15 papers
SkillScaffold: Scaffolding Agentic Reinforcement Learning with Policy-Adaptive Skill Credit Assignment and Selective Injection
Wenlin Zhong, Hao Wang, Kun Yuan +6
Agentic reinforcement learning (RL) enables large language models (LLMs) to learn from multi-turn interactions with verifiable outcomes. Textual skills can improve exploration by d…
AsymSpec: Context-Asymmetric Speculative Decoding for Agentic LLMs
Sheng Liang, Yongyue Zhang, Nathanael Brian +4
Agentic LLM pipelines face escalating inference costs as context accumulates across retrieval, tool use, and multi-turn interactions. To control latency, deployments routinely comp…
AI Learning and Conceptual Transfer in the Game of Hidden Rules
Christo Mathew, Wentian Wang, Jacob Feldman +4
This report summarizes the work conducted on the Game of Hidden Rules (GOHR), focusing on reinforcement learning agents trained to infer hidden rules from trial-and-error feedback,…
Interpretable GOHR Agents via Sparse Autoencoders
Shiwei Tan, Yusong Zhao, Weiyi Qin +6
A central challenge in interpreting learned decision-making systems is to determine whether their internal representations contain concepts that help explain their behavior. We rep…
Cross-Tokenizer On-Policy Distillation via Byte-Prefix Marginalization
Hao Wang, Kun Yuan, Wenlin Zhong +4
Open-weight language models from different families exhibit complementary capabilities, motivating their consolidation into a compact student through on-policy distillation (OPD).…
ClimAgent: LLM as Agents for Autonomous Open-ended Climate Science Analysis
Hao Wang, Jindong Han, Wei Fan +1
Climate research is pivotal for mitigating global environmental crises, yet the accelerating volume of multi-scale datasets and the complexity of analytical tools have created sign…
Incentivizing Parametric Knowledge via Reinforcement Learning with Verifiable Rewards for Cross-Cultural Entity Translation
Jiang Zhou, Xiaohu Zhao, Xinwei Wu +8
Cross-cultural entity translation remains challenging for large language models (LLMs) as literal or phonetic renderings are usually yielded instead of culturally appropriate trans…
Rethinking the Necessity of Adaptive Retrieval-Augmented Generation through the Lens of Adaptive Listwise Ranking
Jun Feng, Jiahui Tang, Zhicheng He +5
Adaptive Retrieval-Augmented Generation aims to mitigate the interference of extraneous noise by dynamically determining the necessity of retrieving supplementary passages. However…
IE as Cache: Information Extraction Enhanced Agentic Reasoning
Hang Lv, Sheng Liang, Hongchao Gu +5
Information Extraction aims to distill structured, decision-relevant information from unstructured text, serving as a foundation for downstream understanding and reasoning. However…
When More Thinking Hurts: Overthinking in LLM Test-Time Compute Scaling
Shu Zhou, Rui Ling, Junan Chen +3
Scaling test-time compute through extended chains of thought has become a dominant paradigm for improving large language model reasoning. However, existing research implicitly assu…
Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents
Hao Wang, Guozhi Wang, Han Xiao +8
Reinforcement learning (RL) has been widely used to train LLM agents for multi-turn interactive tasks, but its sample efficiency is severely limited by sparse rewards and long hori…
Learning from Emptiness: De-biasing Listwise Rerankers with Content-Agnostic Probability Calibration
Hang Lv, Hongchao Gu, Ruiqing Yang +5
Generative listwise reranking leverages global context for superior retrieval but is plagued by intrinsic position bias, where models exhibit structural sensitivity to input order…
Compressing Sequences in the Latent Embedding Space: -Token Merging for Large Language Models
Zihao Xu, John Harvill, Ziwei Fan +3
Large Language Models (LLMs) incur significant computational and memory costs when processing long prompts, as full self-attention scales quadratically with input length. Token com…
Toward a Metrology for Artificial Intelligence: Hidden-Rule Environments and Reinforcement Learning
Christo Mathew, Wentian Wang, Jacob Feldman +4
We investigate reinforcement learning in the Game Of Hidden Rules (GOHR) environment, a complex puzzle in which an agent must infer and execute hidden rules to clear a 66 b…
MMLU-SR: A Benchmark for Stress-Testing Reasoning Capability of Large Language Models
Wentian Wang, Sarthak Jain, Paul Kantor +3
We propose MMLU-SR, a novel dataset designed to measure the true comprehension abilities of Large Language Models (LLMs) by challenging their performance in question-answering task…