内存高效的NLLB-200：大规模多语言机器翻译模型的语言特定专家剪枝
Memory-efficient NLLB-200: Language-specific Expert Pruning of a Massively Multilingual Machine Translation Model
AI总结：
本文提出一种专家剪枝方法，可在不微调的情况下移除NLLB-200中多达80%的专家，翻译质量损失可忽略，使54.5B参数模型能在单块32GB GPU上推理，并能识别语言特定专家。
摘要：
近期发布的NLLB-200是一组覆盖202种语言的多语言神经机器翻译模型。其中最大的模型基于Mixture of Experts架构，在众多语言对上取得了SoTA结果。该模型包含54.5B参数，仅推理就需要至少四块32GB GPU。在本工作中，我们提出了一种剪枝方法，能够在无需进一步微调的情况下移除多达80%的专家，且翻译质量损失可忽略不计，从而使模型能够在单块32GB GPU上运行。进一步分析表明，我们的剪枝度量能够识别语言特定的专家。
英文摘要：
The recently released NLLB-200 is a set of multilingual Neural Machine Translation models that cover 202 languages. The largest model is based on a Mixture of Experts architecture and achieves SoTA results across many language pairs. It contains 54.5B parameters and requires at least four 32GB GPUs just for inference. In this work, we propose a pruning method that enables the removal of up to 80% of experts without further finetuning and with a negligible loss in translation quality, which makes it feasible to run the model on a single 32GB GPU. Further analysis suggests that our pruning metrics can identify language-specific experts.
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。
