CULL-MT：用于机器翻译的基于语言和层剪枝的压缩方法
CULL-MT: Compression Using Language and Layer pruning for Machine Translation
AI总结：
本文提出CULL-MT压缩方法，通过结构化层剪枝和知识蒸馏结合参数高效微调，在多向翻译任务中有效减小模型规模并保持性能。
摘要：
多语言机器翻译模型通常通过利用翻译知识迁移来优于传统的双语模型。近期的进展使得这些模型能够支持数百种语言，并在各种翻译方向上取得最先进的结果。然而，随着这些模型变得越来越大，其推理操作的成本也日益增加。在许多用例中，并不需要支持如此广泛的语言对，因为通常只需要在少数选定的方向上进行翻译。在本文中，我们提出了CULL-MT，一种基于结构化层剪枝和选定语言方向的机器翻译模型压缩方法。我们的方法使用贪婪策略识别并剪枝不重要的层，然后通过应用来自原始模型的知识蒸馏以及参数高效微调来减轻影响。我们将CULL-MT应用于NLLB-3.3B和LLaMA3.1-8B-Instruct模型。在多向翻译场景（波斯语、法语和德语翻译为英语）中，我们发现NLLB-3.3B模型具有鲁棒性，允许剪枝25%的层，spBLEU仅下降0.9。然而，LLaMA3.1-8B-Instruct更为敏感，在剪枝5层后spBLEU下降2.0。
英文摘要：
Multilingual machine translation models often outperform traditional bilingual models by leveraging translation knowledge transfer. Recent advancements have led to these models supporting hundreds of languages and achieving state-of-the-art results across various translation directions. However, as these models grow larger, their inference operations become increasingly costly. In many use cases, there is no need to support such a wide range of language pairs, as translation is typically needed in only a few selected directions. In this paper, we present CULL-MT, a compression method for machine translation models based on structural layer pruning and selected language directions. Our approach identifies and prunes unimportant layers using a greedy strategy, then mitigates the impact by applying knowledge distillation from the original model along with parameter-efficient fine-tuning. We apply CULL-MT to the NLLB-3.3B and LLaMA3.1-8B-Instruct models. In a multi-way translation scenario (Persian, French, and German to English), we find the NLLB-3.3B model to be robust, allowing 25% of layers to be pruned with only a 0.9 spBLEU drop. However, LLaMA3.1-8B-Instruct is more sensitive, with a 2.0 spBLEU drop after pruning 5 layers.
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。
