联合优化乘法器结构与模型参数，显著降低AI加速器功耗。
TRAM: Training Approximate Multiplier Structures for Low-Power AI Accelerators
- 在训练中同步设计低功耗近似乘法器，实现软硬件协同优化。
- 在CIFAR-10上使乘法器功耗降低25.05%，ImageNet上整体功耗降27.09%。
- 适合追求低功耗部署的边缘AI开发者和芯片设计人员。
降低AI加速器的功耗日益重要。近似计算可在保持较小精度损失的前提下减少功耗。由于乘法器是AI模型中功耗较高的组件，本文聚焦于合成低功耗近似乘法器（AxMs）。与以往将近似乘法器设计与AI模型训练分离的方法不同，本文提出TRAM，通过联合优化近似乘法器结构与模型参数，在保持小精度损失的同时降低功耗。实验表明，相比现有最先进方法，TRAM在卷积神经网络（CNNs）与CIFAR-10数据集上实现了最高25.05%的近似乘法器功耗降低，在视觉变换器（vision transformers）与ImageNet数据集上实现最高27.09%的整体功耗降低。
原文摘要 · Abstract (English)
Reducing power consumption in AI accelerators is increasingly important. Approximate computing can reduce power consumption while keeping the accuracy loss small. Since multipliers are power-hungry components in AI models, this paper focuses on synthesizing low-power approximate multipliers (AxMs). Unlike prior works that design AxMs separately from AI model training, we present TRAM, which jointly optimizes the AxM structure and AI model parameters to lower power with small accuracy loss. Experiments show that compared to state-of-the-art AxMs, TRAM achieves up to 25.05% AxM power reduction on CNNs with CIFAR-10, and reduces power by up to 27.09% on vision transformers with ImageNet.
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。