Papers906

ERNIE 5.0 Technical Report

ERNIE 5.0 is a single giant model that can read and create text, images, video, and audio by predicting the next pieces step by step, like writing a story one line at a time.

#ERNIE 5.0#unified autoregressive model#mixture-of-experts

WideSeek-R1: Exploring Width Scaling for Broad Information Seeking via Multi-Agent Reinforcement Learning

Intermediate

Zelai Xu, Zhexuan Xu et al.Feb 4arXiv

WideSeek-R1 teaches a small 4B-parameter language model to act like a well-run team: one leader plans, many helpers work in parallel, and everyone learns together with reinforcement learning.

#width scaling#multi-agent reinforcement learning#orchestration

Vibe AIGC: A New Paradigm for Content Generation via Agentic Orchestration

Beginner

Jiaheng Liu, Yuanxing Zhang et al.Feb 4arXiv

This paper says today's content AIs are great at pretty pictures and videos but often miss what people actually want, creating a big Intent-Execution Gap.

#Vibe AIGC#Agentic Orchestration#Meta Planner

LycheeDecode: Accelerating Long-Context LLM Inference via Hybrid-Head Sparse Decoding

Intermediate

Gang Lin, Dongfang Li et al.Feb 4arXiv

Long texts make language models slow because they must keep and re-check a huge memory called the KV cache for every new word they write.

#long-context LLM#sparse attention#head specialization

EgoActor: Grounding Task Planning into Spatial-aware Egocentric Actions for Humanoid Robots via Visual-Language Models

Intermediate

Yu Bai, MingMing Yu et al.Feb 4arXiv

EgoActor is a vision-language model that turns everyday instructions like 'Go to the door and say hi' into step-by-step, egocentric actions a humanoid robot can actually do.

#EgoActing#vision-language model#humanoid robot

Beyond Unimodal Shortcuts: MLLMs as Cross-Modal Reasoners for Grounded Named Entity Recognition

Intermediate

Jinlong Ma, Yu Zhang et al.Feb 4arXiv

The paper teaches multimodal large language models (MLLMs) to stop guessing from just text or just images and instead check both together before answering.

#GMNER#Multimodal Large Language Models#Modality Bias

Agent-Omit: Training Efficient LLM Agents for Adaptive Thought and Observation Omission via Agentic Reinforcement Learning

Intermediate

Yansong Ning, Jun Fang et al.Feb 4arXiv

Agent-Omit teaches AI agents to skip unneeded thinking and old observations, cutting tokens while keeping accuracy high.

#LLM agents#reinforcement learning#agentic RL

Steering LLMs via Scalable Interactive Oversight

Intermediate

Enyu Zhou, Zhiheng Xi et al.Feb 4arXiv

The paper tackles a common problem: people can ask AI to do big, complex tasks, but they can’t always explain exactly what they want or check the results well.

#scalable oversight#interactive alignment#requirement elicitation

Training Data Efficiency in Multimodal Process Reward Models

Intermediate

Jinyuan Li, Chengsong Huang et al.Feb 4arXiv

Multimodal Process Reward Models (MPRMs) teach AI to judge each step of a picture-and-text reasoning process, not just the final answer.

#Multimodal Process Reward Model#Process Supervision#Monte Carlo Annotation

Likelihood-Based Reward Designs for General LLM Reasoning

Beginner

Ariel Kwiatkowski, Natasha Butt et al.Feb 3arXiv

Binary right/wrong rewards for training reasoning in large language models are hard to design and often too sparse to learn from.

#log-likelihood reward#chain-of-thought (CoT)#reinforcement learning for LLMs

VLS: Steering Pretrained Robot Policies via Vision-Language Models

Intermediate

Shuo Liu, Ishneet Sukhvinder Singh et al.Feb 3arXiv

Robots often learn good hand motions during training but get confused when the scene or the instructions change at test time, even a little bit.

#Vision–Language Steering#Inference-time control#Diffusion policy

AgentArk: Distilling Multi-Agent Intelligence into a Single LLM Agent

Intermediate

Yinyi Luo, Yiqiao Jin et al.Feb 3arXiv

AgentArk teaches one language model to think like a whole team of models that debate, so it can solve tough problems quickly without running a long, expensive debate at answer time.

#multi-agent distillation#process reward model#GRPO

1 2 3 4 5