English
Related papers

Related papers: On-Policy Replay for Continual Supervised Fine-Tun…

200 papers

A prevailing view holds that supervised fine-tuning (SFT) memorizes training data and fails to generalize, whereas reinforcement learning (RL) attains broader robustness. We revisit this claim through a systematic evaluation on two…

Machine Learning · Computer Science 2025-10-02 Xiaofeng Lin , Hejian Sang , Zhipeng Wang , Xuezhou Zhang

On-policy reinforcement learning (RL) algorithms are typically characterized as algorithms that perform policy updates using i.i.d. trajectories collected by the agent's current policy. However, after observing only a finite number of…

Machine Learning · Computer Science 2026-02-11 Nicholas E. Corrado , Josiah P. Hanna

Post-training methods, especially Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), play an important role in improving large language models' (LLMs) complex reasoning abilities. However, the dominant two-stage pipeline (SFT…

Machine Learning · Computer Science 2025-12-22 Mingyu Su , Jian Guan , Yuxian Gu , Minlie Huang , Hongning Wang

Reinforcement Learning from Human Feedback (RLHF) allows us to train models, such as language models (LMs), to follow complex human preferences. In RLHF for LMs, we first train an LM using supervised fine-tuning, sample pairs of responses,…

Machine Learning · Computer Science 2025-07-22 Johannes Ackermann , Takashi Ishida , Masashi Sugiyama

Existing approaches for aligning large language models with human preferences face a trade-off that requires a separate reward model (RM) for on-policy learning. In this paper, we present a novel alignment framework, SELF-JUDGE that (1)…

Machine Learning · Computer Science 2024-06-26 Sangkyu Lee , Sungdong Kim , Ashkan Yousefpour , Minjoon Seo , Kang Min Yoo , Youngjae Yu

Deep reinforcement learning has been able to solve various tasks successfully, however, due to the construction of policy gradient and training dynamics, tuning deep reinforcement learning models remains challenging. As one of the most…

Machine Learning · Computer Science 2026-02-11 Hanyong Wang , Menglong Yang

Post-training processes are essential phases in grounding pre-trained language models to real-world tasks, with learning from demonstrations or preference signals playing a crucial role in this adaptation. We present a unified theoretical…

Machine Learning · Computer Science 2025-07-08 Bo Wang , Qinyuan Cheng , Runyu Peng , Rong Bao , Peiji Li , Qipeng Guo , Linyang Li , Zhiyuan Zeng , Yunhua Zhou , Xipeng Qiu

On-policy distillation (OPD) is a standard tool for transferring teacher behavior to a smaller student, but it implicitly assumes that teacher and student predictions are comparable token by token, an assumption that fails whenever the two…

Computation and Language · Computer Science 2026-05-22 Jie Sun , Mao Zheng , Mingyang Song , Qiyong Zhong , Yilin Cheng , Bichuan Feng , Pengfei Liu , Junfeng Fang , Xiang Wang

On-policy distillation has recently emerged as a promising alternative to standard sequence-level imitation, training a student by scoring its own rollouts with a teacher model. However, we observe ``Off-policy Teacher Decay'' problem in…

Machine Learning · Computer Science 2026-05-27 Zhou Ziheng , Jiaqi Li , Huacong Tang , Ying Nian Wu , Demetri Terzopoulos

Offline reinforcement learning aims to utilize datasets of previously gathered environment-action interaction records to learn a policy without access to the real environment. Recent work has shown that offline reinforcement learning can be…

Machine Learning · Computer Science 2023-08-30 Hanhan Zhou , Tian Lan , Vaneet Aggarwal

Many real-world continuous control problems are in the dilemma of weighing the pros and cons, multi-objective reinforcement learning (MORL) serves as a generic framework of learning control policies for different preferences over…

Machine Learning · Computer Science 2024-07-26 Wei Hung , Bo-Kai Huang , Ping-Chun Hsieh , Xi Liu

Recent advance in deep offline reinforcement learning (RL) has made it possible to train strong robotic agents from offline datasets. However, depending on the quality of the trained agents and the application being considered, it is often…

Robotics · Computer Science 2021-11-02 Seunghyun Lee , Younggyo Seo , Kimin Lee , Pieter Abbeel , Jinwoo Shin

Fine-tuning large language models (LLMs) using zeroth-order optimization (ZO) offers a memory-efficient alternative to gradient-based methods but suffers from slower convergence and unstable optimization due to noisy gradient estimates.…

Machine Learning · Computer Science 2025-06-24 Jikai Long , Zijian Hu , Xiaodong Yu , Jianwen Xie , Zhaozhuo Xu

Pre-trained Large Language Model (LLM) exhibits broad capabilities, yet, for specific tasks or domains their attainment of higher accuracy and more reliable reasoning generally depends on post-training through Supervised Fine-Tuning (SFT)…

Artificial Intelligence · Computer Science 2026-03-17 Haitao Jiang , Wenbo Zhang , Jiarui Yao , Hengrui Cai , Sheng Wang , Rui Song

Safety alignment often improves robustness to harmful queries at the cost of reasoning ability, a tradeoff known as the safety tax. A common cause is distributional mismatch: supervised fine-tuning trains the target model on safety…

Machine Learning · Computer Science 2026-05-19 Yu Fu , Longxuan Yu , Haz Sameen Shahgir , Zhipeng Wei , Hui Liu , N. Benjamin Erichson , Yue Dong

With the rapid advancement of large language models (LLMs) technologies, their application in the domain of autonomous driving has become increasingly widespread. However, existing methods suffer from unstructured reasoning, poor…

Artificial Intelligence · Computer Science 2026-01-09 Chang Zhao , Zheming Yang , Yunqing Hu , Qi Guo , Zijian Wang , Pengcheng Li , Wen Ji

Supervised fine-tuning (SFT) has emerged as one of the most effective ways to improve the performance of large language models (LLMs) in downstream tasks. However, SFT can have difficulty generalizing when the underlying data distribution…

Computation and Language · Computer Science 2025-12-15 Mrinal Rawat , Arkajyoti Chakraborty , Neha Gupta , Roberto Pieraccini

On-policy distillation (OPD) has recently emerged as an effective post-training paradigm for consolidating the capabilities of specialized expert models into a single student model. Despite its empirical success, the conditions under which…

Many existing studies have achieved significant improvements in the reasoning capabilities of large language models (LLMs) through reinforcement learning with verifiable rewards (RLVR), while the enhancement of reasoning abilities in small…

Machine Learning · Computer Science 2025-08-26 Zhong Guan , Likang Wu , Hongke Zhao , Jiahui Wang , Le Wu

Large language models are often post-trained with sparse verifier rewards, which indicate whether a sampled trajectory succeeds but provide limited guidance about where reasoning succeeds or fails. On-policy distillation (OPD) offers denser…

Machine Learning · Computer Science 2026-05-14 Weichen Yu , Xiaomin Li , Yizhou Zhao , Xiaoze Liu , Ruowang Zhang , Haixin Wang , Yinyi Luo , Chen Henry Wu , Gaurav Mittal , Matt Fredrikson , Yu Hu