多语言量化税：边缘小型语言模型的结构性崩溃与类型脆弱性
原标题：The Multilingual Quantization Tax: Structural Collapse and Typological Fragility in Edge SLMs
AI 摘要
本研究首次对4-bit量化在Gemma 4和Qwen 3.5等小型语言模型上的多语言性能损失进行了零样本评估，覆盖八种类型多样的语言。研究发现量化税在不同语言间极不平等，低资源和非拉丁文字语言出现结构性崩溃，而模型的基础语言（如英语、中文）也缺乏免疫性。多步逻辑推理受损严重，但联想记忆领域表现出量化抵抗性。
正文节选
The Multilingual Quantization Tax: Structural Collapse and Typological Fragility in Edge SLMs Abstract While 4-bit weight quantization is critical for deploying Small Language Models (SLMs) on edge devices, evaluations of the resulting performance degradation—the quantization tax—remain overwhelmingly English-centric. We present a zero-shot multilingual evaluation of 4-bit quantization across the Gemma 4 and Qwen 3.5 architectures. Evaluating on eight typologically diverse languages using MMLU P