Enemray：面向哈桑尼亚语的能力型语言模型
Enemray: Toward Capable Language Models for Hassaniya
浏览论文内容
中文总结 AI 辅助
Enemray是一个以哈桑尼亚语为中心的语言模型，通过分离语言习得与行为特化的流程，在保留通用能力的同时，实现了最强的英译哈桑尼亚语翻译和最高的毛里塔尼亚翻译错误检测得分。
中文摘要 AI 辅助
我们推出了Enemray，一个以哈桑尼亚语为中心的语言模型，支持在哈桑尼亚语中进行通用交互。Enemray围绕稳定性-可塑性目标进行训练：在获取强大的哈桑尼亚语言和文化能力的同时，保留一个能力型指令微调模型的通用推理、多语言、指令遵循和安全行为。开发流程将语言习得与行为特化分离。一个单独构建的持续预训练语料库提供了对自然哈桑尼亚语和毛里塔尼亚文本的广泛接触；层选择性持续预训练学习一个紧凑的语言特定参数更新；该更新被转移到指令微调参数空间中；监督后训练则发展出对话、文化、文学、任务导向和跨语言行为。监督语料库整合了选定的公开哈桑尼亚语和毛里塔尼亚资源，以及大量新收集、重建、策展和构建的指令数据，而策略生成的回放则从参考模型自身的行为分布中提供保留信号。由此产生的语料库在规模和目的范围上远超现有的哈桑尼亚语文本资源。在评估中，Enemray在比较的开源和专有模型中取得了最强的英语到哈桑尼亚语翻译性能，并在毛里塔尼亚翻译错误检测上获得了最高总分，同时在其指令微调基础模型的数学推理、知识、代码生成和函数调用等通用能力上保持了大部分性能。本报告描述了Enemray的动机、数据构建、模型设计、训练方法和评估。
英文摘要
We introduce Enemray, a Hassaniya-centric language model that enables general-purpose interaction in Hassaniya. Enemray is trained around a stability--plasticity objective: acquire strong Hassaniya linguistic and cultural competence while preserving the general reasoning, multilingual, instruction-following, and safety behaviors of a capable instruction-tuned model. The development pipeline separates language acquisition from behavioral specialization. A separately assembled continual-pretraining corpus provides broad exposure to natural Hassaniya and Mauritanian text; layer-selective continual pretraining learns a compact language-specific parameter update; that update is transferred into the instruction-tuned parameter space; and supervised post-training develops conversational, cultural, literary, task-oriented, and cross-lingual behavior. The supervised corpus integrates selected public Hassaniya and Mauritanian resources with a substantially larger body of newly collected, reconstructed, curated, and constructed instruction data, while policy-generated replay provides a retention signal from the reference model's own behavior distribution. The resulting collection is substantially larger and broader in purpose than existing Hassaniya text resources. In evaluation, Enemray achieves the strongest English to Hassaniya translation among the compared open and proprietary models and the highest overall score on Mauritanian translation error detection, while retaining most of the general capabilities of its instruction-tuned base model on mathematical reasoning, knowledge, code generation, and function calling. This report describes the motivation, data construction, model design, training methodology, and evaluation of Enemray.
