Peiyan Dong

Zhenglun Kong, Haoyu Ma, Geng Yuan et al. · harvard, meta-ai

Vision transformers (ViTs) have recently obtained success in many applications, but their intensive computation and heavy memory usage at both training and inference time limit their generalization. Previous compression algorithms usually start from the pre-trained dense models and only focus on efficient inference, while time-consuming training is still unavoidable. In contrast, this paper points out that the million-scale training data is redundant, which is the fundamental reason for the tedious training. To address the issue, this paper aims to introduce sparsity into data and proposes an end-to-end efficient training framework from three sparse perspectives, dubbed Tri-Level E-ViT. Specifically, we leverage a hierarchical data redundancy reduction scheme, by exploring the sparsity under three levels: number of training examples in the dataset, number of patches (tokens) in each example, and number of connections between tokens that lie in attention weights. With extensive experiments, we demonstrate that our proposed technique can noticeably accelerate training for various ViT architectures while maintaining accuracy. Remarkably, under certain ratios, we are able to improve the ViT accuracy rather than compromising it. For example, we can achieve 15.2% speedup with 72.6% (+0.4) Top-1 accuracy on Deit-T, and 15.7% speedup with 79.9% (+0.1) Top-1 accuracy on Deit-S. This proves the existence of data redundancy in ViT.

Xuan Shen, Zhenglun Kong, Minghai Qin et al. · harvard

The conventional lottery ticket hypothesis (LTH) claims that there exists a sparse subnetwork within a dense neural network and a proper random initialization method called the winning ticket, such that it can be trained from scratch to almost as good as the dense counterpart. Meanwhile, the research of LTH in vision transformers (ViTs) is scarcely evaluated. In this paper, we first show that the conventional winning ticket is hard to find at the weight level of ViTs by existing methods. Then, we generalize the LTH for ViTs to input data consisting of image patches inspired by the input dependence of ViTs. That is, there exists a subset of input image patches such that a ViT can be trained from scratch by using only this subset of patches and achieve similar accuracy to the ViTs trained by using all image patches. We call this subset of input patches the em winning tickets, which represent a significant amount of information in the input data. We use a ticket selector to generate the winning tickets based on the informativeness of patches for various types of ViT, including DeiT, LV-ViT, and Swin Transformers. The experiments show that there is a clear difference between the performance of models trained with winning tickets and randomly selected subsets, which verifies our proposed theory. We elaborate on the analogical similarity between our proposed Data-LTH-ViTs and the conventional LTH to further verify the integrity of our theory. The Source codes are available at https://github.com/shawnricecake/vit-lottery-ticket-input.

Peiyan Dong, Mengshu Sun, Alec Lu et al. · harvard

While vision transformers (ViTs) have continuously achieved new milestones in the field of computer vision, their sophisticated network architectures with high computation and memory costs have impeded their deployment on resource-limited edge devices. In this paper, we propose a hardware-efficient image-adaptive token pruning framework called HeatViT for efficient yet accurate ViT acceleration on embedded FPGAs. By analyzing the inherent computational patterns in ViTs, we first design an effective attention-based multi-head token selector, which can be progressively inserted before transformer blocks to dynamically identify and consolidate the non-informative tokens from input images. Moreover, we implement the token selector on hardware by adding miniature control logic to heavily reuse existing hardware components built for the backbone ViT. To improve the hardware efficiency, we further employ 8-bit fixed-point quantization, and propose polynomial approximations with regularization effect on quantization error for the frequently used nonlinear functions in ViTs. Finally, we propose a latency-aware multi-stage training strategy to determine the transformer blocks for inserting token selectors and optimize the desired (average) pruning rates for inserted token selectors, in order to improve both the model accuracy and inference latency on hardware. Compared to existing ViT pruning studies, under the similar computation cost, HeatViT can achieve 0.7%$\sim$8.9% higher accuracy; while under the similar model accuracy, HeatViT can achieve more than 28.4%$\sim$65.3% computation reduction, for various widely used ViTs, including DeiT-T, DeiT-S, DeiT-B, LV-ViT-S, and LV-ViT-M, on the ImageNet dataset. Compared to the baseline hardware accelerator, our implementations of HeatViT on the Xilinx ZCU102 FPGA achieve 3.46$\times$$\sim$4.89$\times$ speedup.

Hao Cheng, Changtao Miao, Tianle Song et al.

Autonomous LLM agents increasingly operate in stateful environments where they access tools, files, memory, and external services. While such capabilities enable complex real-world workflows, they also introduce security risks that are difficult to capture with existing evaluations. Current agent security benchmarks often rely on manually curated tasks, provide limited coverage of emerging threats, and focus primarily on final outcomes rather than the execution processes that lead to unsafe behavior. We introduce SeClaw, a framework that combines specification-driven security task synthesis with execution-based security evaluation for Autonomous agents. Spec-driven security task synthesis enables scalable and controllable construction of security tasks from structured risk specifications, while SeClaw docker provides a standardized testbed for evaluating agent behavior under diverse safety-risk scenarios. The benchmark covers risks arising from resources, user tasks, environments, and intrinsic agent behaviors, and supports trajectory-aware assessment of unsafe actions beyond final responses. By bridging systematic task synthesis and reproducible security evaluation, SeClaw provides a practical foundation for measuring, diagnosing, and comparing security failures in autonomous LLM agents. The code is available at https://github.com/seclaw-eval/seclaw-eval.

Zhengang Li, Alec Lu, Yanyue Xie et al. · harvard

Vision transformers (ViTs) have demonstrated their superior accuracy for computer vision tasks compared to convolutional neural networks (CNNs). However, ViT models are often computation-intensive for efficient deployment on resource-limited edge devices. This work proposes Quasar-ViT, a hardware-oriented quantization-aware architecture search framework for ViTs, to design efficient ViT models for hardware implementation while preserving the accuracy. First, Quasar-ViT trains a supernet using our row-wise flexible mixed-precision quantization scheme, mixed-precision weight entanglement, and supernet layer scaling techniques. Then, it applies an efficient hardware-oriented search algorithm, integrated with hardware latency and resource modeling, to determine a series of optimal subnets from supernet under different inference latency targets. Finally, we propose a series of model-adaptive designs on the FPGA platform to support the architecture search and mitigate the gap between the theoretical computation reduction and the practical inference speedup. Our searched models achieve 101.5, 159.6, and 251.6 frames-per-second (FPS) inference speed on the AMD/Xilinx ZCU102 FPGA with 80.4%, 78.6%, and 74.9% top-1 accuracy, respectively, for the ImageNet dataset, consistently outperforming prior works.

Zhirui Hu, Peiyan Dong, Zhepeng Wang et al.

Model compression, such as pruning and quantization, has been widely applied to optimize neural networks on resource-limited classical devices. Recently, there are growing interest in variational quantum circuits (VQC), that is, a type of neural network on quantum computers (a.k.a., quantum neural networks). It is well known that the near-term quantum devices have high noise and limited resources (i.e., quantum bits, qubits); yet, how to compress quantum neural networks has not been thoroughly studied. One might think it is straightforward to apply the classical compression techniques to quantum scenarios. However, this paper reveals that there exist differences between the compression of quantum and classical neural networks. Based on our observations, we claim that the compilation/traspilation has to be involved in the compression process. On top of this, we propose the very first systematical framework, namely CompVQC, to compress quantum neural networks (QNNs).In CompVQC, the key component is a novel compression algorithm, which is based on the alternating direction method of multipliers (ADMM) approach. Experiments demonstrate the advantage of the CompVQC, reducing the circuit depth (almost over 2.5 %) with a negligible accuracy drop (<1%), which outperforms other competitors. Another promising truth is our CompVQC can indeed promote the robustness of the QNN on the near-term noisy quantum devices.

Jun Liu, Pu Zhao, Zhenglun Kong et al. · harvard

Embodied robotic systems increasingly rely on large language model (LLM)-based agents to support high-level reasoning, planning, and decision-making during interactions with the environment. However, invoking LLM reasoning introduces substantial computational latency and resource overhead, which can interrupt action execution and reduce system reliability. Excessive reasoning may delay actions, while insufficient reasoning often leads to incorrect decisions and task failures. This raises a fundamental question for embodied agents: when should the agent reason, and when should it act? In this work, we propose RARRL (Resource-Aware Reasoning via Reinforcement Learning), a hierarchical framework for resource-aware orchestration of embodied agents. Rather than learning low-level control policies, RARRL learns a high-level orchestration policy that operates at the agent's decision-making layer. This policy enables the agent to adaptively determine whether to invoke reasoning, which reasoning role to employ, and how much computational budget to allocate based on current observations, execution history, and remaining resources. Extensive experiments, including evaluations with empirical latency profiles derived from the ALFRED benchmark, show that RARRL consistently improves task success rates while reducing execution latency and enhancing robustness compared with fixed or heuristic reasoning strategies. These results demonstrate that adaptive reasoning control is essential for building reliable and efficient embodied robotic agents.

Zhengang Li, Geng Yuan, Tomoharu Yamauchi et al.

Adiabatic Quantum-Flux-Parametron (AQFP) is a superconducting logic with extremely high energy efficiency. By employing the distinct polarity of current to denote logic `0' and `1', AQFP devices serve as excellent carriers for binary neural network (BNN) computations. Although recent research has made initial strides toward developing an AQFP-based BNN accelerator, several critical challenges remain, preventing the design from being a comprehensive solution. In this paper, we propose SupeRBNN, an AQFP-based randomized BNN acceleration framework that leverages software-hardware co-optimization to eventually make the AQFP devices a feasible solution for BNN acceleration. Specifically, we investigate the randomized behavior of the AQFP devices and analyze the impact of crossbar size on current attenuation, subsequently formulating the current amplitude into the values suitable for use in BNN computation. To tackle the accumulation problem and improve overall hardware performance, we propose a stochastic computing-based accumulation module and a clocking scheme adjustment-based circuit optimization method. We validate our SupeRBNN framework across various datasets and network architectures, comparing it with implementations based on different technologies, including CMOS, ReRAM, and superconducting RSFQ/ERSFQ. Experimental results demonstrate that our design achieves an energy efficiency of approximately 7.8x10^4 times higher than that of the ReRAM-based BNN framework while maintaining a similar level of model accuracy. Furthermore, when compared with superconductor-based counterparts, our framework demonstrates at least two orders of magnitude higher energy efficiency.

Xuan Shen, Peiyan Dong, Lei Lu et al. · harvard

Large Language Models (LLMs) stand out for their impressive performance in intricate language modeling tasks. However, their demanding computational and memory needs pose obstacles for broad use on edge devices. Quantization is then introduced to boost LLMs' on-device efficiency. Recent works show that 8-bit or lower weight quantization is feasible with minimal impact on end-to-end task performance, while the activation is still not quantized. On the other hand, mainstream commodity edge devices still struggle to execute these sub-8-bit quantized networks effectively. In this paper, we propose Agile-Quant, an activation-guided quantization framework for popular Large Language Models (LLMs), and implement an end-to-end accelerator on multiple edge devices for faster inference. Considering the hardware profiling and activation analysis, we first introduce a basic activation quantization strategy to balance the trade-off of task performance and real inference speed. Then we leverage the activation-aware token pruning technique to reduce the outliers and the adverse impact on attentivity. Ultimately, we utilize the SIMD-based 4-bit multiplier and our efficient TRIP matrix multiplication to implement the accelerator for LLMs on the edge. We apply our framework on different scales of LLMs including LLaMA, OPT, and BLOOM with 4-bit or 8-bit for the activation and 4-bit for the weight quantization. Experiments show that Agile-Quant achieves simultaneous quantization of model weights and activations while maintaining task performance comparable to existing weight-only quantization methods. Moreover, in the 8- and 4-bit scenario, Agile-Quant achieves an on-device speedup of up to 2.55x compared to its FP16 counterparts across multiple edge devices, marking a pioneering advancement in this domain. Code: https://github.com/shawnricecake/agile-quant

Xuan Shen, Peiyan Dong, Zhenglun Kong et al. · harvard

A growing trend has emerged in designing high-quality Small Language Models (SLMs) with a few million parameters. This trend is driven by the increasing concerns over cloud costs, privacy, and latency. Considering that full parameter training is feasible for SLMs on mobile devices, Quantization-Aware Training (QAT) is employed to improve efficiency by reducing computational overhead and memory footprint. However, previous QAT works adopt fine-grained quantization methods to compress models with billions of parameters on GPUs, incompatible with current commodity hardware, such as mobile and edge devices, which relies on Single Instruction Multiple Data (SIMD) instructions. Thus, the generalization of these methods to SLMs on mobile devices is limited. In this paper, we propose Squat method, an effective QAT framework with deployable quantization for SLMs on mobile devices. Specifically, we propose entropy-guided and distribution-aligned distillation to mitigate the distortion of attention information from quantization. Besides, we employ sub-8-bit token adaptive quantization, assigning varying bit widths to different tokens based on their importance. Furthermore, we develop a SIMD-based Multi-Kernel Mixed-Precision (MKMP) multiplier to support sub-8-bit mixed-precision MAC on mobile devices. Our extensive experiments verify the substantial improvements of our method compared to other QAT methods across various datasets. Furthermore, we achieve an on-device speedup of up to 2.37x compared with its FP16 counterparts, signaling a great advancement. Code: https://github.com/shawnricecake/squant

Zhenglun Kong, Zheng Zhan, Shiyue Hou et al. · harvard

Large language models (LLMs) have shown remarkable promise but remain challenging to continually improve through traditional finetuning, particularly when integrating capabilities from other specialized LLMs. Popular methods like ensemble and weight merging require substantial memory and struggle to adapt to changing data environments. Recent efforts have transferred knowledge from multiple LLMs into a single target model; however, they suffer from interference and degraded performance among tasks, largely due to limited flexibility in candidate selection and training pipelines. To address these issues, we propose a framework that adaptively selects and aggregates knowledge from diverse LLMs to build a single, stronger model, avoiding the high memory overhead of ensemble and inflexible weight merging. Specifically, we design an adaptive selection network that identifies the most relevant source LLMs based on their scores, thereby reducing knowledge interference. We further propose a dynamic weighted fusion strategy that accounts for the inherent strengths of candidate LLMs, along with a feedback-driven loss function that prevents the selector from converging on a single subset of sources. Experimental results demonstrate that our method can enable a more stable and scalable knowledge aggregation process while reducing knowledge interference by up to 50% compared to existing approaches. Code is avaliable at https://github.com/ZLKong/LLM_Integration

Jun Liu, Zhenglun Kong, Peiyan Dong et al. · harvard

Fine-tuning helps large language models (LLM) recover degraded information and enhance task performance. Although Low-Rank Adaptation (LoRA) is widely used and effective for fine-tuning, we have observed that its scaling factor can limit or even reduce performance as the rank size increases. To address this issue, we propose RoRA (Rank-adaptive Reliability Optimization), a simple yet effective method for optimizing LoRA's scaling factor. By replacing $α/r$ with $α/\sqrt{r}$, RoRA ensures improved performance as rank size increases. Moreover, RoRA enhances low-rank adaptation in fine-tuning uncompressed models and excels in the more challenging task of accuracy recovery when fine-tuning pruned models. Extensive experiments demonstrate the effectiveness of RoRA in fine-tuning both uncompressed and pruned models. RoRA surpasses the state-of-the-art (SOTA) in average accuracy and robustness on LLaMA-7B/13B, LLaMA2-7B, and LLaMA3-8B, specifically outperforming LoRA and DoRA by 6.5% and 2.9% on LLaMA-7B, respectively. In pruned model fine-tuning, RoRA shows significant advantages; for SHEARED-LLAMA-1.3, a LLaMA-7B with 81.4% pruning, RoRA achieves 5.7% higher average accuracy than LoRA and 3.9% higher than DoRA.

Jun Liu, Zhenglun Kong, Changdi Yang et al. · harvard

Multi-agent large language model (LLM) systems have shown strong potential in complex reasoning and collaborative decision-making tasks. However, most existing coordination schemes rely on static or full-context routing strategies, which lead to excessive token consumption, redundant memory exposure, and limited adaptability across interaction rounds. We introduce RCR-Router, a modular and role-aware context routing framework designed to enable efficient, adaptive collaboration in multi-agent LLMs. To our knowledge, this is the first routing approach that dynamically selects semantically relevant memory subsets for each agent based on its role and task stage, while adhering to a strict token budget. A lightweight scoring policy guides memory selection, and agent outputs are iteratively integrated into a shared memory store to facilitate progressive context refinement. To better evaluate model behavior, we further propose an Answer Quality Score metric that captures LLM-generated explanations beyond standard QA accuracy. Experiments on three multi-hop QA benchmarks -- HotPotQA, MuSiQue, and 2WikiMultihop -- demonstrate that RCR-Router reduces token usage (up to 30%) while improving or maintaining answer quality. These results highlight the importance of structured memory routing and output-aware evaluation in advancing scalable multi-agent LLM systems.

Jun Liu, Zhenglun Kong, Peiyan Dong et al. · harvard

Large language models (LLMs) exhibit strong capabilities as decision-making agents by interleaving reasoning and actions, as seen in ReAct-style frameworks. Yet, their practical deployment is constrained by high inference costs and large model sizes. We propose Structured Agent Distillation, a framework that compresses large LLM-based agents into smaller student models while preserving both reasoning fidelity and action consistency. Unlike standard token-level distillation, our method segments trajectories into {[REASON]} and {[ACT]} spans, applying segment-specific losses to align each component with the teacher's behavior. This structure-aware supervision enables compact agents to better replicate the teacher's decision process. Experiments on ALFWorld, HotPotQA-ReAct, and WebShop show that our approach consistently outperforms token-level and imitation learning baselines, achieving significant compression with minimal performance drop. Scaling and ablation results further highlight the importance of span-level alignment for efficient and deployable agents.

Zhenglun Kong, Peiyan Dong, Xiaolong Ma et al.

Recently, Vision Transformer (ViT) has continuously established new milestones in the computer vision field, while the high computation and memory cost makes its propagation in industrial production difficult. Pruning, a traditional model compression paradigm for hardware efficiency, has been widely applied in various DNN structures. Nevertheless, it stays ambiguous on how to perform exclusive pruning on the ViT structure. Considering three key points: the structural characteristics, the internal data pattern of ViTs, and the related edge device deployment, we leverage the input token sparsity and propose a computation-aware soft pruning framework, which can be set up on vanilla Transformers of both flatten and CNN-type structures, such as Pooling-based ViT (PiT). More concretely, we design a dynamic attention-based multi-head token selector, which is a lightweight module for adaptive instance-wise token selection. We further introduce a soft pruning technique, which integrates the less informative tokens generated by the selector module into a package token that will participate in subsequent calculations rather than being completely discarded. Our framework is bound to the trade-off between accuracy and computation constraints of specific edge devices through our proposed computation-aware training strategy. Experimental results show that our framework significantly reduces the computation cost of ViTs while maintaining comparable performance on image classification. Moreover, our framework can guarantee the identified model to meet resource specifications of mobile devices and FPGA, and even achieve the real-time execution of DeiT-T on mobile platforms. For example, our method reduces the latency of DeiT-T to 26 ms (26%$\sim $41% superior to existing works) on the mobile device with 0.25%$\sim $4% higher top-1 accuracy on ImageNet.

Wei Niu, Zhengang Li, Xiaolong Ma et al.

It is appealing but challenging to achieve real-time deep neural network (DNN) inference on mobile devices because even the powerful modern mobile devices are considered as ``resource-constrained'' when executing large-scale DNNs. It necessitates the sparse model inference via weight pruning, i.e., DNN weight sparsity, and it is desirable to design a new DNN weight sparsity scheme that can facilitate real-time inference on mobile devices while preserving a high sparse model accuracy. This paper designs a novel mobile inference acceleration framework GRIM that is General to both convolutional neural networks (CNNs) and recurrent neural networks (RNNs) and that achieves Real-time execution and high accuracy, leveraging fine-grained structured sparse model Inference and compiler optimizations for Mobiles. We start by proposing a new fine-grained structured sparsity scheme through the Block-based Column-Row (BCR) pruning. Based on this new fine-grained structured sparsity, our GRIM framework consists of two parts: (a) the compiler optimization and code generation for real-time mobile inference; and (b) the BCR pruning optimizations for determining pruning hyperparameters and performing weight pruning. We compare GRIM with Alibaba MNN, TVM, TensorFlow-Lite, a sparse implementation based on CSR, PatDNN, and ESE (a representative FPGA inference acceleration framework for RNNs), and achieve up to 14.08x speedup.

Peiyan Dong, Siyue Wang, Wei Niu et al.

Recurrent neural networks (RNNs) based automatic speech recognition has nowadays become prevalent on mobile devices such as smart phones. However, previous RNN compression techniques either suffer from hardware performance overhead due to irregularity or significant accuracy loss due to the preserved regularity for hardware friendliness. In this work, we propose RTMobile that leverages both a novel block-based pruning approach and compiler optimizations to accelerate RNN inference on mobile devices. Our proposed RTMobile is the first work that can achieve real-time RNN inference on mobile platforms. Experimental results demonstrate that RTMobile can significantly outperform existing RNN hardware acceleration methods in terms of inference accuracy and time. Compared with prior work on FPGA, RTMobile using Adreno 640 embedded GPU on GRU can improve the energy-efficiency by about 40$\times$ while maintaining the same inference time.

Ao Ren, Tao Zhang, Yuhao Wang et al.

The rapidly growing parameter volume of deep neural networks (DNNs) hinders the artificial intelligence applications on resource constrained devices, such as mobile and wearable devices. Neural network pruning, as one of the mainstream model compression techniques, is under extensive study to reduce the number of parameters and computations. In contrast to irregular pruning that incurs high index storage and decoding overhead, structured pruning techniques have been proposed as the promising solutions. However, prior studies on structured pruning tackle the problem mainly from the perspective of facilitating hardware implementation, without analyzing the characteristics of sparse neural networks. The neglect on the study of sparse neural networks causes inefficient trade-off between regularity and pruning ratio. Consequently, the potential of structurally pruning neural networks is not sufficiently mined. In this work, we examine the structural characteristics of the irregularly pruned weight matrices, such as the diverse redundancy of different rows, the sensitivity of different rows to pruning, and the positional characteristics of retained weights. By leveraging the gained insights as a guidance, we first propose the novel block-max weight masking (BMWM) method, which can effectively retain the salient weights while imposing high regularity to the weight matrix. As a further optimization, we propose a density-adaptive regular-block (DARB) pruning that outperforms prior structured pruning work with high pruning ratio and decoding efficiency. Our experimental results show that DARB can achieve 13$\times$ to 25$\times$ pruning ratio, which are 2.8$\times$ to 4.3$\times$ improvements than the state-of-the-art counterparts on multiple neural network models and tasks. Moreover, DARB can achieve 14.3$\times$ decoding efficiency than block pruning with higher pruning ratio.