10 papers
OPDArena: Understanding Learning and Forgetting in On-Policy Distillation
Yuanze Xu, Tianzhu Ye, Hui Sun +3
On-policy distillation (OPD) is increasingly used for language model post-training, yet how teacher supervision shapes learning and forgetting remains unclear. We present OPDArena,…
Learning to Coach for Experiential Learning
Guanheng Chen, Tianzhu Ye, Li Dong +3
Language models can learn from experience, but raw solution trajectories are often too long and noisy to provide effective guidance. In this work, we propose Learning to Coach (L2C…
LLM-as-a-Coach: Experiential Learning for Non-Verifiable Tasks
Tianzhu Ye, Li Dong, Guanheng Chen +4
Reinforcement learning (RL) on open-ended tasks compresses an LLM's rubric-based evaluation into a scalar reward, discarding rich textual feedback and conflating responses with dis…
Universal YOCO for Efficient Depth Scaling
Yutao Sun, Li Dong, Tianzhu Ye +3
The rise of test-time scaling has remarkably boosted the reasoning and agentic proficiency of Large Language Models (LLMs). Yet, standard Transformers struggle to scale inference-t…
On-Policy Context Distillation for Language Models
Tianzhu Ye, Li Dong, Xun Wu +2
Context distillation enables language models to internalize in-context knowledge into their parameters. In our work, we propose On-Policy Context Distillation (OPCD), a framework t…
Online Experiential Learning for Language Models
Tianzhu Ye, Li Dong, Qingxiu Dong +3
The prevailing paradigm for improving large language models relies on offline training with human annotations or simulated environments, leaving the rich experience accumulated dur…
Black-Box On-Policy Distillation of Large Language Models
Tianzhu Ye, Li Dong, Zewen Chi +3
Black-box distillation creates student large language models (LLMs) by learning from a proprietary teacher model's text outputs alone, without access to its internal logits or para…
SeerAttention-R: Sparse Attention Adaptation for Long Reasoning
Yizhao Gao, Shuming Guo, Shijie Cao +12
We introduce SeerAttention-R, a sparse attention framework specifically tailored for the long decoding of reasoning models. Extended from SeerAttention, SeerAttention-R retains the…
Reinforcement Pre-Training
Qingxiu Dong, Li Dong, Yao Tang +4
In this work, we introduce Reinforcement Pre-Training (RPT) as a new scaling paradigm for large language models and reinforcement learning (RL). Specifically, we reframe next-token…
Rectified Sparse Attention
Yutao Sun, Tianzhu Ye, Li Dong +6
Efficient long-sequence generation is a critical challenge for Large Language Models. While recent sparse decoding methods improve efficiency, they suffer from KV cache misalignmen…