Transforming complex actions into discrete skill abstractions has demonstrated strong potential for robotic manipulation. Existing approaches mainly leverage latent variable models, e.g., VQ-VAE, to learn skill abstractions through learned vectors (codebooks), while they suffer from codebook collapse and modeling the causal relationship between learned skills. To address these limitations, we present \textbf{S}kill \textbf{T}raining with \textbf{A}ugmented \textbf{R}otation (\textbf{STAR}), a framework that advances both skill learning and composition to complete complex behaviors. Specifically, to prevent codebook collapse, we devise rotation-augmented residual skill quantization (RaRSQ). It encodes relative angles between encoder outputs into the gradient flow by rotation-based gradient mechanism. Points within the same skill code are forced to be either pushed apart or pulled closer together depending on gradient directions. Further, to capture the causal relationship between skills, we present causal skill transformer (CST) which explicitly models dependencies between skill representations through an autoregressive mechanism for coherent action generation. Extensive experiments demonstrate the superiority of STAR on both LIBERO benchmark and realworld tasks, with around 12\% improvement over the baselines.
This paper introduces an innovative error feedback framework designed to mitigate quantization noise in distributed graph filtering, where communications are constrained to quantized messages. It comes from error spectrum shaping techniques from state-space digital filters, and therefore establishes connections between quantized filtering processes over different domains. In contrast to existing error compensation methods, our framework quantitatively feeds back the quantization noise for exact compensation. We examine the framework under three key scenarios: (i) deterministic graph filtering, (ii) graph filtering over random graphs, and (iii) graph filtering with random node-asynchronous updates. Rigorous theoretical analysis demonstrates that the proposed framework significantly reduces the effect of quantization noise, and we provide closed-form solutions for the optimal error feedback coefficients. Moreover, this quantitative error feedback mechanism can be seamlessly integrated into communication-efficient decentralized optimization frameworks, enabling lower error floors. Numerical experiments validate the theoretical results, consistently showing that our method outperforms conventional quantization strategies in terms of both accuracy and robustness.
Personalized federated learning (PFL) offers a solution to balancing personalization and generalization by conducting federated learning (FL) to guide personalized learning (PL). Little attention has been given to wireless PFL (WPFL), where privacy concerns arise. Performance fairness of PL models is another challenge resulting from communication bottlenecks in WPFL. This paper exploits quantization errors to enhance the privacy of WPFL and proposes a novel quantization-assisted Gaussian differential privacy (DP) mechanism. We analyze the convergence upper bounds of individual PL models by considering the impact of the mechanism (i.e., quantization errors and Gaussian DP noises) and imperfect communication channels on the FL of WPFL. By minimizing the maximum of the bounds, we design an optimal transmission scheduling strategy that yields min-max fairness for WPFL with OFDMA interfaces. This is achieved by revealing the nested structure of this problem to decouple it into subproblems solved sequentially for the client selection, channel allocation, and power control, and for the learning rates and PL-FL weighting coefficients. Experiments validate our analysis and demonstrate that our approach substantially outperforms alternative scheduling strategies by 87.08%, 16.21%, and 38.37% in accuracy, the maximum test loss of participating clients, and fairness (Jain's index), respectively.
Efficiently serving large language models (LLMs) under dynamic and bursty workloads remains a key challenge for real-world deployment. Existing serving frameworks and static model compression techniques fail to adapt to workload fluctuations, leading to either service-level objective (SLO) violations under full-precision serving or persistent accuracy degradation with static quantization. We present MorphServe, a dynamic, workload-aware LLM serving framework based on morphological adaptation. MorphServe introduces two asynchronous, token-level runtime mechanisms: quantized layer swapping, which selectively replaces less impactful layers with quantized alternatives during high-load periods, and pressure-aware KV cache resizing, which dynamically adjusts KV cache capacity in response to memory pressure. These mechanisms enable state-preserving transitions with minimum runtime overhead and are fully compatible with modern scheduling and attention techniques. Extensive experiments on Vicuna and Llama family models with real-world workloads demonstrate that MorphServe reduces average SLO violations by 92.45 percent and improves the P95 TTFT latency by 2.2x-3.9x compared to full-precision serving, without compromising generation quality. These results establish MorphServe as a practical and elastic solution for LLM deployment in dynamic environments.
This study uses Jordanian law as a case study to explore the fine-tuning of the Llama-3.1 large language model for Arabic question-answering. Two versions of the model- Llama-3.1-8B-bnb-4bit and Llama-3.1-8B-Instruct-bnb-4bit-were fine-tuned using parameter-efficient fine-tuning (PEFT) with LoRA adapters and 4-bit quantized models, leveraging the Unsloth framework for accelerated and resource-efficient training. A custom dataset of 6000 legal question-answer pairs was curated from Jordanian laws and formatted into structured prompts. Performance was evaluated using the BLEU and the ROUGE metrics to compare the fine-tuned models to their respective base versions. Results demonstrated improved legal reasoning and accuracy while achieving resource efficiency through quantization and optimized fine-tuning strategies. This work underscores the potential of adapting large language models for Arabic legal domains and highlights effective techniques for fine-tuning domain-specific tasks.
Physics-driven artificial intelligence (PD-AI) reconstruction methods have emerged as the state-of-the-art for accelerating MRI scans, enabling higher spatial and temporal resolutions. However, the high resolution of these scans generates massive data volumes, leading to challenges in transmission, storage, and real-time processing. This is particularly pronounced in functional MRI, where hundreds of volumetric acquisitions further exacerbate these demands. Edge computing with FPGAs presents a promising solution for enabling PD-AI reconstruction near the MRI sensors, reducing data transfer and storage bottlenecks. However, this requires optimization of PD-AI models for hardware efficiency through quantization and bypassing traditional FFT-based approaches, which can be a limitation due to their computational demands. In this work, we propose a novel PD-AI computational MRI approach optimized for FPGA-based edge computing devices, leveraging 8-bit complex data quantization and eliminating redundant FFT/IFFT operations. Our results show that this strategy improves computational efficiency while maintaining reconstruction quality comparable to conventional PD-AI methods, and outperforms standard clinical methods. Our approach presents an opportunity for high-resolution MRI reconstruction on resource-constrained devices, highlighting its potential for real-world deployment.
Training SER models in natural, spontaneous speech is especially challenging due to the subtle expression of emotions and the unpredictable nature of real-world audio. In this paper, we present a robust system for the INTERSPEECH 2025 Speech Emotion Recognition in Naturalistic Conditions Challenge, focusing on categorical emotion recognition. Our method combines state-of-the-art audio models with text features enriched by prosodic and spectral cues. In particular, we investigate the effectiveness of Fundamental Frequency (F0) quantization and the use of a pretrained audio tagging model. We also employ an ensemble model to improve robustness. On the official test set, our system achieved a Macro F1-score of 39.79% (42.20% on validation). Our results underscore the potential of these methods, and analysis of fusion techniques confirmed the effectiveness of Graph Attention Networks. Our source code is publicly available.
The Wasserstein metric has become increasingly important in many machine learning applications such as generative modeling, image retrieval and domain adaptation. Despite its appeal, it is often too costly to compute. This has motivated approximation methods like entropy-regularized optimal transport, downsampling, and subsampling, which trade accuracy for computational efficiency. In this paper, we consider the challenge of computing efficient approximations to the Wasserstein metric that also serve as strict upper or lower bounds. Focusing on discrete measures on regular grids, our approach involves formulating and exactly solving a Kantorovich problem on a coarse grid using a quantized measure and specially designed cost matrix, followed by an upscaling and correction stage. This is done either in the primal or dual space to obtain valid upper and lower bounds on the Wasserstein metric of the full-resolution inputs. We evaluate our methods on the DOTmark optimal transport images benchmark, demonstrating a 10x-100x speedup compared to entropy-regularized OT while keeping the approximation error below 2%.
Analog In-Memory Compute (AIMC) can improve the energy efficiency of Deep Learning by orders of magnitude. Yet analog-domain device and circuit non-idealities -- within the analog ``Tiles'' performing Matrix-Vector Multiply (MVM) operations -- can degrade neural-network task accuracy. We quantify the impact of low-level distortions and noise, and develop a mathematical model for Multiply-ACcumulate (MAC) operations mapped to analog tiles. Instantaneous-current IR-drop (the most significant circuit non-ideality), and ADC quantization effects are fully captured by this model, which can predict MVM tile-outputs both rapidly and accurately, as compared to much slower rigorous circuit simulations. A statistical model of PCM read noise at nanosecond timescales is derived from -- and matched against -- experimental measurements. We integrate these (statistical) device and (deterministic) circuit effects into a PyTorch-based framework to assess the accuracy impact on the BERT and ALBERT Transformer networks. We show that hardware-aware fine-tuning using simple Gaussian noise provides resilience against ADC quantization and PCM read noise effects, but is less effective against IR-drop. This is because IR-drop -- although deterministic -- is non-linear, is changing significantly during the time-integration window, and is ultimately dependent on all the excitations being introduced in parallel into the analog tile. The apparent inability of simple Gaussian noise applied during training to properly prepare a DNN network for IR-drop during inference implies that more complex training approaches -- incorporating advances such as the Tile-circuit model introduced here -- will be critical for resilient deployment of large neural networks onto AIMC hardware.
As large language models (LLMs) are increasingly used in legal applications, current evaluation benchmarks tend to focus mainly on factual accuracy while largely neglecting important linguistic quality aspects such as clarity, coherence, and terminology. To address this gap, we propose three steps: First, we develop a regression model to evaluate the quality of legal texts based on clarity, coherence, and terminology. Second, we create a specialized set of legal questions. Third, we analyze 49 LLMs using this evaluation framework. Our analysis identifies three key findings: First, model quality levels off at 14 billion parameters, with only a marginal improvement of $2.7\%$ noted at 72 billion parameters. Second, engineering choices such as quantization and context length have a negligible impact, as indicated by statistical significance thresholds above 0.016. Third, reasoning models consistently outperform base architectures. A significant outcome of our research is the release of a ranking list and Pareto analysis, which highlight the Qwen3 series as the optimal choice for cost-performance tradeoffs. This work not only establishes standardized evaluation protocols for legal LLMs but also uncovers fundamental limitations in current training data refinement approaches. Code and models are available at: https://github.com/lyxx3rd/LegalEval-Q.
As large language models (LLMs) are increasingly deployed in high-stakes applications, robust uncertainty estimation is essential for ensuring the safe and trustworthy deployment of LLMs. We present the most comprehensive study to date of uncertainty estimation in LLMs, evaluating 80 models spanning open- and closed-source families, dense and Mixture-of-Experts (MoE) architectures, reasoning and non-reasoning modes, quantization variants and parameter scales from 0.6B to 671B. Focusing on three representative black-box single-pass methods, including token probability-based uncertainty (TPU), numerical verbal uncertainty (NVU), and linguistic verbal uncertainty (LVU), we systematically evaluate uncertainty calibration and selective classification using the challenging MMLU-Pro benchmark, which covers both reasoning-intensive and knowledge-based tasks. Our results show that LVU consistently outperforms TPU and NVU, offering stronger calibration and discrimination while being more interpretable. We also find that high accuracy does not imply reliable uncertainty, and that model scale, post-training, reasoning ability and quantization all influence estimation performance. Notably, LLMs exhibit better uncertainty estimates on reasoning tasks than on knowledge-heavy ones, and good calibration does not necessarily translate to effective error ranking. These findings highlight the need for multi-perspective evaluation and position LVU as a practical tool for improving the reliability of LLMs in real-world settings.
With the increasing size of frontier LLMs, post-training quantization has become the standard for memory-efficient deployment. Recent work has shown that basic rounding-based quantization schemes pose security risks, as they can be exploited to inject malicious behaviors into quantized models that remain hidden in full precision. However, existing attacks cannot be applied to more complex quantization methods, such as the GGUF family used in the popular ollama and llama$.$cpp frameworks. In this work, we address this gap by introducing the first attack on GGUF. Our key insight is that the quantization error -- the difference between the full-precision weights and their (de-)quantized version -- provides sufficient flexibility to construct malicious quantized models that appear benign in full precision. Leveraging this, we develop an attack that trains the target malicious LLM while constraining its weights based on quantization errors. We demonstrate the effectiveness of our attack on three popular LLMs across nine GGUF quantization data types on three diverse attack scenarios: insecure code generation ($\Delta$=$88.7\%$), targeted content injection ($\Delta$=$85.0\%$), and benign instruction refusal ($\Delta$=$30.1\%$). Our attack highlights that (1) the most widely used post-training quantization method is susceptible to adversarial interferences, and (2) the complexity of quantization schemes alone is insufficient as a defense.
Memristor-based hardware offers new possibilities for energy-efficient machine learning (ML) by providing analog in-memory matrix multiplication. Current hardware prototypes cannot fit large neural networks, and related literature covers only small ML models for tasks like MNIST or single word recognition. Simulation can be used to explore how hardware properties affect larger models, but existing software assumes simplified hardware. We propose a PyTorch-based library based on "Synaptogen" to simulate neural network execution with accurately captured memristor hardware properties. For the first time, we show how an ML system with millions of parameters would behave on memristor hardware, using a Conformer trained on the speech recognition task TED-LIUMv2 as example. With adjusted quantization-aware training, we limit the relative degradation in word error rate to 25% when using a 3-bit weight precision to execute linear operations via simulated analog computation.
Weight binarization has emerged as a promising strategy to drastically reduce the complexity of large language models (LLMs). It is mainly classified into two approaches: post-training binarization and finetuning with training-aware binarization methods. The first approach, while having low complexity, leads to significant loss of information from the original LLMs, resulting in poor performance. The second approach, on the other hand, relies heavily on full-precision latent weights for gradient approximation of binary weights, which not only remains suboptimal but also introduces substantial complexity. In this paper, we introduce a novel framework that effectively transforms LLMs into multi-kernel Boolean parameters, for the first time, finetunes them directly in the Boolean domain, eliminating the need for expensive latent weights. This significantly reduces complexity during both finetuning and inference. Through extensive and insightful experiments across a wide range of LLMs, we demonstrate that our method outperforms recent ultra low-bit quantization and binarization methods.
Climate Finance Bench introduces an open benchmark that targets question-answering over corporate climate disclosures using Large Language Models. We curate 33 recent sustainability reports in English drawn from companies across all 11 GICS sectors and annotate 330 expert-validated question-answer pairs that span pure extraction, numerical reasoning, and logical reasoning. Building on this dataset, we propose a comparison of RAG (retrieval-augmented generation) approaches. We show that the retriever's ability to locate passages that actually contain the answer is the chief performance bottleneck. We further argue for transparent carbon reporting in AI-for-climate applications, highlighting advantages of techniques such as Weight Quantization.
Speculative decoding and quantization effectively accelerate memory-bound inference of large language models. Speculative decoding mitigates the memory bandwidth bottleneck by verifying multiple tokens within a single forward pass, which increases computational effort. Quantization achieves this optimization by compressing weights and activations into lower bit-widths and also reduces computations via low-bit matrix multiplications. To further leverage their strengths, we investigate the integration of these two techniques. Surprisingly, experiments applying the advanced speculative decoding method EAGLE-2 to various quantized models reveal that the memory benefits from 4-bit weight quantization are diminished by the computational load from speculative decoding. Specifically, verifying a tree-style draft incurs significantly more time overhead than a single-token forward pass on 4-bit weight quantized models. This finding led to our new speculative decoding design: a hierarchical framework that employs a small model as an intermediate stage to turn tree-style drafts into sequence drafts, leveraging the memory access benefits of the target quantized model. Experimental results show that our hierarchical approach achieves a 2.78$\times$ speedup across various tasks for the 4-bit weight Llama-3-70B model on an A100 GPU, outperforming EAGLE-2 by 1.31$\times$. Code available at https://github.com/AI9Stars/SpecMQuant.
DiLoCo is a powerful framework for training large language models (LLMs) under networking constraints with advantages for increasing parallelism and accelerator utilization in data center settings. Despite significantly reducing communication frequency, however, DiLoCo's communication steps still involve all-reducing a complete copy of the model's parameters. While existing works have explored ways to reduce communication in DiLoCo, the role of error feedback accumulators and the effect of the inner-optimizer on compressibility remain under-explored. In this work, we investigate the effectiveness of standard compression methods including Top-k sparsification and quantization for reducing the communication overhead of DiLoCo when paired with two local optimizers (AdamW and Muon). Our experiments pre-training decoder-only transformer language models (LMs) reveal that leveraging Muon as the inner optimizer for DiLoCo along with an error-feedback accumulator allows to aggressively compress the communicated delta to 2-bits with next to no performance degradation. Crucially, MuLoCo (Muon inner optimizer DiLoCo) significantly outperforms DiLoCo while communicating 8X less and having identical memory complexity.
The main goal of post-training quantization (PTQ) is to produced a compressed model whose output distribution is as close to the original model's as possible. To do this tractably, almost all LLM PTQ algorithms quantize linear layers by independently minimizing the immediate activation error. However, this localized objective ignores the effect of subsequent layers, so reducing it does not necessarily give a closer model. In this work, we introduce Yet Another Quantization Algorithm (YAQA), an adaptive rounding algorithm that uses Kronecker-factored approximations of each linear layer's Hessian with respect to the \textit{full model} KL divergence. YAQA consists of two components: Kronecker-factored sketches of the full layerwise Hessian that can be tractably computed for hundred-billion parameter LLMs, and a quantizer-independent rounding algorithm that uses these sketches and comes with theoretical guarantees. Across a wide range of models and quantizers, YAQA empirically reduces the KL divergence to the original model by $\approx 30\%$ while achieving state of the art performance on downstream tasks.
Model merging has emerged as a powerful technique for combining task-specific weights, achieving superior performance in multi-target domain adaptation. However, when applied to practical scenarios, such as quantized models, new challenges arise. In practical scenarios, quantization is often applied to target-specific data, but this process restricts the domain of interest and introduces discretization effects, making model merging highly non-trivial. In this study, we analyze the impact of quantization on model merging through the lens of error barriers. Leveraging these insights, we propose a novel post-training quantization, HDRQ - Hessian and distant regularizing quantization - that is designed to consider model merging for multi-target domain adaptation. Our approach ensures that the quantization process incurs minimal deviation from the source pre-trained model while flattening the loss surface to facilitate smooth model merging. To our knowledge, this is the first study on this challenge, and extensive experiments confirm its effectiveness.
High time complexity is one of the biggest challenges faced by $k$-Nearest Neighbors ($k$NN). Although current classical and quantum $k$NN algorithms have made some improvements, they still have a speed bottleneck when facing large amounts of data. To address this issue, we propose an innovative algorithm called Granular-Ball based Quantum $k$NN(GB-Q$k$NN). This approach achieves higher efficiency by first employing granular-balls, which reduces the data size needed to processed. The search process is then accelerated by adopting a Hierarchical Navigable Small World (HNSW) method. Moreover, we optimize the time-consuming steps, such as distance calculation, of the HNSW via quantization, further reducing the time complexity of the construct and search process. By combining the use of granular-balls and quantization of the HNSW method, our approach manages to take advantage of these treatments and significantly reduces the time complexity of the $k$NN-like algorithms, as revealed by a comprehensive complexity analysis.