paper-with-me

홈 › Papers

PositNN: Tapered Precision Deep Learning Inference for the Edge

2018-10-20 · Hamed F. Langroudi, Zachariah Carmichael, John L. Gustafson, Dhireesha Kudithipudi

The performance of neural networks, especially the currently popular form of deep neural networks, is often limited by the underlying hardware. Computations in deep neural networks are expensive, have large memory footprint, and are power hungry. Conventional reduced-precision numerical formats, such as fixed-point and floating point, cannot accurately represent deep neural network parameters with a nonlinear distribution and small dynamic range. Recently proposed posit numerical format with tapered precision represents small values more accurately than the other formats. In this work, we propose an ultra-low precision deep neural network, PositNN, that uses posits during inference. The efficacy of PositNN is demonstrated on a deep neural network architecture with two datasets (MNIST, Fashion MNIST and Cifar-10), where an 8-bit PositNN outperforms other {5-8}-bit low-precision neural networks and a 32-bit floating point baseline network.

📄 PDF Abstract BibTeX

Code (0)

등록된 구현이 없습니다.

Tasks

Deep Learning

Similar Papers 제목 키워드 기반

TENT: Efficient Quantization of Neural Networks on the tiny Edge with Tapered FixEd PoiNT

2021-04-06 · Hamed F. Langroudi, Vedant Karia, Tej Pandit, Dhireesha Kudithipudi

In this research, we propose a new low-precision framework, TENT, to leverage the benefits of a tapered fixed-point numerical format in TinyML models. We introduce a tapered fixed-point quantization algorithm that matche…

Quantization

PositNN: Training Deep Neural Networks with Mixed Low-Precision Posit

2021-04-30 · Gonçalo Raposo, Pedro Tomás, Nuno Roma

Low-precision formats have proven to be an efficient way to reduce not only the memory footprint but also the hardware resources and power consumption of deep learning computations. Under this premise, the posit numerica…

SPADE: A SIMD Posit-enabled compute engine for Accelerating DNN Efficiency

2026-01-24 · Sonu Kumar, Lavanya Vinnakota, Mukul Lokhande, Santosh Kumar Vishvakarma 외 arxiv

The growing demand for edge-AI systems requires arithmetic units that balance numerical precision, energy efficiency, and compact hardware while supporting diverse formats. Posit arithmetic offers advantages over floatin…

Ascend HiFloat8 Format for Deep Learning

2024-09-25 · Yuanyong Luo, Zhongxing Zhang, Richard Wu, Hu Liu 외

This preliminary white paper proposes a novel 8-bit floating-point data format HiFloat8 (abbreviated as HiF8) for deep learning. HiF8 features tapered precision. For normal value encoding, it provides 7 exponent values w…

Deep Learning

Tapered Off-Policy REINFORCE: Stable and efficient reinforcement learning for LLMs

2025-03-18 · Nicolas Le Roux, Marc G. Bellemare, Jonathan Lebensold, Arnaud Bergeron 외

We propose a new algorithm for fine-tuning large language models using reinforcement learning. Tapered Off-Policy REINFORCE (TOPR) uses an asymmetric, tapered variant of importance sampling to speed up learning while mai…

GSM8KMath