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

Mar 12

Submitted by

Lingaaaaaaa

OpenClaw-RL: Train Any Agent Simply by Talking

princeton-ai

Princeton AI Lab

Submitted by

xihc-ucb

Flash-KMeans: Fast and Memory-Efficient Exact K-Means

Berkeley

Submitted by

KangsanKim71

MA-EgoQA: Question Answering over Egocentric Videos from Multiple Embodied Agents

kaist-ai

Submitted by

q-rz

ReMix: Reinforcement routing for mixtures of LoRAs in LLM finetuning

metaresearch

Submitted by

jiyeonkim

Can Large Language Models Keep Up? Benchmarking Online Adaptation to Continual Knowledge Streams

kaist-ai

Submitted by

xyingzhang

RetroAgent: From Solving to Evolving via Retrospective Dual Intrinsic Feedback

Submitted by

TongkunGuan

CodePercept: Code-Grounded Visual STEM Perception for MLLMs

Qwen

Submitted by

YuyaoGe

Prism-Δ: Differential Subspace Steering for Prompt Highlighting in Large Language Models

ict-cas

Institute of Computing Technology, Chinese Academy of Sciences

Submitted by

Luckyyy

Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning

·
9 authors

Submitted by

taesiri

Just-in-Time: Training-Free Spatial Acceleration for Diffusion Transformers

·
3 authors

Submitted by

shwu22

RbtAct: Rebuttal as Supervision for Actionable Review Feedback Generation

yale-nlp

Submitted by

Yiran0924

In-Context Reinforcement Learning for Tool Use in Large Language Models

·
7 authors

Submitted by

taesiri

V_{0.5}: Generalist Value Model as a Prior for Sparse RL Rollouts

meituan-longcat

Submitted by

genjib

V2M-Zero: Zero-Pair Time-Aligned Video-to-Music Generation

adobe-research

Submitted by

taesiri

CLIPO: Contrastive Learning in Policy Optimization Generalizes RLVR

·
9 authors

Submitted by

taesiri

Code-Space Response Oracles: Generating Interpretable Multi-Agent Policies with Large Language Models

deepmind

Submitted by

Harry4582

EmboAlign: Aligning Video Generation with Compositional Constraints for Zero-Shot Manipulation

northwestern-university

Northwestern University

1

Submitted by

zhiyuandaily

Any to Full: Prompting Depth Anything for Depth Completion in One Stage

·
7 authors

1

Submitted by

parishadbehnam

LLM2Vec-Gen: Generative Embeddings from Large Language Models

McGill-NLP

McGill NLP Group