arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.03211cs.CLcs.LG
收藏

量化如何影响多语言 LLMs?

How Does Quantization Affect Multilingual LLMs?

Cohere;Cohere For AI

Kelly Marchisio, Saurabh Dash, Hongyu Chen, Dennis Aumiller, Ahmet Üstün, Sara Hooker, Sebastian Ruder

AI总结:

研究量化对多语言LLMs的影响,发现量化在非拉丁字母语言和数学推理任务上退化最严重,且自动指标严重低估了实际损害,强调需将多语言性能纳入高效模型评估标准。

摘要:

量化技术被广泛用于提升大型语言模型的推理速度和部署能力。尽管大量研究考察了量化对英语 LLMs 的影响,但尚无研究跨语言进行评估。我们对量化的多语言 LLMs 进行了全面分析,重点关注跨语言和不同规模下的性能。我们使用自动基准、LLM-as-a-Judge 和人工评估,发现:(1) 量化的有害影响在人工评估中显而易见,而自动指标严重低估了这一点:日语在自动任务中平均下降 1.7%,而人工评估者在现实提示上报告的下降为 16.0%;(2) 各语言受量化影响程度不同,非拉丁字母语言受影响最严重;(3) 数学推理等挑战性任务退化最快。由于服务低计算模型的能力对于 NLP 技术的广泛全球采用至关重要,我们的结果敦促将多语言性能作为高效模型的关键评估标准。

英文摘要:

Quantization techniques are widely used to improve inference speed and deployment of large language models. While a wide body of work examines the impact of quantization on LLMs in English, none have evaluated across languages. We conduct a thorough analysis of quantized multilingual LLMs, focusing on performance across languages and at varying scales. We use automatic benchmarks, LLM-as-a-Judge, and human evaluation, finding that (1) harmful effects of quantization are apparent in human evaluation, which automatic metrics severely underestimate: a 1.7% average drop in Japanese across automatic tasks corresponds to a 16.0% drop reported by human evaluators on realistic prompts; (2) languages are disparately affected by quantization, with non-Latin script languages impacted worst; and (3) challenging tasks like mathematical reasoning degrade fastest. As the ability to serve low-compute models is critical for wide global adoption of NLP technologies, our results urge consideration of multilingual performance as a key evaluation criterion for efficient models.

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。

↑