木星-N技术报告
Jupiter-N Technical Report
AI总结：
木星-N通过混合推理模型在Nemotron 3 Super基础上进行微调，提升代理能力、文化对齐和威尔士语支持，同时保留基础模型能力。
摘要：
我们介绍了Jupiter-N，一种基于Nemotron 3 Super的混合推理模型，在完全开源的1200亿参数大语言模型基础上进行微调。我们旨在三个目标：（1）通过不确定性校准的轨迹提升代理能力；（2）通过基于文化规范的合成数据实现UK文化对齐；（3）通过平行语料库和LLM翻译的威尔士语对话实现威尔士语支持。我们的数据校准策略精心保留了基础模型的能力：使用我们的Forget-Me-Not框架，我们混合在线策略的合成回放与离线策略的任务数据以缓解灾难性遗忘，并包含推理和非推理轨迹的混合以维持Nemotron的混合推理能力。Jupiter-N在威尔士语（+18在ARC-Easy，+5.25在MMLU-Lite）、终端使用（+9.1在Terminal Bench 2）和指令遵循（+4.4在IFBench）方面均优于Nemotron，同时保留基础模型能力。我们将这项工作框架为一个可重复的主权微调模板：替换文化知识、机构语料库和目标语言会产生任何国家等效的流程。所有模型权重和所有微调数据集均以开源许可公开发布。
英文摘要：
We present Jupiter-N, a hybrid reasoning model post-trained from Nemotron 3 Super, a fully open-source 120 billion parameter LLM. We target three objectives: (1) agentic capability via uncertainty-curated trajectories; (2) UK cultural alignment via synthetic data grounded in cultural norms; and (3) Welsh language support via parallel corpora and LLM-translated Welsh conversations. Our data curation strategy carefully preserves the base model's capabilities: using our Forget-Me-Not framework, we mix on-policy synthetic replay with off-policy task data to mitigate catastrophic forgetting, and include a mixture of reasoning and non-reasoning traces to maintain Nemotron's hybrid reasoning ability. Jupiter-N achieves standout gains over Nemotron in Welsh (+18 on ARC-Easy, +5.25 on MMLU-Lite), terminal-use (+9.1 on Terminal Bench 2) and instruction following (+4.4 on IFBench), while retaining the base model capabilities. We frame this work as a reproducible template for sovereign post-training: substituting cultural knowledge, institutional corpora, and target languages produces an equivalent pipeline for any country. All model weights and all post-training datasets are publicly released under open licences.
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。
