用低比特梯度压缩实现大模型微调中的高效数据筛选
QLESS: A Quantized Approach for Data Valuation and Selection in Large Language Model Fine-Tuning
- 通过低秩投影与梯度量化两阶段压缩，降低数据评估内存开销
- 在多个大模型上实现与原方法相当的数据选择效果，内存减少16倍
- 1比特量化仍保持良好性能，适合资源受限场景
大语言模型微调常受海量数据处理计算成本的限制。本文提出QLESS（量化低秩梯度相似性搜索），将梯度量化与LESS框架结合，实现内存高效的样本价值评估与选择。QLESS采用两步压缩：首先通过基于LoRA的随机投影获取低维梯度表示；随后将这些梯度量化为低比特表示。在LLaMA、Mistral、Qwen等多类模型及MMLU、BBH、TyDiQA等基准上的实验表明，QLESS在数据选择性能上可媲美LESS，同时内存使用最高降低16倍。即使在1比特梯度量化下，数据价值评估质量依然保持良好。这些结果表明，QLESS是一种在严格内存约束下识别有效样本的实用且可扩展的方法。
原文摘要 · Abstract (English)
Fine-tuning large language models (LLMs) is often constrained by the computational costs of processing massive datasets. We propose \textbf{QLESS} (Quantized Low-rank Gradient Similarity Search), which integrates gradient quantization with the LESS framework to enable memory-efficient data valuation and selection. QLESS employs a two-step compression process: first, it obtains low-dimensional gradient representations through LoRA-based random projection; then, it quantizes these gradients to low-bitwidth representations. Experiments on multiple LLM architectures (LLaMA, Mistral, Qwen) and benchmarks (MMLU, BBH, TyDiQA) show that QLESS achieves comparable data selection performance to LESS while reducing memory usage by up to 16x. Even 1-bit gradient quantization preserves data valuation quality. These findings underscore QLESS as a practical, scalable approach to identifying informative examples within strict memory constraints.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。