2 citations · 2 across the 16 of their papers we have counts for
19 papers
Learning to Reason with Selective Internal Feedback
Xun Lu, Shumeng Yang, Xinrui Wang +3
Reinforcement learning from internal feedback can improve language models without ground-truth answers or external verifiers. However, confidence and correctness can diverge during…
MemChain: Learning Interpretable Memory Traces for Memory-Augmented LLM Agents
Yiwen Ma, Songjun Tu, Qichao Zhang +3
Memory-augmented LLM agents typically answer queries by retrieving relevant memories and feeding them directly to an answer model. This retrieval-as-evidence paradigm assumes retri…
UCOB: Learning to Utilize and Evolve Agentic Skills via Credit-Aware On-Policy Bidirectional Self-Distillation
Songjun Tu, Chengdong Xu, Qichao Zhang +6
Skill memories can improve agentic reinforcement learning by reusing past experience as textual guidance, but retrieved skills are not oracular: they may help in one state while mi…
Are Full Rollouts Necessary for On-Policy Distillation?
Yaocheng Zhang, Jiajun Chai, Yuqian Fu +7
On-policy distillation (OPD) provides dense teacher feedback along student-generated rollouts rather than fixed teacher traces and has emerged as a promising post-training paradigm…
One LR Doesn't Fit All: Heavy-Tail Guided Layerwise Learning Rates for LLMs
Di He, Songjun Tu, Keyu Wang +2
Learning rate configuration is a fundamental aspect of modern deep learning. The prevailing practice of applying a uniform learning rate across all layers overlooks the structural…
-Play: Multi-Agent Self-Play via Privileged Self-Distillation without External Data
Yaocheng Zhang, Yuanheng Zhu, Wenyue Chong +7
Deep search agents have emerged as a promising paradigm for addressing complex information-seeking tasks, but their training remains challenging due to sparse rewards, weak credit…
STRIDE: A Self-Reflective Agent Framework for Reliable Automatic Equation Discovery
Jiarui Su, Songjun Tu, Bei Sun +1
LLM-based equation discovery offers a promising route to recovering symbolic laws from data, but many systems still rely on generation-centered loops that propose candidates, fit p…
AutoSearch: Adaptive Search Depth for Efficient Agentic RAG via Reinforcement Learning
Jingbo Sun, Wenyue Chong, Songjun Tu +7
Agentic retrieval-augmented generation (RAG) systems enable large language models (LLMs) to solve complex tasks through multi-step interaction with external retrieval tools. Howeve…
Saliency-Guided Representation with Consistency Policy Learning for Visual Unsupervised Reinforcement Learning
Jingbo Sun, Qichao Zhang, Songjun Tu +5
Zero-shot unsupervised reinforcement learning (URL) offers a promising direction for building generalist agents capable of generalizing to unseen tasks without additional supervisi…
Dynamic Dual-Granularity Skill Bank for Agentic RL
Songjun Tu, Chengdong Xu, Qichao Zhang +5
Agentic RL can benefit substantially from reusable experience, yet existing skill-based methods mainly extract trajectory-level guidance and often lack principled mechanisms for ma…
PaperAudit-Bench: Benchmarking Error Detection in Research Papers for Critical Automated Peer Review
Songjun Tu, Yiwen Ma, Jiahao Lin +6
Large language models (LLMs) can generate fluent peer reviews but often miss sparse, subtle, and document-level scientific errors. We introduce PaperAudit-Bench, a benchmark for ev…
Perception-Consistency Multimodal Large Language Models Reasoning via Caption-Regularized Policy Optimization
Songjun Tu, Qichao Zhang, Jingbo Sun +6
While multimodal large language models excel at tasks that integrate visual perception with symbolic reasoning, their performance is often undermined by a critical vulnerability: p…
SRFT: A Single-Stage Method with Supervised and Reinforcement Fine-Tuning for Reasoning
Yuqian Fu, Tinghong Chen, Jiajun Chai +7
Large language models (LLMs) have achieved remarkable progress in reasoning tasks, yet the optimal integration of Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) remai…
AlphaDecay: Module-wise Weight Decay for Heavy-Tailed Balancing in LLMs
Di He, Songjun Tu, Ajay Jaiswal +4
Weight decay is a standard regularization technique for training large language models (LLMs). While it is common to assign a uniform decay rate to every layer, this approach overl…
Learning When to Think: Shaping Adaptive Reasoning in R1-Style Models via Multi-Stage RL
Songjun Tu, Jiahao Lin, Qichao Zhang +4
Large reasoning models (LRMs) are proficient at generating explicit, step-by-step reasoning sequences before producing final answers. However, such detailed reasoning can introduce…
Enhancing LLM Reasoning with Iterative DPO: A Comprehensive Empirical Investigation
Songjun Tu, Jiahao Lin, Xiangyu Tian +8
Recent advancements in post-training methodologies for large language models (LLMs) have highlighted reinforcement learning (RL) as a critical component for enhancing reasoning. Ho…
Salience-Invariant Consistent Policy Learning for Generalization in Visual Reinforcement Learning
Jingbo Sun, Songjun Tu, Qichao Zhang +2
Generalizing policies to unseen scenarios remains a critical challenge in visual reinforcement learning, where agents often overfit to the specific visual observations of the train…
Online Preference-based Reinforcement Learning with Self-augmented Feedback from Large Language Model
Songjun Tu, Jingbo Sun, Qichao Zhang +2
Preference-based reinforcement learning (PbRL) provides a powerful paradigm to avoid meticulous reward engineering by learning rewards based on human preferences. However, real-tim…
In-Dataset Trajectory Return Regularization for Offline Preference-based Reinforcement Learning
Songjun Tu, Jingbo Sun, Qichao Zhang +4
Offline preference-based reinforcement learning (PbRL) typically operates in two phases: first, use human preferences to learn a reward model and annotate rewards for a reward-free…