让大模型在推理时自动分身并行处理，更高效解题。
General learned delegation by clones
- 用强化学习训练模型自动创建相同权重的并行分支
- 在数学推理和长文本问答上提升准确率且节省计算资源
- 适合需要高效推理的复杂任务应用
前沿语言模型通过增加推理时计算量可提升性能，但串行推理或无协调的并行采样在固定推理预算下效率低下。我们提出SELFCEST，通过智能体强化学习使基础模型能在独立并行上下文中生成相同权重的克隆体。训练采用端到端方式，在共享参数回放下以全局任务奖励优化，得到一个能分配生成与上下文资源的控制器。在多个高难度数学推理基准和长上下文多跳问答任务中，SELFCEST在匹配推理预算条件下，优于单体基线模型的准确率-成本权衡表现，并在两个领域均展现出分布外泛化能力。
原文摘要 · Abstract (English)
Frontier language models improve with additional test-time computation, but serial reasoning or uncoordinated parallel sampling can be compute-inefficient under fixed inference budgets. We propose SELFCEST, which equips a base model with the ability to spawn same-weight clones in separate parallel contexts by agentic reinforcement learning. Training is end-to-end under a global task reward with shared-parameter rollouts, yielding a learned controller that allocates both generation and context budget across branches. Across challenging math reasoning benchmarks and long-context multi-hop QA, SELFCEST improves the accuracy-cost Pareto frontier relative to monolithic baselines at matched inference budget, and exhibits out-of-distribution generalization in both domains.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。