320亿参数模型通过迭代蒸馏提升智能体推理能力
SAGE-32B: Agentic Reasoning via Iterative Distillation
- 用迭代蒸馏训练，强化任务分解与工具调用能力
- 在多工具场景下成功率达78.3%，超越同类模型
- 适合需要长期规划的自主智能体系统开发
我们展示了SAGE-32B，一个320亿参数的语言模型，专注于智能体推理与长程规划任务。不同于追求对话流畅性的聊天模型，SAGE-32B设计为在智能体循环中运行，强调任务分解、工具使用和错误恢复。模型基于Qwen2.5-32B预训练模型初始化，并通过迭代蒸馏进行微调，该过程通过严格测试的反馈环路提升推理性能。SAGE-32B还引入逆向推理方法，利用元认知头在执行前预测规划过程中的潜在失败。在MMLU-Pro、AgentBench和MATH-500等智能体推理基准测试中，相比同规模基线模型，SAGE-32B在多工具使用场景下取得更高的成功率，同时在标准推理评估中保持竞争力。模型权重已在Hugging Face公开：https://huggingface.co/sagea-ai/sage-reasoning-32b
原文摘要 · Abstract (English)
We demonstrate SAGE-32B, a 32 billion parameter language model that focuses on agentic reasoning and long range planning tasks. Unlike chat models that aim for general conversation fluency, SAGE-32B is designed to operate in an agentic loop, emphasizing task decomposition, tool usage, and error recovery. The model is initialized from the Qwen2.5-32B pretrained model and fine tuned using Iterative Distillation, a two stage training process that improves reasoning performance through rigorously tested feedback loops. SAGE-32B also introduces an inverse reasoning approach, which uses a meta cognition head to forecast potential failures in the planning process before execution. On agentic reasoning benchmarks including MMLU-Pro, AgentBench, and MATH-500, SAGE-32B achieves higher success rates in multi tool usage scenarios compared to similarly sized baseline models, while remaining competitive on standard reasoning evaluations. Model weights are publicly released at https://huggingface.co/sagea-ai/sage-reasoning-32b
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。