English
Related papers

Related papers: A Study of BFLOAT16 for Deep Learning Training

200 papers

The state-of-the-art (SOTA) for mixed precision training is dominated by variants of low precision floating point operations, and in particular, FP16 accumulating into FP32 Micikevicius et al. (2017). On the other hand, while a lot of…

Given the current trend of increasing size and complexity of machine learning architectures, it has become of critical importance to identify new approaches to improve the computational efficiency of model training. In this context, we…

Machine Learning · Computer Science 2022-06-08 Badreddine Noune , Philip Jones , Daniel Justus , Dominic Masters , Carlo Luschi

In a large class of deep learning models, including vector embedding models such as word and database embeddings, we observe that floating point exponent values cluster around a few unique values, permitting entropy based data compression.…

Machine Learning · Computer Science 2022-02-04 Rajesh Bordawekar , Bulent Abali , Ming-Hung Chen

Reinforcement learning (RL) fine-tuning of large language models (LLMs) often suffers from instability due to the numerical mismatch between the training and inference policies. While prior work has attempted to mitigate this issue through…

Machine Learning · Computer Science 2025-10-31 Penghui Qi , Zichen Liu , Xiangxin Zhou , Tianyu Pang , Chao Du , Wee Sun Lee , Min Lin

Recently, the posit numerical format has shown promise for DNN data representation and compute with ultra-low precision ([5..8]-bit). However, majority of studies focus only on DNN inference. In this work, we propose DNN training using…

Machine Learning · Computer Science 2019-08-01 Hamed F. Langroudi , Zachariah Carmichael , Dhireesha Kudithipudi

Training deep neural networks (DNNs) is a computationally expensive job, which can take weeks or months even with high performance GPUs. As a remedy for this challenge, community has started exploring the use of more efficient data…

Machine Learning · Computer Science 2022-03-15 Seock-Hwan Noh , Jahyun Koo , Seunghyun Lee , Jongse Park , Jaeha Kung

As recently demonstrated, Deep Neural Networks (DNN), usually trained using single precision IEEE 754 floating point numbers (binary32), can also work using lower precision. Therefore, 16-bit and 8-bit compressed format have attracted…

Large foundation models have become central to modern machine learning, with performance scaling predictably with model size and data. However, training and deploying such models incur substantial computational and memory costs, motivating…

In this paper, we introduce MCTensor, a library based on PyTorch for providing general-purpose and high-precision arithmetic for DL training. MCTensor is used in the same way as PyTorch Tensor: we implement multiple basic, matrix-level…

Machine Learning · Computer Science 2022-08-31 Tao Yu , Wentao Guo , Jianan Canal Li , Tiancheng Yuan , Christopher De Sa

Deep neural networks (DNN) are powerful models for many pattern recognition tasks, yet their high computational complexity and memory requirement limit them to applications on high-performance computing platforms. In this paper, we propose…

Machine Learning · Computer Science 2018-10-24 Lukas Mauch , Bin Yang

Block Floating Point (BFP) can efficiently support quantization for Deep Neural Network (DNN) training by providing a wide dynamic range via a shared exponent across a group of values. In this paper, we propose a Fast First, Accurate Second…

Machine Learning · Computer Science 2021-11-01 Sai Qian Zhang , Bradley McDanel , H. T. Kung

Single-precision floating point (FP32) data format, defined by the IEEE 754 standard, is widely employed in scientific computing, signal processing, and deep learning training, where precision is critical. However, FP32 multiplication is…

Hardware Architecture · Computer Science 2025-10-09 Bindu G Gowda , Yogesh Goyal , Yash Gupta , Madhav Rao

Modern deep neural network (DNN) models generally require a huge amount of weight and activation values to achieve good inference outcomes. Those data inevitably demand a massive off-chip memory capacity/bandwidth, and the situation gets…

Machine Learning · Computer Science 2021-04-27 Cheng-Wei Huang , Tim-Wei Chen , Juinn-Dar Huang

The recent surge of interest in Deep Neural Networks (DNNs) has led to increasingly complex networks that tax computational and memory resources. Many DNNs presently use 16-bit or 32-bit floating point operations. Significant performance…

Distributed, Parallel, and Cluster Computing · Computer Science 2019-01-23 Zachariah Carmichael , Hamed F. Langroudi , Char Khazanov , Jeffrey Lillie , John L. Gustafson , Dhireesha Kudithipudi

As deep learning models grow and deployment becomes more widespread, reducing the storage and transmission costs of neural network weights has become increasingly important. While prior work such as ZipNN has shown that lossless compression…

Machine Learning · Computer Science 2025-08-28 Anat Heilper , Doron Singer

Is multiplication really necessary for deep neural networks? Here we propose just adding two IEEE754 floating-point numbers with an integer-add instruction in place of a floating-point multiplication instruction. We show that ResNet can be…

Machine Learning · Computer Science 2020-12-08 Tsuguo Mogami

Low-precision formats have recently driven major breakthroughs in neural network (NN) training and inference by reducing the memory footprint of the NN models and improving the energy efficiency of the underlying hardware architectures.…

Hardware Architecture · Computer Science 2024-10-28 Luca Bertaccini , Gianna Paulin , Tim Fischer , Stefan Mach , Luca Benini

Low precision training is one of the most popular strategies for deploying the deep model on limited hardware resources. Fixed point implementation of DCNs has the potential to alleviate complexities and facilitate potential deployment on…

Computer Vision and Pattern Recognition · Computer Science 2019-12-30 Zehua Cheng , Weiyang Wang , Yan Pan , Thomas Lukasiewicz

Lowering the precision of neural networks from the prevalent 32-bit precision has long been considered harmful to performance, despite the gain in space and time. Many works propose various techniques to implement half-precision neural…

Machine Learning · Computer Science 2024-05-06 Juyoung Yun , Byungkon Kang , Zhoulai Fu

In the field of deep learning, the prevalence of models initially trained with 32-bit precision is a testament to its robustness and accuracy. However, the continuous evolution of these models often demands further training, which can be…

Machine Learning · Computer Science 2023-12-04 Juyoung Yun