QU-NLP在QIAS 2026中的表现：针对阿拉伯伊斯兰继承推理的多阶段QLoRA微调
QU-NLP at QIAS 2026: Multi-Stage QLoRA Fine-Tuning for Arabic Islamic Inheritance Reasoning
AI总结：
本文提出通过多阶段QLoRA微调策略提升阿拉伯伊斯兰继承推理能力，利用领域适应和任务特定训练，在低资源下实现高精度推理。
摘要：
伊斯兰继承法（ilm al-mawarıth）是一个评估大语言模型结构推理能力的挑战性领域，需要多步骤法律分析、基于规则的决策阻塞和精确的分数计算。我们提出了QU-NLP在QIAS 2026共享任务上的提交。我们的方法在Qwen3-4B上采用多阶段量化低秩适应（QLoRA）微调策略：（1）在3,166条伊斯兰fatwa记录上进行领域适应以获取继承术语和教法推理模式，随后（2）在12,000个结构化继承案例上进行任务特定训练以优化JSON格式输出生成。使用4位NF4量化和秩128 LoRA适配器，我们的模型在测试集上实现了90%的MIR-E（Mawarith Inheritance Reasoning Evaluation）分数，展示了在最小计算资源下具有竞争力的表现。我们的结果表明，结合领域特定预适应和结构化输出训练，小型语言模型能够有效完成复杂的法律推理任务，与商业系统如Gemini-2.5-flash相比表现优异。
英文摘要：
Islamic inheritance law (ilm al-mawarıth) presents a challenging domain for evaluating large language models' structured reasoning capabilities, requiring multi-step legal analysis, rule-based blocking decisions, and precise fractional calculations. We present QU-NLP's submission to the QIAS 2026 shared task on Arabic Islamic inheritance reasoning. Our approach employs a multi-stage Quantized Low-Rank Adaptation (QLoRA) fine-tuning strategy on Qwen3-4B: (1) domain adaptation on 3,166 Islamic fatwa records to acquire inheritance terminology and jurisprudential reasoning patterns, followed by (2) task-specific training on 12,000 structured inheritance cases to optimize JSON-formatted output generation. Using 4-bit NF4 quantization with rank-128 LoRA adapters, our model achieves 90% MIR-E (Mawarith Inheritance Reasoning Evaluation) score on the test set, demonstrating competitive performance while requiring minimal computational resources. Our results show that domain-specific pre-adaptation combined with structured output training enables small language models to perform complex legal reasoning tasks effectively comparing to commercial systems such as Gemini-2.5-flash.
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。
