The equivalent nature of 3D coordinates has posed long term challenges in protein structure representation learning, alignment, and generation. Can we create a compact and invariant language that equivalently represents protein structures? Towards this goal, we propose FoldToken2 to transfer equivariant structures into discrete tokens, while maintaining the recoverability of the original structures. From FoldToken1 to FoldToken2, we improve three key components: (1) invariant structure encoder, (2) vector-quantized compressor, and (3) equivalent structure decoder. We evaluate FoldToken2 on the protein structure reconstruction task and show that it outperforms previous FoldToken1 by 20\% in TMScore and 81\% in RMSD. FoldToken2 probably be the first method that works well on both single-chain and multi-chain protein structures quantization. We believe that FoldToken2 will inspire further improvement in protein structure representation learning, structure alignment, and structure generation tasks.
We present new fundamental results for the mean square error (MSE)-optimal conditional mean estimator (CME) in one-bit quantized systems for a Gaussian mixture model (GMM) distributed signal of interest, possibly corrupted by additive white Gaussian noise (AWGN). We first derive novel closed-form analytic expressions for the Bussgang estimator, the well-known linear minimum mean square error (MMSE) estimator in quantized systems. Afterward, closed-form analytic expressions for the CME in special cases are presented, revealing that the optimal estimator is linear in the one-bit quantized observation, opposite to higher resolution cases. Through a comparison to the recently studied Gaussian case, we establish a novel MSE inequality and show that that the signal of interest is correlated with the auxiliary quantization noise. We extend our analysis to multiple observation scenarios, examining the MSE-optimal transmit sequence and conducting an asymptotic analysis, yielding analytic expressions for the MSE and its limit. These contributions have broad impact for the analysis and design of various signal processing applications.
The deployment of Large Language Models (LLMs) on edge devices is increasingly important to enhance on-device intelligence. Weight quantization is crucial for reducing the memory footprint of LLMs on devices. However, low-bit LLMs necessitate mixed precision matrix multiplication (mpGEMM) of low precision weights and high precision activations during inference. Existing systems, lacking native support for mpGEMM, resort to dequantize weights for high precision computation. Such an indirect way can lead to a significant inference overhead. In this paper, we introduce T-MAC, an innovative lookup table(LUT)-based method designed for efficient low-bit LLM (i.e., weight-quantized LLM) inference on CPUs. T-MAC directly supports mpGEMM without dequantization, while simultaneously eliminating multiplications and reducing additions required. Specifically, T-MAC transforms the traditional data-type-centric multiplication to bit-wise table lookup, and enables a unified and scalable mpGEMM solution. Our LUT-based kernels scale linearly to the weight bit-width. Evaluated on low-bit Llama and BitNet models, T-MAC demonstrates up to 4x increase in throughput and 70% reduction in energy consumption compared to llama.cpp. For BitNet-b1.58-3B, T-MAC delivers a token generation throughput of 30 tokens/s with a single core and 71 tokens/s with eight cores on M2-Ultra, and 11 tokens/s on lower-end devices like Raspberry Pi 5, which significantly exceeds the adult average reading speed. T-MAC with LUT-based computing paradigm, paves the way for the practical deployment of low-bit LLMs on resource-constrained edge devices without compromising computational efficiency. The system is open-sourced at https://github.com/microsoft/T-MAC .
Quantization techniques are widely used to improve inference speed and deployment of large language models. While a wide body of work examines the impact of quantization on LLMs in English, none have evaluated across languages. We conduct a thorough analysis of quantized multilingual LLMs, focusing on performance across languages and at varying scales. We use automatic benchmarks, LLM-as-a-Judge, and human evaluation, finding that (1) harmful effects of quantization are apparent in human evaluation, which automatic metrics severely underestimate: a 1.7% average drop in Japanese across automatic tasks corresponds to a 16.0% drop reported by human evaluators on realistic prompts; (2) languages are disparately affected by quantization, with non-Latin script languages impacted worst; and (3) challenging tasks like mathematical reasoning degrade fastest. As the ability to serve low-compute models is critical for wide global adoption of NLP technologies, our results urge consideration of multilingual performance as a key evaluation criterion for efficient models.
The rapid advancement of large language models (LLMs) has facilitated their transformation into conversational chatbots that can grasp contextual nuances and generate pertinent sentences, closely mirroring human values through advanced techniques such as instruction tuning and reinforcement learning from human feedback (RLHF). However, the computational efficiency required for LLMs, achieved through techniques like post-training quantization (PTQ), presents challenges such as token-flipping that can impair chatbot performance. In response, we propose a novel preference alignment approach, quantization-aware direct preference optimization (QDPO), that aligns quantized LLMs with their full-precision counterparts, improving conversational abilities. Evaluated on two instruction-tuned LLMs in various languages, QDPO demonstrated superior performance in improving conversational abilities compared to established PTQ and knowledge-distillation fine-tuning techniques, marking a significant step forward in the development of efficient and effective conversational LLMs.
Vision Transformers (ViTs) have exhibited exceptional performance across diverse computer vision tasks, while their substantial parameter size incurs significantly increased memory and computational demands, impeding effective inference on resource-constrained devices. Quantization has emerged as a promising solution to mitigate these challenges, yet existing methods still suffer from significant accuracy loss at low-bit. We attribute this issue to the distinctive distributions of post-LayerNorm and post-GELU activations within ViTs, rendering conventional hardware-friendly quantizers ineffective, particularly in low-bit scenarios. To address this issue, we propose a novel framework called Activation-Distribution-Friendly post-training Quantization for Vision Transformers, ADFQ-ViT. Concretely, we introduce the Per-Patch Outlier-aware Quantizer to tackle irregular outliers in post-LayerNorm activations. This quantizer refines the granularity of the uniform quantizer to a per-patch level while retaining a minimal subset of values exceeding a threshold at full-precision. To handle the non-uniform distributions of post-GELU activations between positive and negative regions, we design the Shift-Log2 Quantizer, which shifts all elements to the positive region and then applies log2 quantization. Moreover, we present the Attention-score enhanced Module-wise Optimization which adjusts the parameters of each quantizer by reconstructing errors to further mitigate quantization error. Extensive experiments demonstrate ADFQ-ViT provides significant improvements over various baselines in image classification, object detection, and instance segmentation tasks at 4-bit. Specifically, when quantizing the ViT-B model to 4-bit, we achieve a 10.23% improvement in Top-1 accuracy on the ImageNet dataset.
With the rapid advances in deep learning and smart manufacturing in Industry 4.0, there is an imperative for high-throughput, high-performance, and fully integrated visual inspection systems. Most anomaly detection approaches using defect detection datasets, such as MVTec AD, employ one-class models that require fitting separate models for each class. On the contrary, unified models eliminate the need for fitting separate models for each class and significantly reduce cost and memory requirements. Thus, in this work, we experiment with considering a unified multi-class setup. Our experimental study shows that multi-class models perform at par with one-class models for the standard MVTec AD dataset. Hence, this indicates that there may not be a need to learn separate object/class-wise models when the object classes are significantly different from each other, as is the case of the dataset considered. Furthermore, we have deployed three different unified lightweight architectures on the CPU and an edge device (NVIDIA Jetson Xavier NX). We analyze the quantized multi-class anomaly detection models in terms of latency and memory requirements for deployment on the edge device while comparing quantization-aware training (QAT) and post-training quantization (PTQ) for performance at different precision widths. In addition, we explored two different methods of calibration required in post-training scenarios and show that one of them performs notably better, highlighting its importance for unsupervised tasks. Due to quantization, the performance drop in PTQ is further compensated by QAT, which yields at par performance with the original 32-bit Floating point in two of the models considered.
Fast convolution algorithms, including Winograd and FFT, can efficiently accelerate convolution operations in deep models. However, these algorithms depend on high-precision arithmetic to maintain inference accuracy, which conflicts with the model quantization. To resolve this conflict and further improve the efficiency of quantized convolution, we proposes SFC, a new algebra transform for fast convolution by extending the Discrete Fourier Transform (DFT) with symbolic computing, in which only additions are required to perform the transformation at specific transform points, avoiding the calculation of irrational number and reducing the requirement for precision. Additionally, we enhance convolution efficiency by introducing correction terms to convert invalid circular convolution outputs of the Fourier method into effective ones. The numerical error analysis is presented for the first time in this type of work and proves that our algorithms can provide a 3.68x multiplication reduction for 3x3 convolution, while the Winograd algorithm only achieves a 2.25x reduction with similarly low numerical errors. Experiments carried out on benchmarks and FPGA show that our new algorithms can further improve the computation efficiency of quantized models while maintaining accuracy, surpassing both the quantization-alone method and existing works on fast convolution quantization.
Due to their large size, generative Large Language Models (LLMs) require significant computing and storage resources. This paper introduces a new post-training quantization method, GPTQT, to reduce memory usage and enhance processing speed by expressing the weight of LLM in 3bit/2bit. Practice has shown that minimizing the quantization error of weights is ineffective, leading to overfitting. Therefore, GPTQT employs a progressive two-step approach: initially quantizing weights using Linear quantization to a relatively high bit, followed by converting obtained int weight to lower bit binary coding. A re-explore strategy is proposed to optimize initial scaling factor. During inference, these steps are merged into pure binary coding, enabling efficient computation. Testing across various models and datasets confirms GPTQT's effectiveness. Compared to the strong 3-bit quantization baseline, GPTQT further reduces perplexity by 4.01 on opt-66B and increases speed by 1.24 times on opt-30b. The results on Llama2 show that GPTQT is currently the best binary coding quantization method for such kind of LLMs.
Building generalized models that can solve many computer vision tasks simultaneously is an intriguing direction. Recent works have shown image itself can be used as a natural interface for general-purpose visual perception and demonstrated inspiring results. In this paper, we explore diffusion-based vision generalists, where we unify different types of dense prediction tasks as conditional image generation and re-purpose pre-trained diffusion models for it. However, directly applying off-the-shelf latent diffusion models leads to a quantization issue. Thus, we propose to perform diffusion in pixel space and provide a recipe for finetuning pre-trained text-to-image diffusion models for dense vision tasks. In experiments, we evaluate our method on four different types of tasks and show competitive performance to the other vision generalists.
Long context capability is a crucial competency for large language models (LLMs) as it mitigates the human struggle to digest long-form texts. This capability enables complex task-solving scenarios such as book summarization, code assistance, and many more tasks that are traditionally manpower-intensive. However, transformer-based LLMs face significant challenges with long context input due to the growing size of the KV cache and the intrinsic complexity of attending to extended inputs; where multiple schools of efficiency-driven approaches - such as KV cache quantization, token dropping, prompt compression, linear-time sequence models, and hybrid architectures - have been proposed to produce efficient yet long context-capable models. Despite these advancements, no existing work has comprehensively benchmarked these methods in a reasonably aligned environment. In this work, we fill this gap by providing a taxonomy of current methods and evaluating 10+ state-of-the-art approaches across seven categories of long context tasks. Our work reveals numerous previously unknown phenomena and offers insights - as well as a friendly workbench - for the future development of long context-capable LLMs. The source code is available at https://github.com/henryzhongsc/longctx_bench.
A number of production deep learning clusters have attempted to explore inference hardware for DNN training, at the off-peak serving hours with many inference GPUs idling. Conducting DNN training with a combination of heterogeneous training and inference GPUs, known as hybrid device training, presents considerable challenges due to disparities in compute capability and significant differences in memory capacity. We propose QSync, a training system that enables efficient synchronous data-parallel DNN training over hybrid devices by strategically exploiting quantized operators. According to each device's available resource capacity, QSync selects a quantization-minimized setting for operators in the distributed DNN training graph, minimizing model accuracy degradation but keeping the training efficiency brought by quantization. We carefully design a predictor with a bi-directional mixed-precision indicator to reflect the sensitivity of DNN layers on fixed-point and floating-point low-precision operators, a replayer with a neighborhood-aware cost mapper to accurately estimate the latency of distributed hybrid mixed-precision training, and then an allocator that efficiently synchronizes workers with minimized model accuracy degradation. QSync bridges the computational graph on PyTorch to an optimized backend for quantization kernel performance and flexible support for various GPU architectures. Extensive experiments show that QSync's predictor can accurately simulate distributed mixed-precision training with <5% error, with a consistent 0.27-1.03% accuracy improvement over the from-scratch training tasks compared to uniform precision.
A number of companies recently worked together to release the new Open Compute Project MX standard for low-precision computation, aimed at efficient neural network implementation. In this paper, we describe and evaluate the first open-source FPGA implementation of the arithmetic defined in the standard. Our designs fully support all the standard's concrete formats for conversion into and out of MX formats and for the standard-defined arithmetic operations, as well as arbitrary fixed-point and floating-point formats. Certain elements of the standard are left as implementation-defined, and we present the first concrete FPGA-inspired choices for these elements, which we outline in the paper. Our library of optimized hardware components is available open source, and can be used to build larger systems. For this purpose, we also describe and release an open-source Pytorch library for quantization into the new standard, integrated with the Brevitas library so that the community can develop novel neural network designs quantized with MX formats in mind. We demonstrate the usability and efficacy of our libraries via the implementation of example neural networks such as ResNet-18 on the ImageNet ILSVRC12 dataset. Our testing shows that MX is very effective for formats such as INT5 or FP6 which are not natively supported on GPUs. This gives FPGAs an advantage as they have the flexibility to implement a custom datapath and take advantage of the smaller area footprints offered by these formats.
Gradient aggregation has long been identified as a major bottleneck in today's large-scale distributed machine learning training systems. One promising solution to mitigate such bottlenecks is gradient compression, directly reducing communicated gradient data volume. However, in practice, many gradient compression schemes do not achieve acceleration of the training process while also preserving accuracy. In this work, we identify common issues in previous gradient compression systems and evaluation methodologies. These include excessive computational overheads; incompatibility with all-reduce; and insufficient evaluation methods, such as not using an end-to-end metric or using a 32-bit baseline instead of the stronger 16-bit baseline. We revisit common compression approaches (sparsification, quantization, and low-rank decomposition) and demonstrate how considering the above issues can lead to minor but strategic design changes, resulting in notably better performance. Our goal is to raise awareness of the need for design and evaluation standards that naturally translate to the end-to-end utility of gradient compression.
The resource requirements of deep neural networks (DNNs) pose significant challenges to their deployment on edge devices. Common approaches to address this issue are pruning and mixed-precision quantization, which lead to latency and memory occupation improvements. These optimization techniques are usually applied independently. We propose a novel methodology to apply them jointly via a lightweight gradient-based search, and in a hardware-aware manner, greatly reducing the time required to generate Pareto-optimal DNNs in terms of accuracy versus cost (i.e., latency or memory). We test our approach on three edge-relevant benchmarks, namely CIFAR-10, Google Speech Commands, and Tiny ImageNet. When targeting the optimization of the memory footprint, we are able to achieve a size reduction of 47.50% and 69.54% at iso-accuracy with the baseline networks with all weights quantized at 8 and 2-bit, respectively. Our method surpasses a previous state-of-the-art approach with up to 56.17% size reduction at iso-accuracy. With respect to the sequential application of state-of-the-art pruning and mixed-precision optimizations, we obtain comparable or superior results, but with a significantly lowered training time. In addition, we show how well-tailored cost models can improve the cost versus accuracy trade-offs when targeting specific hardware for deployment.
Intelligent mobile agents (e.g., UGVs and UAVs) typically demand low power/energy consumption when solving their machine learning (ML)-based tasks, since they are usually powered by portable batteries with limited capacity. A potential solution is employing neuromorphic computing with Spiking Neural Networks (SNNs), which leverages event-based computation to enable ultra-low power/energy ML algorithms. To maximize the performance efficiency of SNN inference, the In-Memory Computing (IMC)-based hardware accelerators with emerging device technologies (e.g., RRAM) can be employed. However, SNN models are typically developed without considering constraints from the application and the underlying IMC hardware, thereby hindering SNNs from reaching their full potential in performance and efficiency. To address this, we propose NeuroNAS, a novel framework for developing energyefficient neuromorphic IMC for intelligent mobile agents using hardware-aware spiking neural architecture search (NAS), i.e., by quickly finding an SNN architecture that offers high accuracy under the given constraints (e.g., memory, area, latency, and energy consumption). Its key steps include: optimizing SNN operations to enable efficient NAS, employing quantization to minimize the memory footprint, developing an SNN architecture that facilitates an effective learning, and devising a systematic hardware-aware search algorithm to meet the constraints. Compared to the state-of-the-art techniques, NeuroNAS quickly finds SNN architectures (with 8bit weight precision) that maintain high accuracy by up to 6.6x search time speed-ups, while achieving up to 92% area savings, 1.2x latency improvements, 84% energy savings across different datasets (i.e., CIFAR-10, CIFAR-100, and TinyImageNet-200); while the state-of-the-art fail to meet all constraints at once.
We consider the problem of joint source-channel coding for semantic communication from a rateless perspective, the purpose of which is to settle the balance between reliability (distortion/perception) and effectiveness (rate) of transmission over uncertain channels. In particular, we propose a more general communication objective that minimizes the perceptual distance by incorporating a semantic-level reconstruction objective in addition to the conventional pixel-level reconstruction objective. Based on the proposed objective, we then propose a novel JSCC coding scheme called rateless stochastic coding (RSC) by introducing a generative decoder and dithered quantization. The coding scheme enables reconstruction based on both distortion and perception metrics through rateless transmission. Extensive experiments demonstrate that the proposed RSC can achieve variable rates of transmission maintaining an excellent trade-off between distortion and perception.
As large language models (LLMs) grow in size and deployment scale, quantization has become an essential technique for reducing memory footprint and improving inference efficiency. However, existing quantization toolkits often lack transparency, flexibility, and system-level scalability across GPUs and distributed environments. We present \textbf{LLMEasyQuant}, a modular, system-aware quantization framework designed for efficient, low-bit inference of LLMs on single-node multi-GPU, multi-node, and edge hardware. LLMEasyQuant supports a wide range of quantization methods -- including Symmetric Quantization, ZeroQuant, SmoothQuant, and SimQuant -- with unified interfaces for per-layer calibration, bitwidth assignment, and runtime adaptation. It integrates fused CUDA kernels with NCCL-based distributed synchronization and supports both static and online quantization. Empirical results show that LLMEasyQuant can achieve substantial speedups in GEMM execution, HBM load time, and near-linear multi-GPU scaling. Ablation studies further validate its ability to balance latency, memory, and accuracy under diverse deployment conditions. LLMEasyQuant offers a practical quantization serving system for scalable, hardware-optimized LLM inference.
Extreme data rate scientific experiments create massive amounts of data that require efficient ML edge processing. This leads to unique validation challenges for VLSI implementations of ML algorithms: enabling bit-accurate functional simulations for performance validation in experimental software frameworks, verifying those ML models are robust under extreme quantization and pruning, and enabling ultra-fine-grained model inspection for efficient fault tolerance. We discuss approaches to developing and validating reliable algorithms at the scientific edge under such strict latency, resource, power, and area requirements in extreme experimental environments. We study metrics for developing robust algorithms, present preliminary results and mitigation strategies, and conclude with an outlook of these and future directions of research towards the longer-term goal of developing autonomous scientific experimentation methods for accelerated scientific discovery.
The massive deployment of Machine Learning (ML) models raises serious concerns about data protection. Privacy-enhancing technologies (PETs) offer a promising first step, but hard challenges persist in achieving confidentiality and differential privacy in distributed learning. In this paper, we describe a novel, regulation-compliant data protection technique for the distributed training of ML models, applicable throughout the ML life cycle regardless of the underlying ML architecture. Designed from the data owner's perspective, our method protects both training data and ML model parameters by employing a protocol based on a quantized multi-hash data representation Hash-Comb combined with randomization. The hyper-parameters of our scheme can be shared using standard Secure Multi-Party computation protocols. Our experimental results demonstrate the robustness and accuracy-preserving properties of our approach.