Papers115

#GRPO

Multiplex Thinking: Reasoning via Token-wise Branch-and-Merge

The paper introduces Multiplex Thinking, a new way for AI to think by sampling several likely next words at once and blending them into a single super-token.

#Multiplex Thinking#chain-of-thought#continuous token

Rewarding the Rare: Uniqueness-Aware RL for Creative Problem Solving in LLMs

Intermediate

Zhiyuan Hu, Yucheng Wang et al.Jan 13arXiv

The paper fixes a common problem in training AI reasoners: models get stuck using the same favorite solution style and stop exploring new ways to solve problems.

#Uniqueness-Aware Reinforcement Learning#LLM reasoning#strategy clustering

Your Group-Relative Advantage Is Biased

Intermediate

Fengkai Yang, Zherui Chen et al.Jan 13arXiv

Group-based reinforcement learning for reasoning (like GRPO) uses the group's average reward as a baseline, but that makes its 'advantage' estimates biased.

#Reinforcement Learning from Verifier Rewards#GRPO#GSPO

Dr. Zero: Self-Evolving Search Agents without Training Data

Intermediate

Zhenrui Yue, Kartikeya Upasani et al.Jan 11arXiv

Dr. Zero is a pair of AI agents (a Proposer and a Solver) that teach each other to do web-search-based reasoning without any human-written training data.

#Dr. Zero#self-evolution#proposer-solver

X-Coder: Advancing Competitive Programming with Fully Synthetic Tasks, Solutions, and Tests

Intermediate

Jie Wu, Haoling Li et al.Jan 11arXiv

X-Coder shows that models can learn expert-level competitive programming using data that is 100% synthetic—no real contest problems needed.

#competitive programming#synthetic data generation#feature-based synthesis

LSRIF: Logic-Structured Reinforcement Learning for Instruction Following

Intermediate

Qingyu Ren, Qianyu He et al.Jan 10arXiv

Real instructions often have logic like and first-then and if-else and this paper teaches models to notice and obey that logic.

#instruction following#logical structures#parallel constraints

Chaining the Evidence: Robust Reinforcement Learning for Deep Search Agents with Citation-Aware Rubric Rewards

Intermediate

Jiajie Zhang, Xin Lv et al.Jan 9arXiv

The paper fixes a big problem in training web-searching AI: rewarding only the final answer makes agents cut corners and sometimes hallucinate.

#deep search agents#reinforcement learning#rubric rewards

An Empirical Study on Preference Tuning Generalization and Diversity Under Domain Shift

Intermediate

Constantinos Karouzos, Xingwei Tan et al.Jan 9arXiv

Preference tuning teaches language models to act the way people like, but those habits can fall apart when the topic or style changes (domain shift).

#preference tuning#domain shift#supervised fine-tuning

GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization

Intermediate

Shih-Yang Liu, Xin Dong et al.Jan 8arXiv

When a model learns from many rewards at once, a popular method called GRPO can accidentally squash different reward mixes into the same learning signal, which confuses training.

#GDPO#GRPO#multi-reward reinforcement learning

RelayLLM: Efficient Reasoning via Collaborative Decoding

Intermediate

Chengsong Huang, Tong Zheng et al.Jan 8arXiv

RelayLLM lets a small model do the talking and only asks a big model for help on a few, truly hard tokens.

#token-level collaboration#<call>n</call> command#collaborative decoding

SmartSearch: Process Reward-Guided Query Refinement for Search Agents

Intermediate

Tongyu Wen, Guanting Dong et al.Jan 8arXiv

SmartSearch teaches search agents to fix their own bad search queries while they are thinking, not just their final answers.

#Search agents#Process rewards#Query refinement

AgentOCR: Reimagining Agent History via Optical Self-Compression

Intermediate

Lang Feng, Fuchao Yang et al.Jan 8arXiv

AgentOCR turns an agent’s long text history into pictures so it can remember more using fewer tokens.

#AgentOCR#optical self-compression#visual tokens

3 4 5 6 7