arXiv subjects

Hanqing Zhu

Publications and source records attributed to Hanqing Zhu.

30 records · Page 2Linked to original sources

arXiv2023

Pre-RMSNorm and Pre-CRMSNorm Transformers: Equivalent and Efficient Pre-LN Transformers

Transformers have achieved great success in machine learning applications. Normalization techniques, such as Layer Normalization (LayerNorm, LN) and Root Mean Square Normalization (RMSNorm), play a critical role in accelerating and stabilizing the training of Transformers. While LayerNorm recenters and rescales input vectors, RMSNorm only rescales the vectors by their RMS value. Despite being more computationally efficient, RMSNorm may compromise the representation ability of Transformers. There is currently no consensus regarding the preferred normalization technique, as some models employ LayerNorm while others utilize RMSNorm, especially in recent large language models. It is challenging to convert Transformers with one normalization to the other type. While there is an ongoing disagreement between the two normalization types, we propose a solution to unify two mainstream Transformer architectures, Pre-LN and Pre-RMSNorm Transformers. By removing the inherent redundant mean information in the main branch of Pre-LN Transformers, we can reduce LayerNorm to RMSNorm, achieving higher efficiency. We further propose the Compressed RMSNorm (CRMSNorm) and Pre-CRMSNorm Transformer based on a lossless compression of the zero-mean vectors. We formally establish the equivalence of Pre-LN, Pre-RMSNorm, and Pre-CRMSNorm Transformer variants in both training and inference. It implies that Pre-LN Transformers can be substituted with Pre-(C)RMSNorm counterparts at almost no cost, offering the same arithmetic functionality along with free efficiency improvement. Experiments demonstrate that we can reduce the training and inference time of Pre-LN Transformers by 1% - 10%.

cs.LG↗

arXiv2023

Integrated multi-operand optical neurons for scalable and hardware-efficient deep learning

The optical neural network (ONN) is a promising hardware platform for next-generation neuromorphic computing due to its high parallelism, low latency, and low energy consumption. However, previous integrated photonic tensor cores (PTCs) consume numerous single-operand optical modulators for signal and weight encoding, leading to large area costs and high propagation loss to implement large tensor operations. This work proposes a scalable and efficient optical dot-product engine based on customized multi-operand photonic devices, namely multi-operand optical neurons (MOON). We experimentally demonstrate the utility of a MOON using a multi-operand-Mach-Zehnder-interferometer (MOMZI) in image recognition tasks. Specifically, our MOMZI-based ONN achieves a measured accuracy of 85.89% in the street view house number (SVHN) recognition dataset with 4-bit voltage control precision. Furthermore, our performance analysis reveals that a 128x128 MOMZI-based PTCs outperform their counterparts based on single-operand MZIs by one to two order-of-magnitudes in propagation loss, optical delay, and total device footprint, with comparable matrix expressivity.

physics.optics↗

arXiv2023

Domain Wall-Magnetic Tunnel Junction Analog Content Addressable Memory Using Current and Projected Data

With the rise in in-memory computing architectures to reduce the compute-memory bottleneck, a new bottleneck is present between analog and digital conversion. Analog content-addressable memories (ACAM) are being recently studied for in-memory computing to efficiently convert between analog and digital signals. Magnetic memory elements such as magnetic tunnel junctions (MTJs) could be useful for ACAM due to their low read/write energy and high endurance, but MTJs are usually restricted to digital values. The spin orbit torque-driven domain wall-magnetic tunnel junction (DW-MTJ) has been recently shown to have multi-bit function. Here, an ACAM circuit is studied that uses two domain wall-magnetic tunnel junctions (DW-MTJs) as the analog storage elements. Prototype DW-MTJ data is input into the magnetic ACAM (MACAM) circuit simulation, showing ternary CAM function. Device-circuit co-design is carried out, showing that 8-10 weight bits are achievable, and that designing asymmetrical spacing of the available DW positions in the device leads to evenly spaced ACAM search bounds. Analyzing available spin orbit torque materials shows platinum provides the largest MACAM search bound while still allowing spin orbit torque domain wall motion, and that the circuit is optimized with minimized MTJ resistance, minimized spin orbit torque material resistance, and maximized tunnel magnetoresistance. These results show the feasibility of using DW-MTJs for MACAM and provide design parameters.

cond-mat.mes-hall↗

arXiv2022

NeurOLight: A Physics-Agnostic Neural Operator Enabling Parametric Photonic Device Simulation

Optical computing is an emerging technology for next-generation efficient artificial intelligence (AI) due to its ultra-high speed and efficiency. Electromagnetic field simulation is critical to the design, optimization, and validation of photonic devices and circuits. However, costly numerical simulation significantly hinders the scalability and turn-around time in the photonic circuit design loop. Recently, physics-informed neural networks have been proposed to predict the optical field solution of a single instance of a partial differential equation (PDE) with predefined parameters. Their complicated PDE formulation and lack of efficient parametrization mechanisms limit their flexibility and generalization in practical simulation scenarios. In this work, for the first time, a physics-agnostic neural operator-based framework, dubbed NeurOLight, is proposed to learn a family of frequency-domain Maxwell PDEs for ultra-fast parametric photonic device simulation. We balance the efficiency and generalization of NeurOLight via several novel techniques. Specifically, we discretize different devices into a unified domain, represent parametric PDEs with a compact wave prior, and encode the incident light via masked source modeling. We design our model with parameter-efficient cross-shaped NeurOLight blocks and adopt superposition-based augmentation for data-efficient learning. With these synergistic approaches, NeurOLight generalizes to a large space of unseen simulation settings, demonstrates 2-orders-of-magnitude faster simulation speed than numerical solvers, and outperforms prior neural network models by ~54% lower prediction error with ~44% fewer parameters. Our code is available at https://github.com/JeremieMelo/NeurOLight.

cs.ET↗

arXiv2022

Fuse and Mix: MACAM-Enabled Analog Activation for Energy-Efficient Neural Acceleration

Analog computing has been recognized as a promising low-power alternative to digital counterparts for neural network acceleration. However, conventional analog computing is mainly in a mixed-signal manner. Tedious analog/digital (A/D) conversion cost significantly limits the overall system's energy efficiency. In this work, we devise an efficient analog activation unit with magnetic tunnel junction (MTJ)-based analog content-addressable memory (MACAM), simultaneously realizing nonlinear activation and A/D conversion in a fused fashion. To compensate for the nascent and therefore currently limited representation capability of MACAM, we propose to mix our analog activation unit with digital activation dataflow. A fully differential framework, SuperMixer, is developed to search for an optimized activation workload assignment, adaptive to various activation energy constraints. The effectiveness of our proposed methods is evaluated on a silicon photonic accelerator. Compared to standard activation implementation, our mixed activation system with the searched assignment can achieve competitive accuracy with $>$60% energy saving on A/D conversion and activation.

cs.ET↗

arXiv2022

A compact butterfly-style silicon photonic-electronic neural chip for hardware-efficient deep learning

The optical neural network (ONN) is a promising hardware platform for next-generation neurocomputing due to its high parallelism, low latency, and low energy consumption. Previous ONN architectures are mainly designed for general matrix multiplication (GEMM), leading to unnecessarily large area cost and high control complexity. Here, we move beyond classical GEMM-based ONNs and propose an optical subspace neural network (OSNN) architecture, which trades the universality of weight representation for lower optical component usage, area cost, and energy consumption. We devise a butterfly-style photonic-electronic neural chip to implement our OSNN with up to 7x fewer trainable optical components compared to GEMM-based ONNs. Additionally, a hardware-aware training framework is provided to minimize the required device programming precision, lessen the chip area, and boost the noise robustness. We experimentally demonstrate the utility of our neural chip in practical image recognition tasks, showing that a measured accuracy of 94.16% can be achieved in hand-written digit recognition tasks with 3-bit weight programming precision.

cs.ET↗

arXiv2022

ADEPT: Automatic Differentiable DEsign of Photonic Tensor Cores

Photonic tensor cores (PTCs) are essential building blocks for optical artificial intelligence (AI) accelerators based on programmable photonic integrated circuits. PTCs can achieve ultra-fast and efficient tensor operations for neural network (NN) acceleration. Current PTC designs are either manually constructed or based on matrix decomposition theory, which lacks the adaptability to meet various hardware constraints and device specifications. To our best knowledge, automatic PTC design methodology is still unexplored. It will be promising to move beyond the manual design paradigm and "nurture" photonic neurocomputing with AI and design automation. Therefore, in this work, for the first time, we propose a fully differentiable framework, dubbed ADEPT, that can efficiently search PTC designs adaptive to various circuit footprint constraints and foundry PDKs. Extensive experiments show superior flexibility and effectiveness of the proposed ADEPT framework to explore a large PTC design space. On various NN models and benchmarks, our searched PTC topology outperforms prior manually-designed structures with competitive matrix representability, 2-30x higher footprint compactness, and better noise robustness, demonstrating a new paradigm in photonic neural chip design. The code of ADEPT is available at https://github.com/JeremieMelo/ADEPT using the https://github.com/JeremieMelo/pytorch-onn (TorchONN) library.

cs.ET↗

arXiv2022

Uni-Retriever: Towards Learning The Unified Embedding Based Retriever in Bing Sponsored Search

Embedding based retrieval (EBR) is a fundamental building block in many web applications. However, EBR in sponsored search is distinguished from other generic scenarios and technically challenging due to the need of serving multiple retrieval purposes: firstly, it has to retrieve high-relevance ads, which may exactly serve user's search intent; secondly, it needs to retrieve high-CTR ads so as to maximize the overall user clicks. In this paper, we present a novel representation learning framework Uni-Retriever developed for Bing Search, which unifies two different training modes knowledge distillation and contrastive learning to realize both required objectives. On one hand, the capability of making high-relevance retrieval is established by distilling knowledge from the ``relevance teacher model''. On the other hand, the capability of making high-CTR retrieval is optimized by learning to discriminate user's clicked ads from the entire corpus. The two training modes are jointly performed as a multi-objective learning process, such that the ads of high relevance and CTR can be favored by the generated embeddings. Besides the learning strategy, we also elaborate our solution for EBR serving pipeline built upon the substantially optimized DiskANN, where massive-scale EBR can be performed with competitive time and memory efficiency, and accomplished in high-quality. We make comprehensive offline and online experiments to evaluate the proposed techniques, whose findings may provide useful insights for the future development of EBR systems. Uni-Retriever has been mainstreamed as the major retrieval path in Bing's production thanks to the notable improvements on the representation and EBR serving quality.

cs.IR↗

arXiv2021

ELight: Enabling Efficient Photonic In-Memory Neurocomputing with Life Enhancement

With the recent advances in optical phase change material (PCM), photonic in-memory neurocomputing has demonstrated its superiority in optical neural network (ONN) designs with near-zero static power consumption, time-of-light latency, and compact footprint. However, photonic tensor cores require massive hardware reuse to implement large matrix multiplication due to the limited single-core scale. The resultant large number of PCM writes leads to serious dynamic power and overwhelms the fragile PCM with limited write endurance. In this work, we propose a synergistic optimization framework, ELight, to minimize the overall write efforts for efficient and reliable optical in-memory neurocomputing. We first propose write-aware training to encourage the similarity among weight blocks, and combine it with a post-training optimization method to reduce programming efforts by eliminating redundant writes. Experiments show that ELight can achieve over 20X reduction in the total number of writes and dynamic power with comparable accuracy. With our ELight, photonic in-memory neurocomputing will step forward towards viable applications in machine learning with preserved accuracy, order-of-magnitude longer lifetime, and lower programming energy.

cs.ET↗

arXiv2021

L2ight: Enabling On-Chip Learning for Optical Neural Networks via Efficient in-situ Subspace Optimization

Silicon-photonics-based optical neural network (ONN) is a promising hardware platform that could represent a paradigm shift in efficient AI with its CMOS-compatibility, flexibility, ultra-low execution latency, and high energy efficiency. In-situ training on the online programmable photonic chips is appealing but still encounters challenging issues in on-chip implementability, scalability, and efficiency. In this work, we propose a closed-loop ONN on-chip learning framework L2ight to enable scalable ONN mapping and efficient in-situ learning. L2ight adopts a three-stage learning flow that first calibrates the complicated photonic circuit states under challenging physical constraints, then performs photonic core mapping via combined analytical solving and zeroth-order optimization. A subspace learning procedure with multi-level sparsity is integrated into L2ight to enable in-situ gradient evaluation and fast adaptation, unleashing the power of optics for real on-chip intelligence. Extensive experiments demonstrate our proposed L2ight outperforms prior ONN training protocols with 3-order-of-magnitude higher scalability and over 30X better efficiency, when benchmarked on various models and learning tasks. This synergistic framework is the first scalable on-chip learning solution that pushes this emerging field from intractable to scalable and further to efficient for next-generation self-learnable photonic neural chips. From a co-design perspective, L2ight also provides essential insights for hardware-restricted unitary subspace optimization and efficient sparse training. We open-source our framework at https://github.com/JeremieMelo/L2ight.

cs.LG↗

arXiv2021

Towards Memory-Efficient Neural Networks via Multi-Level in situ Generation

Deep neural networks (DNN) have shown superior performance in a variety of tasks. As they rapidly evolve, their escalating computation and memory demands make it challenging to deploy them on resource-constrained edge devices. Though extensive efficient accelerator designs, from traditional electronics to emerging photonics, have been successfully demonstrated, they are still bottlenecked by expensive memory accesses due to tremendous gaps between the bandwidth/power/latency of electrical memory and computing cores. Previous solutions fail to fully-leverage the ultra-fast computational speed of emerging DNN accelerators to break through the critical memory bound. In this work, we propose a general and unified framework to trade expensive memory transactions with ultra-fast on-chip computations, directly translating to performance improvement. We are the first to jointly explore the intrinsic correlations and bit-level redundancy within DNN kernels and propose a multi-level in situ generation mechanism with mixed-precision bases to achieve on-the-fly recovery of high-resolution parameters with minimum hardware overhead. Extensive experiments demonstrate that our proposed joint method can boost the memory efficiency by 10-20x with comparable accuracy over four state-of-the-art designs, when benchmarked on ResNet-18/DenseNet-121/MobileNetV2/V3 with various tasks.

cs.LG↗

arXiv2017

A Rule-Based Computational Model of Cognitive Arithmetic

Cognitive arithmetic studies the mental processes used in solving math problems. This area of research explores the retrieval mechanisms and strategies used by people during a common cognitive task. Past research has shown that human performance in arithmetic operations is correlated to the numerical size of the problem. Past research on cognitive arithmetic has pinpointed this trend to either retrieval strength, error checking, or strategy-based approaches when solving equations. This paper describes a rule-based computational model that performs the four major arithmetic operations (addition, subtraction, multiplication and division) on two operands. We then evaluated our model to probe its validity in representing the prevailing concepts observed in psychology experiments from the related works. The experiments specifically explore the problem size effect, an activation-based model for fact retrieval, backup strategies when retrieval fails, and finally optimization strategies when faced with large operands. From our experimental results, we concluded that our model's response times were comparable to results observed when people performed similar tasks during psychology experiments. The fit of our model in reproducing these results and incorporating accuracy into our model are discussed.

cs.AI↗