APTQ: Attention-aware Post-Training Mixed-Precision Quantization for Large Language Models
This work addresses the deployment problem for edge devices by providing an incremental improvement in quantization techniques for large language models.
The paper tackles the challenge of deploying large language models on edge devices by proposing APTQ, a mixed-precision quantization method that reduces model size and computational load while maintaining performance, achieving an average of 4-bit width with 5.22 perplexity on C4 and state-of-the-art zero-shot accuracy up to 70.48% at 3.8-bit width.
Large Language Models (LLMs) have greatly advanced the natural language processing paradigm. However, the high computational load and huge model sizes pose a grand challenge for deployment on edge devices. To this end, we propose APTQ (Attention-aware Post-Training Mixed-Precision Quantization) for LLMs, which considers not only the second-order information of each layer's weights, but also, for the first time, the nonlinear effect of attention outputs on the entire model. We leverage the Hessian trace as a sensitivity metric for mixed-precision quantization, ensuring an informed precision reduction that retains model performance. Experiments show APTQ surpasses previous quantization methods, achieving an average of 4 bit width a 5.22 perplexity nearly equivalent to full precision in the C4 dataset. In addition, APTQ attains state-of-the-art zero-shot accuracy of 68.24\% and 70.48\% at an average bitwidth of 3.8 in LLaMa-7B and LLaMa-13B, respectively, demonstrating its effectiveness to produce high-quality quantized LLMs.