32 citations · 93 across the 30 of their papers we have counts for
34 papers
Model What Matters: Decision-Guided Selective World Modeling for Language Agents
Maolin Ran, Minda Chen, Di Huang +3
Reinforcement learning provides a direct mechanism for optimizing task objectives in large language model-driven agents, but task rewards are typically sparse trajectory-level sign…
Past, Future, All at Once: Mitigating Stability-Plasticity Dilemma via Post-hoc JANUS Rectification
Zhilong Zheng, Letian Tao, Yang Guan +7
Fine-tuning foundation models on new tasks inevitably suffer from catastrophic forgetting. While existing works attempt to mitigate this on the basis of parameter-efficient fine-tu…
Beyond Token-Local Imitation: Reward-Compatible Temporal Credit Assignment for On-Policy Distillation
Shiqi Liu, Zeyu He, Letian Tao +9
On-policy distillation (OPD) has emerged as an effective approach for large language model post-training, yet existing objectives face a trade-off between objective fidelity and op…
Coalition-Aware Skill Reliability for Self-Evolving Agents
Qiyan Zhao, Xiaofeng Zhang, Bo Liu +11
Agent skills, structured artifacts distilled from interaction trajectories and dynamically reused from skill banks, have become a central mechanism for enabling large language mode…
Reinforce-Ada: An Adaptive Sampling Framework under Non-linear RL Objectives
Wei Xiong, Chenlu Ye, Baohao Liao +6
Reinforcement learning (RL) for large language model reasoning is frequently hindered by signal loss, a phenomenon where standard uniform sampling with small group sizes fails to u…
StepWiser: Stepwise Generative Judges for Wiser Reasoning
Wei Xiong, Wenting Zhao, Weizhe Yuan +4
As models increasingly leverage multi-step reasoning strategies to solve complex problems, supervising the logical validity of these intermediate steps has become a critical resear…
Optimizing Chain-of-Thought Reasoners via Gradient Variance Minimization in Rejection Sampling and RL
Jiarui Yao, Yifan Hao, Hanning Zhang +4
Chain-of-thought (CoT) reasoning in large language models (LLMs) can be formalized as a latent variable problem, where the model needs to generate intermediate reasoning steps. Whi…
A Minimalist Approach to LLM Reasoning: from Rejection Sampling to Reinforce
Wei Xiong, Jiarui Yao, Yuhui Xu +8
Reinforcement learning (RL) has become a prevailing approach for fine-tuning large language models (LLMs) on complex reasoning tasks. Among recent methods, GRPO stands out for its…
Self-rewarding correction for mathematical reasoning
Wei Xiong, Hanning Zhang, Chenlu Ye +3
We study self-rewarding reasoning large language models (LLMs), which can simultaneously generate step-by-step reasoning and evaluate the correctness of their outputs during the in…
Logarithmic Regret for Online KL-Regularized Reinforcement Learning
Heyang Zhao, Chenlu Ye, Wei Xiong +2
Recent advances in Reinforcement Learning from Human Feedback (RLHF) have shown that KL-regularization plays a pivotal role in improving the efficiency of RL fine-tuning for large…
From Lists to Emojis: How Format Bias Affects Model Alignment
Xuanchang Zhang, Wei Xiong, Lichang Chen +3
In this paper, we study format biases in reinforcement learning from human feedback (RLHF). We observe that many widely-used preference models, including human evaluators, GPT-4, a…
Building Math Agents with Multi-Turn Iterative Preference Learning
Wei Xiong, Chengshuai Shi, Jiaming Shen +10
Recent studies have shown that large language models' (LLMs) mathematical problem-solving capabilities can be enhanced by integrating external tools, such as code interpreters, and…
Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts
Haoxiang Wang, Wei Xiong, Tengyang Xie +2
Reinforcement learning from human feedback (RLHF) has emerged as the primary method for aligning large language models (LLMs) with human preferences. The RLHF process typically sta…
RLHF Workflow: From Reward Modeling to Online RLHF
Hanze Dong, Wei Xiong, Bo Pang +7
We present the workflow of Online Iterative Reinforcement Learning from Human Feedback (RLHF) in this technical report, which is widely reported to outperform its offline counterpa…
Strengthening Multimodal Large Language Model with Bootstrapped Preference Optimization
Renjie Pi, Tianyang Han, Wei Xiong +4
Multimodal Large Language Models (MLLMs) excel in generating responses based on visual inputs. However, they often suffer from a bias towards generating responses similar to their…
DPO Meets PPO: Reinforced Token Optimization for RLHF
Han Zhong, Zikang Shan, Guhao Feng +6
In the classical Reinforcement Learning from Human Feedback (RLHF) framework, Proximal Policy Optimization (PPO) is employed to learn from sparse, sentence-level rewards -- a chall…
Arithmetic Control of LLMs for Diverse User Preferences: Directional Preference Alignment with Multi-Objective Rewards
Haoxiang Wang, Yong Lin, Wei Xiong +5
Fine-grained control over large language models (LLMs) remains a significant challenge, hindering their adaptability to diverse user needs. While Reinforcement Learning from Human…
Online Iterative Reinforcement Learning from Human Feedback with General Preference Model
Chenlu Ye, Wei Xiong, Yuheng Zhang +3
We investigate Reinforcement Learning from Human Feedback (RLHF) in the context of a general preference oracle. In particular, we do not assume the existence of a reward function a…
Iterative Preference Learning from Human Feedback: Bridging Theory and Practice for RLHF under KL-Constraint
Wei Xiong, Hanze Dong, Chenlu Ye +5
This paper studies the alignment process of generative models with Reinforcement Learning from Human Feedback (RLHF). We first identify the primary challenges of existing popular m…
Mitigating the Alignment Tax of RLHF
Yong Lin, Hangyu Lin, Wei Xiong +14
LLMs acquire a wide range of abilities during pre-training, but aligning LLMs under Reinforcement Learning with Human Feedback (RLHF) can lead to forgetting pretrained abilities, w…
Provably Efficient Offline Reinforcement Learning with Perturbed Data Sources
Chengshuai Shi, Wei Xiong, Cong Shen +1
Existing theoretical studies on offline reinforcement learning (RL) mostly consider a dataset sampled directly from the target task. In practice, however, data often come from seve…
LMFlow: An Extensible Toolkit for Finetuning and Inference of Large Foundation Models
Shizhe Diao, Rui Pan, Hanze Dong +4
Foundation models have demonstrated a great ability to achieve general human-level intelligence far beyond traditional approaches. As the technique keeps attracting attention from…
Reward Teaching for Federated Multi-armed Bandits
Chengshuai Shi, Wei Xiong, Cong Shen +1
Most of the existing federated multi-armed bandits (FMAB) designs are based on the presumption that clients will implement the specified design to collaborate with the server. In r…
RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment
Hanze Dong, Wei Xiong, Deepanshu Goyal +7
Generative foundation models are susceptible to implicit biases that can arise from extensive unsupervised training data. Such biases can produce suboptimal samples, skewed outcome…