无需微调，自动提炼金融推理技能并动态注入。
ASDA: Automated Skill Distillation and Adaptation for Financial Reasoning
- 通过错误修正迭代生成结构化技能文件
- 在FAMMA上提升17.33%算术与5.95%非算术推理
- 适合无模型权重访问的机构做可审计领域适配
将大语言模型（LLMs）适配至专业金融推理通常需昂贵的微调，导致模型锁定专长。训练无关方法虽已出现，但实验表明主流方法（GEPA和ACE）在FAMMA金融推理基准上仅带来微弱提升，暴露了无结构文本优化在复杂多步领域推理中的局限性。我们提出自动化技能蒸馏与适配框架ASDA，通过迭代纠错学习自动生成结构化技能成果，无需修改模型权重。教师模型分析学生模型在金融推理任务中的失败，按子领域和错误类型聚类，并合成包含推理流程、代码模板和示范案例的技能文件，在推理时动态注入。在FAMMA上评估，ASDA在算术推理上实现最高+17.33%提升，非算术推理提升+5.95%，显著优于所有训练无关基线。生成的技能文件具备人类可读性、版本控制能力，且兼容Agent Skills开源标准，使任何拥有标注领域数据集的组织都能在无权重访问或重训练的前提下，获得实用且可审计的领域适配路径。
原文摘要 · Abstract (English)
Adapting large language models (LLMs) to specialized financial reasoning typically requires expensive fine-tuning that produces model-locked expertise. Training-free alternatives have emerged, yet our experiments show that leading methods (GEPA and ACE) achieve only marginal gains on the FAMMA financial reasoning benchmark, exposing the limits of unstructured text optimization for complex, multi-step domain reasoning. We introduce Automated Skill Distillation and Adaptation (ASDA), a framework that automatically generates structured skill artifacts through iterative error-corrective learning without modifying model weights. A teacher model analyzes a student model's failures on financial reasoning tasks, clusters errors by subfield and error type, and synthesizes skill files containing reasoning procedures, code templates, and worked examples, which are dynamically injected during inference. Evaluated on FAMMA, ASDA achieves up to +17.33% improvement on arithmetic reasoning and +5.95% on non-arithmetic reasoning, substantially outperforming all training-free baselines. The resulting skill artifacts are human-readable, version-controlled, and compatible with the Agent Skills open standard, offering any organization with a labeled domain dataset a practical and auditable path to domain adaptation without weight access or retraining.
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。