new

Get trending papers in your email inbox once a day!

Get trending papers in your email inbox!

Daily Papers

byAK and the research community

Oct 3

Submitted by

YerbaPage

LongCodeZip: Compress Long Context for Code Language Models

Stanford-University

Stanford University

Submitted by

cuijiaxing

Self-Forcing++: Towards Minute-Scale High-Quality Video Generation

ByteDance-Seed

Submitted by

rzzhan

ExGRPO: Learning to Reason from Experience

·
8 authors

Submitted by

yulunliu

StealthAttack: Robust 3D Gaussian Splatting Poisoning via Density-Guided Illusions

NYCU

National Yang Ming Chiao Tung University

Submitted by

yuntian-deng

Interactive Training: Feedback-Driven Neural Network Optimization

yuntian-group

Submitted by

taesiri

StockBench: Can LLM Agents Trade Stocks Profitably In Real-world Markets?

·
7 authors

Submitted by

ahatamiz

RLP: Reinforcement as a Pretraining Objective

nvidia

Submitted by

manu

ModernVBERT: Towards Smaller Visual Document Retrievers

ModernVBERT

2

Submitted by

ruohao

Tree-based Dialogue Reinforced Policy Optimization for Red-Teaming Attacks

·
6 authors

2

Submitted by

weiminwang

Ovi: Twin Backbone Cross-Modal Fusion for Audio-Video Generation

characterai

Submitted by

invokerliang

CLUE: Non-parametric Verification from Experience via Hidden-State Clustering

tencent

Submitted by

therem

The Rogue Scalpel: Activation Steering Compromises LLM Safety

·
6 authors

2

Submitted by

lr10260

VOGUE: Guiding Exploration with Visual Uncertainty Improves Multimodal Reasoning

tencent

2

Submitted by

xw-eric

The Unreasonable Effectiveness of Scaling Agents for Computer Use

simular-ai

Submitted by

songw-zju

RewardMap: Tackling Sparse Rewards in Fine-grained Visual Reasoning via Multi-Stage Reinforcement Learning

WestlakeUniversity

Westlake University

Submitted by

Geralt-Targaryen

F2LLM Technical Report: Matching SOTA Embedding Performance with 6 Million Open-Source Data

codefuse-ai

Submitted by

taesiri

A Rigorous Benchmark with Multidimensional Evaluation for Deep Research Agents: From Answers to Reports

·
12 authors

Submitted by

jacksukk

Learning to Reason for Hallucination Span Detection

·
10 authors

4

Submitted by

zhangchenxu

TOUCAN: Synthesizing 1.5M Tool-Agentic Data from Real-World MCP Environments

ibm

IBM

Submitted by

Shilin-LU

DragFlow: Unleashing DiT Priors with Region Based Supervision for Drag Editing

·
7 authors

2

Submitted by

AdamF92

Sparse Query Attention (SQA): A Computationally Efficient Attention Mechanism with Query Heads Reduction

ReactiveAI

Submitted by

jthomm

Aristotle: IMO-level Automated Theorem Proving

·
22 authors

Submitted by

Enxin

VideoNSA: Native Sparse Attention Scales Video Understanding

·
8 authors

Submitted by

Harold328

Go with Your Gut: Scaling Confidence for Autoregressive Image Generation

·
7 authors

Submitted by

zorik

Fine-Grained Detection of Context-Grounded Hallucinations Using LLMs

Technion

Technion Israel institute of technology

2

Submitted by

YuZeng260

Agentic Jigsaw Interaction Learning for Enhancing Visual Perception and Reasoning in Vision-Language Models

·
12 authors

Submitted by

yxl66666

Visual Multi-Agent System: Mitigating Hallucination Snowballing via Visual Flow

·
11 authors

Submitted by

erjui

Automated Structured Radiology Report Generation with Rich Clinical Context

·
6 authors

Submitted by

enisimsar

Optimal Control Meets Flow Matching: A Principled Route to Multi-Subject Fidelity

·
3 authors

Submitted by

taesiri

Just Do It!? Computer-Use Agents Exhibit Blind Goal-Directedness

microsoft

3

Submitted by

SteveZeyuZhang

VLA-R1: Enhancing Reasoning in Vision-Language-Action Models

·
6 authors

Submitted by

Ksgk-fy

RLAD: Training LLMs to Discover Abstractions for Solving Reasoning Problems

·
7 authors

Submitted by

tkreiman

Transformers Discover Molecular Structure Without Graph Priors

Berkeley

Submitted by

Wyattz23

TimeSeriesScientist: A General-Purpose AI Agent for Time Series Analysis

·
7 authors

2

Submitted by

SwyWang

VIRTUE: Visual-Interactive Text-Image Universal Embedder

·
5 authors

Submitted by

yanxi-chen

Group-Relative REINFORCE Is Secretly an Off-Policy Algorithm: Demystifying Some Myths About GRPO and Its Friends

·
8 authors

Submitted by

tetrisd

Drawing Conclusions from Draws: Rethinking Preference Semantics in Arena-Style LLM Evaluation

UniversityCollegeLondon

University College London

Submitted by

James-WYang

Parallel Scaling Law: Unveiling Reasoning Generalization through A Cross-Linguistic Perspective

CASIA

Chinese Academic of Science Institute of Automation

Submitted by

FengTing

Rethinking the shape convention of an MLP

MediaTek-Research

MediaTek Research

2

Submitted by

Ksgk-fy

Generalized Parallel Scaling with Interdependent Generations

·
8 authors

Submitted by

Yunzhen

Rethinking Thinking Tokens: LLMs as Improvement Operators

·
9 authors

Submitted by

Yalimu

One-Token Rollout: Guiding Supervised Fine-Tuning of LLMs with Policy Gradient

·
5 authors

2

Submitted by

Xiaoye08

FrameThinker: Learning to Think with Long Videos via Multi-Turn Frame Spotlighting

·
6 authors

Submitted by

taesiri

SKYLENAGE Technical Report: Mathematical Reasoning and Contest-Innovation Benchmarks for Multi-Level Math Evaluation

·
18 authors

Submitted by

patricebechard

Optimizing What Matters: AUC-Driven Learning for Robust Neural Retrieval

ServiceNow-AI

2

Submitted by

zzhao0104

Controlled Generation for Private Synthetic Text

jhu-clsp

Center for Language and Speech Processing @ JHU

2

Submitted by

taesiri

MedQ-Bench: Evaluating and Exploring Medical Image Quality Assessment Abilities in MLLMs

·
20 authors

Submitted by

nandan523

Spectral Scaling Laws in Language Models: How Effectively Do Feed-Forward Networks Use Their Latent Space?

New York University

2

Submitted by

whats2000

SQUARE: Semantic Query-Augmented Fusion and Efficient Batch Reranking for Training-free Zero-Shot Composed Image Retrieval

·
3 authors

2

Submitted by

dinobby

Think Right: Learning to Mitigate Under-Over Thinking via Adaptive, Attentive Compression

·
6 authors

Submitted by

pranamanam

AReUReDi: Annealed Rectified Updates for Refining Discrete Flows with Multi-Objective Guidance

·
3 authors

Submitted by

therem

IoT-MCP: Bridging LLMs and IoT Systems Through Model Context Protocol

·
10 authors