning in LLMs

Omatharv Bharat Vaidya, Connor Thomas Jerzak, Zayne Rea Sprague, Fangcong Yin, Nhat Ho

Jerzak Labs

Reasoning
ReflectRL: Learning from Golden Negative Trajectories via Reflective-to-Direct Reasoning
HuggingFace
1

ReflectRL: Learning from Golden Negative Trajectories via Reflective-to-Direct Reasoning

Jinhe Bi, Chennan Zhou, Zengjie Jin, Aniri, Shuo Lu

National University of Singapore

Computer Vision
CURV: Enhancing Chart Understanding Through Curriculum Visual Grounded Reasoning
HuggingFace
0

CURV: Enhancing Chart Understanding Through Curriculum Visual Grounded Reasoning

Xuehang Guo, Pingyue Zhang, Ruiyi Zhang, Zhenhailong Wang, Hanrui Lyu