arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2512.12967cs.CL
收藏

QwenLong-L1.5: 长上下文推理与内存管理的训练配方

QwenLong-L1.5: Post-Training Recipe for Long-Context Reasoning and Memory Management

Weizhou Shen, Ziyi Yang, Chenliang Li, Zhiyuan Lu, Miao Peng, Huashan Sun, Yingcheng Shi, Shengyi Liao, Shaopeng Lai, Bo Zhang, Dayiheng Liu, Fei Huang, Jingren Zhou, Ming Yan

AI总结:

QwenLong-L1.5通过长上下文数据合成、稳定强化学习和内存增强架构,提升长上下文推理能力及超长任务处理性能。

摘要:

我们介绍了QwenLong-L1.5,一种通过系统性训练创新实现卓越长上下文推理能力的模型。QwenLong-L1.5的关键技术突破包括:(1)长上下文数据合成流水线:我们开发了一个系统性合成框架,生成需要多跳支撑的复杂推理任务,通过将文档分解为原子事实及其底层关系,然后编程生成可验证的推理问题,从而生成高质量的训练数据,显著超越简单的检索任务,实现真正的长距离推理能力。(2)稳定强化学习用于长上下文训练:为克服长上下文强化学习中的关键不稳定性,我们引入任务平衡采样与任务特定优势估计以缓解奖励偏差,并提出自适应熵控制策略优化(AEPO)以动态调节探索-利用平衡。(3)内存增强架构用于超长上下文:认识到即使扩展的上下文窗口也无法容纳任意长的序列,我们开发了具有多阶段融合强化学习训练的内存管理框架,无缝整合单次推理与迭代内存处理,适用于超过400万token的任务。基于Qwen3-30B-A3B-Thinking,QwenLong-L1.5在长上下文推理基准测试中表现与GPT-5和Gemini-2.5-Pro相当,在平均上超越基线9.90点。在超长任务(100万~400万token)中,QwenLong-L1.5的内存代理框架相对于代理基线获得9.48点的提升。此外,获得的长上下文推理能力也转化为科学推理、内存工具使用和扩展对话等通用领域的增强性能。

英文摘要:

We introduce QwenLong-L1.5, a model that achieves superior long-context reasoning capabilities through systematic post-training innovations. The key technical breakthroughs of QwenLong-L1.5 are as follows: (1) Long-Context Data Synthesis Pipeline: We develop a systematic synthesis framework that generates challenging reasoning tasks requiring multi-hop grounding over globally distributed evidence. By deconstructing documents into atomic facts and their underlying relationships, and then programmatically composing verifiable reasoning questions, our approach creates high-quality training data at scale, moving substantially beyond simple retrieval tasks to enable genuine long-range reasoning capabilities. (2) Stabilized Reinforcement Learning for Long-Context Training: To overcome the critical instability in long-context RL, we introduce task-balanced sampling with task-specific advantage estimation to mitigate reward bias, and propose Adaptive Entropy-Controlled Policy Optimization (AEPO) that dynamically regulates exploration-exploitation trade-offs. (3) Memory-Augmented Architecture for Ultra-Long Contexts: Recognizing that even extended context windows cannot accommodate arbitrarily long sequences, we develop a memory management framework with multi-stage fusion RL training that seamlessly integrates single-pass reasoning with iterative memory-based processing for tasks exceeding 4M tokens. Based on Qwen3-30B-A3B-Thinking, QwenLong-L1.5 achieves performance comparable to GPT-5 and Gemini-2.5-Pro on long-context reasoning benchmarks, surpassing its baseline by 9.90 points on average. On ultra-long tasks (1M~4M tokens), QwenLong-L1.5's memory-agent framework yields a 9.48-point gain over the agent baseline. Additionally, the acquired long-context reasoning ability translates to enhanced performance in general domains like scientific reasoning, memory tool using, and extended dialogue.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑