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

Feb 12

Submitted by

taesiri

Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters

stepfun-ai

Submitted by

mickyhimself

GENIUS: Generative Fluid Intelligence Evaluation Suite

·
11 authors

Submitted by

txiong23

PhyCritic: Multimodal Critic Models for Physical AI

nvidia

Submitted by

wangyoujin

ASA: Training-Free Representation Engineering for Tool-Calling Agents

·
9 authors

Submitted by

taesiri

When to Memorize and When to Stop: Gated Recurrent Memory for Long-Context Reasoning

ByteDance-Seed

Submitted by

taesiri

Towards Autonomous Mathematics Research

google

Submitted by

yaolily

TimeChat-Captioner: Scripting Multi-Scene Videos with Time-Aware and Structural Audio-Visual Captions

PekingUniversity

Peking University

Submitted by

ZBoyn

G-LNS: Generative Large Neighborhood Search for LLM-Based Automatic Heuristic Design

·
3 authors

Submitted by

Abyssaledge

FeatureBench: Benchmarking Agentic Coding for Complex Feature Development

·
12 authors

Submitted by

yichengchen24

DataChef: Cooking Up Optimal Data Recipes for LLM Adaptation via Reinforcement Learning

·
5 authors

Submitted by

ammarali32

ROCKET: Rapid Optimization via Calibration-guided Knapsack Enhanced Truncation for Efficient Model Compression

MTSAIR

Submitted by

YuZeng260

Internalizing Meta-Experience into Memory for Guided Reinforcement Learning in Large Language Models

·
10 authors

Submitted by

ChengqianMa

Ex-Omni: Enabling 3D Facial Animation Generation for Omni-modal Large Language Models

·
4 authors

Submitted by

taesiri

CLI-Gym: Scalable CLI Task Generation via Agentic Environment Inversion

·
7 authors

Submitted by

langfeng01

Online Causal Kalman Filtering for Stable and Effective Policy Optimization

NanyangTechnologicalUniversity

Nanyang Technological University

Submitted by

zhaors00

Stroke3D: Lifting 2D strokes into rigged 3D model via latent diffusion models

Zhejiang University

Submitted by

huxueyu

EcoGym: Evaluating LLMs for Long-Horizon Plan-and-Execute in Interactive Economies

·
16 authors

Submitted by

Hiiamein

QP-OneModel: A Unified Generative LLM for Multi-Task Query Understanding in Xiaohongshu Search

·
12 authors

Submitted by

Awiny

When the Prompt Becomes Visual: Vision-Centric Jailbreak Attacks for Large Image Editing Models

CSU-JPG

Submitted by

dakopi

Data Repetition Beats Data Scaling in Long-CoT Supervised Fine-Tuning

·
4 authors

Submitted by

0xFarzad

Benchmarking Large Language Models for Knowledge Graph Validation

FactCheck-AI

Submitted by

djalexj

Blockwise Advantage Estimation for Multi-Objective RL with Verifiable Rewards

nebius

1

Submitted by

hyunseoki

Beyond Correctness: Learning Robust Reasoning via Transfer

·
4 authors

1

Submitted by

libertywing

Free(): Learning to Forget in Malloc-Only Reasoning Models

tencent

Submitted by

djstrong

Bielik Guard: Efficient Polish Language Safety Classifiers for LLM Content Moderation

speakleash

SpeakLeash | Spichlerz

Submitted by

ymyy307

ArcFlow: Unleashing 2-Step Text-to-Image Generation via High-Precision Non-Linear Flow Distillation

Fudan-University

Fudan University

Submitted by

XianyangLiu

AgenticPay: A Multi-Agent LLM Negotiation System for Buyer-Seller Transactions

Berkeley

Submitted by

GMFTBY

UMEM: Unified Memory Extraction and Management Framework for Generalizable Memory

AIDC-AI

Submitted by

xuefeng-xu

FedPS: Federated data Preprocessing via aggregated Statistics

·
2 authors

Submitted by

taesiri

Spend Search Where It Pays: Value-Guided Structured Sampling and Optimization for Generative Recommendation

·
7 authors

Submitted by

nnnyt

When Actions Go Off-Task: Detecting and Correcting Misaligned Actions in Computer-Use Agents

osunlp

Submitted by

hlzhang109

Weight Decay Improves Language Model Plasticity

·
4 authors

Submitted by

woorkhaarder

GoodVibe: Security-by-Vibe for LLM-Based Code Generation

is-tuda

Technical University of Darmstadt - Information Systems

1

Submitted by

woorkhaarder

Large Language Lobotomy: Jailbreaking Mixture-of-Experts via Expert Silencing

RadboudUniversity

Radboud University Nijmegen