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 9

Submitted by

Mystery

On the Entropy Dynamics in Reinforcement Fine-Tuning of Large Language Models

·
7 authors

Submitted by

amphora

Judging What We Cannot Solve: A Consequence-Based Approach for Oracle-Free Evaluation of Research-Level Math

·
8 authors

Submitted by

master-lan

Self-Improving Multilingual Long Reasoning via Translation-Reasoning Integrated Training

·
9 authors

1

Submitted by

fairyang

Baichuan-M3: Modeling Clinical Inquiry for Reliable Medical Decision-Making

baichuan-inc

Baichuan Intelligent Technology

Submitted by

lgy0404

MemGUI-Bench: Benchmarking Memory of Mobile GUI Agents in Dynamic Environments

zju

Zhejiang University

Submitted by

xufangzhi

OdysseyArena: Benchmarking Large Language Models For Long-Horizon, Active and Inductive Interactions

·
19 authors

1

Submitted by

taesiri

InftyThink+: Effective and Efficient Infinite-Horizon Reasoning via Reinforcement Learning

·
10 authors

Submitted by

taesiri

DreamDojo: A Generalist Robot World Model from Large-Scale Human Videos

nvidia

Submitted by

Duke-de-Artois

PlanViz: Evaluating Planning-Oriented Image Generation and Editing for Computer-Use Tasks

·
10 authors

Submitted by

taesiri

POINTS-GUI-G: GUI-Grounding Journey

·
10 authors

Submitted by

LiPengyi29

Back to Basics: Revisiting Exploration in Reinforcement Learning for LLM Reasoning via Generative Probabilities

·
4 authors

Submitted by

lx865712528

MSign: An Optimizer Preventing Training Instability in Large Language Models via Stable Rank Restoration

microsoft

Submitted by

fmmarkmq

SEMA: Simple yet Effective Learning for Multi-Turn Jailbreak Attacks

microsoft

Submitted by

d7chong

RaBiT: Residual-Aware Binarization Training for Accurate and Efficient LLMs

SamsungResearch

Samsung Research

1

Submitted by

FengTing

Revisiting the Shape Convention of Transformer Language Models

MediaTek-Research

MediaTek Research

Submitted by

taesiri

EgoAVU: Egocentric Audio-Visual Understanding

·
12 authors

Submitted by

JingzeShi

OmniMoE: An Efficient MoE by Orchestrating Atomic Experts at Scale

BAAI

Beijing Academy of Artificial Intelligence

Submitted by

leon0923

Avoiding Premature Collapse: Adaptive Annealing for Entropy-Regularized Structural Inference

·
1 authors

1

Submitted by

taesiri

SeeUPO: Sequence-Level Agentic-RL with Convergence Guarantees

Tongyi-MAI

Submitted by

Jinyang23

Exploring Knowledge Purification in Multi-Teacher Knowledge Distillation for LLMs

·
8 authors