Reference-Free Post-Training of Open Large Language Models for Multilingual Machine Translation

• 类型：arxiv
• 标识：2608.10812
• 链接：https://arxiv.org/abs/2608.10812
• 主分类：engineering
• 形态：method
• 被引：0
• 被引来源：Semantic Scholar
• S2被引：0
• 影响力被引：0
• TLDR：This work studies reference-free post-training for multilingual machine translation with open large language models and finds that on-policy distillation reaches, but does not surpass, the quality frontier achieved by RL with checkpoint interpolation.
• 待LLM分类：否
• 标题中文：面向多语言机器翻译的开源大语言模型无参考后训练
• TLDR中文：本文研究基于开源大语言模型的无参考多语言机器翻译后训练，发现 on-policy 蒸馏能够达到但无法超越结合 checkpoint 插值的强化学习所确立的质量前沿。
• 来源文件：
• /inbox/tom/_candidates/2026-08-12-agent-rag-longcontext-candidates.json
• [S2 enrich]